68060 vs Pentium
ANDERS BAKKEVOLD
-Anders Bakkevold.
John Crookshank
wrote:
> Does anyone have some comparable speed tests for the 68060 vs Pentium?
Benchmarks printed in Video Toaster User magazine show the '060 to be
about half the speed of a P90.
JC
--
--------------------------------------------------------------
| John Crookshank | jo...@mcs.com |
| MicroTech Solutions, Inc. | http://www.mcs.net/~johnc/ |
|--------------------------------------------------------------|
| Desktop video systems dealer - finger/http for more info. |
--------------------------------------------------------------
Mr Gavin Moran
: On 16-Jul-95 21:33:00, ANDERS BAKKEVOLD (anders.b...@norway.bbs.no)
: wrote:
: > Does anyone have some comparable speed tests for the 68060 vs Pentium?
: Benchmarks printed in Video Toaster User magazine show the '060 to be
: about half the speed of a P90.
I believe that mhz for mhz the 060 is very slightly slower than the Pentium,
wrt integer operations. Dunno about floating point though. Therefore a
P90 should have nearly twice the power of a 50mHz 060. Of course, by the
time you've ploughed through the overhead imposed by NT your render
performance will be a bit slower than predicted by processor benchmarks.
gavan
--
email: G.M...@ee.qub.ac.uk | 'There can be only one!'
or gmo...@nyx.cs.du.edu | - The Highlander
tem...@netins.net
>Does anyone have some comparable speed tests for the 68060 vs Pentium?
>
>
Also get someone to give you the REAL LIFE speed tests of a pentium after
the crap os's on ibm slow it down by 50-60%. hehehehe.
68060 should be quite good in comparsion.
Glenn Saunders
: On 16-Jul-95 21:33:00, ANDERS BAKKEVOLD (anders.b...@norway.bbs.no)
: wrote:
: > Does anyone have some comparable speed tests for the 68060 vs Pentium?
: Benchmarks printed in Video Toaster User magazine show the '060 to be
: about half the speed of a P90.
Were these render benchmarks? If so were they comparing 3.5 SA to 4.0
prerelease? Remember, there is no 4.0 prerelease for the Amiga.
Mat Bettinson
In a message dated 17 Jul 95 you wrote to All :
>> Does anyone have some comparable speed tests for the 68060 vs Pentium?
JC> Benchmarks printed in Video Toaster User magazine show the '060 to be
JC> about half the speed of a P90.
What was the 68060 running at? 50Mhz then sure it's about half the speed of the
a P90!
Mat Bettinson - CU Amiga Magazine.
The Prophet
In article <3uevts$t...@Mercury.mcs.com> jo...@MCS.COM (John Crookshank) writes:
> On 16-Jul-95 21:33:00, ANDERS BAKKEVOLD (anders.b...@norway.bbs.no)
> wrote:
>
>
>
> --
> --------------------------------------------------------------
> | John Crookshank | jo...@mcs.com |
> | MicroTech Solutions, Inc. | http://www.mcs.net/~johnc/ |
> |--------------------------------------------------------------|
> | Desktop video systems dealer - finger/http for more info. |
> --------------------------------------------------------------
They only tested a 50Mhz 060 and I don't think Lightwave has been optimized
for the 060 yet. With an 80Mhz 060 and optimized code, you'll be closer
to the 100Mhz Pentium, probably not faster but closer.
Jaeson K.
--
____ ____ _ _
( | \ ( / \ ( ) _ / )
)| )_ __ / /_ _ __ / __ ( X_)
( | /~ \ /\_) /---~/ ) / )/ )/~\ /\_) / _
)ll/ l/ \__ (/ (/ (_//__// / \__ (___)
(____________) (___/ (___)
Jaeson Koszarsky Amiga 3000+
---------------- -----------
cyberp...@psu.edu 68040/30Mhz
jb...@email.psu.edu 24Megs-1GIG
ja...@chaos.ezgate.com OS3.1
Joanne Dow
anders.b...@norway.bbs.no (ANDERS BAKKEVOLD) wrote:
>Does anyone have some comparable speed tests for the 68060 vs Pentium?
>
>
Depending on what is "around" the processor in terms of cache ram, peripheral
bus speed, memory speed, and the like they appear to be in the same "class"
with each other for most things. With AGA chips on AGA screen sizes it should
appear quite zippy compared to competing pentium systems with "ordinary" video
systems. But that is mostly differences in the OS and Workbench. (The former
development crew performed miracles with well thought out windowing support on
AmigaDOS.)
{^_^} Joanne Dow, Amiga Exchange Editor on BIX
jd...@bix.com The Wizardess
J.T.Vermeulen
Mr Gavin Moran <gavinm@vsprsun_14> wrote:
> I believe that mhz for mhz the 060 is very slightly slower than the Pentium,
Make that `040' and I might believe you. A test I ran was floating-point
with array accesses; MHz-for-MHz my A4000 was just as fast, under a preemptive
OS, and in higher precision too. This was code that didn't make much use of
the extra registers the 68k series has, which is advantagous to the Pentium.
The Pentium does less cache accesses per cycle than the 060, is still partially
microcoded which the 060 isn't, its scheduling rules are more restrictive, it
has less registers and a less powerful instruction set. L1 cache performance
are roughly similar; TLB performance may be slightly better because Intel code
needs a lot of stack accesses that a Motorola chip doesn't need but which make
replacement algorithms look good because they're highly localized.
The 060 has a four-stage pipeline, making branch misprediction penalty
potentially less expensive. Branch prediction accuracy of the P6 is "over 90%"
according to Intel; AFAIK the 060 does 91% to 92% and presumably the P5 does
not do better than that. Although the P6 seems worse than the P5 in some other
respects.
>wrt integer operations. Dunno about floating point though. Therefore a
>P90 should have nearly twice the power of a 50mHz 060. Of course, by the
Didn't somebody here report Lightwave results that were 20 to 30 percent
better for the 50MHz 060 (as compared to a P90)?
One reaction was that "the Pentium probably didn't have enough L2 cache" but
at the time of the test the CyberStorm L2 cache board wasn't even ready yet,
so the 060 probably ran with no second-level cache at all.
>time you've ploughed through the overhead imposed by NT your render
>performance will be a bit slower than predicted by processor benchmarks.
A MicroSoft Sales Assassin Droid (tm) told me that Windows 45 used NT's core
code. That might explain why it's so incredibly slow at 100MHz...
My PC-using uncle says it runs fine, but Widows 3.1 ran "fine" too. OTOH he
still doesn't believe us when we tell him we've been 32-bit for ten years,
just like he didn't believe we could multitask when we told him seven years
ago. "Gibt's nicht!" ("Can't be. No such thing"). Didn't even believe it
when he saw it. "That's not real multitasking! It's a fake".
Don't worry--things aren't as bad as they would have as believe!
>gavan
>--
>email: G.M...@ee.qub.ac.uk | 'There can be only one!'
> or gmo...@nyx.cs.du.edu | - The Highlander
Jeroen
John Crookshank
>In article <3uevts$t...@Mercury.mcs.com> jo...@MCS.COM (John Crookshank)
>writes:
>> On 16-Jul-95 21:33:00, ANDERS BAKKEVOLD (anders.b...@norway.bbs.no)
>> wrote:
>>
>> > Does anyone have some comparable speed tests for the 68060 vs Pentium?
>>
>> Benchmarks printed in Video Toaster User magazine show the '060 to be
>> about half the speed of a P90.
>>
>> JC
>> --
>> --------------------------------------------------------------
>> | John Crookshank | jo...@mcs.com |
>> | MicroTech Solutions, Inc. | http://www.mcs.net/~johnc/ |
>> |--------------------------------------------------------------|
>> | Desktop video systems dealer - finger/http for more info. |
>> --------------------------------------------------------------
>They only tested a 50Mhz 060 and I don't think Lightwave has been optimized
>for the 060 yet. With an 80Mhz 060 and optimized code, you'll be closer
>to the 100Mhz Pentium, probably not faster but closer.
>Jaeson K.
Yeah, we can only hope, eh? I remember when Lightwave was optimized for the
'040. It made an immediate 300% increase in rendering speed, the only change
was the updated Lightwave software.
JC
Christian Brandt
|> Also get someone to give you the REAL LIFE speed tests of a pentium after
|> the crap os's on ibm slow it down by 50-60%. hehehehe.
|>
|> 68060 should be quite good in comparsion.
In fact a 50Mhz 060 is 10% slower than a 60Mhz Pentium, therefore its
plausible to say, they show nearly the same speed when running at the
same frequency.
The 060 does quite well in Floating Point and random Memory-Access,
Integer and other are clearly better on Pentium
And now the Problem:
An LC060/50 costs the same as a Pentium 100.
An 060/66 costs the same as a Pentium 133...
An PPC604/120 costs less than a Pentium 60...
--
Christian Brandt
"I suspect if the Amiga chipset goes back into production,
it'll be the slowest graphics chipset for a
general purpose computer still in production."
Dave Haynie in <1995Jun28....@scala.scala.com>
Mike Noreen
Thusly Glenn Saunders spake unto All:
GS> : Benchmarks printed in Video Toaster User magazine show the '060 to be
GS> : about half the speed of a P90.
GS> Were these render benchmarks? If so were they comparing 3.5 SA to 4.0
GS> prerelease? Remember, there is no 4.0 prerelease for the Amiga.
I don't know about that, but the '060 being about half the speed of a
90Mhz Pentium seems about right, since the '060 is about as fast as
a Pentium at the same clockspeed (and the fastest '060s today are 50Mhz)
MVH: Mike Noreen
Internet: rad...@p14.anet.canit.se
Fido: 2:201/411.14
___ Terminate 1.50/Pro
John Hendrikx
MGM> I believe that mhz for mhz the 060 is very slightly slower than the
MGM> Pentium, wrt integer operations. Dunno about floating point though.
MGM> Therefore a P90 should have nearly twice the power of a 50mHz 060. Of
MGM> course, by the time you've ploughed through the overhead imposed by NT
MGM> your render performance will be a bit slower than predicted by
MGM> processor benchmarks.
I simply can't believe that. The 680x0 line of chips have always had a *huge*
advantage when it comes to flexibility of the instruction set. Even if some
instructions are now less usable because of pipe-lining, there should still be
more usefull instructions left in the 68060 than the Pentium could ever dream
of having. Not to mention that the 680x0 line has twice the number of general
purpose registers.
I think (no, make that: I know) that there is no compiler yet which can
generate code optimized for 68060, while there are probably 10+ for the
Pentium. Let's program these chips in Assembly and I'll bet that the 68060
beats the Pentium everytime.
Grtz John
-- Via Xenolink 1.96, XenolinkUUCP 1.1
Kevin Paul Kubish
>
>Benchmarks printed in Video Toaster User magazine show the '060 to be
What speed is the 060? 50mhz?
If so, then that isn't TOO bad. Of course, in actually everyday use it might seem to be different.
I am waiting to see what an 060 accel. for my 2000 will cost. (I will probably
just end up buying an 040 since it should drop in price when the 060 is widely
available.)
Kk...
--
Kevin Kubish - (A2000HD/020/2meg CHIP/4meg FAST/CD/280 meg HD space/3.1)
Jyrki Saarinen
> In fact a 50Mhz 060 is 10% slower than a 60Mhz Pentium, therefore its
> plausible to say, they show nearly the same speed when running at the
> same frequency.
Prove it.
> And now the Problem:
>
> An LC060/50 costs the same as a Pentium 100.
>
> An 060/66 costs the same as a Pentium 133...
>
> An PPC604/120 costs less than a Pentium 60...
What an earth are these for prices?
Glenn Saunders
: Yeah, we can only hope, eh? I remember when Lightwave was optimized for the
: '040. It made an immediate 300% increase in rendering speed, the only change
: was the updated Lightwave software.
It's a losing battle for any Amiga to win a numbercrunching benchmark
against a clone these days. Motorola is out of the race with the 680x0
line. If they even release a 66mhz 060 I will be pleasantly surprised.
80mhz is wishful thinking and 120 is just never going to happen yet we
have higher clocks with the Pentiums already.
The thing is that any app that relies on numbercrunching is going to run
slower on an Amiga but any app that isn't particularly CPU-hogging will
probably be crisper on an Amiga due to the more efficient user interface.
So a clone makes a great render slave but may seem sluggish trying to
manage a lot of lesser apps running preemptively.
Hence the Amiga becomes a more effective PERSONAL computer rather than a
dedicated computer surrounding a commercial (i.e. on the job) application
like image manipulation or rendering where you would likely want to just
start one program and let it hog the whole machine's resources.
Computers for the latter may as well run DOS/Windows since the
multitasking is rarely taken advantage-of.
Then just imagine how AmigaOS would fly on a PPC compared to other OSs!
You probably have never seen such speed.
Jyrki Saarinen
> line. If they even release a 66mhz 060 I will be pleasantly surprised.
I though 68060 was available now in 50MHz and 66MHz?
J.T.Vermeulen
Philipp Boerker <rawn...@hydra.zrz.TU-Berlin.DE> wrote:
>On the other hand the P5 has already a 64 bit databus while 060 still has 32-bit.
