AMD K5-133 FP performance
Henrik Schmiediche
Hello
does anyone know anything about the Floating Point performance of the
K5-133 compared to an Intel Pentium 133. Any benchmark numbers
available?
I crosspost to Quake not because I care much about Quake, but since
Quake is FP intensive, perhaps someone there knows?
Thanks,
- Henrik
--
Henrik Schmiediche, Dept. of Statistics, Texas A&M, College Station, TX 77843
E-mail: hen...@stat.tamu.edu | Tel: (409) 862-1764 | Fax: (409) 845-3144
Finger for pgp 2.6 key, fingerprint: E867 D9DB 9616 5DAC 0F67 FE98 77FE 8583
Michael D. Boulanger
From my experience in benchmarking I have done, the AMD performs about
25% worse in most FPU operations. This is on a P100 vs an AMD PR-100.
I don't know if anything will change with the 133 Mhz version yet.
Haven't been able to get my hands on one.
--
************************************************************************
* Michael D. Boulanger Wankel Rotary Dragon *
* mboul...@ids.net -==[UDIC]==- *
* 1986 Mazda RX-7 Sport Package *
************************************************************************
Piston engine goes "Boing-ditty-boing-ditty-boing" but the Mazda goes
"Hmmmm"
Henrik Schmiediche <hen...@stat.tamu.edu> wrote in article
<56hv0p$i...@news.tamu.edu>...
Kevin Podszebka
Henrik Schmiediche wrote:
>
> Hello
> does anyone know anything about the Floating Point performance of the
> K5-133 compared to an Intel Pentium 133. Any benchmark numbers
> available?
>
> I crosspost to Quake not because I care much about Quake, but since
> Quake is FP intensive, perhaps someone there knows?
>
> Thanks,
>
> - Henrik
>
> --
> Henrik Schmiediche, Dept. of Statistics, Texas A&M, College Station, TX 77843
> E-mail: hen...@stat.tamu.edu | Tel: (409) 862-1764 | Fax: (409) 845-3144
> Finger for pgp 2.6 key, fingerprint: E867 D9DB 9616 5DAC 0F67 FE98 77FE 8583
It would be best to compare the K5-133 to a P75. Though the K5 is
SLIGHTLY faster. I should know, I have one.. As for the FPS, that
would still depend on more than just the processor speed.
--
/ ____
/ / _ . /__)_ _ / _ / / / _
_/ \ _(/_/|/_/_/)_ __/ _(_)_(_/_/)_/)_(/_/_)_/ \_(_|_
\_______________________________/)________________
___
_/_ _ _ /) . _ _
_/__ ><_(__(/_//_/)_/_(_)/(_
Wayne Smith
Kevin Podszebka wrote:
>
> Henrik Schmiediche wrote:
> >
> > Hello
> > does anyone know anything about the Floating Point performance of the
> > K5-133 compared to an Intel Pentium 133. Any benchmark numbers
> > available?
> >
> > I crosspost to Quake not because I care much about Quake, but since
> > Quake is FP intensive, perhaps someone there knows?
> >
> > Thanks,
> >
> > - Henrik
> >
> > --
> > Henrik Schmiediche, Dept. of Statistics, Texas A&M, College Station, TX 77843
> > E-mail: hen...@stat.tamu.edu | Tel: (409) 862-1764 | Fax: (409) 845-3144
> > Finger for pgp 2.6 key, fingerprint: E867 D9DB 9616 5DAC 0F67 FE98 77FE 8583
>
> It would be best to compare the K5-133 to a P75. Though the K5 is
> SLIGHTLY faster. I should know, I have one.. As for the FPS, that
> would still depend on more than just the processor speed.
> --
>
much higher than a p75!...........Wayne
Francesco Di Tolla
Kevin Podszebka wrote:
> It would be best to compare the K5-133 to a P75.
post some data please to support your statement
regards
Franz
--
Francesco Di Tolla, Center for Atomic-scale Materials Physics
Physics Departement, Build. 307, Technical Univesity of Denmark,
DK-2800 Lyngby, Denmark, Tel.: (+45) 4525 3208 Fax: (+45) 4593 2399
mailto:dit...@fysik.dtu.dk http://www.fysik.dtu.dk/persons/ditolla.html
Bergen Friundervisningen
Kevin Podszebka (kpds...@dreamscape.com) wrote:
: Henrik Schmiediche wrote:
: >
: > Hello
: > does anyone know anything about the Floating Point performance of the
: > K5-133 compared to an Intel Pentium 133. Any benchmark numbers
: > available?
: >
: > I crosspost to Quake not because I care much about Quake, but since
: > Quake is FP intensive, perhaps someone there knows?
: >
: > Thanks,
: >
: > - Henrik
: >
: > --
: > Henrik Schmiediche, Dept. of Statistics, Texas A&M, College Station, TX 77843
: > E-mail: hen...@stat.tamu.edu | Tel: (409) 862-1764 | Fax: (409) 845-3144
: > Finger for pgp 2.6 key, fingerprint: E867 D9DB 9616 5DAC 0F67 FE98 77FE 8583
: It would be best to compare the K5-133 to a P75. Though the K5 is
: SLIGHTLY faster. I should know, I have one.. As for the FPS, that
: would still depend on more than just the processor speed.
: --
I think you have the AMD 5x86 133 (486 on steroids) and the AMD K5-133
mixed up. The latter is a Pentium clone that has by many been compared
to Intels Pentium 166 in performance.
Best regards.
Kjartan Maraas
: / ____
Jeff Terrill
Wayne Smith wrote:
>
> Kevin Podszebka wrote:
> >
> > Henrik Schmiediche wrote:
> > >
> > > Hello
> > > does anyone know anything about the Floating Point performance of the
> > > K5-133 compared to an Intel Pentium 133. Any benchmark numbers
> > > available?
> > >
> > > I crosspost to Quake not because I care much about Quake, but since
> > > Quake is FP intensive, perhaps someone there knows?
> > >
> > > Thanks,
> > >
> > > - Henrik
> > >
> > > --
> > > Henrik Schmiediche, Dept. of Statistics, Texas A&M, College Station, TX 77843
> > > E-mail: hen...@stat.tamu.edu | Tel: (409) 862-1764 | Fax: (409) 845-3144
> > > Finger for pgp 2.6 key, fingerprint: E867 D9DB 9616 5DAC 0F67 FE98 77FE 8583
> >
> > It would be best to compare the K5-133 to a P75. Though the K5 is
> > SLIGHTLY faster. I should know, I have one.. As for the FPS, that
> > would still depend on more than just the processor speed.
> > --
> >
> much higher than a p75!...........Wayne
AMD sucks. Buy the Intel if your looking for a new chipset. AMD is like
the Cyrix 6x86 -> FRY, FRY, FRY!!!!
Tom
I believe your thinking of the AMD 5X86 133 not the AMD K5-PR133
>
> > It would be best to compare the K5-133 to a P75.
>
Tom
Tom <td...@ix.netcom.com> wrote in article
<01bbd5ad$50fc6f80$8dc7...@Tom.ix.netcom.com>...
> I believe your thinking of the AMD 5X86 133 not the AMD K5-PR133
>
>
> >
> > > It would be best to compare the K5-133 to a P75.
> >
> > post some data please to support your statement
>
>
begin 600 PR133NU.jpg
<uuencoded_portion_removed>
=SJ]Q>Q:3:W$5V)[22$H6:,K]X#(.#T]/I0(`_]D`
`
end
Pete Olson
Jeff Terrill (jt...@Intrex.net) wrote:
:
: AMD sucks. Buy the Intel if your looking for a new chipset. AMD is like
: the Cyrix 6x86 -> FRY, FRY, FRY!!!!
You're gonna have to back that one up if you want to
convince anybody. I've got years of experience with
AMD chips and have never even heard of a heat problem
with them, ever. _EVER_
Tom
I've been using AMD since the 80286 and find them to be stable and
compatible. Cyrix doesn't even belong in the same league as AMD. My K5-90
didn't even have to use thermal compound it ran so cool.
tom
Pete Olson <nor...@bigdipper.umd.edu> wrote in article
<56ret7$t...@hecate.umd.edu>...
Dr Martin Atkinson-Barr
Radioman wrote:
>
> Jeff Terrill <jt...@Intrex.net> wrote:
>
> >:
> >:AMD sucks. Buy the Intel if your looking for a new chipset. AMD is like
> >:the Cyrix 6x86 -> FRY, FRY, FRY!!!!
>
> Not much hotter than a steel mill....
>
Interestingly this may not be the case. Evidently Compaq is looking at
the K6 for notebooks. see:
http://techweb.cmp.com/crn/issues/710_711/
Suggests that the K6 may have a low power consumption.
LGN
In article <01bbd636$b2659380$52c7...@Tom.ix.netcom.com>, "Tom" <td...@ix.netcom.com> writes:
> I've been using AMD since the 80286 and find them to be stable and
> compatible. Cyrix doesn't even belong in the same league as AMD. My K5-90
> didn't even have to use thermal compound it ran so cool.
>
> tom
Nither do my K5-100, i just have a cheapo crap fan and my cpu is as cold as it's not even beeing use'd...
/L-GN
Charlie Savko
> AMD sucks. Buy the Intel if your looking for a new chipset. AMD is like
> the Cyrix 6x86 -> FRY, FRY, FRY!!!!
always used AMD. AMD chips do not get hot, I have a 586/133 with no CPU
fan and a real tiny heatsink, it is always slightly warm to touch.
You're just another ignorant, mindless freak who never tried an AMD.
-Charlie Savko
sa...@epix.net
Chris Quirke
Jeff Terrill <jt...@Intrex.net> wrote:
>AMD sucks. Buy the Intel if your looking for a new chipset. AMD is like
>the Cyrix 6x86 -> FRY, FRY, FRY!!!!
Nope.
AMD is fabbed at .35, it's the .50 Cyrix 686 that fry, although as of
stepping 2.7 it's now .44 - maybe one day IBM will let 'em use a
decent fab?
jc
Bergen Friundervisningen wrote:
>
> Kevin Podszebka (kpds...@dreamscape.com) wrote:
> : Henrik Schmiediche wrote:
> : >
> : > Hello
> : > does anyone know anything about the Floating Point performance of the
> : > K5-133 compared to an Intel Pentium 133. Any benchmark numbers
> : > available?
> : >
> : > I crosspost to Quake not because I care much about Quake, but since
> : > Quake is FP intensive, perhaps someone there knows?
> : >
> : > Thanks,
> : >
> : > - Henrik
> : >
> : > --
> : > Henrik Schmiediche, Dept. of Statistics, Texas A&M, College Station, TX 77843
> : > E-mail: hen...@stat.tamu.edu | Tel: (409) 862-1764 | Fax: (409) 845-3144
> : > Finger for pgp 2.6 key, fingerprint: E867 D9DB 9616 5DAC 0F67 FE98 77FE 8583
>
> : SLIGHTLY faster. I should know, I have one.. As for the FPS, that
> : would still depend on more than just the processor speed.
> : --
>
> mixed up. The latter is a Pentium clone that has by many been compared
> to Intels Pentium 166 in performance.
Stephen Knilans
In article <329000...@dreamscape.com> kpds...@dreamscape.com writes:
>Henrik Schmiediche wrote:
>>
>> Hello
>> does anyone know anything about the Floating Point performance of the
>> K5-133 compared to an Intel Pentium 133. Any benchmark numbers
>> available?
>>
>> I crosspost to Quake not because I care much about Quake, but since
>> Quake is FP intensive, perhaps someone there knows?
>>
>> Thanks,
>>
>> - Henrik
>>
>> --
>> Henrik Schmiediche, Dept. of Statistics, Texas A&M, College Station, TX 77843
>> E-mail: hen...@stat.tamu.edu | Tel: (409) 862-1764 | Fax: (409) 845-3144
>> Finger for pgp 2.6 key, fingerprint: E867 D9DB 9616 5DAC 0F67 FE98 77FE 8583
>
>
>It would be best to compare the K5-133 to a P75. Though the K5 is
>SLIGHTLY faster. I should know, I have one.. As for the FPS, that
>would still depend on more than just the processor speed.
>--
>
The K5-133 is SLIGHTLY faster than a P75?!?!?!?!?!?!?!? I just saw a K5-166
side by side to a P5-166. GUESS which one was faster! The K5-166 is MUCH
faster! I had to wait MINUTES to see the same pictures on the P5-166!
BTW, the K5-166 runs at 133Mhz!
Perhaps you are confusing the NEW technology P5 compatible
(software AND hardware) 5K86(aka K5) with the enhanced 486 compatible
(software AND hardware) 5X86! The 5X86 is in EVERY way a TOTALLY different
chip! Some idiot in marketting gave them similar names, but that is IT!
Steve
Daniel Nash
In article <stephenkE...@netcom.com>, step...@netcom.com (Stephen
Knilans) wrote:
: In article <329000...@dreamscape.com> kpds...@dreamscape.com writes:
: >Henrik Schmiediche wrote:
: >It would be best to compare the K5-133 to a P75. Though the K5 is
: >SLIGHTLY faster. I should know, I have one.. As for the FPS, that
: >would still depend on more than just the processor speed.
