Hey, UAE-JIT is pretty neat.
SG
For those who may be interested in how UAE-JIT really stacks up
to a seriously overclocked CS-PPC.
--
-Steve.
SG
> On Fri, 18 Jan 2002 10:21:42 GMT, "Bjørnar Bolsøy"
> <bbo...@hotmail.nospam> wrote:
>
> >"SG" <s...@erols.com> wrote in news:3C476B86.M...@erols.com:
> >
> >> http://amigapro.com
> >
> > Got my Cel400 right here, anything new you
> > want me to race you at?
>
> Just today i got my hands on a Laptop P2-266mhz 64mb, which is happy
> running Amithlon at the moment.
> quick performance judgement is in the 040 range,
Thats too bad, I expected it to be a tiny bit faster than AF5.
Even AF5's lowly 68K-only performance is faster than that. With
UI performance well below the unaccel'd A500 range.
> it have no
> problems playing the OS3.9 included video's at full frame
> rate.
How do you know that, the player automaticallys skip frames.
> and it's MUCH faster than the 030@25mhz standing next to it.
> (both with P96 running)
>
> So Steve, up for some benchmarks, running your Win-UAE (not UAE-JIT)
> vs bjørnar (Winuae) and me running Amithlon?
How does your Amithlon mess stack up to my UAE-JIT Amiga's
1141.21MB/s sustained disk performance? Point is, benchmarking
both is an utter waste of time--their timing is obviously (likely
intentionally) utter garbage.
> According to you, they all use the same 68k engine, right?
>
>
> Lasse Bodilsen <-> Sca 'at' BodyCount.dk
> AMItlhon@1300/MatroxG100/256mb/13gig/Os3.9
--
-Steve.
Stefan Kleinheinrich
"SG" <s...@erols.com> schrieb im Newsbeitrag
news:3C47FD55.M...@erols.com...
It depends on what software your running.
The JIT is optimiced for new CPUs (Athlon(XP)/P3/P4) it will be much slower
on a
Duron or Celeron running at the same clockspeed. Same goes for the mobile
versions
who have smaller and slower caches, and it won't even run on my old K6/400.
So try to compare you >$2000 Amiga with a ~$1000
(AthlonXP/DDR-Ram/GeForce2),
maybe even try Amithlon on it.
I got A2000/60 at 50mhz and a Athlon1600XP with Amithlon and i tell you i
haven't used
the A2000 eversince.
Oh, before you begin with this 16bit-crap: The ram is on the blizzard so is
the SCSI and
the GFXCard doesn't matter when i run a compiler.
A2000->WinUAE : factor 4-5
A2000->Amithlon : factor 8-20
Thats the reality yours are just dreams
Stefan Kleinheinrich
SG
> Oh, before you begin with this 16bit-crap: The ram is on the blizzard so is
> the SCSI and
> the GFXCard doesn't matter when i run a compiler.
> A2000->WinUAE : factor 4-5
> A2000->Amithlon : factor 8-20
> Thats the reality yours are just dreams
Sure, as long as you don't care about memory, disk, or gfx speed,
and working through a UI thats slower than A500, the emulation is
fine, thank you.
--
-Steve.
Bjørnar Bolsøy
> Lasse Bodilsen <s...@bodycount.dk> wrote:
>> it have no
>> problems playing the OS3.9 included video's at full frame rate.
>
> How do you know that, the player automaticallys skip frames.
You mean you don't notice something as noticable as missed
frames? Gosh. :)
>> So Steve, up for some benchmarks, running your Win-UAE (not
>> UAE-JIT) vs bjørnar (Winuae) and me running Amithlon?
>
> How does your Amithlon mess stack up to my UAE-JIT Amiga's
> 1141.21MB/s sustained disk performance? Point is, benchmarking
> both is an utter waste of time--their timing is obviously (likely
> intentionally) utter garbage.
That's why we'll use our stopwatches instead. Syntethic
benchmarks are not interesting anyway.
What shall we have a go at?
Regards...
Lasse Bodilsen
<bbo...@hotmail.nospam> wrote:
>> How does your Amithlon mess stack up to my UAE-JIT Amiga's
>> 1141.21MB/s sustained disk performance? Point is, benchmarking
>> both is an utter waste of time--their timing is obviously (likely
>> intentionally) utter garbage.
>
> That's why we'll use our stopwatches instead. Syntethic
> benchmarks are not interesting anyway.
>
> What shall we have a go at?
I've done some fast tests, just to get the picture of this little
thing.
but of curse some better test could be made, for instance the visage
test. but i fear the 9mb i have left after booting won't be enough
for the larger pictures :-)
P2-266mhz.
-----------
RC5 - 291.280
OGR - 539.033
Sysspeed Mips - 100.45
Sysspeed Mflops - 186.37
Voxelspace (warpup)
WB window 320*256 - 30fps
WB window 640*480 - 8fps
Fullscreen 640*480 - 22fps
CandyFactory pro (aminet demo)
PurpleGoo - 2.45
ShinyMetal - 2.46
TinFoil - 0.47
Transparent - 0.38
-----------
Looks like 060 performence to me.