What I've heard from a Motorola employee is that cache-memory bandwidth is "no
longer the greatest bottleneck". He couldn't say any more about it, but it's
easy to figure out that Intel chips have more need for such brute-force aid.
>Both CPU are superscalar which means that they can execute more than one opperation
>per clockcycle. The average superscalar factors for 1-cycle-ops are:
>P5 : 1.2-1.3 ops/cycle
>060: 1.6 ops/cycle !
Are you sure? The figure I remember for the 060 is a sustained 0.7 CPI rate
(1.43 ops/cycle) on standard SPEC benchmark code. OTOH that wasn't optimized
for the 060 and was generated by an existing compiler. :-)
Also of course the 060 does a lot more useful work per instruction.
>>Didn't somebody here report Lightwave results that were 20 to 30 percent
>>better for the 50MHz 060 (as compared to a P90)?
>
>A friend of mine compared his 25 MHz 040 A4000 to a 60 MHz P5, both computers
>using the newest version of Imagine (3.1 ?). They were nearly equally fast!
>(F. i. when the A4000 took 5 min the P5 took 4:30 min).
That must have felt great! Intel advocates often say that the chips are very
fast but *all* compilers are very bad, so it isn't "fair" to compare
performance with C programs. Of course C is even more common on PCs because
the processors are so hideously irregular that assembly is a nightmare.
On comp.arch, one Intel CPU designer wrote that he was frustrated about the
low quality of PC compilers because he knew "that our chips could run just as
fast as the competition"--to me that means that he admits that their chips
aren't faster than the Motorola ones at all.
>>One reaction was that "the Pentium probably didn't have enough L2 cache" but
>>at the time of the test the CyberStorm L2 cache board wasn't even ready yet,
>>so the 060 probably ran with no second-level cache at all.
>
>L2 cache is not optimal for multitasking computers. That's the reason why they
>start to use EDO-RAM in Pentiumn-boards. I think they will do so for future Amiga-
>accelerators too.
What's EDO-RAM?
>Summing up an Amiga with an 060 will be as fast as a P90-clone!
And the 060 probably has more room for growth. If you look at the P6, the
architects must be really desperate. Fifteen-cycle branch misprediction
penalties are no exception I hear. And because of what they call "dynamic
execution", the processor is already trying to execute code some 30
instructions ahead; on the average those contain maybe five or six branches,
so the chance that the foremost instruction should be executed is only about
fifty or sixty percent.
>----------------------------
>rawn...@sp.zrz.tu-berlin.de
>----------------------------
Jeroen
Philipp Boerker
>In article <3ug70m$1i...@info4.rus.uni-stuttgart.de>,
>Mr Gavin Moran <gavinm@vsprsun_14> wrote:
>> I believe that mhz for mhz the 060 is very slightly slower than the Pentium,
>The Pentium does less cache accesses per cycle than the 060, is still partially
>microcoded which the 060 isn't, its scheduling rules are more restrictive, it
>has less registers and a less powerful instruction set. L1 cache performance
>are roughly similar; TLB performance may be slightly better because Intel code
>needs a lot of stack accesses that a Motorola chip doesn't need but which make
>replacement algorithms look good because they're highly localized.
Also you have to consider that P5 may have to access misaligned addresses after
a instructioncache-miss because of the byte-organized x86-code.
On the other hand the P5 has already a 64 bit databus while 060 still has 32-bit.
>The 060 has a four-stage pipeline, making branch misprediction penalty
>potentially less expensive. Branch prediction accuracy of the P6 is "over 90%"
>according to Intel; AFAIK the 060 does 91% to 92% and presumably the P5 does
>not do better than that. Although the P6 seems worse than the P5 in some other
>respects.
Both CPU are superscalar which means that they can execute more than one opperation
per clockcycle. The average superscalar factors for 1-cycle-ops are:
P5 : 1.2-1.3 ops/cycle
060: 1.6 ops/cycle !
>Didn't somebody here report Lightwave results that were 20 to 30 percent
>better for the 50MHz 060 (as compared to a P90)?
A friend of mine compared his 25 MHz 040 A4000 to a 60 MHz P5, both computers
using the newest version of Imagine (3.1 ?). They were nearly equally fast!
(F. i. when the A4000 took 5 min the P5 took 4:30 min).
>One reaction was that "the Pentium probably didn't have enough L2 cache" but
>at the time of the test the CyberStorm L2 cache board wasn't even ready yet,
>so the 060 probably ran with no second-level cache at all.
L2 cache is not optimal for multitasking computers. That's the reason why they
start to use EDO-RAM in Pentiumn-boards. I think they will do so for future Amiga-
accelerators too.
Summing up an Amiga with an 060 will be as fast as a P90-clone!
With kind regards,
Phil.
----------------------------
rawn...@sp.zrz.tu-berlin.de
----------------------------
Michael van Elst
> And now the Problem:
> An LC060/50 costs the same as a Pentium 100.
> An 060/66 costs the same as a Pentium 133...
> An PPC604/120 costs less than a Pentium 60...
Completeley irrelevant. You cannot just buy CPU chips and plug them in.
Regards,
--
Michael van Elst
Internet: mle...@serpens.rhein.de
"A potential Snark may lurk in every tree."
Maxwell Daymon
: On the other hand the P5 has already a 64 bit databus while 060 still
: has 32-bit.
And MHz to MHz is STILL performs as well as or better. That says
something pretty loudly, doesn't it?
: Both CPU are superscalar which means that they can execute more than one
: opperation per clockcycle. The average superscalar factors for 1-cycle-ops
: are:
: P5 : 1.2-1.3 ops/cycle
: 060: 1.6 ops/cycle !
Based on what? The code has a lot to do with whether these numbers are
important or not. This is like MIPs - only useful between identical
instruction sets and mixes. Give me real world, not averages of 1 cycle ops.
: A friend of mine compared his 25 MHz 040 A4000 to a 60 MHz P5, both computers
: using the newest version of Imagine (3.1 ?). They were nearly equally fast!
: (F. i. when the A4000 took 5 min the P5 took 4:30 min).
Ugh, the 25MHz 68040 in the A4000 is pretty slow.
: >One reaction was that "the Pentium probably didn't have enough L2 cache" but
If the 040 runs in a bad memory system and the Pentium wants an even
bigger L2 cache and has a larger ON CHIP cache, that screams something to
me and not in favor of the P5.
: L2 cache is not optimal for multitasking computers. That's the reason why they
: start to use EDO-RAM in Pentiumn-boards. I think they will do so for future Amiga-
: accelerators too.
: Summing up an Amiga with an 060 will be as fast as a P90-clone!
I think this is a leap, but I want to know more.
Fred Heitkamp
>
>Thusly Glenn Saunders spake unto All:
>
> GS> : Benchmarks printed in Video Toaster User magazine show the '060 to be
> GS> : about half the speed of a P90.
>
> GS> Were these render benchmarks? If so were they comparing 3.5 SA to 4.0
> GS> prerelease? Remember, there is no 4.0 prerelease for the Amiga.
>
>I don't know about that, but the '060 being about half the speed of a
>90Mhz Pentium seems about right, since the '060 is about as fast as
>a Pentium at the same clockspeed (and the fastest '060s today are 50Mhz)
>
Cyberstorm on a couple benchmarks, it looks like the Cyberstorm is
not performing as well as it could. I'd be interested to see how the
WarpEngine 060 does when it finally comes out.
-Fred Heitkamp
fhe...@ibm.net
John Hendrikx
>>> Does anyone have some comparable speed tests for the 68060 vs Pentium?
JC>> Benchmarks printed in Video Toaster User magazine show the '060 to be
JC>> about half the speed of a P90.
MB> What was the 68060 running at? 50Mhz then sure it's about half the
MB> speed of the a P90!
I'd bet that Lightwave as optimized for the Pentium, while Lightwave for the
68060 probably was optimized for the 68040. Also the Pentium system is likely
to have 256K or more 2nd level cache, while the 060 probably didn't.
tem...@netins.net
>
>In article <009938A4...@netins.net>, tem...@netins.net writes:
>
>|> Also get someone to give you the REAL LIFE speed tests of a pentium after
>|> the crap os's on ibm slow it down by 50-60%. hehehehe.
>|>
>|> 68060 should be quite good in comparsion.
>
>plausible to say, they show nearly the same speed when running at the
>same frequency.
>
hehhehehee... Comparing mhz I suppose, and subtracting mhz to come up with
a speed difference perhaps? Yeah you really know your processor's dont ya.
> The 060 does quite well in Floating Point and random Memory-Access,
>Integer and other are clearly better on Pentium
>
> And now the Problem:
>
> An LC060/50 costs the same as a Pentium 100.
>
> An 060/66 costs the same as a Pentium 133...
>
> An PPC604/120 costs less than a Pentium 60...
>
Amiga still rules over any ibm no matter how much the Amiga costs. IBM's
struggle to make a real machine out of the total crap they have been based
on. They come out with new os's and faster processors but still they remain
the same crap as when they came out. Mac's? I won't even bother to comment
on them with their total crap os.
>--
>Christian Brandt
>
> "I suspect if the Amiga chipset goes back into production,
> it'll be the slowest graphics chipset for a
> general purpose computer still in production."
> Dave Haynie in <1995Jun28....@scala.scala.com>
Take out your graphics card in your crap ibm and tell me how great your
computer's graphics are and how fast they are.. hehehehe.. lamer!
To give you a clue: Graphic cards exist on Amiga as well. Just that we
have better machines without expansion than your crap ibm.
tem...@netins.net
>
>> line. If they even release a 66mhz 060 I will be pleasantly surprised.
>
>I though 68060 was available now in 50MHz and 66MHz?
and in 80 mhz this fall.. don't listen to idiot ibm users... they have shown
their stupidity in the first place by buying ibm crap. They know the Amiga
is 1000 times the computer of an ibm that is why they are in these areas
posting such crap. Before they did this, they posted that 060s would never
come out in the first place. After this statement by them gets proven
wrong they move on to the next bullshit subject and hack at it. One
thing never changes though, they always show their stupidity.
Mr Gavin Moran
: In a message of 18 Jul 95 Mr Gavin Moran wrote to All:
: MGM> I believe that mhz for mhz the 060 is very slightly slower than the
: MGM> Pentium, wrt integer operations. Dunno about floating point though.
: MGM> Therefore a P90 should have nearly twice the power of a 50mHz 060. Of
: MGM> course, by the time you've ploughed through the overhead imposed by NT
: MGM> your render performance will be a bit slower than predicted by
: MGM> processor benchmarks.
: I simply can't believe that. The 680x0 line of chips have always had a *huge*
: advantage when it comes to flexibility of the instruction set. Even if some
: instructions are now less usable because of pipe-lining, there should still be
: more usefull instructions left in the 68060 than the Pentium could ever dream
: of having. Not to mention that the 680x0 line has twice the number of general
: purpose registers.
Beleive it! I was shocked at first as I've always considered the 68k
chips to be a lot faster mhz for mhz than their x86 rivals. But Intel really
have made big progress with the pentium and if you look at the sysinfo
MIPS ( OK, OK, I know its not a perfect test) of a 50mHz 060 vs a 25mHz 040
the 060 has TWICE as many MIPS but it runs at TWICE the clockspeed - very
dodgy...... Plus the pentium has a 64-bit interface to memory (the 060
has 32) and I think that a 90 mhz pentium can access memory at up to 60 mhz
but the 060 can only access at 33? (can anyone comment?)
: I think (no, make that: I know) that there is no compiler yet which can
: generate code optimized for 68060, while there are probably 10+ for the
: Pentium. Let's program these chips in Assembly and I'll bet that the 68060
: beats the Pentium everytime.
I'd bet an Amiga with 060 could beat the pants off a p90 running NT for
multitasking etc, but for number-cruncing the p90 has the edge....
Mr Gavin Moran
: Philipp Boerker (rawn...@hydra.zrz.TU-Berlin.DE) wrote:
: : On the other hand the P5 has already a 64 bit databus while 060 still
: : has 32-bit.
: And MHz to MHz is STILL performs as well as or better. That says
: something pretty loudly, doesn't it?
It doesn't matter HOW it achieves its performance - it DOES and at lower
cost and with faster models than the 060.
jtv
Philipp Boerker <rawn...@hydra.zrz.TU-Berlin.DE> wrote:
>
>>Are you sure? The figure I remember for the 060 is a sustained 0.7 CPI rate
>>(1.43 ops/cycle) on standard SPEC benchmark code. OTOH that wasn't optimized
>>for the 060 and was generated by an existing compiler. :-)
>
>will use optimized code for testing purposes. But this will be true for Intel
>too.
Motorola's figures were for existing code. Optimization for the 060 can add
more performance.
>>Also of course the 060 does a lot more useful work per instruction.
>
>Yes! Think of the many, many mve x,Rx -instructions they have to do because
>of the tiny set of registers!
...And because many instructions are limited to certain combinations of
registers, they probably have to do more register-register moves too, so
they effectively lose even more registers.
>>What's EDO-RAM?