: >--
: >
:
: The K5-133 is SLIGHTLY faster than a P75?!?!?!?!?!?!?!? I just saw a K5-166
: side by side to a P5-166. GUESS which one was faster! The K5-166 is MUCH
: faster! I had to wait MINUTES to see the same pictures on the P5-166!
: BTW, the K5-166 runs at 133Mhz!
:
: Perhaps you are confusing the NEW technology P5 compatible
: (software AND hardware) 5K86(aka K5) with the enhanced 486 compatible
: (software AND hardware) 5X86! The 5X86 is in EVERY way a TOTALLY different
: chip! Some idiot in marketting gave them similar names, but that is IT!
:
: Steve
Ummm, there is no 5k86 P166, unless you are watching some pre-release demo
or something. AMD has announced K5-P75, P90, P100, P120 and P133. The
K5-P133 is supposedly as fast or faster than an Intel P133 for integer
operations. The K5-P133 runs at 100 MHz. AMD has said they will release a
K5-P166 by the end of the year, I think, but there has not been an
announcement yet.
- Daniel Nash
David Wight
According to AMD's Website.
The K5 processors (not to be confused with 5x86 processors) are a pin
for pin pentium alternative. The 133 is not the speed of the chip,
rather it is a pentium comparison benchmark. Supposedly, the tested a
computer with a pentium 133, removed the pentium chip and installed
their chip and ran the same tests. The 133 designation means that the
chip delivered the same performance as the pentium of that speed. The
results of the tests are posted on their website.
For more info: http://www.amd.com
BTW does anyone know where I can buy the K5 PR133?
Stephen Knilans
It WAS a prerelease, scheduled for 1Q/97! They HAVE announced it!
Steve
Daniel Nash
: In article <danash-ya0232800...@news.doit.wisc.edu>
dan...@students.wisc.edu (Daniel Nash) writes:
: >Ummm, there is no 5k86 P166, unless you are watching some pre-release demo
: >or something. AMD has announced K5-P75, P90, P100, P120 and P133. The
: >K5-P133 is supposedly as fast or faster than an Intel P133 for integer
: >operations. The K5-P133 runs at 100 MHz. AMD has said they will release a
: >K5-P166 by the end of the year, I think, but there has not been an
: >announcement yet.
: >
: > - Daniel Nash
:
: It WAS a prerelease, scheduled for 1Q/97! They HAVE announced it!
:
: Steve
Where's the announcement? I double-checked around their web site, and
there's not a peep about it. Where did you see this, by the way, at Comdex
or something like that?
- Daniel Nash
Responsible Mikael Ronstrom
Is there anyone who know's if i can find it at a descent price in sweden ??
The best i've heard here is 1100:- skr, something like $166.50, or is there any american company that ship stuff all around the world ?? or at least to the ECC ??
Guillermo Gonzalez
All these bickering assholes, can't someone out there with some REAL
knowledge post a set of Quake benchmark numbers using the AMD K5-Pr133?
I suspect its FP performance is about on par with a regular P5-100. Its
interger performance is probably at least as good as a P5-133 , therefore
I would expect its Quake performance to be somewhere inbetween a P5-100 and
a P5-133.
Stephen Knilans
In PC Currents page 3, they show it. BTW, I DID see it at comdex.
Steve
Marco S. Passafaro
>The K5-133 is SLIGHTLY faster than a P75?!?!?!?!?!?!?!? I just saw a K5-166
>side by side to a P5-166. GUESS which one was faster! The K5-166 is MUCH
>faster! I had to wait MINUTES to see the same pictures on the P5-166!
>BTW, the K5-166 runs at 133Mhz!
>
>Perhaps you are confusing the NEW technology P5 compatible
>(software AND hardware) 5K86(aka K5) with the enhanced 486 compatible
>(software AND hardware) 5X86! The 5X86 is in EVERY way a TOTALLY different
>chip! Some idiot in marketting gave them similar names, but that is IT!
>
>Steve
Ok give us some numbers. If the K5 is so much supperior to the P5 why doen't
AMD post numbers like SpecInts and SpecFlops. As long as I don't see any
numbers I'll assume that the P5 is better.
(BTW the processor is not the only component, if a PPro 200 runs windows on
2MB you can wait forever to see any pictures)
Just my two cents of bandwith.
Marco
Stephen Knilans
In article <573tr3$o...@euas20.eua.ericsson.se> n...@eua.ericsson.se writes:
>Is there anyone who know's if i can find it at a descent price in sweden ??
>The best i've heard here is 1100:- skr, something like $166.50, or is there any american company that ship stuff all around the world ?? or at least to the ECC ??
>
>
>
I would EXPECT the Swedish price to be AT LEAST $144. The rest could be a duty,
and extra profit.
Heck, My danish aunt/uncle are having me send them things so they won't pay
the VERY high duties! That can REALLY add up!
Steve
Tom
yes the have announced it and it is posted on the AMD web site. its under
roadmap for the K5 in the K5 processor section.
tom
Stephen Knilans
Actually, FEW companies do this! Intel gives Icomp numbers that are SO
unreliable that they TOTALLY changed them! I asked them to tell me how
they arrived at them, and they told me "We are not in the benchmark
business, and we use this just to test ours."
The K5 would have better Specint numbers, maybe 20%+ better, and worse
Specflop numbers(about 10-25% worse). This is comparing the P5/X to
the AMD K5PRX. In other words, AMDs 133 product to intels 166.
BTW, I HAVE seen win95 run in 8MB on the P5. It isn't pretty. The P5
at the show was NOT that bad! Also, both systems were supposed to be
IDENTICAL! I'll tell you one thing. The defrag was slower on the AMD!
If they cheated, then why not with that?
Steve
Frankly, Intel may have something to worry about even if AMD was 30% slower,
which I guarantee it is NOT! The AMD chip is WIN95 licensed(M/S says it runs
Win95 WELL), and costs HALF what the comparable Intel product does!
Adam Williamson
In article <575jl9$i...@ds2.acs.ucalgary.ca>, "Marco S. Passafaro"
<pass...@acs.ucalgary.ca> writes
>Ok give us some numbers. If the K5 is so much supperior to the P5 why doen't
>AMD post numbers like SpecInts and SpecFlops. As long as I don't see any
>numbers I'll assume that the P5 is better.
>(BTW the processor is not the only component, if a PPro 200 runs windows on
>2MB you can wait forever to see any pictures)
>
there are for CPUs? I tried, and got bored somewhere around 23. And they
_all_ _give_ _different_ _results_. The only decent test is a real-world
one. I don't have any Quake numbers for an AMD 5k86/133 (give you one
guess why) but in a test someone did a while ago the 5k86/90 beat a
Cyrix P166+ and came below a P133 (it got about 16fps at the start of
the game).
--
Adam Williamson
Stephen Moore
That's because you haven't got your computer switched on. Switch it on
and watch
that mother burst into flames... :)
BTW As official r.g.c.q.m English police officer, I have to point out
that your
spelling of "used" is terrible. Sort it out.
Steve.
--
+-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-+
| Steve Moore - aka Granny Killer |
| Punishment Squad Quake Clan |
| mailto:*st...@shpcorp.dnet.co.uk* |
| WWW: http://www.niweb.com/dnet/dnetKhbw |
+-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-+
| Remove *'s before replying by email!! |
| SEX MURDER ART |
+-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-+
LGN
> > Nither do my K5-100, i just have a cheapo crap fan and my cpu is as cold as it's not even beeing use'd...
> >
> > /L-GN
>
> That's because you haven't got your computer switched on. Switch it on
> and watch
> that mother burst into flames... :)
Well well, i've had it running for houer's and it still doesn't get war (at least not so warm that you can feel it's hot) so bugger off, you're another of the Intel ppl who's sobing cause u bougt u'r self an expensive cpu...
> BTW As official r.g.c.q.m English police officer, I have to point out
> that your
> spelling of "used" is terrible. Sort it out.
>
>
> Steve.
> --
So what, bugger off... it's not that easy when it's not u'r birth languish...
And what the heck, i had the next best grade u can get here in sweden in english at school... and i just hate britt's cause they got to be so freak'n correct and all...
Tell me of someone that never ever do a spelling error...
I don't think u never do that.. and btw can u speak swedish ?? i can and i do understand norwegian to do u ?? and i do understand and speek a bit german but u brit's are too good to learn any other languish i presume or am i wrong ??
Please bugger off, or maybee pick that bugger out of u'r nose and well see if it help's....
And hey can u pronounce my name ?? i can pronounce u'r's...
/Lars-Göran Nilsson
Dennis, WI
THREE CHEERS FOR MR. NILSSON!
To coin a phrase given to one of our better Presidents,
"Give'em Hell Lars"
From a Yank in Wisconsin
Dennis
Jim Basilio
I don't have numbers. I read all this stuff before carefully deciding
what precessor to buy. I needed it for quake (and other applications ...
imagine that). But anyway I got an AMD K90 combod with a Matrox
Millenium. The K90 isn't all that .. benched around a p100 in integer ...
and maybe 25% lower in float. However, with the kickin card it reducing
at least one of the common bottle necks.
The processor has a *lot* to do with quake and other applications. But
don't kid yourself into thinking that $40 trident isn't hurting much.
Again, I don't have numbers, only experience, but a good video card goes a
long way.
Jim
David Gillespie
On 25 Nov 1996, Dennis, WI wrote:
> THREE CHEERS FOR MR. NILSSON!
> To coin a phrase given to one of our better Presidents,
> "Give'em Hell Lars"
>
To use a phrase from the original Star Trek series....
I WANT ANSWERS, MISTER!!!!!
I think the topic is self-expalnitory... Who has actual AMD k5-133 fp
numbers ALONGSIDE an Intel-133???? Logical and on topic Re's only!
David L. Gillespie \ /\ /\ /\ /\ /\ /\ Lab:
dg...@unm.edu |\ /||\||\ /||\||\ /||\||\ (505) 277-5774
University of New Mexico \||\||/ \||\||/ \||\||/ \| FAX:
Dept. of Biochemistry \/ \/ \/ \/ \/ \/ \ (505) 277-5389
www.unm.edu/~dgill
Bjorn Lovoll
I'm not writing this from where my k5-133 is located, but it gave quite low
benchmarks.
I have it in a FuguTech m520 MB (VX). It gave about 12/3 the mips and 1/2
the mflops of a P133.
If there is any interest I could post some specifics from home later on.
Also I tried to OC it a bit.
No luck there. I set the bus 75 (1.5x75) it came up as a AMDK5 at 110mhz,
while at 66x1.5 is comes up as an AMD133 at 100mhz. The text was not a
problem since it's clearly just a case of AMD not suppling an id string at
a clock rate they didn't anticipate. Win95(osr2) was stable, but Monster
Truck Madness was not, while it's perfectly stable and fairly snappy at
100mhz. The problem is not some other card, since the motherboard runs
fine at 75mhz (tested at 187.5), so I have to assume the problem is the k5.
I might test further and figure out what the break point is, but seems a
bit of work.
Jim Basilio <jab...@cac.psu.edu> wrote in article
<Pine.SUN.3.93.961125161343.16263A-100000@wilbur>...
> I don't have numbers. I read all this stuff before carefully deciding
> what precessor to buy. I needed it for quake (and other applications ...
> imagine that). But anyway I got an AMD K90 combod with a Matrox
> Millenium. The K90 isn't all that .. benched around a p100 in integer
..
Will Kim
In article <3293A0...@epix.net>, sa...@epix.net wrote:
>> AMD sucks. Buy the Intel if your looking for a new chipset. AMD is like
>> the Cyrix 6x86 -> FRY, FRY, FRY!!!!
>always used AMD. AMD chips do not get hot, I have a 586/133 with no CPU
>fan and a real tiny heatsink, it is always slightly warm to touch.
>You're just another ignorant, mindless freak who never tried an AMD.
Well, you're just another ignorant, mindless freak who never tried an Intel.
:P~
Wk
--
Will Kim MediaLight Inc.
wk...@medialight.com 20 Queen St W, Suite 208
416.598.3200 / 1.888.999.ADSL x222 Toronto, ON M5H 3R3 Canada
Designers Of The 1st ADSL PC Card http://www.medialight.com
J. Andrew Culpepper
In article <Pine.A32.3.91.961125...@pegasus.unm.edu>,
dg...@unm.edu says...
>
>On 25 Nov 1996, Dennis, WI wrote:
>
>> THREE CHEERS FOR MR. NILSSON!
>> To coin a phrase given to one of our better Presidents,
>> "Give'em Hell Lars"
>>
>To use a phrase from the original Star Trek series....
>I WANT ANSWERS, MISTER!!!!!
>
>I think the topic is self-expalnitory... Who has actual AMD k5-133 fp
>numbers ALONGSIDE an Intel-133???? Logical and on topic Re's only!
>
Shouldn't it be compared to processors similar in price ($115-$125)?
KKK
On 25 Nov 1996, LGN wrote:
> > BTW As official r.g.c.q.m English police officer, I have to point out
> > that your
> > spelling of "used" is terrible. Sort it out.