Stefan Kleinheinrich
Memory? Let me see the CSPPC has FP-RAM at 50ns, no 2nd-levelcache. The
Athlon has
DDR-266 at cl2. That just about 5 years of new technology plus the cache.
Disk-Speed ?
UW-SCSI of 1997 against AtA100 with current 60Gb IBM-drive, cmon forget it.
GFX-Speed ? Max. on a A4000: GRex+Voodoo3PCI. Max. on Amithlon GeForce2 with
GeForce3 coming soon.
Oh i forgot Amithlon loses 99% of the power !! Get real, it loses about 60%
in a worst
case scenario and thats still faster then your 5 year old patchwork-system.
As i said try Amithlon on a current PC and you'll see. I've compared to my
060 and i know
what the P5-PPC can do or can't do. Like running software in pure PPC-Code
(because
ther isn't any)
al
> How does your Amithlon mess stack up to my UAE-JIT Amiga's
> 1141.21MB/s sustained disk performance? Point is, benchmarking
> both is an utter waste of time--their timing is obviously (likely
> intentionally) utter garbage.
>
>
>
> -Steve.
Sorry to butt in here, but if benchmarking is a waste of time, why do
you seem to be so obsessed by it?
And why is it you refuse to open your eyes to the fact that other OSs
are also just as good and as usable as AOS?
Your obsession with trying to rubbish the multitasking of WinXP is also
rather worrying, I suppose you're going to try and rubbish the multitasking
of POSIX systems next? Windows XP, although not perfect (it is designed so
even an idiot can use it), is an incredibly advanced piece of software
design with multitasking performance way above what AOS can manage.
The PPC is NOT the be all and end all of processor design (taken a look
at AMDs Palomino core Athlons or the new Hammer architecture), there are
plenty of other processors available that are much more advanced.
Don't give me any of that shit like "you are clueless" or "you have
obviously never used an Amiga" like you seem to end all your arguments with.
I'm not clueless, I have a degree in software engineering and systems
integration, and I have had an Amiga (and used it regularly up until a
couple of years ago when it became too limited for my purposes and upgrade
options too expensive) since 1987.
Rant over.
Alex.
Tony Mulvihill
LB> P2-266mhz.
LB> -----------
LB> RC5 - 291.280
LB> OGR - 539.033
LB>
LB> Sysspeed Mips - 100.45
LB> Sysspeed Mflops - 186.37
LB>
LB> Voxelspace (warpup)
LB> WB window 320*256 - 30fps
LB> WB window 640*480 - 8fps
LB> Fullscreen 640*480 - 22fps
LB>
LB> CandyFactory pro (aminet demo)
LB> PurpleGoo - 2.45
LB> ShinyMetal - 2.46
LB> TinFoil - 0.47
LB> Transparent - 0.38
LB> -----------
LB>
LB> Looks like 060 performence to me.
Actually it's quite a bit better than 50mhz 060 performance.
Cheers :-)
--
Tony Mulvihill
Happy Amiga User. A1200 060 scsi ide ram rom blah blah
Member & Secretary of the AUG <http://www.aug.org.au>
"Workbench" Editor. The magazine of the AUG
ICQ # 20578798
Petri Airio
What do you think happens when you type "make" to command promt and
compiling starts ?
Bjørnar Bolsøy
What do you mean, UI and Gfx slower than A500?
Regards...
Bjørnar Bolsøy
news:fa1h4u8oa7cbn7pum...@4ax.com:
> P2-266mhz.
> -----------
> RC5 - 291.280
> OGR - 539.033
>
> Sysspeed Mips - 100.45
> Sysspeed Mflops - 186.37
>
> Voxelspace (warpup)
> WB window 320*256 - 30fps
> WB window 640*480 - 8fps
> Fullscreen 640*480 - 22fps
Which archive? I'd like to test that.
Regards...
Lasse Bodilsen
<bbo...@hotmail.nospam> wrote:
>> Voxelspace (warpup)
>> WB window 320*256 - 30fps
>> WB window 640*480 - 8fps
>> Fullscreen 640*480 - 22fps
>
>
> Which archive? I'd like to test that.
OS3.9 Cd /contribution drawer /warpup_v5 :-)
Also available in this archive:
ftp://ftp.haage-partner.com/WarpUP/WarpUP_Release4.0.lha
Bjørnar Bolsøy
news:a2a0gu$7rb$03$1...@news.t-online.com:
> UW-SCSI of 1997 against AtA100 with current 60Gb IBM-drive, cmon
> forget it. GFX-Speed ? Max. on a A4000: GRex+Voodoo3PCI. Max. on
> Amithlon
Looks like Amithlon is already considerably faster than
Voodoo3'ed Amigas in gfx output. My 640Mhz WinUAE-JIT
setup is not far behind either.