>
>EDO-RAM is RAM that has a 64K-block of cache for every 1M inside itself! The advantage
>of this is:
>Using L2-cache in a multitasking environment actually means flushing this cache
>with every task-switch. Before you have "filled" it with the new data your task is
>reading from another task will be running and therefor no cache-hits will occur.
>With EDO-RAM this is far more unlikely because the second task will have its code and
>data (most likely) in an other block of RAM!
Ah, cache in the memory bank!
Damn, I thought of that years ago and didn't sell the idea... :-)
>The price of this cannot be much higher of that of ordinary RAM plus L2-cache:
>A big german PC-store-chain gives you the option of taking 8MB EDO-RAM instead of
>8MB ordinary RAM plus 256K L2-cache for the same price! 256K L2-cache cost 100 DM,
>that is ca. 65 US$ !
"A big German PC store chain"... That used to mean Escom. Now I suppose you're
talking about Vobis?
>BTW does anybody know anything about the MC 88000 ? I have seen figures that this proc
>does 34000 Dhrystones at 20MHz. That is 4.5 times as much as a 030 at the same MHz!
>Maybe there is also a 88060 ;-) ?
The 88000 is pretty old now. Its successor is the 88100.
>Phil.
>----------------------------
>rawn...@sp.zrz.tu-berlin.de
>----------------------------
Jeroen
Philipp Boerker
>In article <3uoikg$8...@brachio.zrz.tu-berlin.de>,
>Philipp Boerker <rawn...@hydra.zrz.TU-Berlin.DE> wrote:
>>On the other hand the P5 has already a 64 bit databus while 060 still has 32-bit.
>What I've heard from a Motorola employee is that cache-memory bandwidth is "no
>longer the greatest bottleneck". He couldn't say any more about it, but it's
>easy to figure out that Intel chips have more need for such brute-force aid.
Also, this 64-bit bus is the reason why in a P5-board you always have to use two
of the four PS/2-slots! To me that's a bit of an disadvantage too...
>>Both CPU are superscalar which means that they can execute more than one opperation
>>per clockcycle. The average superscalar factors for 1-cycle-ops are:
>>P5 : 1.2-1.3 ops/cycle
>>060: 1.6 ops/cycle !
>Are you sure? The figure I remember for the 060 is a sustained 0.7 CPI rate
>(1.43 ops/cycle) on standard SPEC benchmark code. OTOH that wasn't optimized
>for the 060 and was generated by an existing compiler. :-)
Well, this will depend on the software. People that want to sell a processor
will use optimized code for testing purposes. But this will be true for Intel
too.
>Also of course the 060 does a lot more useful work per instruction.
Yes! Think of the many, many mve x,Rx -instructions they have to do because
of the tiny set of registers!
>>>One reaction was that "the Pentium probably didn't have enough L2 cache" but
>>>at the time of the test the CyberStorm L2 cache board wasn't even ready yet,
>>>so the 060 probably ran with no second-level cache at all.
>>
>>L2 cache is not optimal for multitasking computers. That's the reason why they
>>start to use EDO-RAM in Pentiumn-boards. I think they will do so for future Amiga-
>>accelerators too.
>What's EDO-RAM?
EDO-RAM is RAM that has a 64K-block of cache for every 1M inside itself! The advantage
of this is:
Using L2-cache in a multitasking environment actually means flushing this cache
with every task-switch. Before you have "filled" it with the new data your task is
reading from another task will be running and therefor no cache-hits will occur.
With EDO-RAM this is far more unlikely because the second task will have its code and
data (most likely) in an other block of RAM!
The price of this cannot be much higher of that of ordinary RAM plus L2-cache:
A big german PC-store-chain gives you the option of taking 8MB EDO-RAM instead of
8MB ordinary RAM plus 256K L2-cache for the same price! 256K L2-cache cost 100 DM,
that is ca. 65 US$ !
>>Summing up an Amiga with an 060 will be as fast as a P90-clone!
>And the 060 probably has more room for growth.
The only growth of the 680x0-series we can expect is higher MHz for the 060
because Motorola will not develop an 070 or anything of this kind :-(.
BTW does anybody know anything about the MC 88000 ? I have seen figures that this proc
does 34000 Dhrystones at 20MHz. That is 4.5 times as much as a 030 at the same MHz!
Maybe there is also a 88060 ;-) ?
>Jeroen
Greets,
Phil.
----------------------------
rawn...@sp.zrz.tu-berlin.de
----------------------------
jtv
Mr Gavin Moran <gavinm@vsprsun_14> wrote:
>
>chips to be a lot faster mhz for mhz than their x86 rivals. But Intel really
>have made big progress with the pentium and if you look at the sysinfo
>MIPS ( OK, OK, I know its not a perfect test) of a 50mHz 060 vs a 25mHz 040
It's nowhere *near* a perfect test. I've measured speed increases of up to
TEN times going from a stock A4000 to a 10Mb 68040 WarpEngine at 40 MHz
plus SCSI-II HD. Comparing to PC's I've seen so far, I doubt any Pentium
could give me that kind of improvement. Yet our WarpEngine isn't exactly
the fastest Amiga in town anymore.
Also most comparisons with PCs are done in single-tasking environments.
So before you compare, run an OS that will allow you to do simple work
without going crazy. Maybe OS/2 or Linux will get the job done, but
Widows 95 certainly can't and it's still dead slow.
> I'd bet an Amiga with 060 could beat the pants off a p90 running NT for
>multitasking etc, but for number-cruncing the p90 has the edge....
If it's safe, and the manufacturer doesn't more skeletons in his closet
than the one that fell out last year.
>gavan
>--
>email: G.M...@ee.qub.ac.uk | 'There can be only one!'
> or gmo...@nyx.cs.du.edu | - The Highlander
Jeroen
Charles E Taylor IV
:Take out your graphics card in your crap ibm and tell me how great your
:computer's graphics are and how fast they are.. hehehehe.. lamer!
Errm, actually the graphics chips in my Compaq 486 are on the motherboard. No
"card". There goes *that* argument. :)
And the gpaphics built in to the Compaq are pretty damn fast compared to
ECS or AGA. :)
:To give you a clue: Graphic cards exist on Amiga as well. Just that we
:have better machines without expansion than your crap ibm.
And graphics chips on the motherboard as in the Amiga also exist in the PC
world. Try an *informed* argument next time.
--
__ ___ _ _ _ | >>>>> cha...@hubcap.clemson.edu <<<<<
|_)o _ |/ | |_|\_/| / \|_) |'94 BS ChE, Clemson ; "Will work for food!"
| \||_ |\ | | | | |_\_/| \ |"So this is it, we're going to die!" --A.D.
---> My WWW home is now at http://www.clemson.edu/~charlet/ <---
Maxwell Daymon
: Beleive it! I was shocked at first as I've always considered the 68k
: chips to be a lot faster mhz for mhz than their x86 rivals. But Intel really
: have made big progress with the pentium and if you look at the sysinfo
: MIPS ( OK, OK, I know its not a perfect test) of a 50mHz 060 vs a 25mHz 040
: dodgy...... Plus the pentium has a 64-bit interface to memory (the 060
: has 32) and I think that a 90 mhz pentium can access memory at up to 60 mhz
: but the 060 can only access at 33? (can anyone comment?)
That does it. I'm writing a MIPS test that only uses NOP instructions.
MIPS MEANS NOTHING. Not "I know it's not a perfect test" it's a
/completely/ irrelevant test. Except in SOME isolated cases of processor
family comparison, and even THEN is barely a guideline.
: I'd bet an Amiga with 060 could beat the pants off a p90 running NT for
: multitasking etc, but for number-cruncing the p90 has the edge....
If you want JUST number crunching get a PPC620 or a DEC Alpha. If you're
interested in a general purpose machine, multi-tasking is the future.
Also, MIPS means nothing between processor families (unless they just
happen to have the EXACT same instruction set, but even then it only
tells you how fast the MIPS test will execute - little to do with real life)
Maxwell Daymon
: In article <00993CF2...@netins.net> tem...@netins.net writes:
: :Take out your graphics card in your crap ibm and tell me how great your
: :computer's graphics are and how fast they are.. hehehehe.. lamer!
: Errm, actually the graphics chips in my Compaq 486 are on the motherboard. No
: "card". There goes *that* argument. :)
True. Many Compaqs I've seen have a CL5428 on board. (Same as the Picasso
II/Spectrum/etc.)
tem...@netins.net
>John Hendrikx (john.h...@grafix.xs4all.nl) wrote:
>: In a message of 18 Jul 95 Mr Gavin Moran wrote to All:
>
>: MGM> I believe that mhz for mhz the 060 is very slightly slower than the
>: MGM> Pentium, wrt integer operations. Dunno about floating point though.
>: MGM> Therefore a P90 should have nearly twice the power of a 50mHz 060. Of
>: MGM> course, by the time you've ploughed through the overhead imposed by NT
>: MGM> your render performance will be a bit slower than predicted by
>: MGM> processor benchmarks.
>
>: I simply can't believe that. The 680x0 line of chips have always had a *huge*
>: advantage when it comes to flexibility of the instruction set. Even if some
>: instructions are now less usable because of pipe-lining, there should still be
>: more usefull instructions left in the 68060 than the Pentium could ever dream
>: of having. Not to mention that the 680x0 line has twice the number of general
>: purpose registers.
>
>chips to be a lot faster mhz for mhz than their x86 rivals. But Intel really
>have made big progress with the pentium and if you look at the sysinfo
>MIPS ( OK, OK, I know its not a perfect test) of a 50mHz 060 vs a 25mHz 040
>the 060 has TWICE as many MIPS but it runs at TWICE the clockspeed - very
>dodgy...... Plus the pentium has a 64-bit interface to memory (the 060
>has 32) and I think that a 90 mhz pentium can access memory at up to 60 mhz
>but the 060 can only access at 33? (can anyone comment?)
>
Now I know you don't know what you are talking about.. 060 50mhz only twice
as fast as the 040 25mhz, mip wise?? Where on earth did you come up with
this and how on earth would you expect anybody to believe such crap?
Try 4-5 times faster mips wise, at least! You think an 060 50mhz runs at
only 38 mips or so? Think again guy. I know you must hate that Amiga's
are good computers and ibm's are crap but lets get things a bit more
realistic and not rediculously stupid.
Pentiums might be fast processors, but placed on an ibm with their crap
os's, forget about it. They loose most of their speed instantly.
>
>
>: I think (no, make that: I know) that there is no compiler yet which can
>: generate code optimized for 68060, while there are probably 10+ for the
>: Pentium. Let's program these chips in Assembly and I'll bet that the 68060
>: beats the Pentium everytime.
>
> I'd bet an Amiga with 060 could beat the pants off a p90 running NT for
>multitasking etc, but for number-cruncing the p90 has the edge....
>
Unless I stacked 9 Risc processors together for rendering, then I bet that
p90 would look pretty lame compared to my Amiga, seeing that just one of the
risc processors would beat it, let alone the 9 together.
Philipp Boerker
>Philipp Boerker (rawn...@hydra.zrz.TU-Berlin.DE) wrote:
>: On the other hand the P5 has already a 64 bit databus while 060 still
>: has 32-bit.
>And MHz to MHz it STILL performs as well as or better. That says
>something pretty loudly, doesn't it?
Sure!
>: Both CPU are superscalar which means that they can execute more than one
>: opperation per clockcycle. The average superscalar factors for 1-cycle-ops
>: are:
>: P5 : 1.2-1.3 ops/cycle
>: 060: 1.6 ops/cycle !
>Based on what? The code has a lot to do with whether these numbers are
>important or not. This is like MIPs - only useful between identical
>instruction sets and mixes. Give me real world, not averages of 1 cycle ops.
Of course, you can only compare two processors like this if you use identical
code. If you look at the overhead of x86 code compared to 680x0-code you will
certainly receive something like 2.5 intel-ops/cycle on a 060!
>: A friend of mine compared his 25 MHz 040 A4000 to a 60 MHz P5, both computers
>: using the newest version of Imagine (3.1 ?). They were nearly equally fast!
>: (F. i. when the A4000 took 5 min the P5 took 4:30 min).
>Ugh, the 25MHz 68040 in the A4000 is pretty slow.
And the 50 MHz 060-cyberstorm is five times faster than an A4000...
Maxwell Daymon
: Motorola doesn't seem to agree with you. They said quite the opposite! They
: said the had no plans to discontinue development of the 680x0 series. Where
Motorola also said that the move to RISC would now let them focus on
designing 680x0 processors for embedded applications (toasters,
microwaves, cellphones, etc.) Hardly a processor for a computer.
: was to be the last processor, it certainly gives the Amiga enough speed to
: stay ahead of ibm's for years. IBM's will suffer from shit os's for years
For about six months. The PowerPC 604 is already out @ ~130 MHz, and even
higher speeds slated for the future.
: opposite. Look at all the idiots that said the 060 would never come out.
: Now that the 060 is out they go on to something else by saying that the 060
: is the last in the series.
The ColdFire is already evidence of their intention to design processors
for limited embedded applications.