> >
> > Steve.
>
> And what the heck, i had the next best grade u can get here in sweden in english at school... and i just hate britt's cause they got to be so freak'n correct and all...
> Tell me of someone that never ever do a spelling error...
> I don't think u never do that.. and btw can u speak swedish ?? i can and i do understand norwegian to do u ?? and i do understand and speek a bit german but u brit's are too good to learn any other languish i presume or am i wrong ??
> Please bugger off, or maybee pick that bugger out of u'r nose and well see if it help's....
> And hey can u pronounce my name ?? i can pronounce u'r's...
>
> /Lars-Göran Nilsson
>
I guess this really isn't the channel to discuss linguistic matters...or
to stuck-up with one's skills. But I still must to some extent agree with
Steven. Lasse's spelling isn't that good - at least if considering the
'second best grade in English in Sweden' he has. I wonder... and guess we
had pretty strict grades then...
- KKK
Guillermo Gonzalez
On 25 Nov 1996 23:44:05 GMT, bi...@crosslink.net (J. Andrew Culpepper)
wrote:
>>I think the topic is self-expalnitory... Who has actual AMD k5-133 fp
>>numbers ALONGSIDE an Intel-133???? Logical and on topic Re's only!
>>
>
>Shouldn't it be compared to processors similar in price ($115-$125)?
Lets compare it to anything, but we need some FREAKIN NUMBERS!!!
'scuse me..
Richard Keith Patrick
I have an AMD 5x86-133 as well. On several occasions, the CPU has fallen
off the chip, often not to be discovered until weeks later. The chip has
never failed, although it does get hot.
One thing I will say in favor of Intel is this: I asked MIchael Abrash some
time back whether or not the optimizing that is done for Pentiums will also
optimize on Cyrix and AMD clones. His response was something along the
lines of 'probably not'. Whether or not this is true, there is something
to be said for paying a little more for a chip that is guaranteed to work
exactly as intended, optimizations and all. Those other chips will always
be just clones(very good ones, but still clones)
Stephen Moore <*st...@shpcorp.dnet.co.uk*> wrote in article
<57bvil$1...@bcrkh13.bnr.ca>...
> LGN wrote:
> >
> > In article <01bbd636$b2659380$52c7...@Tom.ix.netcom.com>, "Tom"
<td...@ix.netcom.com> writes:
> > > I've been using AMD since the 80286 and find them to be stable and
> > > compatible. Cyrix doesn't even belong in the same league as AMD. My
K5-90
> > > didn't even have to use thermal compound it ran so cool.
> > >
> > > tom
> >
> > Nither do my K5-100, i just have a cheapo crap fan and my cpu is as
cold as it's not even beeing use'd...
> >
> > /L-GN
>
> That's because you haven't got your computer switched on. Switch it on
> and watch
> that mother burst into flames... :)
>
> BTW As official r.g.c.q.m English police officer, I have to point out
> that your
> spelling of "used" is terrible. Sort it out.
>
>
> Steve.
Chris Quirke
Adam Williamson <Ad...@scss.demon.co.uk> wrote:
>In article <575jl9$i...@ds2.acs.ucalgary.ca>, "Marco S. Passafaro"
><pass...@acs.ucalgary.ca> writes
>>Ok give us some numbers. If the K5 is so much supperior to the P5 why doen't
>>AMD post numbers like SpecInts and SpecFlops. As long as I don't see any
>>numbers I'll assume that the P5 is better.
>Because numbers are pointless. You ever counted how many benchmarks
>there are for CPUs? I tried, and got bored somewhere around 23. And they
>_all_ _give_ _different_ _results_. The only decent test is a real-world
ROTFL!! If you think there are a lot of benchmarks that behave diff,
any idea how many diff apps/instruc.mixes/situations there are? Even
the same app will skew perf this way and that when you switch chip
innards (some fall off the table altogether, due to unfounded timing
assumptions)...
But, along with info on clock speeds, core multipliers, L1 cache etc.,
benchmarks are useful, and the more the better, *exactly* because they
give diff results. If you just want to feel good about your pet chip,
choose the ones with the numbers you like - the "throw the darts
first, call whatever you hit the target" approach ;-)
But if you really want an idea as to what's what, use low-level
benches that narrow the focus to certain aspects of design (e.g. FPU,
Integer, RAM access, perhaps even particular instruction categories)
to understand the chips better, then sanity-check that with
"real-world" and seat-of-the-pants results, and an understanding on
what keeps you waiting and what needs your particular sware have.
Otherwise, you can just go by whose ads boasts the most ;-)
bobby
On Fri, 22 Nov 1996 23:06:19 GMT, gonz...@bridge.net (Guillermo
Gonzalez) wrote:
I only have one benchmark i can remeber since i'm at work right now,
vid_mode 0 timerefresh = 20.6 and timedemo demo2 25.3
k5-133 on a ga586 512cache rev 1.55 matrox mystique
Mike Ching
In article <01bbdb6b$9bd48380$6fa720cc@IX>, Richard Keith Patrick wrote:
>I have an AMD 5x86-133 as well. On several occasions, the CPU has fallen
>off the chip, often not to be discovered until weeks later. The chip has
>never failed, although it does get hot.
>
>One thing I will say in favor of Intel is this: I asked MIchael Abrash some
>time back whether or not the optimizing that is done for Pentiums will also
>optimize on Cyrix and AMD clones. His response was something along the
>lines of 'probably not'. Whether or not this is true, there is something
>to be said for paying a little more for a chip that is guaranteed to work
>exactly as intended, optimizations and all. Those other chips will always
>be just clones(very good ones, but still clones)
You may be misinterpreting the 'probably not' comment. The Pentium
optimizations make code run faster by working around LIMITATIONS in
the Pentium architecture. The clones got designed later and were
able to avoid implementing the same limitations so they don't need
the optimizations and run as fast on unoptimized code as the Pentium
does with the optimization. You're paying more to get second best.
Mike Ching
Chris Quirke
step...@netcom.com (Stephen Knilans) wrote:
>In article <575jl9$i...@ds2.acs.ucalgary.ca> pass...@acs.ucalgary.ca (Marco S. Passafaro) writes:
>>In article <stephenkE...@netcom.com>,
>> step...@netcom.com (Stephen Knilans) wrote:
>>>In article <329000...@dreamscape.com> kpds...@dreamscape.com writes:
>>>>Henrik Schmiediche wrote:
>The K5 would have better Specint numbers, maybe 20%+ better, and worse
>Specflop numbers(about 10-25% worse). This is comparing the P5/X to
>the AMD K5PRX. In other words, AMDs 133 product to intels 166.
But the "133" is actually BetterDesignedK5-100.
So you aren't comparing P5/X with K5PRX.
Best case: Hooray for better innards!
Worse case: All the way down to P-100 fundamentals.
>BTW, I HAVE seen win95 run in 8MB on the P5. It isn't pretty. The P5
>at the show was NOT that bad! Also, both systems were supposed to be
>IDENTICAL! I'll tell you one thing. The defrag was slower on the AMD!
>If they cheated, then why not with that?
By the time you get down to 8M, OS setup and running software
dominates the effect on performance.
Rex
cqu...@iafrica.com (Chris Quirke) wrote:
>Best case: Hooray for better innards!
>Worse case: All the way down to P-100 fundamentals.
I'm curous. As the owner of a P-100. What are P-100 fundamentals,
and why are they bad?
--
Rex
Email address witheld due to advertising abuse.
Please reply to Newsgroup.
Chris Quirke
>>Best case: Hooray for better innards!
>>Worse case: All the way down to P-100 fundamentals.
>I'm curous. As the owner of a P-100. What are P-100 fundamentals,
>and why are they bad?
Well, it's all relative. In fact, the iP-100 and K5-100 fundamentals
are pretty good; nice 66MHz external clock speed, giving full 33MHz
PCI, and fabbed at a good ucomp size.
<detail>
The Cyrix/IBM 686-100 "P120" is another story, with a slow external
50MHz clock giving a lame 25MHz PCI clock, but that's offset by more
efficient interger performance for the given clock speed. It's fabbed
large so it pulls current and runs a bit hot, so your mobo needs a
good voltage regulator and heatsink.
The K5-100's internal design just doesn't get the rubber to the road,
so they improved it, but they called the new one "K5 PR133" which is
misleading to the point of retchworthiness, as it's still a 100MHz
66MHz x 1.5 design. They can boast thier improved internal design
efficiency, but the name suggests it should be 133MHz 66MHz x 2, and
it is not.
It is still running externally as efficiently as before, but
internally it is clocked like a -100, not a -133; those are the
fundamentals that determine one's baseline expectations, and whereas
they are fine if you think it's a -100 (which it is) but suck if you
thought you were getting a chip clocked as a -133 and expected whaever
AMD can cook up in the efficiency dept. to improve on the iP-133, with
the comfort of knowing it couldn't get much worse than the old K5
design were that to be clocked as a -133.
<enddetail>
In short; the P100 has good fundementals for a -100 chip, but they are
not the same fundamentals one expects from a -133.
Manassas
Richard Keith Patrick wrote:
>
> I have an AMD 5x86-133 as well. On several occasions, the CPU has fallen
> off the chip, often not to be discovered until weeks later. The chip has
> never failed, although it does get hot.
"The CPU has fallen off the chip"?!? This alone, I could say, would
have stirred the greatest bewilderment in me if it were not for what he
says next, "often not to be discovered until weeks later." My question
is, did you continue to use it while "The CPU [had] fallen off the
chip"? Perhaps all of this is answered if he's talking about an
overdrive upgrade, but that not being mentioned made his post rather
humorous to me. In any case, having worked on AMD 5x86s, and numerous
overdrive upgrades, I am not familiar with what he's talking about.
"May the force be with you. . . . and the rocket launcher
along with the strength rune."
Later,
Manassas,
"You are a very fine person, Mr. Baggins, and I am very fond of you;
but you are only quite a little fellow in a wide world after all!"
-- Gandalf, The Hobbit, J. R. R. Tolkien
Manassas
A good post, but perhaps a little too esoteric for those not versed in
the subject. Allow me to simplify (I'm only going to talk Intel chips
as they are the model).
Basically, whole number over-clocking is one of the best things to have
in a CPU fundamental. Therefore, the two best Pentium chips are the 133
and the 200. Why? they both contain the best BUS of the Pentium family,
a 66MHz (another key fundamental), and are then over-clocked with a
whole number (x2 for the Intel 133, and x3 for Intel 200).
So, when buying a chip (and we're talking Pentium types here) one should
look for, primarily,
1. 66MHz BUS.
2. Whole number over-clocking.
Although the .5s in the over-clockings of such chips as the 100 and 166
made them faster than previous family members, the differences are not
that great and do not justify the average price increase. Many people
call these small increments in over-clocking "baby-steps" in PC
performance.
Now, here are some Intel, Pentium chips one should try to stay away from
when wanting power. However, do go with them if price is a problem, but
try to meet the reasons listed above if possible.
The Intel 75MHz: the worst chip of the family. It fails both criteria
above miserably, the worst BUS of all (a 50MHz), with puny over-clocking
(1.5).
The Intel 100MHz: although it has the best BUS speed (66MHz) its puny
over-clocking of 1.5 does not do it justice.
The Intel 120MHz: its BUS is less than what one would want in a powerful
Pentium (a 60MHz BUS), however, the whole number over-clocking of x2
makes it faster, on average, than the first over-clocked Pentium 66MHz
BUS (the 100). However, it does pay a price for the weaker BUS.
The Intel 150MHz: the numbers impress many people (as the titled, sale
numbers always do), but its BUS (of 60MHz) hurts its promising
performance, plus the fact that it is not a whole number over-clocked
(only by 2.5). On average, a 133 matches the 150 and in some marks
out-performs it. Eric Knorr of PC World confirms this,
"Take an identically configured P-133 and P-150, and price aside, I defy
you to tell the difference. The latter's slower 60-MHz bus speed makes
its performance virtually indistinguishable from that of the former."
Well, there you have it. A layman's take on the subject. Please go
easy on me. I don't claim to have the knowledge that many do out there.
I only know what I only know.
David Keeler
In article <32A410...@wingnet.net> Manassas <mana...@wingnet.net> writes:
>Richard Keith Patrick wrote:
>>
>> I have an AMD 5x86-133 as well. On several occasions, the CPU has fallen
>> off the chip, often not to be discovered until weeks later. The chip has
>> never failed, although it does get hot.
>
>"The CPU has fallen off the chip"?!? This alone, I could say, would
>have stirred the greatest bewilderment in me if it were not for what he
>says next, "often not to be discovered until weeks later." My question
>is, did you continue to use it while "The CPU [had] fallen off the
>chip"? Perhaps all of this is answered if he's talking about an
>overdrive upgrade, but that not being mentioned made his post rather
>humorous to me. In any case, having worked on AMD 5x86s, and numerous
>overdrive upgrades, I am not familiar with what he's talking about.
>
>along with the strength rune."