If you look here:
http://www.amithlon.com/product-amithlon-bench.html
I've put the V3/Mediator figures in behind my own
here (800x600@16bit):
Cel640-JIT Amiga060/V3
RectFill()................ 2582 op/s 17,794
RectFill() Pattern........ 2420 op/s 16,801
WritePixel().............. 386349 op/s 194,676
WriteChunkyPixels()....... 365 op/s 288
WritePixelArray8()........ 365 op/s 285
WritePixelLine8()......... 34440 op/s 16,158
DrawEllipse()............. 31570 op/s 11,050
DrawCircle().............. 36211 op/s 11,610
Draw().................... 3635 op/s 22,784
Draw() Hor/Ver............ 10107 op/s 26,385
ScrollRaster() X.......... 579 op/s 971
ScrollRaster() Y.......... 607 op/s 1,002
PutText()................. 11456 op/s 12,236
BlitBitMap().............. 5610 op/s 17,194
BlitBitMapRastPort()...... 4944 op/s 13,037
BitMapScale()............. 216 op/s 29
--------------- Intuition -------------
OpenWindow().............. 251 op/s 158
MoveWindow().............. 312 op/s 893
SizeWindow().............. 144 op/s 196
CON-Output................ 271 op/s 1,176
ScreenToFront()........... 59 op/s 60
Regards...
Bjørnar Bolsøy
> On Sat, 19 Jan 2002 14:47:23 GMT, "Bjørnar Bolsøy"
><bbo...@hotmail.nospam> wrote:
>>> Voxelspace (warpup)
>>> WB window 320*256 - 30fps
>>> WB window 640*480 - 8fps
>>> Fullscreen 640*480 - 22fps
>>
>>
>> Which archive? I'd like to test that.
>
> OS3.9 Cd /contribution drawer /warpup_v5 :-)
>
> Also available in this archive:
> ftp://ftp.haage-partner.com/WarpUP/WarpUP_Release4.0.lha
Thank's. My Cel640-JIT:
Fullscreen CybGfx 640*480 - 19fps
Fullscreen P96 640*480 - 8fps
WB window 640*480 - 7fps
Regards...
Bjørnar Bolsøy
My Cel640-JIT:
WarpRace V1.1 1997 by Sam Jordan
A performance test program for 68K and PPC processors
CPU: MC68040
----------------------------------------
Module: ByteWrite V1.0
Result: Memory performance: 132.05 MB/s
----------------------------------------
Module: LongWrite V1.0
Result: Memory performance: 141.00 MB/s
----------------------------------------
Module: WordWrite V1.0
Result: Memory performance: 141.05 MB/s
----------------------------------------
Module: ByteRead V1.0
Result: Memory performance: 88.02 MB/s
----------------------------------------
Module: LongRead V1.0
Result: Memory performance: 209.04 MB/s
----------------------------------------
Module: WordRead V1.0
Result: Memory performance: 110.02 MB/s
----------------------------------------
Module: ByteCopy V1.0
Result: Memory performance: 62.00 MB/s
----------------------------------------
Module: LongCopy V1.0
Result: Memory performance: 65.04 MB/s
----------------------------------------
Module: TurboCopy V1.0
Result: Memory performance: 73.02 MB/s
----------------------------------------
Module: WordCopy V1.0
Result: Memory performance: 62.03 MB/s
----------------------------------------
Module: PixelOMania V1.0
Result: Elapsed time: 0.032768 seconds
----------------------------------------
Module: C2P V1.0
Result: Elapsed time: 2.088 ms
----------------------------------------
Module: CyberMand V1.0
Result: Elapsed time: 0.214527 seconds
----------------------------------------
Module: Landscape V1.0
Result: Elapsed time: 0.965626 seconds
Regards...
Lasse Bodilsen
<bbo...@hotmail.nospam> wrote:
Cel640-WINUAE JIT vs P2 266mhz AMIthlon
Winuae numbers 1st, and Amithlon numbers 2nd.
>WarpRace V1.1 1997 by Sam Jordan
>A performance test program for 68K and PPC processors
>
>CPU: MC68040
>----------------------------------------
>Module: ByteWrite V1.0
>Result: Memory performance: 132.05 MB/s - 144.06 MB/s
>----------------------------------------
>Module: LongWrite V1.0
>Result: Memory performance: 141.00 MB/s - 220.00 MB/s
>----------------------------------------
>Module: WordWrite V1.0
>Result: Memory performance: 141.05 MB/s - 144.04 MB/s
>----------------------------------------
>Module: ByteRead V1.0
>Result: Memory performance: 88.02 MB/s - 92.04 MB/s
>----------------------------------------
>Module: LongRead V1.0
>Result: Memory performance: 209.04 MB/s - 361.02 MB/s
>----------------------------------------
>Module: WordRead V1.0
>Result: Memory performance: 110.02 MB/s - 108.01 MB/s
>----------------------------------------
>Module: ByteCopy V1.0
>Result: Memory performance: 62.00 MB/s - 78.03 MB/s
>----------------------------------------
>Module: LongCopy V1.0
>Result: Memory performance: 65.04 MB/s - 144.01 MB/s
>----------------------------------------
>Module: TurboCopy V1.0
>Result: Memory performance: 73.02 MB/s - 157.04 MB/s
>----------------------------------------
>Module: WordCopy V1.0
>Result: Memory performance: 62.03 MB/s - 81.01 MB/s
>----------------------------------------
>Module: PixelOMania V1.0
>Result: Elapsed time: 0.032768 seconds - 0.056455 seconds
>----------------------------------------
>Module: C2P V1.0
>Result: Elapsed time: 2.088 ms - 2.907 ms
>----------------------------------------
>Module: CyberMand V1.0
>Result: Elapsed time: 0.214527 seconds - 0.347554 seconds
>----------------------------------------
>Module: Landscape V1.0
>Result: Elapsed time: 0.965626 seconds - 1.878472 seconds
>
It should be noted, that the reason the P2 266mhz might perform quite
good, is properly due to its 512KB L2 cache, vs the Celerons 128KB L2
cache.