: I also remember people saying the a4000 was a pipe dream, amiga would never
: come back, etc etc...
It really HASN'T come back except in limited areas in limited quantities.
: How people can state things they THINK will happen in the future and always
: be dead wrong when the time comes is stupid. Until things are known as fact,
: don't come out with stories or opinons without saying it is a personal
: opnion of yours and not true.
Sorry, but there's other information out there that YOU are ignoring. You
are presenting just as much an opinion as anybody else here.
tem...@netins.net
>>>Summing up an Amiga with an 060 will be as fast as a P90-clone!
>>And the 060 probably has more room for growth.
>
>The only growth of the 680x0-series we can expect is higher MHz for the 060
>because Motorola will not develop an 070 or anything of this kind :-(.
>BTW does anybody know anything about the MC 88000 ? I have seen figures that this proc
>does 34000 Dhrystones at 20MHz. That is 4.5 times as much as a 030 at the same MHz!
>Maybe there is also a 88060 ;-) ?
>
>>Jeroen
>
>Phil.
>----------------------------
>rawn...@sp.zrz.tu-berlin.de
>----------------------------
Motorola doesn't seem to agree with you. They said quite the opposite! They
said the had no plans to discontinue development of the 680x0 series. Where
people come up with the idea that 060 is the last processor of the 680x0
series? Motorola has publicly stated quite the opposite. Even if the 060
was to be the last processor, it certainly gives the Amiga enough speed to
stay ahead of ibm's for years. IBM's will suffer from shit os's for years
to come yet, if they ever are able to develop a REAL OS that is. As far
as rendering, Amiga can have risc procesors added for rendering. I am talking
about the everday use of the computers in general, they run many times faster
than any crap ibm on the market today. I just love it when people say something
is not going to come out when the company making the product states the
opposite. Look at all the idiots that said the 060 would never come out.
Now that the 060 is out they go on to something else by saying that the 060
is the last in the series.
I also remember people saying the a4000 was a pipe dream, amiga would never
come back, etc etc...
Maxwell Daymon
: > I'd bet an Amiga with 060 could beat the pants off a p90 running NT for
: >multitasking etc, but for number-cruncing the p90 has the edge....
: Unless I stacked 9 Risc processors together for rendering, then I bet that
: p90 would look pretty lame compared to my Amiga, seeing that just one of the
: risc processors would beat it, let alone the 9 together.
That's a fantasy. Didn't you just say people shouldn't be talking if
their statements weren't based on fact?
Your extreme hypocracy is noted.
Mr Gavin Moran
: >MIPS ( OK, OK, I know its not a perfect test) of a 50mHz 060 vs a 25mHz 040
: >the 060 has TWICE as many MIPS but it runs at TWICE the clockspeed - very
: >dodgy...... Plus the pentium has a 64-bit interface to memory (the 060
: >has 32) and I think that a 90 mhz pentium can access memory at up to 60 mhz
: >but the 060 can only access at 33? (can anyone comment?)
: >
: Now I know you don't know what you are talking about.. 060 50mhz only twice
: as fast as the 040 25mhz, mip wise?? Where on earth did you come up with
: this and how on earth would you expect anybody to believe such crap?
I got that from the MIPS figures published in Amiga Format and CU Amiga.
My mate reviewed the 060 for Format and told me he was shocked on the MIPS
findings himself. He was heartened to find that renders and suchlike benefitted
from 3x to 5x increases though.
:
: Try 4-5 times faster mips wise, at least! You think an 060 50mhz runs at
: only 38 mips or so? Think again guy. I know you must hate that Amiga's
: are good computers and ibm's are crap but lets get things a bit more
: realistic and not rediculously stupid.
MIPS tests have always been a bit dodgy - but when you get a MIPS test
on two different models of the same CPU familty its a _bit_ more accurate.
Unless of course the sysinfo test picks some stupid instruction like NOP
as the 'I' in MIPS.
: Pentiums might be fast processors, but placed on an ibm with their crap
: os's, forget about it. They loose most of their speed instantly.
No, Pentiums are still fast. I use both 60 and 90 mhz models running on
Windows NT machines and they _are_ fast for computationally heavy tasks.
: Unless I stacked 9 Risc processors together for rendering, then I bet that
: p90 would look pretty lame compared to my Amiga, seeing that just one of the
: risc processors would beat it, let alone the 9 together.
You have an Amiga running 9 RISC processors? WOW!
Mat Bettinson
In a message dated 24 Jul 95 you wrote to All :
t> Motorola doesn't seem to agree with you. They said quite the opposite!
t> They said the had no plans to discontinue development of the 680x0
t> series. Where people come up with the idea that 060 is the last
t> processor of the 680x0 series? Motorola has publicly stated quite the
Because they have said it will be the last in the line. Publically. On more
than one occasion.
Mat Bettinson - CU Amiga Magazine.
Glenn Saunders
: Motorola doesn't seem to agree with you. They said quite the opposite! They
: said the had no plans to discontinue development of the 680x0 series. Where
Considering the long delay between the completion of the 060 design and
its availability, even a P6like 080 would not be worth the wait if it
takes 2-3 years to be completed and then another year and a half to
trickle into production.
Charles E Taylor IV
:The point is *all* Amigas come with graphics chips on the motherboard. PC's
:dont.
It's not even a *valid* point considering the cost of a good graphics card
for the PC compared to the cost of getting good graphics on the Amiga. (AGA
doesn't cut it anymore ... actually it never did).
Think about what you're trying to say about the Amiga. Having an
antiquated graphics chipset built in to new Amigas is *NOT* an advantage!
Mr Gavin Moran
: The mortal tem...@netins.net wrote:
: : and in 80 mhz this fall.. don't listen to idiot ibm users... they have shown
: Don't get our hopes up unless you can cite sources for your claims that
: Moto will have an 80mhz 060 this fall.
When are the 120 and 133 mhz 060s coming out?????
There are pentiums of that speed available TODAY.
Mr Gavin Moran
: Wouldnt one raptor kick the pentiums arse in rendering? I thought it had
: a 200MHz DEC Alpha hip inside of it...
But we are discussing the 060 vs pentium (read title of thread).
Wouldn't a CRAY kick the 200mHz Alphas arse............
Glenn Saunders
J.T.Vermeulen
Maxwell Daymon <mda...@rainbow.rmii.com> wrote:
>
>Also, the post said "pretty lame compared to my Amiga" (which doesn't
>have a RISC processor, it is based on the 680x0)
Based on the 680x0 it may be, but it also has a RISC processor!
The Copper. In dire straits you learn to count your blessings...
It's more than just a joke: Hans Guijt uses the copper to command the blitter
in his MSX emulator, so you get fluid screen redraws without taking up any CPU
cycles.
>The Amiga is a very nice machine IMHO, but it doesn't pay to be
>unrealistic.
Some artists I know swear by it. :-)
Jeroen
--
============================================================================
# Jeroen T. Vermeulen \"How are we doing kid?"/ Yes, we use Amigas. #
#--- j...@xs4all.nl ---\"Oh, same as always."/-- ... --#
#jver...@wi.leidenuniv.nl \ "That bad, huh?" / Got a problem with that? #
James McArthur
: In article <00993CF2...@netins.net> tem...@netins.net writes:
: :Take out your graphics card in your crap ibm and tell me how great your
: :computer's graphics are and how fast they are.. hehehehe.. lamer!
: Errm, actually the graphics chips in my Compaq 486 are on the motherboard. No
: "card". There goes *that* argument. :)
: And the gpaphics built in to the Compaq are pretty damn fast compared to
: ECS or AGA. :)
: :To give you a clue: Graphic cards exist on Amiga as well. Just that we
: :have better machines without expansion than your crap ibm.
: And graphics chips on the motherboard as in the Amiga also exist in the PC
: world. Try an *informed* argument next time.
The point is *all* Amigas come with graphics chips on the motherboard. PC's
dont.
: --
James McArthur
: tem...@netins.net wrote:
: : > I'd bet an Amiga with 060 could beat the pants off a p90 running NT for
: : >multitasking etc, but for number-cruncing the p90 has the edge....
: : Unless I stacked 9 Risc processors together for rendering, then I bet that
: : p90 would look pretty lame compared to my Amiga, seeing that just one of the
: : risc processors would beat it, let alone the 9 together.
: That's a fantasy. Didn't you just say people shouldn't be talking if
: their statements weren't based on fact?
: Your extreme hypocracy is noted.
Wouldnt one raptor kick the pentiums arse in rendering? I thought it had
Maxwell Daymon
: In article <3v1183$5...@natasha.rmii.com> mda...@rainbow.rmii.com (Maxwell Daymon) writes:
: > That does it. I'm writing a MIPS test that only uses NOP instructions.
: From Motorola's "Programmer's Reference Manual" 4-147,
: "The NOP instruction does not begin execution until all pending
: bus cycles have completed. This synchronizes the pipeline and
: prevents instruction overlap."
Yep. Dammit. Was the first thing that came to mind. I didn't bother to
think or anything.
: Doesn't this cause a bad effect on MIPS testing?
Yes, so obviously we're going to have to use a better instruction. I'm
just sick of all these MIPS comparisons against other processor families,
so we might as well make the best LOOKING test. Heck, we can use the
vertical refresh in 72Hz and time it from a PAL base freq. of 50Hz. It's
be about as useful as any other MIPS test.
KUNISAWA Ryota
>
> That does it. I'm writing a MIPS test that only uses NOP instructions.
>
From Motorola's "Programmer's Reference Manual" 4-147,
"The NOP instruction does not begin execution until all pending
bus cycles have completed. This synchronizes the pipeline and
prevents instruction overlap."
Doesn't this cause a bad effect on MIPS testing?
--
-----------------------------------------------------------------
KUNISAWA Ryota University of Tokyo
Postgraduate Student Graduate School of Science
email:kuni...@is.s.u-tokyo.ac.jp Dept of Information Science
ED BARCIK
JC> @FROM :jo...@MCS.COM N
JC> @UMSGID :<3uevts$t...@Mercury.mcs.com> N
JC> @UNEWSGR:01comp.sys.amiga.misc N
JC> From: jo...@MCS.COM (John Crookshank)
JC> Newsgroups: comp.sys.amiga.misc
JC> Message-ID: <3uevts$t...@Mercury.mcs.com>
JC> On 16-Jul-95 21:33:00, ANDERS BAKKEVOLD (anders.b...@norway.bbs.no)
JC> wrote:
JC> > Does anyone have some comparable speed tests for the 68060 vs Pentium?
JC> Benchmarks printed in Video Toaster User magazine show the '060 to be
JC> about half the speed of a P90.
I'm not really up on this and don't proclaim to know but have seen
numerous posts regarding the above and they all say that the 060 is
faster than the Pentium. The 040 is supposed to be as fast or
faster than the 486/66.
Just my .02
Ed Barcik in Cleveland, OH at ed.b...@amcom.com
... What's worse, a part missing or a part left over?
* Q-Blue 2.0 *
Maxwell Daymon
: : : p90 would look pretty lame compared to my Amiga, seeing that just one of the
: : : risc processors would beat it, let alone the 9 together.
: : That's a fantasy. Didn't you just say people shouldn't be talking if
: a 200MHz DEC Alpha hip inside of it...
Yes, but I saw no reference to a "200MHz DEC Alpha" - I saw "just one of
the RISC processors would beat it" with no qualification, no evidence.
Also, the post said "pretty lame compared to my Amiga" (which doesn't
have a RISC processor, it is based on the 680x0)
The Amiga is a very nice machine IMHO, but it doesn't pay to be
unrealistic.
James McArthur
The Amiga needs POWER! The sooner we get PPC Amigas or whatever the better..
James McArthur
: James McArthur (jam...@morinda.it.ntu.edu.au) wrote:
: : Wouldnt one raptor kick the pentiums arse in rendering? I thought it had
: : a 200MHz DEC Alpha hip inside of it...
: But we are discussing the 060 vs pentium (read title of thread).
: Wouldn't a CRAY kick the 200mHz Alphas arse............
Yeah but the question was something like, wouldnt an "Amiga with 9 RISC
processors beat a Pentium with 1". Off topic - Yes. Sensible Question? Who
knows...
: gavan
Michael van Elst
>have made big progress with the pentium and if you look at the sysinfo
>MIPS ( OK, OK, I know its not a perfect test) of a 50mHz 060 vs a 25mHz 040
>the 060 has TWICE as many MIPS but it runs at TWICE the clockspeed - very
>dodgy......
That's why SysInfo's speed test has no real meaning. It just checks a
very tiny part of the chip that is only twice as fast. If you use
instructions that execute in a single cycle you can only scale linearily
with the clock speed.
>Plus the pentium has a 64-bit interface to memory (the 060
>has 32) and I think that a 90 mhz pentium can access memory at up to 60 mhz
>but the 060 can only access at 33? (can anyone comment?)
The 060 can transfer a longword per clock. This doesn't mean that any
060 board will do that, RAM at that speed would be pretty expensive.
> I'd bet an Amiga with 060 could beat the pants off a p90 running NT for
>multitasking etc, but for number-cruncing the p90 has the edge....