>
>Later,
>
>Manassas,
>
dave
rj
In article <32A416...@wingnet.net>, Manassas <mana...@wingnet.net> wrote:
>Basically, whole number over-clocking is one of the best things to have
>in a CPU fundamental. Therefore, the two best Pentium chips are the 133
Er, not to be too rude here, but who fed you this line of BS about
'whole number over-clocking' as 'one of the best things to have
in a CPU fundamental'? Bus speed is _way_ more important than
'whole number over-clocking' which has 0 effect on a chip's performance.
I have never even seen this 'whole #' reference until I read your
post. This includes Intel's web site, white papers and USENET.
Right now if you want bang for your buck, buy a Pro 150 or a Pro 180
and overclock them. Both can be had for a decent price, with the
150 being a steal right now. It is around $110 less than a Pentium 166
CPU, which would cover the difference in cost of the motherboards.
A Pro 150 will beat a Pentium 166 running Quake or any 32 bit program
for that matter. (This was confirmed comparing my 166 to a friends
Pro 200 that we underclocked to 150 and 180)
rj of the Quake clan IHoS
--
sig? I don't need no stinkin sig
Manassas
Er, PC World, "Getting the Best PC Price/Performance" by Mr. Knorr in
the "Here's How: Hardware Q&A" section, PC World, July '96 pages
268-270,
"--'Look for CPUs that multiply the bus speed by whole numbers'
You lost here [the guy Mr. Knorr is responding to had bought a Pentium
166]. Your CPU multiplies the bus speed by 2.5, so the
synchronization between internal CPU processing and the bus isn't as
clean as with whole-number multipliers -- one reason the P-166s we've
tested average only 8 percent faster than P-133s, which multiply by 2.
By the same token, the P-133s average a solid 15 percent faster than
P-100s, which multiply by a wimpy 1.5. Pentium-133s are the best buys
at this writing: For example, Dell was selling a P-133 for only $20 more
than a P-100 in an identical configuration."
I'm sure that someone will desire to continue to butt heads with me on
this, but at this point it needs to be taken up with Eric Knorr. If
this isn't true then I'd like to know myself. Also, the reason I
remembered this article and the whole # over-clocking point in general
was due to the original post that I read & responded to by Chris Quirke
(which, by the way, I found good and interesting) where it sounded like
he was saying the same thing:
> The K5-100's internal design just doesn't get the rubber to the road,
> so they improved it, but they called the new one "K5 PR133" which is
> misleading to the point of retchworthiness, as it's still a 100MHz
> 66MHz x 1.5 design. They can boast thier improved internal design
> efficiency, but the name suggests it should be 133MHz 66MHz x 2, and
> it is not.
Note the "x 1.5 design" part and the "x 2" part. I suppose that Chris
needs to reply to your "Er. . ." response too. And as I mentioned, we
were talking Pentium level chips and not Pros. I didn't set the agenda.
I only responded to it, and Pentiums were the agenda. A Pro? Darn
tootin' I'd like to talk Pros, but I found the AMD K5 thread interesting
in its comparison to Pentium level chips, FPs, etc. (WHICH, IF YOU'LL
LOOK, IS THE TITLE OF THE ARTICLE). Do you mind if we talk Pentiums?
Is that OK? Allowed? OK with you? I specifically mentioned 2 things
about Pentiums -- BUS & Whole number over-clocking, but these went along
with Mr. Quirke's post. I thought that I also made it clear that I
didn't know everything. Did you even see this part,
> Well, there you have it. A layman's take on the subject. Please go
> easy on me. I don't claim to have the knowledge that many do out
> there. I only know what I only know.
Try having the same attitude. You do seem rather presumptuous.
--
"May the force be with you. . . . and the rocket launcher
along with the strength rune."
Later,
Manassas,
"You are a very fine person, Mr. Baggins, and I am very fond of you;
Tom Fitzgerald
I'll bet he meant the fan fell off, not the CPU.
Tom
Manassas
Pete Olson wrote:
>
> Manassas (mana...@wingnet.net) wrote:
> :
> : Er, PC World, "Getting the Best PC Price/Performance" by Mr. Knorr in
> : the "Here's How: Hardware Q&A" section, PC World, July '96 pages
> : 268-270,
> : I'm sure that someone will desire to continue to butt heads with me on
> : this, but at this point it needs to be taken up with Eric Knorr. If
> : this isn't true then I'd like to know myself.
> : I didn't set the agenda.
> : I only responded to it, and Pentiums were the agenda.
> : > Well, there you have it. A layman's take on the subject. Please go
> : > easy on me. I don't claim to have the knowledge that many do out
> : > there. I only know what I only know.
> :
> : Try having the same attitude. You do seem rather presumptuous.
>
> and recycled (Knorr) doesn't save it from being bullshit, bullshit
> you peddled.
So you do not believe in using sources either to support a fact or help
out in a question? It is the stuff of modern (even medieval) academia.
Also, if the source is what you are saying it is, then I would like to
know. In this I am being honest and am not trying to "peddle" anything
as I thought I made clear. Yes, it was an old article. You could not
be more right about that, but I was sure that that fact did not make it
totally obsolete. Is an article on the technology of an XT, written
during the XT's heyday, obsolete? Maybe it is, or maybe it is still
correct as the XTs it addressed haven't changed.
> "I know 1.5x multipliers suck!"
> "I don't know shit; go easy on me!"
> "Hey, don't blame me, I didn't say I knew whether or not I knew
> what the hell I was talking about, I only parrotted this guy!"
I am sorry for being so forward. I realize that I did sound
belligerent, but did it warrant such a harsh response?
Mr. Quirke seemed to be bringing up the same fact: that there is better
performance from whole-number over-clocking, especially for the money,
and Pentium chips still cost several hundreds of dollars (over $500 for
a 200, and at least $200 for a 133 -- I felt that pricing and
smart-buying in getting a Pentium might still be an issue with many as
it is with me) which to many (myself included) is a lot of money. A 166
is faster than a 133, but not equally as fast as the 133 was over its
predecessors. .5 just isn't as good as a whole number. This is my only
point, and one that I thought would still be true today as it was in
July. Is it not?
> Your quoted source spoke of cost effectiveness as of almost
> SIX MONTHS AGO. Get with the program: Intel has been through
> at least two rounds of price cuts in that time, and the performance
> numbers quoted only apply to motherboard designs available then.
I realized this too, but I didn't wish to draw out a post with my own
explanations of the source. I simply gave the source and let it stand.
I could give the entire source if you like. I expected intelligent
people to understand the source's date, context, etc., none of which
would necessarily nulify whether the main point in the source I was
using was relevant or not -- that a 66MHz BUS is better, and
whole-number over-clocking is better. Do you still disagree with these?
Of course any pricing mentioned in the source would have changed by now
-- I thought this to be obvious not meriting redundant explanation.
> Six months is an eternity in the PC market.
concur
> (I can't even decide whether or not to mention that the positions
> you advocate seem to assume the Pentium issues intructions either
> every cycle or in magically synchronized patterns. Nah, what could
> you possibly get out of it?)
Do explain what you mean as I am interested. Again, I am not claiming
superior knowledge in this subject (I don't say such a thing in order to
be base, as you implied). By trade, I am more of a liberal arts person.
Let's do go easy here. It is a fascinating subject.
Pete Olson
Manassas (mana...@wingnet.net) wrote:
:
: Er, PC World, "Getting the Best PC Price/Performance" by Mr. Knorr in
: the "Here's How: Hardware Q&A" section, PC World, July '96 pages
: 268-270,
:
: I'm sure that someone will desire to continue to butt heads with me on
: this, but at this point it needs to be taken up with Eric Knorr. If
: this isn't true then I'd like to know myself.
:
: I didn't set the agenda.
: I only responded to it, and Pentiums were the agenda.
:
: > Well, there you have it. A layman's take on the subject. Please go
: > easy on me. I don't claim to have the knowledge that many do out
: > there. I only know what I only know.
:
: Try having the same attitude. You do seem rather presumptuous.
Sorry, bud, that doesn't cut it. Bullshit being old (July '96)
and recycled (Knorr) doesn't save it from being bullshit, bullshit
you peddled.
"I know 1.5x multipliers suck!"
"I don't know shit; go easy on me!"
"Hey, don't blame me, I didn't say I knew whether or not I knew
what the hell I was talking about, I only parrotted this guy!"
Your quoted source spoke of cost effectiveness as of almost
SIX MONTHS AGO. Get with the program: Intel has been through
at least two rounds of price cuts in that time, and the performance
numbers quoted only apply to motherboard designs available then.
Six months is an eternity in the PC market.
(I can't even decide whether or not to mention that the positions
j.r. pierce [see .sig]
Manassas <mana...@wingnet.net> wrote:
>"--'Look for CPUs that multiply the bus speed by whole numbers'
>You lost here [the guy Mr. Knorr is responding to had bought a Pentium
>166]. Your CPU multiplies the bus speed by 2.5, so the
>synchronization between internal CPU processing and the bus isn't as
>clean as with whole-number multipliers -- one reason the P-166s we've
>tested average only 8 percent faster than P-133s, which multiply by 2.
>By the same token, the P-133s average a solid 15 percent faster than
>P-100s, which multiply by a wimpy 1.5. Pentium-133s are the best buys
>at this writing: For example, Dell was selling a P-133 for only $20 more
>than a P-100 in an identical configuration."
Well, actually P-100's CAN multiply by 2... You can run the bus at
50MHz and the processor at 2X. Um, this is SLOWER than 1.5X and
66MHz.
My guess is the diminishing gains between P100, P133, and P166 is due
to the fact that all 3 are running on the same 66MHz memory bus. As
the processor core gets faster, its going to be more memory starved.
This has NOTHING to do with integer vs non-integer bus multipliers.
-jrp
--------------------------------------------------------------------
This posting has a invalid email address to discourage bulk emailers
if you need to mail me try "pierce at hogranch dot com"
--------------------------------------------------------------------
rj
In article <32A49A...@wingnet.net>, Manassas <mana...@wingnet.net> wrote:
>rj wrote:
>Er, PC World, "Getting the Best PC Price/Performance" by Mr. Knorr in
>the "Here's How: Hardware Q&A" section, PC World, July '96 pages
>268-270,
Ahh, that explains it. Do you always believe the BS rags spew
at you? A 100, 133, 166 do not show significant improvements moving
from one to another because they all use a 66 mhz bus. (note:
there is a huge jump from 100 to 166, I meant from 100 to 133,
and 133 to 166) So the only gain in performance you are getting
is CPU based and this is always limited due to the fact of the
CPU to memory limitation in speed. "whole number" sounds like
something Mr. Knorr dreamed up for some buzz word.
>clean as with whole-number multipliers -- one reason the P-166s we've
>tested average only 8 percent faster than P-133s, which multiply by 2.
Funny, when I upgraded from 133 to 166, I gained 15%, darn my
'half-numbers' must have worked better for my machine. Of course
that is less than the 25% increase in clock speed, but like I
mentioned above, memory to cpu is a rather large limitation.
>I'm sure that someone will desire to continue to butt heads with me on
>this, but at this point it needs to be taken up with Eric Knorr. If
>this isn't true then I'd like to know myself. Also, the reason I
Ditto, and you should have mentioned your source in your first note.
>> The K5-100's internal design just doesn't get the rubber to the road,
>> so they improved it, but they called the new one "K5 PR133" which is
>> misleading to the point of retchworthiness, as it's still a 100MHz
>> 66MHz x 1.5 design. They can boast thier improved internal design
>> efficiency, but the name suggests it should be 133MHz 66MHz x 2, and
>> it is not.
>
>Note the "x 1.5 design" part and the "x 2" part. I suppose that Chris
>needs to reply to your "Er. . ." response too. And as I mentioned, we
And the reason for this would be? I do not see him complaining about
this mysterious 'whole number' theory Knorr dreamed up, I see him
saying AMD flat out deceived people. Kinda like Cyrix does with their
"P" rating BS on their pentium clones.
>were talking Pentium level chips and not Pros. I didn't set the agenda.
>I only responded to it, and Pentiums were the agenda. A Pro? Darn
>tootin' I'd like to talk Pros, but I found the AMD K5 thread interesting
>in its comparison to Pentium level chips, FPs, etc. (WHICH, IF YOU'LL
>LOOK, IS THE TITLE OF THE ARTICLE). Do you mind if we talk Pentiums?
>Is that OK? Allowed? OK with you? I specifically mentioned 2 things
Do you mind if I talk about Pro's? Hypocrite. Is it against your law
to bring in other subjects to a post?
>about Pentiums -- BUS & Whole number over-clocking, but these went along
>with Mr. Quirke's post. I thought that I also made it clear that I
>didn't know everything. Did you even see this part,
Yes I saw that part, that is why I was easy on you.
Stephen Knilans
In article <k1Xpykw8...@cet.com*> r...@cet.com* (rj) writes:
>In article <32A49A...@wingnet.net>, Manassas <mana...@wingnet.net> wrote:
>>rj wrote:
>>Er, PC World, "Getting the Best PC Price/Performance" by Mr. Knorr in
>>the "Here's How: Hardware Q&A" section, PC World, July '96 pages
>>268-270,
Some of the "pc gurus" are the MOST inept! Like one person here that said that
bi-directional printer cables were so that the printer could print both ways.