And also please bear in mind that the P2 is a laptop.
But enough syntectic benchmarks:
Bernoit v2.0 - 1024*768 16bit
Default settings : 1.58sec
Exponent = 2, Max Iter. = 99999 : 10:43.81 mins
Exponent = 16, Max Iter. = 9999 : 8:41.78 mins
(verified with stopwatch.)
Casper
SK> Duron or Celeron running at the same clockspeed. Same goes for the mobile
SK> versions
SK> who have smaller and slower caches, and it won't even run on my old K6/400.
What? You mean I'm not going to be able to give this stuff a try on
the K6 2/500 that someone gave me yesterday! :0
regards,
--
------------------------------------------------------------------------
Posted with Amiga NewsRog
------------------------------------------------------------------------
Stefan Kleinheinrich
"Casper" <Cas...@netnet.net> schrieb im Newsbeitrag
news:SB.GB.iH279mW...@netnet.net...
> "SK" == "Stefan Kleinheinrich" writes:
>
>
> SK> Duron or Celeron running at the same clockspeed. Same goes for the
mobile
> SK> versions
> SK> who have smaller and slower caches, and it won't even run on my old
K6/400.
>
> What? You mean I'm not going to be able to give this stuff a try on
> the K6 2/500 that someone gave me yesterday! :0
My board wasn't really designed for the K6/3 (Intel TX Chips) so you may
have more
luck. 060/50 using WinUAE or 100mhz 060 using Amithlon is my guess. A 500mhz
Athlon would be at ~200mhz (Amithlon).
Bjørnar Bolsøy
news:a2hn2l$bqu$07$1...@news.t-online.com:
> news:SB.GB.iH279mW...@netnet.net...
>> "SK" == "Stefan Kleinheinrich" writes:
>>
>>
>> SK> Duron or Celeron running at the same clockspeed. Same goes
>> for the mobile SK> versions
>> SK> who have smaller and slower caches, and it won't even run on
>> my old K6/400.
>>
>> What? You mean I'm not going to be able to give this stuff a try
>> on the K6 2/500 that someone gave me yesterday! :0
>
> My board wasn't really designed for the K6/3 (Intel TX Chips) so
> you may have more
> luck. 060/50 using WinUAE or 100mhz 060 using Amithlon is my
> guess.
Meassured how, exactly? Over the past year I've posted about a
dozen different tests here in this group, ranging from games,
graphics and productivity, and the speed meassured on my home
640Mhz Celeron WinUAE-JIT setup paints a pretty conclusive
picture -- it's from equal up to 4-5 times an 060, right up
to fast PPC speeds in some cases.
This Beniot fractal run is typical.
Benoit startup fractal (1024x768 @ 16bit)
-----------------------------------------
A3k/060-66 2.62 seconds
A3k/604e-280 0.31 seconds
JIT640 0.65 seconds
Regards...
Casper
SK> "Casper" <Cas...@netnet.net> schrieb im Newsbeitrag
SK> news:SB.GB.iH279mW...@netnet.net...
SK> > "SK" == "Stefan Kleinheinrich" writes:
SK> > SK> Duron or Celeron running at the same clockspeed. Same goes for
SK> > SK> the mobile versions
SK> > SK> who have smaller and slower caches, and it won't even run on my
SK> > SK> old
SK> < K6/400.
SK> > What? You mean I'm not going to be able to give this stuff a try on
SK> > the K6 2/500 that someone gave me yesterday! :0
SK> My board wasn't really designed for the K6/3 (Intel TX Chips) so you
SK> may have more luck. 060/50 using WinUAE or 100mhz 060 using Amithlon
SK> is my guess. A 500mhz Athlon would be at ~200mhz (Amithlon).
Does anyone know if it would even worth the effort to try either
WinUAE JIT or Amithlon on a pretty well loaded out K6 2/500?
It only has 60MB's of RAM btw...but that could easily be remedied.