Yes. Although it looks as if you'd need a very smart compiler for the p90
most often. I had no problems to get good results for the 060 with SAS C.
Regards,
--
Michael van Elst
Internet: mle...@serpens.rhein.de
"A potential Snark may lurk in every tree."
Maxwell Daymon
^^^^^^^^^^^^
: My mate reviewed the 060 for Format and told me he was shocked on the MIPS
: findings himself. He was heartened to find that renders and suchlike benefitted
: from 3x to 5x increases though.
No wonder you have such a contorted idea about the 68060. When will you
get it through your head about MIPS?
: MIPS tests have always been a bit dodgy - but when you get a MIPS test
: on two different models of the same CPU familty its a _bit_ more accurate.
A bit, but hardly. MIPS take a serious fall when the instruction set
changes (030 -> 040, 040 -> 060) because you CAN use better instructions
(like MOVE16) that didn't exist when the test was devised.
Result? Bogus figures.
: Unless of course the sysinfo test picks some stupid instruction like NOP
: as the 'I' in MIPS.
A MIPS test is supposed to contain a mix of instructions that would be
found "in the real world" (but I really question the choices). NOP is
slow, as pointed out, in a pipelined architecture so your results would
be very slanted.
: No, Pentiums are still fast. I use both 60 and 90 mhz models running on
: Windows NT machines and they _are_ fast for computationally heavy tasks.
Yes. I agree with that, but an 060 running code designed with the 060 in
mind will certainly give that P60 a run, and without a lot of L2 cache
may even beat it in PURELY COMPUTATIONAL tasks. The P90 is certainly a
work horse, but at a 30MHz bus, isn't it?
: You have an Amiga running 9 RISC processors? WOW!
Impressive huh? And he was just criticising people for making things up.
Glenn Saunders
: It's not even a *valid* point considering the cost of a good graphics card
: for the PC compared to the cost of getting good graphics on the Amiga. (AGA
: doesn't cut it anymore ... actually it never did).
No, but whenever I see a little too much one-sided ECS/AGA bashing I feel
compelled to remind people that this is an apple and oranges argument.
AGA still can do things which takes a lot more CPU power and often STILL
looks rougher on clones. It just so happens that _most_ of the things we
want from graphics today, the clones can do faster.
It is true that C= should have released AAA when it could have made a
difference. But C= should never have abandoned the chipset for clone
chips. The reason this is necessary from Escom is because there is no
viable chipset that can carry on the legacy and compete in all areas.
Ideally C= would have been strong enough to complete enough R&D to
produce a chipset which could beat the clones AND do its own unique things.
It's nice being able to RTG to any 3rd party card, but it's nicer if the
company is state of the art enough to be able to boast about having the
best and most unique cards for their own machines.
John Crookshank
>: Wouldnt one raptor kick the pentiums arse in rendering? I thought it
>: had a 200MHz DEC Alpha hip inside of it...
Saying someone has a DEC Alpha CPU is about as descriptive as saying that
they have an Intel CPU. Which one, and at what speed? There are several
generations of Alpha processors, some of the older ones are not too
much faster than the latest Pentiums. The latest 300Mhz DEC Alpha CPUs as
used in the Raptor3, the 21164 CPU, average about 8x the speed of a P100.
The older 21064 CPU, which is currently being sold in systems like Carrera
and Aspen for example, is half the speed of the 21164 used in the
Raptor3, when clock speeds are the same. Very much like the difference
between a 25Mhz 030 vs a 25Mhz 040.
And of course, the Intel crowd is working on the P6, but then
again, the DEC crowd is working on a 600-900Mhz version, so the leap-
frog continues.
--------------------------------------------------------------
| John Crookshank | jo...@mcs.com |
| MicroTech Solutions, Inc. | http://www.mcs.net/~johnc/ |
|--------------------------------------------------------------|
| Desktop video systems dealer - finger/http for more info. |
--------------------------------------------------------------
Mat Bettinson
In a message dated 25 Jul 95 you wrote to All :
MGM> : Now I know you don't know what you are talking about.. 060 50mhz only
MGM> twice : as fast as the 040 25mhz, mip wise?? Where on earth did you
MGM> come up with : this and how on earth would you expect anybody to
MGM> believe such crap?
Sysinfo wise, something is obviously wrong when it thinks that the 50Mhz
Cyberstorm is only twice as fast as a 25Mhz 68040 with a crippled memory bus...
MGM> I got that from the MIPS figures published in Amiga Format and CU
MGM> Amiga. My mate reviewed the 060 for Format and told me he was shocked
MGM> on the MIPS findings himself. He was heartened to find that renders
MGM> and suchlike benefitted from 3x to 5x increases though.
MIPS really are fairly pointless though it's interesting to see what SysInfo
comes up with given the same test on different CPUs. It is the only benchmark
that every Amiga owners has and so we have to use it. However, CU Amiga didn't
ONLY use that. You'll see from render tests etc that that rating isn't quite
indicative of real work speeds.
James McArthur
: :The point is *all* Amigas come with graphics chips on the motherboard. PC's
: :dont.
: It's not even a *valid* point considering the cost of a good graphics card
: for the PC compared to the cost of getting good graphics on the Amiga. (AGA
: doesn't cut it anymore ... actually it never did).
: Think about what you're trying to say about the Amiga. Having an
: antiquated graphics chipset built in to new Amigas is *NOT* an advantage!
Hey, I wasnt saying it was an advantage, I was just replying to the original
part of a post that said that no PC's come with standard graphics chips.
Thats all. I dont think that I'd ever be able to advocate the use of AGA
in a modern computer...
Michael van Elst
>code. If you look at the overhead of x86 code compared to 680x0-code you will
>certainly receive something like 2.5 intel-ops/cycle on a 060!
Have a look at the performance of existing Pentium and 060 systems and
you will find that your claim has not much to do with reality.
>>Ugh, the 25MHz 68040 in the A4000 is pretty slow.
>And the 50 MHz 060-cyberstorm is five times faster than an A4000...
More like 3.5-4 times as fast on average.
Philipp Boerker
>>
>>>>Summing up an Amiga with an 060 will be as fast as a P90-clone!
>>>And the 060 probably has more room for growth.
>>
>>The only growth of the 680x0-series we can expect is higher MHz for the 060
>>because Motorola will not develop an 070 or anything of this kind :-(.
>>
>Motorola doesn't seem to agree with you. They said quite the opposite! They
>said the had no plans to discontinue development of the 680x0 series. Where
>people come up with the idea that 060 is the last processor of the 680x0
>series? Motorola has publicly stated quite the opposite.
Well, I admit that I relied on the statements of the amiga-developers who must
be victims of this global ibm-conspiracy too.
>As far as rendering, Amiga can have risc procesors added for rendering.
This is infact a very interesting idea! I can even think of a possibility to
deal with this: just use an 68LC060 and an PPC604. Everytime that an
Unknown Opcode Exception occurs use the autovector-code for emulation. Or try
to do an "hardware emulation" of the coprocessor protocol (is there such a thing
on >030 ?). Hey, you hardware wizzards! Can you comment this? (BTW,I'm not a
moron, I know that it is at least "unlikely" to work)
>How people can state things they THINK will happen in the future and always
>be dead wrong when the time comes is stupid. Until things are known as fact,
>don't come out with stories or opinons without saying it is a personal
>opnion of yours and not true.
It wasn't a mere guess and not an opinion. I believed what people were saying.
Now I'm more careful: I don't believe you until I have seen this statement myself!
(But I hope you are right!)
Regards,
Phil.
----------------------------
rawn...@sp.zrz.tu-berlin.de
----------------------------
jtv
Maxwell Daymon <mda...@rainbow.rmii.com> wrote:
A bit, but hardly. MIPS take a serious fall when the instruction set
>changes (030 -> 040, 040 -> 060) because you CAN use better instructions
>(like MOVE16) that didn't exist when the test was devised.
>
>Result? Bogus figures.
I don't think that's the problem we're seeing here: the old instruction set
and mix are still as fair a representation of existing applications as they
were before.
A more important factor would be IMO that these kinds of test try to use a
large subset of the instruction set, so that rarely-used instructions are
overrepresented; some of those just didn't improve much because they weren't a
priority, and some were even moved to software so they are many times slower.
>A MIPS test is supposed to contain a mix of instructions that would be
>found "in the real world" (but I really question the choices). NOP is
>slow, as pointed out, in a pipelined architecture so your results would
>be very slanted.
Maybe we should be measuring performance as the *inverse* of the speed at
which NOP excecutes then. At least it's original!
Michael van Elst
>Wouldnt one raptor kick the pentiums arse in rendering? I thought it had
>a 200MHz DEC Alpha hip inside of it...
The Raptor uses MIPS CPUs (R4400 or similar) and it runs Windows NT.
Gavan Moran
>
>bra...@informatik.tu-muenchen.de (Christian Brandt) writes:
>
>> And now the Problem:
>
>> An LC060/50 costs the same as a Pentium 100.
>> An 060/66 costs the same as a Pentium 133...
>> An PPC604/120 costs less than a Pentium 60...
>
>Completeley irrelevant. You cannot just buy CPU chips and plug them in.
The cheaper the chips - the cheaper a computer built from them can be.
Theres lots of people making Intel compatible chips and prices are a LOT lower
than the equivalent 680x0 chip.
Glenn Saunders
: If you had any clue to begin with you would know that Amiga's can use
: risc processors for rendering. I suppose you should just learn more about
That's fine if you like spending $2000 on the networking software
(screamernet). And pros will tell you about some of the glitches that
happen with this network due to some kind of incompatibility or bad
tolerance between Amiga and Mips ethernet.
The Raptor phenomenon evolved out of not having LW ported to other
machines and having a need for speed. MIPS was the first platform to be
tackled (for rendering slaves only). Screamernet with Amigas will be of
far less importance in an age of a LW standalone that has the builtin
distributive rendering (which only works for LIKE platforms). People will
then string up multiple Pentiums. That is, if Newtek indeed offers
distributive rendering... I've heard too much conflicting rumors to know
for sure.
Economically speaking the amiga makes little sense anymore for rendering.
For Toaster/Flyer applications, yes, but no longer is it THE platform for
rendering. It's too expensive and too slow. Anyone who is into rendering
as a business will choose the cheaper route and will put up with the
slings and arrows of DOS/Windows, or NT. Those who use their Amigas out
of personal preference and are more hobbyists who don't mind having less
throughput will stick with the Amiga. If there were a CPU available that
was about 5-10 times as fast as the 060 and half the price of the current
060 then the Amiga would again be popular for rendering but that isn't
going to happen within the remaining CISC lifecycle of the Amiga.
Rendering people are all-consumed with raw speed. They don't care as
much about nice operating systems. Render slaves are essentially
single-tasking machines anyway. There are, of course, benefits for
having your main machine with image processor et. al. on a nice OS with
an IPC language and such, but this is secondary.
Glenn Saunders
: Well what you seem to fail to understand is that since the Amiga's OS is
: so much better than the crap os's on ibm. We do not require as fast a cpu
: to do the same work. IBM's require tons of memory, cpu power, hd space, etc
This is fine for getting multiple low-CPU-drain programs churning
together to be productive. Ironically, this would have been very useful
in the business community where a lack of multitasking means wasting time
and wasting money, but that's a different issue.
When dealing with apps that are infinitely taxing (renderers or image
processors) the Amiga starts to show its limitations because no amount of
integration will get beyond the fact that it has to do x number of
calculations and this is just going to take LONGER than other platform.
Maybe a LOT longer.
: If I had a choice between an 040 Amiga and a P6 IBM of the Future.. I would
: take the Amiga 040 anyday! IBm's need to scrap everything and start over
: from scratch. Making faster processors to cover up the crap foundation is
: not going to fix the problems with ibms. Just hides it better.
I would not encourage anyone who doesn't require big-taxing-apps to buy a
machine with a shit OS even if it were 10 times as fast as the 060
because like I said, the ability to get a bunch of
less-processor-intensive apps working together can result in greater
productivity on the Amiga than a faster platform.
Unless something comes along down the road that becomes very popular that
amigas simply can not do even with an 060 then it will be time to realize
that platform is truly archaic, the same way 8-bit platforms are on a
lower level due to their 64K limit, lack of 80 collumns (usually),
inability to compile ANSI C, etc...
The amiga enjoyed many years of OS superiority but this won't necessarily
last forever. At whatever point another machine does everything the Amiga
can and more and for a lower price, then this will test the loyalties of
Amigophiles everywhere. You'll have to live by your words if you can't
find anything to whine about concerning this hypothetical OS. To many,
Windows NT is already this OS. Its only problem seems to be resource
hungriness. To others, they see TAOS as a giant in waiting and how can
you argue with the specifications? You can't expect computer companies
to keep their heads in the sand forever concerning OS improvements. The
longer the Amiga languishes in CISC the more time other platforms will
have to overtake it on each and every issue.
Glenn Saunders
: mhz between 2 different processor families doesn't mean a thing and
: can not be compared.