He went on to say that only dot matrix printers do that, so it isn't needed on
laserjet printers!
The FACTS are that logic seeking(aka bidirectional) DMPs DON'T need a
bidirectional cable to do so, that many INKJET and other printers do logic
seeking(and I would hope laser printers do too), and that the bidirectional
cable merely allows communication in BOTH DIRECTIONS. Incidently, the
windows driver for laserjets requires that, so you DO need a bidirectional
cable.
Also, one ZIFF DAVIS publication reported that C&Ts chipset fully supported
EMS 4.0, when all it did, by their EVENTUAL admission, was allow smaller
units under the 3.2 spec with errors for all the NEEDED stuff that their
chip DIDN'T DO!
There are MANY others, but this is a sample.
>
>Ahh, that explains it. Do you always believe the BS rags spew
>at you? A 100, 133, 166 do not show significant improvements moving
>from one to another because they all use a 66 mhz bus. (note:
>there is a huge jump from 100 to 166, I meant from 100 to 133,
>and 133 to 166) So the only gain in performance you are getting
>is CPU based and this is always limited due to the fact of the
>CPU to memory limitation in speed. "whole number" sounds like
>something Mr. Knorr dreamed up for some buzz word.
>
>>clean as with whole-number multipliers -- one reason the P-166s we've
>>tested average only 8 percent faster than P-133s, which multiply by 2.
The BEST multiplier is one that is near ONE, because you can then hook it
up to memory that can run at full speed. Of course, no chipset, that I
have ever heard of, would allow that with the pentium, and no memory can
run that fast(outside of cache memory). Incidently, the cache memory
used today is VERY slow to a P166.
The reason for this is simply that things done outside the chip are done
at the slower speed. Thus, a 166 is less efficient than a 100, but it is
STILL a good deal faster. It is just that the 166 runs more like a 160.
Frankly, I would like a 200Mhz chip with a 200Mhz bus, but I don't think that will happen too soon.
steve
Scott Hess
In article <32A49A...@wingnet.net> Manassas <mana...@wingnet.net> writes:
rj wrote:
>
> In article <32A416...@wingnet.net>,
> Manassas <mana...@wingnet.net> wrote:
> >Basically, whole number over-clocking is one of the best things
> >to have in a CPU fundamental. Therefore, the two best Pentium
> >chips are the 133
>
> Er, not to be too rude here, but who fed you this line of BS
> about 'whole number over-clocking' as 'one of the best things to
> have in a CPU fundamental'?
Er, PC World, "Getting the Best PC Price/Performance" by Mr. Knorr
in the "Here's How: Hardware Q&A" section, PC World, July '96 pages
268-270,
"--'Look for CPUs that multiply the bus speed by whole numbers' You
lost here [the guy Mr. Knorr is responding to had bought a Pentium
166]. Your CPU multiplies the bus speed by 2.5, so the
synchronization between internal CPU processing and the bus isn't
as clean as with whole-number multipliers -- one reason the P-166s
we've tested average only 8 percent faster than P-133s, which
multiply by 2. By the same token, the P-133s average a solid 15
percent faster than P-100s, which multiply by a wimpy 1.5.
Pentium-133s are the best buys at this writing: For example, Dell
was selling a P-133 for only $20 more than a P-100 in an identical
configuration."
I'm sure that someone will desire to continue to butt heads with me on
this, but at this point it needs to be taken up with Eric Knorr. If
this isn't true then I'd like to know myself.
Well, Mr Knorr isn't available, so I'll just butt heads with you :-).
x.5 clocking _does_ cause a slight mismatch. On 2x66Mhz, 5 waits for
DRAM result in an even 10 cycles for the CPU. At 2.5x66Mhz, those 5
waits will translate to 12.5 cycles in the CPU, which probably means
13 waits for that initial word. For EDO, the next words come at 2
waits, so that's 5 CPU cycles, no problem.
OTOH, that's a pretty minor problem. To demonstrate the concept, they
should have also compared a P200 to a P166, since the P200 is a
whole-number multiplier again. Unfortunately, if you compare
P200/P166 to P166/P133, you'll find that the increase from P166 to
P200 is smaller, too!
Look at the numbers. Moving from 100Mhz to 133Mhz is a 33% increase,
yet it's only 15 percent faster. Moving from 133Mhz to 166Mhz is a
25% increase, and it's only 8% faster. The fact that the raw Mhz
increase is lower accounts for _most_ of the difference.
For the most part the rest can be accounted for by factoring in the
increased cost of cache misses for the higher speed. As noted above,
each cache miss will cost at least 25% more CPU cycles on a 166Mhz CPU
than on a 133Mhz CPU, and it might be 30%. Even if it is 30%, though,
most of the loss is in the minimum 25%, the extra 5% cache miss cost
is where the "synchronization" cost mentioned.
Of course, you need to factor all of that in. So, if we go from P100
to P133, the maximum increase is 33%, the observed increase is 15%, so
let's assume that the missing 18% is due to cache misses. With the
P133 to P166 upgrade, the maximum increase is 25%, the observed
increase is 8%, let's assume the missing 17% is cache misses. Only
5/30 of the miss cost can be attributed to synchronization, so just
under 3% of the performance is lost to synchronization.
Of course, with all of the raw data at three steps removed (someone
generated them, rounded them off, Mr Knorr reported them, and then you
reported them), you do have to question my results. In all
likelihood, the effect of the synchronization is _less_ than %3.
[You might think you'll get back that 3% when you go to P200. Ha!
You might, but on the other hand P166 to P200 is only a 20% Mhz
increase, and the cache miss cost is up another %20, or %50 over the
P133, so you lose in the end.]
Later,
--
scott hess <sh...@one.net> (606) 578-0412 http://w3.one.net/~shess/
<I plan to become so famous that people buy tapes of me reading source code>
Clayton Gewin
> Er, PC World, "Getting the Best PC Price/Performance" by Mr. Knorr
> in the "Here's How: Hardware Q&A" section, PC World, July '96 pages
> 268-270,
Gee, the PC world has changed a lot since July.....
> "--'Look for CPUs that multiply the bus speed by whole numbers' You
> lost here [the guy Mr. Knorr is responding to had bought a Pentium
> 166]. Your CPU multiplies the bus speed by 2.5, so the
> synchronization between internal CPU processing and the bus isn't
> as clean as with whole-number multipliers -- one reason the P-166s
> we've tested average only 8 percent faster than P-133s, which
> multiply by 2. By the same token, the P-133s average a solid 15
> percent faster than P-100s, which multiply by a wimpy 1.5.
> Pentium-133s are the best buys at this writing: For example, Dell
> was selling a P-133 for only $20 more than a P-100 in an identical
> configuration."
He was comparing word processing, spreadsheets etc. for the most part. PC
WORLD, and Mr. Knorr in particular have a tendency to underestimate the
performance of faster IA32 CPU's. Their benchmarks are the weakest of all
the magazines, significantly worse than even Ziff Davis's Winstoned suite.
Their benchmark won't tell you how autocad runs, how video runs, how quake
runs or really how any CPU intensive program runs. In fact, I believe they
have never made their benchmark publicly available(though they claimed they
would) so it is a question as to how they actually get their results. This
bit about "look for whole number multipliers" is nonsense. Buy as a fast a
CPU as you can afford. A Pentium 133 is entry level....If you want the
biggest leap, get a PRO. Sounds like Mr. Knorr does believe in witchcraft
and voodoo though. A closet numerologist????
Clayton Gewin
My opinions, not my employers...
Stephen Knilans
In article <01bbe70d$fc92f0a0$7dd0...@Earthlink.earthlink.net> "Clayton Gewin" <Clay...@earthlink.net> writes:
>
>
>> Er, PC World, "Getting the Best PC Price/Performance" by Mr. Knorr
>> in the "Here's How: Hardware Q&A" section, PC World, July '96 pages
>> 268-270,
>Gee, the PC world has changed a lot since July.....
>
>> "--'Look for CPUs that multiply the bus speed by whole numbers' You
>> lost here [the guy Mr. Knorr is responding to had bought a Pentium
>> 166]. Your CPU multiplies the bus speed by 2.5, so the
>> synchronization between internal CPU processing and the bus isn't
>> as clean as with whole-number multipliers -- one reason the P-166s
>> we've tested average only 8 percent faster than P-133s, which
>> multiply by 2. By the same token, the P-133s average a solid 15
>> percent faster than P-100s, which multiply by a wimpy 1.5.
>> Pentium-133s are the best buys at this writing: For example, Dell
>> was selling a P-133 for only $20 more than a P-100 in an identical
>> configuration."
WAIT a second! A P5/100 that is going at 85% of the speed of the 133 is
FAST!!!!!!!!!! That is true of ANY circuit! WHY? Because the P5/133
shows, according to this guy, only a 15% improvement! News to me, I saw
about 33% improvements in the pcmag benchmark! BTW, as the CPU gets
"faster" in theory it HAS to go slower in relation to the previous speed
chip! WHY? Because the 200Mhz has the SAME speed connection to memory/
etc as the 100Mhz, and the 100Mhz is ALREADY exceeding the speed of the
memory! Hence, as the speed goes above 66Mhz(the speed of cache), the
computer must, in part, become LESS effecient, due to memory bottlenecks.
I would say the P5/133 has the best price/performance ratio. Want more
speed? Try opting for faster memory/peripherals, and maybe dual CPUs.
>
>He was comparing word processing, spreadsheets etc. for the most part. PC
>WORLD,
Gee, that is what PC MAG does! Yet it showed BIG improvements. Then again,
my P5/133 put some P5/166 to SHAME! Clearly, they were NOT setup very well,
and may have used 80ns memory with NO cache!
Doug Zabel
Clayton Gewin wrote:
>
> > Er, PC World, "Getting the Best PC Price/Performance" by Mr. Knorr
> > in the "Here's How: Hardware Q&A" section, PC World, July '96 pages
> > 268-270,
> Gee, the PC world has changed a lot since July.....
>
> > "--'Look for CPUs that multiply the bus speed by whole numbers' You
> > lost here [the guy Mr. Knorr is responding to had bought a Pentium
> > 166]. Your CPU multiplies the bus speed by 2.5, so the
> > synchronization between internal CPU processing and the bus isn't
> > as clean as with whole-number multipliers -- one reason the P-166s
> > we've tested average only 8 percent faster than P-133s, which
> > multiply by 2. By the same token, the P-133s average a solid 15
> > percent faster than P-100s, which multiply by a wimpy 1.5.
> > Pentium-133s are the best buys at this writing: For example, Dell
> > was selling a P-133 for only $20 more than a P-100 in an identical
> > configuration."
>
> performance of faster IA32 CPU's. Their benchmarks are the weakest of all
> the magazines, significantly worse than even Ziff Davis's Winstoned suite.
> Their benchmark won't tell you how autocad runs, how video runs, how quake
> runs or really how any CPU intensive program runs. In fact, I believe they
> have never made their benchmark publicly available(though they claimed they
> would) so it is a question as to how they actually get their results. This
> bit about "look for whole number multipliers" is nonsense. Buy as a fast a
> CPU as you can afford. A Pentium 133 is entry level....If you want the
> biggest leap, get a PRO. Sounds like Mr. Knorr does believe in witchcraft
> and voodoo though. A closet numerologist????
> Clayton Gewin
> My opinions, not my employers...
Well, obviously there's a larger percentage difference between 133 and
100 than there is with 166 and 133. Also, there is the almost-as-obvious
fact that all machines have essectially that same limitations in memory,
harddrive speed.
This does not in any way imply that "whole numbers" are better.
Ron Humphrey
I am wanting to upgrade from a 486 MB&cpu to a pent level one...
I have been wondering how the AMD K5-133 compares to the competition.
I play a lot of games and do some graphics rendering. I have seen
the K5-133 $100.00 cheaper than the Intel P-133 but if performance
is much lower, it may be worth the extra $100.00. anyone with info of
performance of these chips?? a comparision of the 3 contrnders would be
nice.
ron...@magibox.net
On Thu, 12 Dec 1996 07:27:01 -0600, Ron Humphrey <rh...@webmart.net>
wrote:
I have an AMD K5 133 on an ASUS P55T2P4 motherboard and am quite
pleased. I have a terrible video card and still get 23 fps in Quake
320x200. I expect to increace that when I get a REAL video card. I
haven't done any benchmarks but most people want to know how a chip
runs Quake so that is how it runs Quake. BTW that is full screen {no
status bar} running timedemo demo2 with no sound. Considering I got
less than 10 fps on my old 486 I am happy. Even if it doesn't run
Quake as well as an Intel P133 (I don't know if it does or doesn't) at
$129 delivered with fan I can't complain. I've only had it a few days
so I haven't tweaked it.
BTW my videocard is an old Diamond Speedstar w 1meg DRAM Cirrus Logic
chip. It is slower in Windows than my 486. My 486 had a GOOD card
though.