SG
Again, the benchmarks are completely screwed up. With a 50MIPS
or so, average (its wildly different everytime you hit the test
button), SysSpeed performance, I'd peg this emulation at stock
020 speed overall, tops. Somethings are indeed faster than that,
somethings are indeed slower.
--
-Steve.
SG
You guys should really check what's been on the web for years...
http://amigapro.com/cvppc.html
That screenshot was before overclocking all three CS-PPC clocks,
its around 130fps now, my CS-PPC is a mere 7x faster than UAE-JIT
at 640MHz. I'll have to send WarpUp on over to the laptop,
standby for more...
--
-Steve.
SG
> "SG" <s...@erols.com> wrote in news:3C481807.M...@erols.com:
> > Lasse Bodilsen <s...@bodycount.dk> wrote:
>
> >> it have no
> >> problems playing the OS3.9 included video's at full frame rate.
> >
> > How do you know that, the player automaticallys skip frames.
>
> You mean you don't notice something as noticable as missed
> frames? Gosh. :)
If you can verify all 20 each sec, you are quicker than me.
> >> So Steve, up for some benchmarks, running your Win-UAE (not
> >> UAE-JIT) vs bjørnar (Winuae) and me running Amithlon?
> >
> > How does your Amithlon mess stack up to my UAE-JIT Amiga's
> > 1141.21MB/s sustained disk performance? Point is, benchmarking
> > both is an utter waste of time--their timing is obviously (likely
> > intentionally) utter garbage.
>
> That's why we'll use our stopwatches instead. Syntethic
> benchmarks are not interesting anyway.
>
> What shall we have a go at?
My CS-PPC's 130fps to your Cel640's 18fps is interesting.
--
-Steve.
SG
> SG <s...@erols.com> wrote:
>
> > Bjørnar Bolsøy <bbo...@hotmail.nospam> wrote:
> > > Thank's. My Cel640-JIT:
> > >
> > > Fullscreen CybGfx 640*480 - 19fps
> > > Fullscreen P96 640*480 - 8fps
> > > WB window 640*480 - 7fps
> >
> > You guys should really check what's been on the web for years...
> >
> > http://amigapro.com/cvppc.html
> >
> > That screenshot was before overclocking all three CS-PPC clocks,
> > its around 130fps now, my CS-PPC is a mere 7x faster than UAE-JIT
>
> Ooops, I read that wrong the first time, the CS-PPC is actually
> 19x faster than your Cel640 to the WB.
>
> > at 640MHz. I'll have to send WarpUp on over to the laptop,
> > standby for more...
>
> UAE-JIT in a Pic96 16-bit WB window = 3.1 fps.
Ahh... I didn't realize you down'd your res to 640x480x8 for
that test, that res is 5.3fps here.
> Fullscreen 800x600x8-bit = 3.6fps
Interestingly, UAE-JIT on my laptop is 60% faster per MHz than
your Celeron. I wonder why a K6 with crappy laptop-gfx is so
much faster?
--
-Steve.
SG
> Bjørnar Bolsøy <bbo...@hotmail.nospam> wrote:
> > Thank's. My Cel640-JIT:
> >
> > Fullscreen CybGfx 640*480 - 19fps
> > Fullscreen P96 640*480 - 8fps
> > WB window 640*480 - 7fps
>
> You guys should really check what's been on the web for years...
>
> http://amigapro.com/cvppc.html
>
> That screenshot was before overclocking all three CS-PPC clocks,
> its around 130fps now, my CS-PPC is a mere 7x faster than UAE-JIT
Ooops, I read that wrong the first time, the CS-PPC is actually
19x faster than your Cel640 to the WB.
> at 640MHz. I'll have to send WarpUp on over to the laptop,
> standby for more...
UAE-JIT in a Pic96 16-bit WB window = 3.1 fps.
Fullscreen 800x600x8-bit = 3.6fps
--
-Steve.
Stefan Kleinheinrich
my good old d K6/3 with its disfunctional FPU running WinUAE0.6??? without
JIT had about the same speed as my
overclocked A2630 at 32Mhz. That on everyday compiler runs and believe me i
would have noticed even the smallest
difference. Hey but who cares as you can buy a brand new AthlonXP-maschine
for the price of a used lowend PPC-
Card. You would newer know what the previous owner did to his card (these
old things aren't really reliable), and
you would only feel the speed in special PPC-Progs all other are done on the
dog-slow 040 or the still quite slow 060.
After that you can either go after a used overpriced and outdated gfx-card
or a expensive PCI-Bridge which will bring
you even more trouble. But maybe fixing one HW-trouble after another us what
you like most about your Amiga.
For me it's the small fast and understandable OS, and not the CPU it runs
on.
Bjørnar Bolsøy
> Bjørnar Bolsøy <bbo...@hotmail.nospam> wrote:
>> Fullscreen CybGfx 640*480 - 19fps
>> Fullscreen P96 640*480 - 8fps
>> WB window 640*480 - 7fps
>
> You guys should really check what's been on the web for years...