You are using this as a copout excuse. The fact is that the 060 ihas a
bottleneck that the Pentium doesn't and it is notable that the 060
compares as well as it does, but overall, despite the more "elegant"
design, the Pentium is available in incarnations that outdo it simply
through brute-force (which is, of course, Intel's solution to everything).
: Also who really cares if pentiums have higher mhz or even faster chips.
: The Amiga systems will always seem faster to the user that is using the
: computer. My 040 Amiga seems alot faster than the pentiums when using it
: for misc. tasks and everyday use. IBm's have such crap os's that they will
: need processors 10x faster to match what it is like using an Amiga on
: Amiga OS.
:
: Besides, Motorola processors have always had better quality and speed when
: compared to its equivlant from intel. Take a 486 50mhz compared to an 040 50mhz.
: Compare a 386 50mhz to 030 50mhz, etc etc.. Motorola are always faster.
Not anymore. Brute force wins over elegance. If the world only had
Pentiums up to 90mhz maybe the 060 could compete (with apps that use 060
optimizations).
: But see on Amiga we don't require the speed like ibm people do, processor speed
: is not a factor that we need badly, unlike on ibm.
Tell that to rendering people.
tem...@netins.net
>Maxwell Daymon (mda...@rainbow.rmii.com) wrote:
>: tem...@netins.net wrote:
>: : >multitasking etc, but for number-cruncing the p90 has the edge....
>
>: : p90 would look pretty lame compared to my Amiga, seeing that just one of the
>: : risc processors would beat it, let alone the 9 together.
>
>: That's a fantasy. Didn't you just say people shouldn't be talking if
>
>: Your extreme hypocracy is noted.
>
>a 200MHz DEC Alpha hip inside of it...
Yup that was what I was refering to.. You can stack 9! Raptor's together.
Rendering a hell of alot faster than any pentium. The guy just didn't like
the post and didn't know anything about the Amiga. He of course must be just
a simple ibm lamer that spews out crap.
To the ibm guy who didn't know, it was based on facts, just that you don't
have a clue about the Amiga..
tem...@netins.net
>tem...@netins.net wrote:
>: >MIPS ( OK, OK, I know its not a perfect test) of a 50mHz 060 vs a 25mHz 040
>: >the 060 has TWICE as many MIPS but it runs at TWICE the clockspeed - very
>: >has 32) and I think that a 90 mhz pentium can access memory at up to 60 mhz
>: >but the 060 can only access at 33? (can anyone comment?)
>
>: Now I know you don't know what you are talking about.. 060 50mhz only twice
>: as fast as the 040 25mhz, mip wise?? Where on earth did you come up with
>: this and how on earth would you expect anybody to believe such crap?
>
> I got that from the MIPS figures published in Amiga Format and CU Amiga.
>My mate reviewed the 060 for Format and told me he was shocked on the MIPS
>findings himself. He was heartened to find that renders and suchlike benefitted
>from 3x to 5x increases though.
>
Well your mate is an idiot. The mips are 4-5 times higher. Not twice as
much. You and your mate are wrong and stupid for even thinking it would be
only 2x. hehehe. the 060 50mhz is 80+ mips. the 040 25mhz is maybe 19mips
or close to it. Can you do the simple math now?
>
>:
>: Try 4-5 times faster mips wise, at least! You think an 060 50mhz runs at
>: only 38 mips or so? Think again guy. I know you must hate that Amiga's
>: are good computers and ibm's are crap but lets get things a bit more
>: realistic and not rediculously stupid.
>
>
> MIPS tests have always been a bit dodgy - but when you get a MIPS test
>on two different models of the same CPU familty its a _bit_ more accurate.
>Unless of course the sysinfo test picks some stupid instruction like NOP
>as the 'I' in MIPS.
>
>
Exactly, your mate used some stupid util probably.
>: Pentiums might be fast processors, but placed on an ibm with their crap
>: os's, forget about it. They loose most of their speed instantly.
>
>
> No, Pentiums are still fast. I use both 60 and 90 mhz models running on
>Windows NT machines and they _are_ fast for computationally heavy tasks.
>
Yeah pentiums processors are fast, ibm os's are not.
>
>
>: Unless I stacked 9 Risc processors together for rendering, then I bet that
>: p90 would look pretty lame compared to my Amiga, seeing that just one of the
>: risc processors would beat it, let alone the 9 together.
>
>
> You have an Amiga running 9 RISC processors? WOW!
>
I supose you didn't actually read the message did you? I said unless I
added a 9 risc processor system for rending. This does not mean I am using
the system does it? But if I wanted to I could yes. Ever heard of
Raptor System for the Amiga? Probably not, you don't seem to be that well
informed.
tem...@netins.net
>Glenn Saunders (kri...@primenet.com) wrote:
>: The mortal tem...@netins.net wrote:
>: : and in 80 mhz this fall.. don't listen to idiot ibm users... they have shown
>
>: Don't get our hopes up unless you can cite sources for your claims that
>: Moto will have an 80mhz 060 this fall.
>
> When are the 120 and 133 mhz 060s coming out?????
>
> There are pentiums of that speed available TODAY.
>
>gavan
>
>--
>email: G.M...@ee.qub.ac.uk | 'There can be only one!'
> or gmo...@nyx.cs.du.edu | - The Highlander
mhz between 2 different processor families doesn't mean a thing and
can not be compared.
Also who really cares if pentiums have higher mhz or even faster chips.
The Amiga systems will always seem faster to the user that is using the
computer. My 040 Amiga seems alot faster than the pentiums when using it
for misc. tasks and everyday use. IBm's have such crap os's that they will
need processors 10x faster to match what it is like using an Amiga on
Amiga OS.
Besides, Motorola processors have always had better quality and speed when
compared to its equivlant from intel. Take a 486 50mhz compared to an 040 50mhz.
Compare a 386 50mhz to 030 50mhz, etc etc.. Motorola are always faster.
tem...@netins.net
>James McArthur (jam...@morinda.it.ntu.edu.au) wrote:>: : : Unless I stacked 9 Risc processors together for rendering, then I bet that
>: : : p90 would look pretty lame compared to my Amiga, seeing that just one of the
>: : : risc processors would beat it, let alone the 9 together.
>: Wouldnt one raptor kick the pentiums arse in rendering? I thought it had
>: a 200MHz DEC Alpha hip inside of it...
>
>the RISC processors would beat it" with no qualification, no evidence.
>
>Also, the post said "pretty lame compared to my Amiga" (which doesn't
>have a RISC processor, it is based on the 680x0)
>
>The Amiga is a very nice machine IMHO, but it doesn't pay to be
>unrealistic.
If you had any clue to begin with you would know that Amiga's can use
risc processors for rendering. I suppose you should just learn more about
the Amiga before trying to call something fantasy.
One thing ibm probably will never have is a decent os like the Amiga though.
Ibm users will just have to keep suffering like they have from day one. .hehe
tem...@netins.net
>The mortal tem...@netins.net wrote:
>
>: Motorola doesn't seem to agree with you. They said quite the opposite! They
>: said the had no plans to discontinue development of the 680x0 series. Where
>
>its availability, even a P6like 080 would not be worth the wait if it
>takes 2-3 years to be completed and then another year and a half to
>trickle into production.
>
>
Well what you seem to fail to understand is that since the Amiga's OS is
so much better than the crap os's on ibm. We do not require as fast a cpu
to do the same work. IBM's require tons of memory, cpu power, hd space, etc
just to do sluggish work. Amiga's do not require even half the cpu power to
do more. The 66mhz 060 is out already and the 80mhz is comming this fall.
I personally feel that the 040 35mhz I have now is more than enough for anything
I need it to do. If I had an 060 50mhz I don't know what I would do with the
extra speed. It would be overkill I suppose, on ibm however you always can
use more cpu power since the crap os's and crap software runs so poorly.
tem...@netins.net
>In article <3unhnl$f...@serpens.rhein.de>, mle...@serpens.rhein.de says...
>>
>>bra...@informatik.tu-muenchen.de (Christian Brandt) writes:
>>
>>> And now the Problem:
>>
>>> An LC060/50 costs the same as a Pentium 100.
>>> An 060/66 costs the same as a Pentium 133...
>>> An PPC604/120 costs less than a Pentium 60...
>>
>>Completeley irrelevant. You cannot just buy CPU chips and plug them in.
>
>
> The cheaper the chips - the cheaper a computer built from them can be.
>Theres lots of people making Intel compatible chips and prices are a LOT lower
>than the equivalent 680x0 chip.
>
>--
>email: G.M...@ee.qub.ac.uk | "There can be only one!"
> or gmo...@nyx.cs.du.edu | - The Highlander
>
You get what you pay for. Buy cheap ibm's you get a cheap computer. Buy
a nice Computer, like the Amiga, you get a nice machine.
Motorola has always been know to be the best chips and the more reliable as
well.
tem...@netins.net
>The mortal tem...@netins.net wrote:
>
>: can not be compared.
>
>You are using this as a copout excuse. The fact is that the 060 ihas a
>bottleneck that the Pentium doesn't and it is notable that the 060
>compares as well as it does, but overall, despite the more "elegant"
>design, the Pentium is available in incarnations that outdo it simply
>through brute-force (which is, of course, Intel's solution to everything).
>
heheheheheehe...
Answer this: Why does the 060 50mhz! run faster than a Pentium 60mhz!
If mhz are equal and the Pentium is a better processor with no bottenecks,
why would a slower mhz 060 beat the faster mhz pentium? Intel chips have
alays been slower on all the ranges when compared mhz to mhz.
030 50mhz is faster than 386 50mhz
040 50mhz is faster than 486 50mhz, by a mile I might add.
060 50mhz is faster than Pentium 60mhz.
It may be true that Intel comes out with faster mhz before motorola does, but
it is also true that the Motorola's of euqual mhz are always faster, it is
also true that motorola chips are more reliable chips. With the Amiga computer
however, we do not require the brute force, ibm's do.
>: Also who really cares if pentiums have higher mhz or even faster chips.
>: The Amiga systems will always seem faster to the user that is using the
>: computer. My 040 Amiga seems alot faster than the pentiums when using it
>: for misc. tasks and everyday use. IBm's have such crap os's that they will
>: need processors 10x faster to match what it is like using an Amiga on
>: Amiga OS.
>:
>: Besides, Motorola processors have always had better quality and speed when
>: compared to its equivlant from intel. Take a 486 50mhz compared to an 040 50mhz.
>: Compare a 386 50mhz to 030 50mhz, etc etc.. Motorola are always faster.
>
>Not anymore. Brute force wins over elegance. If the world only had
>Pentiums up to 90mhz maybe the 060 could compete (with apps that use 060
>optimizations).
>
Really? When I use an 040 Amiga and a Pentium 60mhz, I see the Amiga running
much nicer than the pentium.. Let alone an 060 Amiga.. could of fooled me
>: But see on Amiga we don't require the speed like ibm people do, processor speed
>: is not a factor that we need badly, unlike on ibm.
>
>Tell that to rendering people.
>
hehehe.. Rendering people have other choices for processing power and make up
a small percentage of computer users. the other 99% of the computer users would
see a much faster computer when using the Amiga as opposed to the pentium ibm's.
I think you are wasting your time thinking in terms of cpu mips and less on the
reality of how ibm's suck even with the fastest of processors.
>
C.L. Simard
Well.. I find it strange that Ibm users are comparing their p60-90 to
the 060.. Pick up a magazine with an ESCOM interview. Escom sold the most
pentiums overseas, when they say that a 060 is as fast (or faster) as a
p90 then you'd better listen (They had 50% of the european market last year
selling PC) Its also strange that one of the biggest Ibm developers
(Escom) says that the Amiga has a yet to be matched multi-tasking..!
Don't believe me? Go buy AMAZING COMPUTING/Amiga.
Its also starnge that I see Vector lines and plotters (have you ever seen
18930 dots on a PC screen? plotting?) running at faster speed on my
old Amiga500 7mhz (I now have A4000) then on my brother 486dx2 66mhz?
Strange? Naww I call it Amiga.. Ibm can have all the pentiums it wants
(needs) it still won't match an 060 in real usage (multi-tasking)! Ibm
doesn't just lack multi-tasking.. No sprites, smooth scrolling.........
Sorry about the long post..!
C.L. Simard
P5(6?) 64-bit? Maybe 64 bit like the Neo-GEo was 24 bit and the sega was
16 bit right? Neo-Geo had a 68000 and a z80, the sega genesis had a 6800
and a z80. Tell me what was difference between the 2 that made the neo-geo 24
bit and the sega only 16? Easy.. Marketing! To a average consumers 64 bit
sounds better then 32 bit! which is exactly what most 64bit systems
are.. I've seen video cards for Ibm that say they are 64bit, but did not
perform any better then my brother old 32 bit, you know why? because it
only dumps 32-bit chunks at one time. the only speed increase that happens is
while one 32 bit chunk is being dumped the other is being read next chunk..
And if the p5 (or p6?) is like this it doesn't make that big a
difference, and to point out another fact that was mention. Have you ever
noticed that Ibm is trying to look and behave more like an Amiga.. 256
charactor file names coming in Win95? Heck A1000 had that, When IBM had
286 monochrome amiga had 4096 colors! And please don't say you have 16
million colors on screen? because you don't. You have a pallate of 16
million but 1024x768 = 786432. You can only have as many colors on screen
as pixels. Maybe you can tell me how you are doing 16 million colors with
only 786432 pixels??