Darin Genereux
I just upgraded from a 5x86-160 to an ASUS T2P4 and AMD K5-PR133, and
am *extremely* pleased. I also have an Intergraph Reactor video card
(3D Verite chip), which makes a huge difference in Quake.
Here's my benchmarks: Full screen (no status bars), nosound, nojoy,
Timedemo 2, Win95
RESOLUTION VQUAKE, Beta9 REGULAR QUAKE
320X200 40.1 23.1
512X384 26.8 13.2
640X480 20.8 9.9
As you can see, the Verite accelerated Quake practically doubled my
framerates--something I was not prepared for as I don't believe it
offers this much of an improvement with Intel chips.
All my other games and apps run great in high-res now--Duke3d,
Shattered Steel, ATF, Whiplash, DescentII, etc. I imagine about the
only game I have that would require tweaking in SVGA is Grand Prix 2,
though I haven't had a chance to try it yet.
Hope this helps,
Darin
Elvis Presley
In article <32B008...@webmart.net>, Ron Humphrey <rh...@webmart.net>
writes
>I am wanting to upgrade from a 486 MB&cpu to a pent level one...
>I have been wondering how the AMD K5-133 compares to the competition.
>I play a lot of games and do some graphics rendering. I have seen
>the K5-133 $100.00 cheaper than the Intel P-133 but if performance
>is much lower, it may be worth the extra $100.00. anyone with info of
>performance of these chips?? a comparision of the 3 contrnders would be
>nice.
Can't really do a fair comparison cos I've never had the pent, but I
have recently upgraded from a 486 to an AMDK5-133. The thing is about 10
times faster at everything. I also upgraded from 8 to 16 megs of RAM at
the same time though.
--
The King <el...@presley.demon.co.uk>
Moped Racer Online Magazine.
Moped Mayhem Results Service, and comprehensive moped racing news
and info pages.
<http://www.presley.demon.co.uk>Last (extensive) update:09.12.96
Mike Fyodoroff
I dont know about amd, but i have intel P133 and like it very much.
The important thing is 512k cache and good video with 2mb edo (at least)
and i think with any proc you'll have goot refresh
--
Sincerly yours
Mike Fyodoroff
Ron Humphrey <rh...@webmart.net> wrote in article
<32B008...@webmart.net>...
Bernard Sy
Ron Humphrey wrote:
>
> I am wanting to upgrade from a 486 MB&cpu to a pent level one...
> I have been wondering how the AMD K5-133 compares to the competition.
> I play a lot of games and do some graphics rendering. I have seen
> the K5-133 $100.00 cheaper than the Intel P-133 but if performance
> is much lower, it may be worth the extra $100.00. anyone with info of
> performance of these chips?? a comparision of the 3 contrnders would be
> nice.
I have a Pentium 133 for my main computer (SCSI, big monitor, etc.) and
just recently upgraded my play computer from a AMD486-133 to a Cyrix
P150+.
I, too, play a lot of games. I new about Cyrix's lagging FPU, but
figured
a P150+ would still be fast enough for Quake at the lowest resolution (I
have a cheap PCI video card on that machine). It was. I don't know if
I would recommend it for your main computer. Runs hot, too.
I would be interested in what people have to say about the K5-PR133.
Bern
Captain Buggery
Bernard Sy <b...@nortel.com> wrote in article <32BAF3...@nortel.com>...
All K5 chips outperform their Pentium counterparts running at the same
clock speed in all areas including floating point. They are great value for
money.
Gary
their chip has been rated right to the edge, where Intels chip is not.
The reason is simple, Intel wants their chip to work in every board it's
plugged into, and retain it's boasted performance. The K5 will not. The
K5 also runs much hotter, uses more power, and sells less then the Intel
(wonder why that is).The other thing to consider is the fact that Intels
chip, being the standard, is likely to have more software fixes in
applications and programs than the AMD.
Stephen Knilans
Any chip that needs "software fixes" is, by its nature, DEFECTIVE! I wouldn't
want it!
BTW, the K5 DOES have 30% more transistors, but it uses the same technology.
I doubt it is that hot, or requires that much power. As for pushing it to
the edge? They STILL haven't done all they plan to do. The K6 is just a
taste of things to come!
Steve
BTW, Intels chips ARE pushed to the edge! My 75Mhz, which old conventions would
require to run RELIABLY at over 86Mhz didn't really run AT ALL at 90Mhz! Any
AMD chip that can't run at its stated speed is simply defectieve.
P. Sanders
In article <32BE1C...@tiac.net>, ra...@tiac.net says...
That's horseshit. The AMD does not run hotter, it does not use more power, and
it's only certified on certain boards primarily because of the BIOS needed to
recognize it. It certainly does what I need it to do and just becouse you may
not like it, does not mean you should make such negative blanket statements
that might make others think twice about purchasing it.
BTW, I also own a Cyrix 6x86-PR166+. It's fast as hell, too, and guess what
-- I've had no problems with it either.
Tom
Gary wrote:
> The K5 May seem faster judging by the specs in their benchmarks, but
> their chip has been rated right to the edge, where Intels chip is not.
> The reason is simple, Intel wants their chip to work in every board it's
> plugged into, and retain it's boasted performance. The K5 will not. The
> K5 also runs much hotter, uses more power, and sells less then the Intel
> (wonder why that is).The other thing to consider is the fact that Intels
> chip, being the standard, is likely to have more software fixes in
> applications and programs than the AMD.
>
the K5 will run in any pentium mobo with the proper bios, and will run on
some that don't have bios support. The K5 doesn't run any hotter at the
same frequency than Intels both are 0.35 micron. the K5 is 100% x86
instruction compatible itdoesn't ned any software-fix .
brih...@sfsu.edu
Ron Humphrey <rh...@webmart.net> wrote:
>I am wanting to upgrade from a 486 MB&cpu to a pent level one...
(rest of article snipped)
Thanks for asking the question, Ron. I, too, have a 486 DX2-66, and
I've been considering getting a Kingston AMD 166 triple-clock upgrade
CPU. My motherboard burned out a few months ago, and my new (but still
486) motherboard was only $75, so I am concerned about plugging in a
non-Intel CPU. I also upgraded from a Diamond Viper to a Hercules
Terminator Pro, and while I see an improvement, I realize that my
system is still VLB, not PCI. The Hercules card, too, was about $250,
but it does not have 3D capabilities. My system only has 8MB RAM, so
I'm thinking of ugrading to 16 MB as well.
As of now, Doom 2 runs superb, but Quake won't move anywhere ;)
Any information about comparing the K AMD 166 chip to an Intel
Overdrive equivalent would be most appreciated.
Thanks in advance,
Brian
brih...@sfsu.edu
Chris Edwards
Price Shoppers http://www.top7.com finds the best prices for the top
rated games, hardware, gaming accessories and 100 other categories.
Tell a friend, they'll save money. Also, Check out Computer Players'
March CD which will have Price Shoppers entire internet site on it.
Chris Edwards
future
>Ron Humphrey <rh...@webmart.net> wrote:
>
>>I am wanting to upgrade from a 486 MB&cpu to a pent level one...
>
>(rest of article snipped)
>
>Thanks for asking the question, Ron. I, too, have a 486 DX2-66, and
>I've been considering getting a Kingston AMD 166 triple-clock upgrade
>CPU. My motherboard burned out a few months ago, and my new (but still
>486) motherboard was only $75, so I am concerned about plugging in a
>non-Intel CPU.
As far as 486-class chips go, don't worry about whether it's from
Intel or AMD. I've never experienced an incompatibility of any kind
with my AMD 486DX4-120. The floating point operation may be slightly
slower than Intel's (don't know for sure) but if so, it's rarely
noticeable.
>I also upgraded from a Diamond Viper to a Hercules
>Terminator Pro, and while I see an improvement, I realize that my
>system is still VLB, not PCI.
VLB should be adequately fast for a 486 class system.
>The Hercules card, too, was about $250,
>but it does not have 3D capabilities. My system only has 8MB RAM, so
>I'm thinking of ugrading to 16 MB as well.
Now THAT is a good idea.
Richard Boykin
>
I have both a p100 (intel) and a k5-100 (amd) running on similar systems.
Overall performance is very similar. The k5 benches ~10% slower than the
intel, and the fpu is even slower still. However, for most applications, the
$100 savings is worth it, save that money and buy a better video card. I play
quake on the 'net with both boxes, and they both play just fine.
Rich Boykin
Satisfied amd and intel owner
Basil Evrenidis
In article <32BE1C...@tiac.net>, Gary <ra...@tiac.net> wrote:
<SNIP>
>> All K5 chips outperform their Pentium counterparts running at the same
>> clock speed in all areas including floating point. They are great value for
>> money.
>The K5 May seem faster judging by the specs in their benchmarks, but
>their chip has been rated right to the edge, where Intels chip is not.
>The reason is simple, Intel wants their chip to work in every board it's
>plugged into, and retain it's boasted performance. The K5 will not. The
>K5 also runs much hotter, uses more power, and sells less then the Intel
>(wonder why that is).The other thing to consider is the fact that Intels
>chip, being the standard, is likely to have more software fixes in
>applications and programs than the AMD.
The "edge"??? Which "edge" would that be? Edge of a cliff? Edge of a razor?
Honestly, what do you base this "edge" hypothesis on? Have you ever used an
AMD K5 chip; and are you basing your assertions that AMD chips are less
compatible than Intel chips on this experience. OR, are your statements some
form of generalized bullshit used to cover up the fact that you don''t know
what you're talking about.
I've used both and guess what, they both perform. However, I've experience
less problems with my K5 based systems that Pentium systems in terms of
system stability for similar configurations. Performance wise, the K5 edges
out an equivalent Pentium in general use but really comes up short when
comparisons of FPU performance are made. Other than that, systems based on
either configuration will perform the same functions.
Basil...
Dennis Moran
Richard Boykin <boyk...@cnw.com> wrote in article
<59vf1e$7...@lassen.cnw.com>...
Playing Quake over the internet has nothing to do with your system, it has
to do with your connection speed. A guy with a 486DX4-100 and a T1
connection will most likely beat the guy with a 200Mhz Pentium Pro with
128MB RAM, who has a 33.6 connection.
--
Dennis Moran (aka Coolio)
coo...@coolio9.com
http://www.coolio9.com/
Jim Scannell
Stephen Knilans <step...@netcom.com> wrote in article
<stephenkE...@netcom.com>...
> Any chip that needs "software fixes" is, by its nature, DEFECTIVE! I
wouldn't
> want it!
The keyword was "software". I didn't read that to say hardware fixes.
Software problems pertaining to AMD or Cyrix chips only are less likely to
be fixed was the point.
Jim Scannell
Madison, Wisconsin
mailto:scan...@concentric.net
http://www.concentric.net/~scannell/win95/
Arnold Lee
dan, my hotheaded, ignorant friend...
he was just noting that the previous posting had made several blatant,
serious errors in his post that's all.
If anyone wants benchmarks (run on my system) comparing the 6x86 133+,
150+ and 166+ to the AMD K5 Pr133 and AMD K5 Pr150 (133 oc'd) email
me...i'll send it to you. Basically, the AMD chips performed as well as
and slightly better than their Cyrix counterparts for the most
part (i.e., disk access, ect)...but lost by slight margins in 16 and 32
bit processing.
Dan Hemphill wrote:
>
> In article <59mu1a$m...@camel5.mindspring.com>, psan...@mindspring.com
> says...
> >
> >In article <32BE1C...@tiac.net>, ra...@tiac.net says...
> >>
> >>Captain Buggery wrote:
> >>>
> >>> Bernard Sy <b...@nortel.com> wrote in article
> <32BAF3...@nortel.com>...
> >>> > Ron Humphrey wrote:
> >>> > >
> >>> > > I am wanting to upgrade from a 486 MB&cpu to a pent level one...
> >>> All K5 chips outperform their Pentium counterparts running at the
> same
> >>> clock speed in all areas including floating point. They are great
> value for
> >>> money.
> >>The K5 May seem faster judging by the specs in their benchmarks, but
> >>their chip has been rated right to the edge, where Intels chip is not.
> >>The reason is simple, Intel wants their chip to work in every board
> it's
> >>plugged into, and retain it's boasted performance. The K5 will not.
> The
> >>K5 also runs much hotter, uses more power, and sells less then the
> Intel
> >>(wonder why that is).The other thing to consider is the fact that
> Intels
> >>chip, being the standard, is likely to have more software fixes in
> >>applications and programs than the AMD.
> >
> >That's horseshit. The AMD does not run hotter, it does not use more
> power, and
> >it's only certified on certain boards primarily because of the BIOS
> needed to
> >recognize it. It certainly does what I need it to do and just becouse
> you may
> >not like it, does not mean you should make such negative blanket
> statements
> >that might make others think twice about purchasing it.
> >
> >BTW, I also own a Cyrix 6x86-PR166+. It's fast as hell, too, and guess
> what
> >-- I've had no problems with it either.
> >
> This is a newsgroup you fucking idiot..it's supposed to be about people
> voicing their opinions. Just because you obviously aren't happy with
> your lousy AMD chip, don't start complaining when someone puts it down.