>
> http://amigapro.com/cvppc.html
>
> That screenshot was before overclocking all three CS-PPC clocks,
> its around 130fps now, my CS-PPC is a mere 7x faster than UAE-JIT
> at 640MHz.
Seems you are testing 320x240, and I'm testing 640x480.
But my figures were a bit off. All my tests were run in
32bit, windowed on the 32bit Win desktop.
These are more correct:
Fullscreen CybGfx 640*480@8 - 26fps
Fullscreen P96 640*480@8 - 31fps
Fullscreen CybGfx 800*600@8 - 11fps
Fullscreen P96 800*600@8 - 8fps
WB window 320*240@8bit - 85fps (my max freq)
WB window 320*240@16bit - 35fps
WB window 800*600@8bit - 20fps
So my Celly is 3-4x your K6, and right up there with
your CSPPC before overclocking.
Regards...
Bjørnar Bolsøy
> Bjørnar Bolsøy <bbo...@hotmail.nospam> wrote:
>> Meassured how, exactly? Over the past year I've posted about a
>> dozen different tests here in this group, ranging from games,
>> graphics and productivity, and the speed meassured on my home
>> 640Mhz Celeron WinUAE-JIT setup paints a pretty conclusive
>> picture -- it's from equal up to 4-5 times an 060, right up to
>> fast PPC speeds in some cases.
>>
>> This Beniot fractal run is typical.
>>
>> Benoit startup fractal (1024x768 @ 16bit)
>> -----------------------------------------
>> A3k/060-66 2.62 seconds
>> A3k/604e-280 0.31 seconds
>> JIT640 0.65 seconds
>
> Again, the benchmarks are completely screwed up.
It's consistent thruought closer to a dozen different
tests Steve.
Why do you simply ignore everything?
Regards...
SG
> "SG" <s...@erols.com> schrieb im Newsbeitrag
> news:3C4CD70A.M...@erols.com...
> > Again, the benchmarks are completely screwed up. With a 50MIPS
> > or so, average (its wildly different everytime you hit the test
> > button), SysSpeed performance, I'd peg this emulation at stock
> > 020 speed overall, tops. Somethings are indeed faster than that,
> > somethings are indeed slower.
> > --
> >
> > -Steve.
> Oh Steve
> my good old d K6/3 with its disfunctional FPU running WinUAE0.6??? without
> JIT had about the same speed as my
> overclocked A2630 at 32Mhz. That on everyday compiler runs and believe me i
> would have noticed even the smallest
> difference. Hey but who cares as you can buy a brand new AthlonXP-maschine
> for the price of a used lowend PPC-
Thats might be a more interesting point if (1) the emulation was
seriously usable--its not, and (2) it was true.
New PPC cards cost around $650 with an 060, PPC, and UWSCSI, and
are amazingly fast, in general use by direct comparision
(reality, not just UAE-goofy benchmarks. CS-PPC cards don't
provide 1.1GB/s sustained disk performance from IDE drives, for
example).
--
-Steve.
Bjørnar Bolsøy
> CS-PPC cards don't
> provide 1.1GB/s sustained disk performance from IDE drives, for
> example).
You are claiming WinUAE does?
Regards...
SG
Problem is, I'm not ignoring anyting, for example 1.1GB/s
syustained disk i/o. Most speed test vary wildly even from
instance to instance. Most of the benchmarks are 100% worthless.
You are also fudging the emulation settings from test to test,
which simply doesn't cut it, you have to be 100% compatible and
consistent (which I'm not even doing becasue UAE is unusably slow
that way).
--
-Steve.
SG
> "SG" <s...@erols.com> wrote in news:3C4CDAAD.M...@erols.com:
> > Bjørnar Bolsøy <bbo...@hotmail.nospam> wrote:
>
> >> Fullscreen CybGfx 640*480 - 19fps
> >> Fullscreen P96 640*480 - 8fps
> >> WB window 640*480 - 7fps
> >
> > You guys should really check what's been on the web for years...
> >
> > http://amigapro.com/cvppc.html
> >
> > That screenshot was before overclocking all three CS-PPC clocks,
> > its around 130fps now, my CS-PPC is a mere 7x faster than UAE-JIT
> > at 640MHz.
>
> Seems you are testing 320x240, and I'm testing 640x480.
>
> But my figures were a bit off. All my tests were run in
> 32bit, windowed on the 32bit Win desktop.
I can't run them in anything but 8-bit, WarpUp 5 doesn't allow
it. How did you do it?
> These are more correct:
>
> Fullscreen CybGfx 640*480@8 - 26fps
> Fullscreen P96 640*480@8 - 31fps
>
> Fullscreen CybGfx 800*600@8 - 11fps
> Fullscreen P96 800*600@8 - 8fps
>
> WB window 320*240@8bit - 85fps (my max freq)
> WB window 320*240@16bit - 35fps
> WB window 800*600@8bit - 20fps
>
> So my Celly is 3-4x your K6, and right up there with
> your CSPPC before overclocking.
>
>
> Regards...
--
-Steve.