To much said.. On something so lame as an IBM!
Michael van Elst
>to the beginning. Inside the branch of course only register-register ops...
>That would be the way I would time it if I was in the ad-biz ;-)
All instructions that can be issued every clock (and that's not just
register moves) can run superscalar.
Gavan Moran
>Hello Mr,
> In a message dated 25 Jul 95 you wrote to All :
> MGM> : Now I know you don't know what you are talking about.. 060 50mhz only
> MGM> twice : as fast as the 040 25mhz, mip wise?? Where on earth did you
> MGM> come up with : this and how on earth would you expect anybody to
> MGM> believe such crap?
>Sysinfo wise, something is obviously wrong when it thinks that the 50Mhz
>Cyberstorm is only twice as fast as a 25Mhz 68040 with a crippled memory bus...
Your quoting system on your mailer is a bit mixed up, the above
comments weren't made by me. I'd guess that the sysinfo MIPS test is
running totally in the cache so the crappy A4000/040 memory access
doesn't come into it.
A more meaningful comparison would be to run various benchmarks on
an A4000 with Warp Engine 040/28 against the 060/50. Maybe someone
has converted the SpecINT and SpecFP suites to the Amiga?
> MGM> I got that from the MIPS figures published in Amiga Format and CU
> MGM> Amiga. My mate reviewed the 060 for Format and told me he was shocked
> MGM> on the MIPS findings himself. He was heartened to find that renders
> MGM> and suchlike benefitted from 3x to 5x increases though.
>MIPS really are fairly pointless though it's interesting to see what SysInfo
>comes up with given the same test on different CPUs. It is the only benchmark
>that every Amiga owners has and so we have to use it. However, CU Amiga didn't
>ONLY use that. You'll see from render tests etc that that rating isn't quite
>indicative of real work speeds.
Its a pity AIBB doesn't work on the 060 at present - it gives a
good 'feel' for how fast the machine is.
One thing I didn't like about the CU amiga review was that it
compared the 060 results to those from an 030/25 which made it look at
LOT faster - it should really have been compared to the 040,
preferably in a 'warp engine' guise.
gavan
--
email: G.M...@ee.qub.ac.uk | "There can be only one"
OR gmo...@nyx.cs.du.edu | - the Highlander
Gavan Moran
>In article <3v293c$4...@pellew.ntu.edu.au>, jam...@morinda.it.ntu.edu.au (James McArthur) writes:
>>Maxwell Daymon (mda...@rainbow.rmii.com) wrote:
>>: tem...@netins.net wrote:
>>: : > I'd bet an Amiga with 060 could beat the pants off a p90 running NT for
>>: : >multitasking etc, but for number-cruncing the p90 has the edge....
>>
>>: : Unless I stacked 9 Risc processors together for rendering, then I bet that
>>: : p90 would look pretty lame compared to my Amiga, seeing that just one of the
>>: : risc processors would beat it, let alone the 9 together.
>>
>>: That's a fantasy. Didn't you just say people shouldn't be talking if
>>: their statements weren't based on fact?
>>
>>: Your extreme hypocracy is noted.
>>
>>Wouldnt one raptor kick the pentiums arse in rendering? I thought it had
>>a 200MHz DEC Alpha hip inside of it...
>Yup that was what I was refering to.. You can stack 9! Raptor's together.
>Rendering a hell of alot faster than any pentium. The guy just didn't like
>the post and didn't know anything about the Amiga. He of course must be just
>a simple ibm lamer that spews out crap.
>To the ibm guy who didn't know, it was based on facts, just that you don't
>have a clue about the Amiga..
Ahem, the raptor ISN'T an Amiga - its a seperate Dec Alpha based
NT machine that connects via ethernet. You just use the Amiga as a
front end. Its just like the Mac and Windows Toasters were originally
- a seperate machine in a box connected to a front end. Plus of
course all the raptor can do at the moment is act as a render-engine
for lightwave, you can run a huge array of software on a pentium
machine. The two aren't comparable in any way and this has nothing to
do with 060 vs pentium either.......
Gavan Moran
>In article <3v31ea$3u...@info4.rus.uni-stuttgart.de>, gavinm@vsprsun_14 ( Mr Gavin Moran) writes:
>>tem...@netins.net wrote:
>>
>>: Unless I stacked 9 Risc processors together for rendering, then I bet that
>>: p90 would look pretty lame compared to my Amiga, seeing that just one of the
>>: risc processors would beat it, let alone the 9 together.
>>
>>
>> You have an Amiga running 9 RISC processors? WOW!
>>
>I supose you didn't actually read the message did you? I said unless I
>added a 9 risc processor system for rending. This does not mean I am using
>the system does it? But if I wanted to I could yes. Ever heard of
>Raptor System for the Amiga? Probably not, you don't seem to be that well
>informed.
>
That Raptor system is NOT an Amiga. They are NOT an Amiga
subsystem, they do NOT act as Amiga processors. The raptor is a
SEPERATE machine that you can stick an Amiga front end on.
What other things can you do on a Raptor 'Amiga' system? Can you
run other 3-D rendering packages faster? Does SimCity2000 run at a
decent speed? Does it speed up your compilations?
I think not.......
Gavan Moran
>In article <3vbs04$p...@nnrp2.primenet.com>, kri...@primenet.com (Glenn Saunders) writes:
>>The mortal tem...@netins.net wrote:
>>
>>: mhz between 2 different processor families doesn't mean a thing and
>>: can not be compared.
>>
>>You are using this as a copout excuse. The fact is that the 060 ihas a
>>bottleneck that the Pentium doesn't and it is notable that the 060
>>compares as well as it does, but overall, despite the more "elegant"
>>design, the Pentium is available in incarnations that outdo it simply
>>through brute-force (which is, of course, Intel's solution to everything).
>>
>heheheheheehe...
>Answer this: Why does the 060 50mhz! run faster than a Pentium 60mhz!
As far as I know, it doesn't. I've seen mags say that the Pentium is
faster in integer opertations very slightly, mhz for mhz with the 060.
Until someone posts some SpecINT and SpecFP tests the arguments are going
to continue.
>If mhz are equal and the Pentium is a better processor with no bottenecks,
>why would a slower mhz 060 beat the faster mhz pentium? Intel chips have
>alays been slower on all the ranges when compared mhz to mhz.
>030 50mhz is faster than 386 50mhz
There is no 50mhz 386, 40 mhz was the fastest 386 speed.
>040 50mhz is faster than 486 50mhz, by a mile I might add.
There was no 50mhz 040, just a 40mhz model.
>060 50mhz is faster than Pentium 60mhz.
The court is still out on that one.
>It may be true that Intel comes out with faster mhz before motorola does, but
>it is also true that the Motorola's of euqual mhz are always faster,
You are just so used to the notion of 680x0 chips outperforming x86
chips that you can't open your mind to the possibility that the Pentium
outperforms the 68060.
The 060 is only ever due to go to 80mhz, 133 mhz pentiums are out NOW,
a 150mhz model will be out soon. Then the 686 is due out, the 060 will
never be superceded by a new 680x0 model as its the end of the line in
terms of more powerful models.
> it is
>also true that motorola chips are more reliable chips. With the Amiga computer
>however, we do not require the brute force, ibm's do.
For RENDERING it does, and Image Processing - the Amiga is a lot more
efficient for everyday tasks but the sort of tasks that require heavy CPU
power are just as demanding on an Amiga as they are on a PC.
>>: Also who really cares if pentiums have higher mhz or even faster chips.
>>: The Amiga systems will always seem faster to the user that is using the
>>: computer.
Not for the sort of people who need powerful machines - 3d artists,
photo-retouching, mathematical modelling ........
>>: My 040 Amiga seems alot faster than the pentiums when using it
>>: for misc. tasks and everyday use. IBm's have such crap os's that they will
>>: need processors 10x faster to match what it is like using an Amiga on
>>: Amiga OS.
Thats why I'm not too worried personally - I get by on an 030/28mhz on
my Amiga and for everyday use its as nice to use as the Pentium 60mhz I'm
typing this on.
Maxwell Daymon
: tem...@netins.net wrote:
: >040 50mhz is faster than 486 50mhz, by a mile I might add.
: There was no 50mhz 040, just a 40mhz model.
Correct, and the 486DX2-50 is a 25MHz based bus. They had the Pentium by
the time the 68040 got to 40MHz, and now you can get a 486DX4-100
: >It may be true that Intel comes out with faster mhz before motorola does, but
: >it is also true that the Motorola's of euqual mhz are always faster,
: The 060 is only ever due to go to 80mhz, 133 mhz pentiums are out NOW,
: a 150mhz model will be out soon. Then the 686 is due out, the 060 will
: never be superceded by a new 680x0 model as its the end of the line in
: terms of more powerful models.
FWIW, the P6 isn't going to be all that much better than the P5. There
are no architectural improvements, and the main change is to put a 256K
cache on the chip. I'd say it's a definite sign that intel has come to
the end of the line with CISC.
However, with L2 cache and EDO-RAM, it will certainly outperform any
68060 made available. The PowerPC will be far beyond it (IMHO) by that
time, and the PPC will have a future.
: >also true that motorola chips are more reliable chips. With the Amiga computer
: >however, we do not require the brute force, ibm's do.
: For RENDERING it does, and Image Processing - the Amiga is a lot more
: efficient for everyday tasks but the sort of tasks that require heavy CPU
: power are just as demanding on an Amiga as they are on a PC.
Once you spend a lot of time processing, the brute force wins. Especially
in an environment where you can hog processor time.
: >>: Also who really cares if pentiums have higher mhz or even faster chips.
: >>: The Amiga systems will always seem faster to the user that is using the
: >>: computer.
: Not for the sort of people who need powerful machines - 3d artists,
: photo-retouching, mathematical modelling ........
It depends completely on the clone and configuration of it.
: Thats why I'm not too worried personally - I get by on an 030/28mhz on
: my Amiga and for everyday use its as nice to use as the Pentium 60mhz I'm
: typing this on.
For me it comes down to what works. I don't need a penis extension MHz or
MIPS figure to put on my car, shirt, underwear, etc. to convince people
that I'm special. If I was rendering, I'd buy a DEC Alpha or MIPS4400 or
4600 based system.
Philipp Boerker
>rawn...@virgo.zrz.TU-Berlin.DE (Philipp Boerker) writes:
>>to the beginning. Inside the branch of course only register-register ops...
>>That would be the way I would time it if I was in the ad-biz ;-)
>All instructions that can be issued every clock (and that's not just
>register moves) can run superscalar.
Yes, and most of these instructions are register-register-ops. When I say this I mean:
OP.x Rx,Ry
and not
OP.x ADDRESS,Rx.
Zsolt Szabo
C.L. Simard <u921...@muss.cis.McMaster.CA> wrote:
>
> P5(6?) 64-bit? Maybe 64 bit like the Neo-GEo was 24 bit and the sega was
>16 bit right? Neo-Geo had a 68000 and a z80, the sega genesis had a 6800
>and a z80. Tell me what was difference between the 2 that made the neo-geo 24
The Genesis had a Motorola 68000 16 bit CPU. It was binary compatible
with the Amiga--that's why many games were ported.
--
__
-/_)_ ( _ __) __)__
_/ \(_)_)_)_/(_/_/(-__
Brian King
>At 7:40 PM on 17 Jul 95, John Crookshank said to All:
>JC> Benchmarks printed in Video Toaster User magazine show the '060 to be
>JC> about half the speed of a P90.
>I'm not really up on this and don't proclaim to know but have seen
>numerous posts regarding the above and they all say that the 060 is
>faster than the Pentium. The 040 is supposed to be as fast or
>faster than the 486/66.
Yes, this applies when talking about the SAME clock speed. If you take a
60MHz Pentium, and a nice 60MHz 060 system, they will be comparable.
Just like the 25MHz 040 is comparable to the 25MHz 486. But not to the
486DX2/66/80 etc.
--
+-------------------------------------+-----------------------------------+
| Brian D. King - Computer Engineer | Fourth Dimension Software, author |
| PBX/ISDN Software Development Group | of LhA-GUI, and the Fusion-40 FAQ |
| Mitel Corporation, Kanata, Ontario | ***** Brian...@Mitel.com ***** |
Jeroen T. Vermeulen
Brian King <ki...@software.mitel.com> wrote:
>ed.b...@amcom.com (ED BARCIK) writes:
>
>>At 7:40 PM on 17 Jul 95, John Crookshank said to All:
>>JC> Benchmarks printed in Video Toaster User magazine show the '060 to be
>>JC> about half the speed of a P90.
>
>>I'm not really up on this and don't proclaim to know but have seen
>>numerous posts regarding the above and they all say that the 060 is
>>faster than the Pentium. The 040 is supposed to be as fast or
>>faster than the 486/66.