> You should have bought Intel.
Francisco Gochez
Arnold Lee <meat...@wam.umd.edu> wrote:
<snip>
> Basically, the AMD chips performed as well as
>and slightly better than their Cyrix counterparts for the most
>part (i.e., disk access, ect)...but lost by slight margins in 16 and 32
>bit processing.
>
What do processors have to do with disk access speed? Shouldn't this
be determined solely by the hard drive (or hard drive related
peripherals)? As for the comment about 16 and 32 bit processing, that's
all that processors are intended to do... process. Saying that the AMD
is worse at 16 and 32 bit processing (which is all that IBM compatible
PCs do) is saying that it is directly worse than the Cyrix.
tom
Francisco Gochez <arc...@erols.com> wrote in article
<5b4gbm$8...@boursy.news.erols.com>...
NU95 benchmark uses 32 bit processing , the K5-PR133 I'm running scores
45.5 to 46.5 . I've gotten e-mail from various people stateing that this K5
is performing the same as their Cyrix p166+ or a bit better.
Chris Quirke
step...@netcom.com (Stephen Knilans) wrote:
>In article <01bbe70d$fc92f0a0$7dd0...@Earthlink.earthlink.net> "Clayton Gewin" <Clay...@earthlink.net> writes:
<stuff on whole-number multipliers skipped>
I look for:
1) Comfortably fast external RAM bus and PCI (66/33)
before anything else, so I'd rather have 1.5 x 66 than 2 x 50.
However, my testing has shown iP-133 faster relative to iP-100 than
expected, as the law of diminishing returns sets in as you push core
multiplication and use the resultant MHz as a guide to performance
expectations. If the iP-133 was cheaper (here, it still costs a lot
more than the iP-100, and we still have iP-75s on the market) it would
be a no-brainer. By Intel's own figures, the iP-150 sux, unless
marketed at a similar price to the iP-133.
Lame-base-speed (60, or <yuck> 50MHz) chips may have a place in
laptops, if slowing down the externals improves battery life, however.
>>and Mr. Knorr in particular have a tendency to underestimate the
>>performance of faster IA32 CPU's.
Uh-huh. You are going to tell me iP-200 is nearly twice as fast as an
iP-133, are you? Maybe we should believe iComp? Like, a 486SX-25 is
much faster than a 386DX-40? Sorry to exaggerate, but.
>>Buy as a fast a CPU as you can afford. A Pentium 133 is entry level...
Left out the bit about "Not speaking for Intel..." <grin>
Choose the best point on the price/performance curve, and temper that
choice with what you actually need. Right now, the best point round
here is a AMD 586DX4-133 on a PCI 486DXn mobo, as the difference in
mobo price magnifies the overall effect. Want more speed? Next happy
points are 686-133 "P166", iP-100 if you want Intel, and iP-133 (not
that happy-priced, but nice speed).
Thereafter, the madness begins... I woudn't bother with anything
faster until my RAM, HD and video subsystems were pretty maxed out,
and would perhaps rather spend the money on extra functionality
(scanner, printer etc.) instead. Or <gasp> not spend it at all !
Clayton Gewin
Chris Quirke <cqu...@iafrica.com> wrote in article
<5b8pf2$ft1$7...@proxy01.iafrica.com>...
> step...@netcom.com (Stephen Knilans) wrote:
>
> >In article <01bbe70d$fc92f0a0$7dd0...@Earthlink.earthlink.net> "Clayton
Gewin" <Clay...@earthlink.net> writes:
>
> <stuff on whole-number multipliers skipped>
>
> I look for:
>
> 1) Comfortably fast external RAM bus and PCI (66/33)
>
> before anything else, so I'd rather have 1.5 x 66 than 2 x 50.
I
I'd rather have a 3x myself....at 66MHZ....with MMX technology and those
nice 16k data and code caches(even a Pro would be great!). MHZ is only one
component of performance...
> However, my testing has shown iP-133 faster relative to iP-100 than
> expected, as the law of diminishing returns sets in as you push core
> multiplication and use the resultant MHz as a guide to performance
> expectations. If the iP-133 was cheaper (here, it still costs a lot
> more than the iP-100, and we still have iP-75s on the market) it would
> be a no-brainer. By Intel's own figures, the iP-150 sux, unless
> marketed at a similar price to the iP-133.
Fine. If all you can afford is a 133 great. But a Pentium 166 is faster
and an MMX technology Pentium is even faster than that. Fit it to your
budget. I have a Pentium 166, I had a 90 and have shown terrific
improvements. And yes I did get a fast subsystem etc. Not sure what your
point is.
> Lame-base-speed (60, or <yuck> 50MHz) chips may have a place in
> laptops, if slowing down the externals improves battery life, however.
I never advocated 50 or 60 MHZ buses. Although depending on what you are
doing a 150/60 is up to 10% faster than a 133/66. The reverse could also
be true for bus intensive apps.
> >>and Mr. Knorr in particular have a tendency to underestimate the
> >>performance of faster IA32 CPU's.
> Uh-huh. You are going to tell me iP-200 is nearly twice as fast as an
> iP-133, are you? Maybe we should believe iComp? Like, a 486SX-25 is
> much faster than a 386DX-40? Sorry to exaggerate, but.
I believe I never claimed 2x performance increase. Even intel's icomp does
not claim that. The 200 is ~40-50% faster than a 133. Could be more,
could be less depending on what you are doing. .
> >>Buy as a fast a CPU as you can afford. A Pentium 133 is entry level...
>
> Left out the bit about "Not speaking for Intel..." <grin>
I'm not. These are my opinions based on using 133-200MHZ intel based
systems(standard and with MMX technology(only 166MHZ and 200's).
> Choose the best point on the price/performance curve, and temper that
> choice with what you actually need. Right now, the best point round
> here is a AMD 586DX4-133 on a PCI 486DXn mobo, as the difference in
> mobo price magnifies the overall effect. Want more speed? Next happy
> points are 686-133 "P166", iP-100 if you want Intel, and iP-133 (not
> that happy-priced, but nice speed).
1. I guess prices must be pretty bad where you are. I can get a nice 133
system for around $1300 or less in the US. I would rather have more
headroom for the future. As for the 6x86, I play a lot of 3d games
including Quake. Not an option.
2. 486 level products are not a bargain at all(5x86 whatever). You claim
you would want a 66MHZ bus above, 486 level products top out at
40-50MHZ(very dependent on the board and not all cards will work). These
products are only viable if they are extremely inexepensive and even then,
there is a significant list of new products that will not run very well or
at all i.e. Quake and any other FP intensive app, any bus intensive app,
any superscalar app...
> Thereafter, the madness begins... I woudn't bother with anything
> faster until my RAM, HD and video subsystems were pretty maxed out,
> and would perhaps rather spend the money on extra functionality
> (scanner, printer etc.) instead. Or <gasp> not spend it at all !
Depends on what you are doing. If you play the 1992 version of
civilization all day, a 386 is fine. If you run Monster truck madness you
need latest greatest hardware. Even office 97 is a bit big:)
Clayton Gewin
Speaking for myself, not my employer....
Chris Quirke
>> 1) Comfortably fast external RAM bus and PCI (66/33)
>> before anything else, so I'd rather have 1.5 x 66 than 2 x 50.
>I'd rather have a 3x myself....at 66MHZ....with MMX technology and those
>nice 16k data and code caches(even a Pro would be great!). MHZ is only one
>component of performance...
Sure, and if they weren't so expensive (or going to halve in price
within a year) I'd go for that too. But as it is, I'd rather spend
the balance on more RAM, HD capacity, functionality items etc.
instead. I don't buy into Intel's CPU-uber-alles vision ;-)
>> By Intel's own figures, the iP-150 sux, unless
>> marketed at a similar price to the iP-133.
>Fine. If all you can afford is a 133 great. But a Pentium 166 is faster
>and an MMX technology Pentium is even faster than that. Fit it to your
>budget. I have a Pentium 166, I had a 90 and have shown terrific
>improvements. And yes I did get a fast subsystem etc. Not sure what your
>point is.
Most of the iP-90 to iP-166 gain would have been from iP-90 to iP-133.
Much of the cost of the move was probably from iP-133 to iP-166 rather
than iP-90 to iP-133. And was that a CPU-only upgrade, or a whole new
system? If so, what portion of the boost came from other components,
how much did they cost relative to the iP-133 - iP-166 premium? I've
seen posted that iP-133 + 512k L2 matches an iP-166 + 256k L2 in
performance, and that cache upgrade cost 10% of the CPU diff.
>> Lame-base-speed (60, or <yuck> 50MHz) chips may have a place in
>> laptops, if slowing down the externals improves battery life, however.
>I never advocated 50 or 60 MHZ buses. Although depending on what you are
>doing a 150/60 is up to 10% faster than a 133/66. The reverse could also
>be true for bus intensive apps.
If I'm waiting, I'm usually waiting for HD. HD speed tends to be HD
limited rather than PCI limited, but. And if you assessed your
requirements as VGA-limited and spent $$$$ on a SVGA card, you are
hobbling it with a slow PCI - follow thru and let it breathe!
>> Uh-huh. You are going to tell me iP-200 is nearly twice as fast as an
>> iP-133, are you? Maybe we should believe iComp? Like, a 486SX-25 is
>> much faster than a 386DX-40? Sorry to exaggerate, but.
>I believe I never claimed 2x performance increase. Even intel's icomp does
>not claim that.
Note the "Sorry to exaggerate, but." line
>The 200 is ~40-50% faster than a 133.
iComp?
>> Choose the best point on the price/performance curve, and temper that
>> choice with what you actually need. Right now, the best point round
>> here is a AMD 586DX4-133 on a PCI 486DXn mobo, as the difference in
>> mobo price magnifies the overall effect. Want more speed? Next happy
>> points are 686-133 "P166", iP-100 if you want Intel, and iP-133 (not
>> that happy-priced, but nice speed).
>1. I guess prices must be pretty bad where you are.
Well, they aren't great... and there's another diff; I look at
component level prices, not systems. IOW an iP-133 (when I speak of
it) is just that, not some vendor's system based on what they think
that chip should go with. And it's the diff in an extra 20% CPU
power, rather than extra 50% RAM or 40% HD etc., that I am looking at.
>2. 486 level products are not a bargain at all(5x86 whatever). You claim
>you would want a 66MHZ bus above, 486 level products top out at
>40-50MHZ(very dependent on the board and not all cards will work).
RAM at 33MHz, PCI at same 33MHz.
Sure, I'd like 66MHz etc. but at Pent prices, it's an expectation.
>These products are only viable if they are extremely inexepensive and even then,
>there is a significant list of new products that will not run very well or
>at all i.e. Quake and any other FP intensive app, any bus intensive app,
>any superscalar app...
Haven't found much that doesn't run pretty well, and 0 that doesn't
run at all. IMO I'm more likely to hit sware probs on a K5 or 686
than a 586DX4-133
If you think you want USB and other mobo enhancements, and aren't
prepared to bet current boards will do this, it makes sense to throw
out a cheaper 486DXn mobo than a Pent board.
>> Thereafter, the madness begins... I woudn't bother with anything
>> faster until my RAM, HD and video subsystems were pretty maxed out,
>> and would perhaps rather spend the money on extra functionality
>> (scanner, printer etc.) instead. Or <gasp> not spend it at all !
>Depends on what you are doing. If you play the 1992 version of
>civilization all day, a 386 is fine. If you run Monster truck madness you
>need latest greatest hardware. Even office 97 is a bit big:)
Not quite the point I was making. Point I was making is that a budget
mobo, tiny HD, 8M RAM, Trident SlowVGA and iP-200 does not make sense.
Daniel Ts'o
:2. 486 level products are not a bargain at all(5x86 whatever). You claim
:you would want a 66MHZ bus above, 486 level products top out at
:products are only viable if they are extremely inexepensive and even then,
:there is a significant list of new products that will not run very well or
:at all i.e. Quake and any other FP intensive app, any bus intensive app,
:any superscalar app...
I must disagree here. A 5x86/160 PCI motherboard + CPU chip can be
purchased for $100-110 and gives roughly P5-100 performance. There is no
Pentium CPU+motherboard deal I know of that is that cheap and performs that
well.
The 66Mhz bus speed you quote, typical for Pentium systems, actually
runs the PCI bus at only 33Mhz, the same as a 486 PCI motherboard.
A Verite 3D video for $150 on a 5x86/160 runs Quake as well as a
P5-133 with a good 2D video card and is prettier to boot.
There are very few apps beyond games that require more CPU than a
5x86/160 can give.
So, if all you can spend is $100 (and you have sufficient memory), a
5x86 ain't bad at all. If you can spend $200+, sure, get a Pentium.
The 5x86 route is a bargain. A Pentium P5-133 or above is better, but
will cost you $300 or more.