Bjørnar Bolsøy
>> > That screenshot was before overclocking all three CS-PPC
>> > clocks, its around 130fps now, my CS-PPC is a mere 7x faster
>> > than UAE-JIT at 640MHz.
>>
>> Seems you are testing 320x240, and I'm testing 640x480.
>>
>> But my figures were a bit off. All my tests were run in 32bit,
>> windowed on the 32bit Win desktop.
>
> I can't run them in anything but 8-bit, WarpUp 5 doesn't allow
> it. How did you do it?
No that's correct, same happens here. I meant to say that all
my tests were running windowed on the 32bit Win desktop.
Voxelspace WB window run was too as my default Amiga screenmode
is 1024x768@32bit), but not the fullscreen ones of course.
And although 85fps is the max vblank limit for my lowres
screenmodes (because my monitor has a tendency of producing
garbage if I go anything higher) it seems to max out around
there in 320x240 anyway.
Regards...
Bjørnar Bolsøy
> Bjørnar Bolsøy <bbo...@hotmail.nospam> wrote:
>> Why do you simply ignore everything?
>
> Problem is, I'm not ignoring anyting, for example 1.1GB/s
> syustained disk i/o.
Here from my sept 30th post, which you are ignoring as
we speak.
1. A2k/040-33/PIV, SCSI
2. A3k/CVPPC:060-50/604e-200, UWSCSI
3. JIT640/GF2Mx IDE-UDMA33
Drive ind1 Op/s 1296 2955 3612/1156
Drive ind2 MB/s 1.79 12.64 42.34/338
NB. First figure is using a hardfile, second is the
Windows filesystem wrapper. But obviously the results
are not particularly interesting as drive I/O is cached
under Windows in both cases.
Where do you see any mention of a 1.1GB/s figure?
> Most of the benchmarks are 100% worthless.
Are they worthless because they put WinUAE right up
there with the fastest Amigas?
Regards...
Lasse Bodilsen
>> You mean you don't notice something as noticable as missed
>> frames? Gosh. :)
>
>If you can verify all 20 each sec, you are quicker than me.
I really hate to drag this old comment up.
but:
http://groups.google.com/groups?selm=3B07999A.MD-1.4.4.sgio%40erols.com&output=gplain
<SG>
You misunuderstood me, what I'm saying is that with my decade and a
half of intense animation experience, I can tell you any frame rate at
at instant in time, +/- .065 sec, within .0043 fps accuracy, just by
looking at it.
</SG>
(sorry :-)
Rick Hodger
Meaningless Intrsuctions per Second are wildly different on every
different processor/memory architecture. Hence why no serious
benchmark will judge speed purely on MIPS, let alone even use them.
> SysSpeed performance, I'd peg this emulation at stock
> 020 speed overall, tops.
How can you? The bench for an A3000/060 is 2.62 seconds. UAEJIT gets
.65 seconds. The only way that can possibly equal 020 speeds is if the
060 equals 000 with half the clock-speed.
--
Rick Hodger
SG
It depends on what day it is. AF5 ranges from 200-1100MB/s sec
here, changing wildly from test to test. SysSpeed on my CS-PPC
is rock solid in this respect, so it isn't the code.
--
-Steve.
SG
> "SG" <s...@erols.com> wrote in news:3C4D3F31.M...@erols.com:
> > Bjørnar Bolsøy <bbo...@hotmail.nospam> wrote:
>
> >> Why do you simply ignore everything?
> >
> > Problem is, I'm not ignoring anyting, for example 1.1GB/s
> > syustained disk i/o.
>
> Here from my sept 30th post, which you are ignoring as
> we speak.
>
> 1. A2k/040-33/PIV, SCSI
> 2. A3k/CVPPC:060-50/604e-200, UWSCSI
> 3. JIT640/GF2Mx IDE-UDMA33
>
> Drive ind1 Op/s 1296 2955 3612/1156
> Drive ind2 MB/s 1.79 12.64 42.34/338
I'm not sure what those nums stand for, sorry, there is little
to no continuity to even reason it out.
> NB. First figure is using a hardfile, second is the
> Windows filesystem wrapper. But obviously the results
> are not particularly interesting as drive I/O is cached
> under Windows in both cases.
>
> Where do you see any mention of a 1.1GB/s figure?
Here.
> > Most of the benchmarks are 100% worthless.
>
> Are they worthless because they put WinUAE right up
> there with the fastest Amigas?
1100MB/s is a lot more than just "up there."
--
-Steve.
SG
> On Tue, 22 Jan 2002 09:50:44 GMT, "SG" <s...@erols.com> wrote:
>
> >> You mean you don't notice something as noticable as missed
> >> frames? Gosh. :)
> >
> >If you can verify all 20 each sec, you are quicker than me.
>
> I really hate to drag this old comment up.
>
> but:
> http://groups.google.com/groups?selm=3B07999A.MD-1.4.4.sgio%40erols.com&output=gplain
>
> <SG>
> You misunuderstood me, what I'm saying is that with my decade and a
> half of intense animation experience, I can tell you any frame rate at
> at instant in time, +/- .065 sec, within .0043 fps accuracy, just by
> looking at it.