>
>Yes, this applies when talking about the SAME clock speed. If you take a
>60MHz Pentium, and a nice 60MHz 060 system, they will be comparable.
>
>Just like the 25MHz 040 is comparable to the 25MHz 486. But not to the
>486DX2/66/80 etc.
I suppose it still depends on what you measure--in FP work the 040 and the
Pentium seem to be on the same level for the same clock speed. I found the
100MHz 486DX4 quite disappointing as compared to my A4000/40 (unless you
work in text mode of course <snigger> like all those Linux enthousiasts).
I've also heard some "interesting" comparisons between the A4000 and a
50MHz 486DX... <Hans--was that the story you told me?>
One famous comparison (in Byte, I believe) measured 68k vs. x86 performance
by writing a Z80 emulation (a chip emulating its older cousin vs. a chip
emulating a radically different design) and not registerizing its variables
so the power of the 68000 was hardly used while the Intel code couldn't be
improved that much anymore..
Guess what--the Intel won! Hooray. Superior technology. Such power.
>+-------------------------------------+-----------------------------------+
>| Brian D. King - Computer Engineer | Fourth Dimension Software, author |
>| PBX/ISDN Software Development Group | of LhA-GUI, and the Fusion-40 FAQ |
>| Mitel Corporation, Kanata, Ontario | ***** Brian...@Mitel.com ***** |
Jeroen
Jyrki Saarinen
> Just like the 25MHz 040 is comparable to the 25MHz 486. But not to the
> 486DX2/66/80 etc.
No it is not. The 040/25 is almost 2x faster in integer and
about 3x in floating point
Maxwell Daymon
: Brian King posted the following:
: : Yes, this applies when talking about the SAME clock speed. If you take a
: : 60MHz Pentium, and a nice 60MHz 060 system, they will be comparable.
: the 60 060 is approximately the same integer speed as a P90.
: the 060 is faster at FP operations though ;)
I thought there were only 060@50MHz and 060@66MHz planned/produced.
--
+------------------------------------+
| Maxx Daymon -+- mda...@rmii.com |
+------------------------------------+
Alan L.M. Buxey
: >
: > P5(6?) 64-bit? Maybe 64 bit like the Neo-GEo was 24 bit and the sega was
: >16 bit right? Neo-Geo had a 68000 and a z80, the sega genesis had a 6800
: >and a z80. Tell me what was difference between the 2 that made the neo-geo 24
clever advertiseing. NeoGeo had the idea that 16-bit + 8 bit is 24.
SEGA had just technical advertising and stated that the 68000 was 16-bit
;)
alan
Alan L.M. Buxey
: Yes, this applies when talking about the SAME clock speed. If you take a
: 60MHz Pentium, and a nice 60MHz 060 system, they will be comparable.
the 60 060 is approximately the same integer speed as a P90.
the 060 is faster at FP operations though ;)
alan
J_PL...@amtrash.comlink.de
(04.08.1995)
> ed.b...@amcom.com (ED BARCIK) writes:
>
> >At 7:40 PM on 17 Jul 95, John Crookshank said to All:
> >JC> Benchmarks printed in Video Toaster User magazine show the '060 to be
> >JC> about half the speed of a P90.
>
> >I'm not really up on this and don't proclaim to know but have seen
> >numerous posts regarding the above and they all say that the 060 is
> >faster than the Pentium. The 040 is supposed to be as fast or
> >faster than the 486/66.
>
> 60MHz Pentium, and a nice 60MHz 060 system, they will be comparable.
That might be correct for the chip itself...(SPECs)
The complete system is much faster...it doesn`t render with the speed of a
Pentium but it SEEMS to be EXTREMLY FAST while working and it is running a
Quadra060:-) powerful in the background..
Thanks to MicroSoft for NO AmigaWintendo!
I spoke to a few Cybervision+Cyberstorm060 user...I`ll order this 060_beast
when there is virtual memory for it...hopefully in near future.
/ J_Pl...@Amtrash.comlink.de \
/ Plew...@Informatik.fh-hamburg.de \
+-------------------------------------+
\ A3000/030/6/52 A3000T/040/18/512 /
Mr Gavin Moran
: : Yes, this applies when talking about the SAME clock speed. If you take a
: : 60MHz Pentium, and a nice 60MHz 060 system, they will be comparable.
: the 60 060 is approximately the same integer speed as a P90.
: the 060 is faster at FP operations though ;)
I doubt it - that makes the 060 considerably faster than a PPC601 at the
same clockspeed.
The 060 is slightly _slower_ integer wise than the Pentium at the same
clockspeed.
gavan
--
email: G.M...@ee.qub.ac.uk | 'There can be only one!'
Jeroen T. Vermeulen
In article <407di7$t...@info4.rus.uni-stuttgart.de> gavinm@vsprsun_14 ( Mr Gavin Moran) writes:
> Alan L.M. Buxey (kc...@central.susx.ac.uk) wrote:
>
> : the 60 060 is approximately the same integer speed as a P90.
> : the 060 is faster at FP operations though ;)
>
> I doubt it - that makes the 060 considerably faster than a PPC601 at the
> same clockspeed.
The 601 is a lot faster than the 040, but it's not the fastest thing in the
world. It's supposed to be a cheap low-end chip; the 060 is apparently
two-thirds the speed of a 604 at the same clock speed. (Damn--I've forgotten
the name of the person who mailed me this tidbit! You know who you are)
The 604 is supposedly faster than the 601; it's from the same generation but
came out a lot later than the 601 so presumably it took more work to design.
> The 060 is slightly _slower_ integer wise than the Pentium at the same
> clockspeed.
Don't believe a word of it. Even counting in MIPS the 060 is faster--and it
needs a lot less instructions for the same amount of work.
> gavan
> --
> email: G.M...@ee.qub.ac.uk | 'There can be only one!'
> or gmo...@nyx.cs.du.edu | - The Highlander
Jeroen
tem...@netins.net
>Alan L.M. Buxey (kc...@central.susx.ac.uk) wrote:
>: : Yes, this applies when talking about the SAME clock speed. If you take a
>: : 60MHz Pentium, and a nice 60MHz 060 system, they will be comparable.
>
>: the 060 is faster at FP operations though ;)
>
>
>--
>+------------------------------------+
>| Maxx Daymon -+- mda...@rmii.com |
>+------------------------------------+
You forgot about the 80mhz version as well, which is comming.
060 66mhz beats the Pentium 90mhz. The 80mhz version will surely beat the
pentium 1xx mhz versions. This and the fact that ibm's have crap os's, the
Amiga looks better and better all the time.. I personally have no need for
an 060 however, the 040 is more than enough on an Amiga. 060's will be nice
for those that render though. Poor ibm people suffer even on pentiums with
normal operations.
Amiga person has to ask if they really need 060 speed, ibm user doesn't
need to think, they of course need as much speed as they can get to run their
crap.
Mr Gavin Moran
: In article <407di7$t...@info4.rus.uni-stuttgart.de>, gavinm@vsprsun_14 ( Mr Gavin Moran) writes:
: >
: > The 060 is slightly _slower_ integer wise than the Pentium at the same
: >clockspeed.
: >
: hehehe.. where do you get this idea? Intel processors are all the sudden
: faster? When they have always been considerabley slower at same clock speeds
: for over a decade? Motorola processors have always been faster at the same
: clock speed and still are. 060's at the same clock speed as pentiums are
: in fact faster. Even more so when you consider the AMiga OS compared to ibm
: junk.
What we need are some specINT and specFP benchmarks for Pentiums and
060s at various clockspeeds. Have Motorola published figures?
Just relying on 'the 680x0 series is _always_ faster than the x86' isn't
a valid argument that the 060 will be faster than the Pentium.
tem...@netins.net
>: Brian King posted the following:
>
>: : Yes, this applies when talking about the SAME clock speed. If you take a
>: : 60MHz Pentium, and a nice 60MHz 060 system, they will be comparable.
>
>: the 60 060 is approximately the same integer speed as a P90.
>: the 060 is faster at FP operations though ;)
>
>same clockspeed.
>
>clockspeed.
>
>--
>email: G.M...@ee.qub.ac.uk | 'There can be only one!'
> or gmo...@nyx.cs.du.edu | - The Highlander
hehehe.. where do you get this idea? Intel processors are all the sudden
Espen Berntsen
>: hehehe.. where do you get this idea? Intel processors are all the sudden
>: faster? When they have always been considerabley slower at same clock speeds
>: for over a decade? Motorola processors have always been faster at the same
>: clock speed and still are. 060's at the same clock speed as pentiums are
>: in fact faster. Even more so when you consider the AMiga OS compared to ibm
>: junk.
>
> What we need are some specINT and specFP benchmarks for Pentiums and
>060s at various clockspeeds. Have Motorola published figures?
>
> Just relying on 'the 680x0 series is _always_ faster than the x86' isn't
>a valid argument that the 060 will be faster than the Pentium.
The 060 and he Pentium did a render test on Lightwave (same scene) and the 060
finished somewhere between 5-15 seconds faster. To just check the NOP
instructions the publish is not very likely to be a fair test. You should
always test it in a user environment, where you have the burden of the OS and
other tasks (hehehe) to slow down the processor. What is the use for a
processor that can make 300 MIPS when the instruction is a NOP?? I like to know
what is left for ME when the system gets its share.
--
Espen.B...@hiMolde.no <> irc: Nameless <> http://www.himolde.no/~espen
===========================================================================
Fighting for peace is like fucking for virginity.
James C C Darling
My MSc report is due in soon, so it is getting rather large. Too large for my
6Meg A4000/030 in fact. So I thought I'd experiment with using other systems.
Word 6 became the program of choice, since it is a bit similar to Final Writer
3
(a lot nastier to use, though !). At the Uni, we have 90Mhz Pentiums, with
16Megs RAM. I tried them. Cool. Word 6 ran OK for a change. However, it was
only about twice as fast at rendering imported graphics than my 030 Amiga.
Then I noticed
that the Word6 document was over 10Megs in size, where the same one in Final
Writer was 800K . Huh ?
I also then noticed that an ftp session transferring this 10Meg doc from the
network to the PC's hard disk more than halved in speed when I was using Word.
Some multitasking.
Being used to proper computers which multitask (Sun, Amiga, DEC) I thought I'd
try the shiny new 66Mhz PowerMacs, with 18Megs RAM, in the Mac lab next door.
Oh dear. Ooooooooooh dear. They are ___shite____. Word 6 was running slower
than it does on my girlfriends' 4Meg '486 33. I just feel sorry for all the
people that have been going round the Maths department here singing their
praises. And guess what ? Still no proper multitasking. While waiting for a
document to print, I thought I'd have a quick sesh on Solitaire. Then I
twigged that printing virtually stopped when I was using any other program than
Word. Maybe the Macs are good. Perhaps it is just Microsqwish Word that is the
pile of shite.
So I have ended up buying some more RAM and an '040 for my A4000.
The moral is : a fast processor is all very well - but if it is in a shite
computer with a poor OS, it isn't worth a light !
James
--
----------------------------------------------------------------------------
(####) James C C Darling BEng(hons) AMIEE AUS
(#######)
"Don't have a cow, (#########) Tel: +44 (0)1483-300800x2212
man !" (#########) +44 (0)1483-60756
(#########) email: J.Da...@ee.surrey.ac.uk
(#########)
__&__ (#########)
/ \ (#########) |\/\/\/| /\ /\ /\ /\
| | (#########) | | | V \/ \---. .----/ \----.
| (o)(o) (o)(o)(##) | | \_ / \ /
C .---_) ,_C (##) | (o)(o) (o)(o) <__. .--\ (o)(o) /__.
| |.___| /____, (##) C _) _C / \ () /
| \__/ \ (#) | ,___| /____, ) \ > (C_) <
/_____\
Brian Kamman
>> The 060 is slightly _slower_ integer wise than the Pentium at the same
>> clockspeed.
>Don't believe a word of it. Even counting in MIPS the 060 is faster--and it
>needs a lot less instructions for the same amount of work.
Sorry Jeroen. MIPS is an artifical measurement. It's hard to
compare processors in a family by MIPS. It is not possible to compare
different processors such as Intel vs HP vs Motorola by MIPS.
See MIPS means millions of instructions per second. This example
is extended way beyond reality to make a point but in finer measures it works
just the same. Chip A does 2 million instructions per second.
Therefore, two mips. Chip B does 1 million instructions per second.
Therefore, one mip. It takes Chip A 4 million instructions to grab two
chunks of memory and add them togther and print them out to the screen.
This means Chip A will perform this tasks in two seconds. It takes Chip
B 1.5 million instructions do do the same task. That means Chip B will
perform this task in 1.5 seconds. Look below.
Chip A Chip B
2MIP chip 1MIP chip
Time for our task 2 seconds 1.5 seconds.
Thus, Chip B with a lower number of MIPS actually out performs
Chip A for this task. This is taken from the second day of my chip
design class as to why MIPS is not a useful comparison.
Now double CHIP B to be a 2 MIP chip. It now takes .75 seconds
to do the same task CHIP A does and each is a 2 MIP chip.
Different chips have different ways on how to do things. MIPS is
useless for cross chip comparisons.