Cheers,
Dan Ts'o
The Rockefeller University
Box 138
1230 York Avenue
New York, NY 10021
dan...@cris.com
212-327-7671
212-327-7671 FAX
Downlo4392
well i am happy with my Cyrix 133 chip and i can bitch about this tread
because i also own a pentium 133 and ill tell you that in every aspect the
speed difference in games is noticable(exept quake, it runs at the same
fps)and i know that my cyrix 133 compared to my pent133 is a much better
buy my cyrix is from CyberMax my pent is from BestBuy i think that the
manufactuer plays one of the biggest roles in computer speed not the
advertisements or the benchmarks
a.j.
Downlo4392 wrote:
>
> well i am happy with my Cyrix 133 chip
Oh REALLY?? BULL.
> and i can bitch about this tread because
You can bitch bout anything u want. You can
also be a total asshole if ya want.
> I also own a pentium 133 and ill tell you that in every aspect the
> speed difference in games is noticable
Oh REALLY??? LIAR!!
> (exept quake,
Oh, well you see the lies are admitted even by YOU.
If you check further, there are other things that
don't work as well either. Not that it matters, if
you truly understand modern computers to be like
model t's were to cars, you'll know in the future
BOTH and evewn the pentium pro's are total garbage.
> it runs at the same fps)
They are very similar, but if you play the ONLY computer
application that requires much at all, (Quake) you'll go
with the pentium pro, cause you'll know how little that
is to work with. Anything less is totally ridicluos. I
boot up my p133 and remember when I thought it was
awesome. Now I know both it and my ppro are lame.
> and i know that my cyrix 133 compared to my pent133
> is a much better buy
HELLOOOOoooOOOOoooo. Better buy!! YA!!! Better??? NOOOO.
Even the fastest ppro is lame. Future games will require
twice as much as the fastest ppro has today.
> my cyrix is from CyberMax my pent is from BestBuy i think that the
> manufactuer plays one of the biggest roles in computer speed
You are an idiot. The manufacturer, IF he does his job, makes
the product as designed. PERIOD. Got it?? HE DOESN'T design
ANYTHING!!! Therefore how could HE improve the speed???
> not the advertisements
Of course not!! God you're stupid. Advertising is another word
for lying.
> or the benchmarks
Again STUPID. Benchmarks ARE truthfull. But they lie bout the
results just like advertising. It confuses consumers into
thinking they are getting a good deal. So this one SHOULD be
a stupid thing to say also, but they squeeze the numbers.(lie)
j...@acsu.buffalo.edu
Dan
You seem to be fairly familiar with the AMD 5x86 chips, I'm thinking of
buying one, since the P6 market has degenerated into a slop of new form
factors and waiting for decent motherboard chip sets. Besides, I LOVE what
AMD is doing with the K6 and I can't wait to check that chip out.
I have a single P90 on an ASUS dual processor motherboard (Neptune
chipset). I need a second machine in the next two months, and this one
could really use a speed boost for its new task as a server. More RAM will
be added, for one. I was thinking about grabbing a second P90 on their way
out and trying the dual processor route (NT4), but I was wondering what you
thought about using an AMD chip as a single faster chip instead. I have no
real experience to suggest whether a single faster processor or two 90Mhz
ones will better service my particular machine. Have you ever seen an AMD
used as the two processors in a dual processor setup? I know of course that
you can't mix the two.
As for the second machine, I've been looking at building an Intel P133 from
the parts up, and now considering a decent AMD chip while I wait for the K6
to hit a mainstream price and availability. I want a decent motherboard, I
assume the newest ones will all handle the AMDs.
How's Quake and other Dos games on an AMD 5x86?
Scott Lyon
hahahaha....SPAM!
Who are you kidding.....Cyrix may run alot of software faster than a
comparable Pentium, but saying that both 133's get the same FPS in Quake
is a load CRAP. Having done the benchmarks the Cyrix is at least 40%
slower in Quake than a Pentuim with the exact same equipment.
The only way you could possibly get a Pentium to run as slow as a Cyrix
in Quake is to put it on a lame motherboard with a lame ISA video card!@
--
********************************************
sco...@cyberlynk.com
GAMERS CORNER -> http://www.cyberlynk.com/~scottl
Handle -> MUGGER. NET QUAKE Rules !!!!!
********************************************
Pete Olson
Downlo4392 (downl...@aol.com) wrote:
: well i am happy with my Cyrix 133 chip and i can bitch about this tread
: because i also own a pentium 133 and ill tell you that in every aspect the
: speed difference in games is noticable(exept quake, it runs at the same
: fps)and i know that my cyrix 133 compared to my pent133 is a much better
: buy my cyrix is from CyberMax my pent is from BestBuy i think that the
: manufactuer plays one of the biggest roles in computer speed not the
: advertisements or the benchmarks
What a fucking lie. The people in this group are not going
to be duped by a dimwit like you into thinking that Cyrix is
anywhere near competitive with the Pentium for Quake performance.
Go peddle your bullshit somewhere else.
-from the owner of P150+ and a Pentium 166
Tim Allen
"a.j." <ho...@electriciti.com> wrote:
>Oh, well you see the lies are admitted even by YOU.
>If you check further, there are other things that
>don't work as well either. Not that it matters, if
>you truly understand modern computers to be like
>model t's were to cars, you'll know in the future
>BOTH and evewn the pentium pro's are total garbage.
Arguing that things will be garbage in the future is completely
useless. It doesn't prove a thing. *Everything* that we have will
*eventually* become obsolete.
>HELLOOOOoooOOOOoooo. Better buy!! YA!!! Better??? NOOOO.
>Even the fastest ppro is lame. Future games will require
>twice as much as the fastest ppro has today.
If you decide to wait till performance in a given price range
improves, you'll be waiting forever. It's always gonna be better. All
you can do is buy as far up the range as you can afford, then save for
the next upgrade (or decide to abandon technology altogether :)
Tim Allen
The Microsoft Network is prohibited from redistributing this work in any form,
in whole or in part. Copyright (C) Thristian 1995. Liscence to distribute this
post is available to Microsoft for $800. Appearance without permission
constitutes agreement to these terms.
==============================================================================
* Mail: ti...@uuscss.cs.su.oz.au Homepage: http://uuscss.cs.su.oz.au/~tim_a *
RalphEvans
I have a 5x86 and it runs Quake somewhat slow. I think it's 10 fps at
that first screen that Intel P133s get 23-28 in. It runs Duke at full
speed (30+) and everything else but the 5x86 floating point processor
doesn't work well with Quake.
Ralph "Hat Trick" Evans at H2H at 572-3666
Best California Duke/Warcraft/Doom board (Hey! I don't own it: I'm just a subscriber!)
Maker of Fragii, Fragiii, Frag4, Frag5, Ralph, R, Ralph3, Ralph4, Ralph5, Ralph6 (God, when will it end!)---Combo deathmatch wads.
Adam Deutsch
Please help.
I would like to know
1) what video settings I should use in Quake with my 17 inch monitor--
when I try to change them, the screen looks weird . I'm just using
whatever default is used, but some others look much sharper.
2) Is there any way to tweak the program to run even better on my PPro200
with 2 MB 3-d SGRAM card. There is something in the manual about a 3rd
party site that offers some kinda program or something.
thanks
***********************************************
Adam Deutsch, M.D.
Columbia Presbyterian Medical Center
New York, New York
***********************************************
Tim Allen
Adam Deutsch <ad...@columbia.edu> wrote:
>1) what video settings I should use in Quake with my 17 inch monitor--
>when I try to change them, the screen looks weird . I'm just using
>whatever default is used, but some others look much sharper.
Use whatever gives you a good frame rate. How do you mean, funny?
Colours weird? Does it look like it's lost its H-hold? Somethings are
the fault of the vidcard, some are the monitor, some are just general
lousiness of the system.
>2) Is there any way to tweak the program to run even better on my PPro200
>with 2 MB 3-d SGRAM card. There is something in the manual about a 3rd
>party site that offers some kinda program or something.
You obviously know that the manual says there *is* a way to tweak the
PPro, so why do you ask us? They even give you the location, so what
are you waiting for?
>thanks
It's OK, don't worry.
>***********************************************
>Adam Deutsch, M.D.
>Columbia Presbyterian Medical Center
>New York, New York
>***********************************************
Tim Allen
Arnold Lee
just a couple things clayton:
1) ever hear of overclocking? i know you may disagree with the procedure,
but since processor life these days is no more than a year you might as
well push your CPU as far as you can. a P133 is easily overclocked (with
the proper stepping) to a P150 and with luck, to a P166.
2) I don't know what "FPU" intensive apps you are using, but go ahead and
list them. quake, btw, is not really FPU intensive. otherwise, the AMD K5
Pr133 would show some truly pathetic frame rates in quake. however, it is
architecturily (argh...i'm sorry about the misspelling) similar to the
intel Pentiums because it can pipeline integer and floating processes at
the same time, wheras the 6x86 can't...i use lightwave 5.0 and previously
3d Studio v4.0, which are "3d" apps.
3) to get truly impressive improvement in frame rate and performance in
"3d" games and "3d" apps, there is no doubt that advanced VIDEO cards
create the biggest performance gains. ever hear of the orchid righteous
3d? with that card, i get far, far better performance from a 6x86 166+
than a single MMX cpu would give me. there is no doubt about that, trust
me. (can you play GL quake with MMX cpus? no? heheheh. too bad.)
Eric Gisin
I don't know why people are saying the 486 is economical.
In Canada, the AMD K5/90 sells for less than the AMD 486/133 (I refuse to
say 5x86), and performs 10% better. How does FP performance compare?
486 MBs sell for more than cheap P5 MBs (FX, SiS), same as no-name VX
MBs, and 20-30$ less that a name brand VX MB.
Considering the 64b vs 32b memory bus, better cache, better chipset
support in Windows for P5s, I can't see any reason to go with a 486
today. I also expect AMD and Cyrix to drop production of 486 clones when
they release MMX P5 clones.
--
Eric Gisin, London.on.ca -- Windows/UNIX/Internet Consulting
http://www.webhaven.com/ericg/ mailto:er...@techie.com
Yousuf Khan
Arnold Lee <meat...@wam.umd.edu> wrote in article
<01bc08d3$5e99f140$51160880@obelisk>...
> just a couple things clayton:
> 2) I don't know what "FPU" intensive apps you are using, but go ahead and
> list them. quake, btw, is not really FPU intensive. otherwise, the AMD
K5
> Pr133 would show some truly pathetic frame rates in quake. however, it
is
> architecturily (argh...i'm sorry about the misspelling) similar to the
> intel Pentiums because it can pipeline integer and floating processes at
> the same time, wheras the 6x86 can't...i use lightwave 5.0 and previously
> 3d Studio v4.0, which are "3d" apps.
Sure it does. The Cyrix can pipeline integer and floating point together
simultaneously just like any other x86 process has been able to since the
286 and 287 math coprocessor. That's just basic FPU design on an x86.
The Cyrix just does it a lot slower, that's all.
Yousuf Khan
Clayton Gewin
Arnold Lee <meat...@wam.umd.edu> wrote in article
<01bc08d3$5e99f140$51160880@obelisk>...
> just a couple things clayton:
>
> 1) ever hear of overclocking? i know you may disagree with the
procedure,
> but since processor life these days is no more than a year you might as
> well push your CPU as far as you can. a P133 is easily overclocked (with
> the proper stepping) to a P150 and with luck, to a P166.
Ever hear of Chip flambe?
> 2) I don't know what "FPU" intensive apps you are using, but go ahead and
> list them. quake, btw, is not really FPU intensive. otherwise, the AMD
K5
> Pr133 would show some truly pathetic frame rates in quake. however, it
is
> architecturily (argh...i'm sorry about the misspelling) similar to the
> intel Pentiums because it can pipeline integer and floating processes at
> the same time, wheras the 6x86 can't...i use lightwave 5.0 and previously
> 3d Studio v4.0, which are "3d" apps.
Maybe. But the pentium has a lookup table that gives it significantly
better floating point. Quake IS very floating point intensive, why do you
think the Pentium Pro does so well with it(it's running in dos)? The Amd
chip does not get very good scores. If you're going to say "I run it just
fine" turn up the resolution to 640x480 on the current version. We'll see
what the GL version does when it gets out of beta...
> 3) to get truly impressive improvement in frame rate and performance in
> "3d" games and "3d" apps, there is no doubt that advanced VIDEO cards
> create the biggest performance gains. ever hear of the orchid righteous
> 3d? with that card, i get far, far better performance from a 6x86 166+
> than a single MMX cpu would give me. there is no doubt about that, trust
> me. (can you play GL quake with MMX cpus? no? heheheh. too bad.)
I have the Diamond Monster 3d card(same 3dfx chip). You want both. 3d
cards do not enhance audio, video or any of the things that MMX does. It
just accelerates 3d rendering. And I can get great performance on my
Pentium 166 with the diamond monster 3d . But You want both. Someone
earlier seems to be under the notion that I preach 200MHZ Pentiums with 16
bit Isa Video cards. Nonsense. But there are lot more things you can do
with mmx technology than just textures and graphics. Go do some research
and you may be suprised. In fact most current systems come with some type
of s3 or ATI or Matrox 3d accelerator. Ever hear of DOLBY AC3? You wont
on current systems....
Clayton Gewin
My Opinions, not my employers...