> </SG>
>
> (sorry :-)
YOu replied to my humor with my humor. That still leaves your
ridiculous claim that a 266 doesn't drop any frames. It does.
Big time. Fact is, Action is so jumpy on this 266 its unusably
bad (and thats with no sound).
--
-Steve.
SG
> "SG" <s...@erols.com> wrote in message news:<3C4CD70A.M...@erols.com>...
> > Bjørnar Bolsøy <bbo...@hotmail.nospam> wrote:
> > Again, the benchmarks are completely screwed up. With a 50MIPS
> > or so, average (its wildly different everytime you hit the test
> > button)
>
> Meaningless Intrsuctions per Second are wildly different on every
> different processor/memory architecture.
Thats why I compared UAE-JIT 68K to 68K, using the same test
code.
Obviously.
> Hence why no serious
> benchmark will judge speed purely on MIPS, let alone even use them.
>
> > SysSpeed performance, I'd peg this emulation at stock
> > 020 speed overall, tops.
>
> How can you? The bench for an A3000/060 is 2.62 seconds. UAEJIT gets
> ..65 seconds. The only way that can possibly equal 020 speeds is if the
> 060 equals 000 with half the clock-speed.
It also benches 1120MB/s sustained from a 12MB/s IDE disk (maybe
1MB/s actual Amiga side performance). I'm going on Amiga
experience when the numbers are obviously screwed up, you have
none so you can't.
Benoit is also highly parameter and gfx system dependent, these
configs aren't even close to matching. I can easily get an 060
startup render under .1 secs by playing with the app/OS config.
--
-Steve.
Matthew Garrett
> Rick Hodger <ri...@simpleftp.co.uk> wrote:
>> Meaningless Intrsuctions per Second are wildly different on every
>> different processor/memory architecture.
>
> Thats why I compared UAE-JIT 68K to 68K, using the same test
> code.
The ability of an emulator to run one specific small piece of code at a
given speed is not on its own representative of the overall speed of the
emulator.
> Obviously.
Obviously.
>> How can you? The bench for an A3000/060 is 2.62 seconds. UAEJIT gets
>> ..65 seconds. The only way that can possibly equal 020 speeds is if the
>> 060 equals 000 with half the clock-speed.
>
> It also benches 1120MB/s sustained from a 12MB/s IDE disk (maybe
> 1MB/s actual Amiga side performance). I'm going on Amiga
> experience when the numbers are obviously screwed up, you have
> none so you can't.
The benchmark is almost certainly failing to take the Windows drive
caching into account, which would explain the ridiculousness of it.
Shouldn't this suggest to you that running benchmarking tools that make
certain assumptions about the underlying hardware is rather pointless?
> Benoit is also highly parameter and gfx system dependent, these
> configs aren't even close to matching. I can easily get an 060
> startup render under .1 secs by playing with the app/OS config.
I can't comment on this.
--
Matthew Garrett | mjg59-comp.sys...@srcf.ucam.org
Stefan Kleinheinrich
> Stefan Kleinheinrich <SKleinh...@t-online.de> wrote:
>
> > "SG" <s...@erols.com> schrieb im Newsbeitrag
> > news:3C4CD70A.M...@erols.com...
> > > Again, the benchmarks are completely screwed up. With a 50MIPS
> > > or so, average (its wildly different everytime you hit the test
> > > button), SysSpeed performance, I'd peg this emulation at stock
> > > 020 speed overall, tops. Somethings are indeed faster than that,
> > > somethings are indeed slower.
> > > --
> > >
> > > -Steve.
> > Oh Steve
> > my good old d K6/3 with its disfunctional FPU running WinUAE0.6???
without
> > JIT had about the same speed as my
> > overclocked A2630 at 32Mhz. That on everyday compiler runs and believe
me i
> > would have noticed even the smallest
> > difference. Hey but who cares as you can buy a brand new
AthlonXP-maschine
> > for the price of a used lowend PPC-
>
> Thats might be a more interesting point if (1) the emulation was
> seriously usable--its not, and (2) it was true.
(1)I've developed for over a year on this maschine so it was usable.
(2)YOU are calling somebody a liar GOOD JOKE
> New PPC cards cost around $650 with an 060, PPC, and UWSCSI, and
> are amazingly fast, in general use by direct comparision
> (reality, not just UAE-goofy benchmarks. CS-PPC cards don't
> provide 1.1GB/s sustained disk performance from IDE drives, for
> example).
>
Hmm that about 800Euro. My AthlonXP1600 had cost me 1500DM last september
that about 780Euro and
already included 256mb ddr-ram 60gb hd fast gfx ...
Today it would be at around 650Euro.
BtW: Why do always hack on those benchmarks NO ONE ever said there were
real. Try using an old hd-bench-
mark from Fish-Disk or that was made in 198x and see what resonable result
you get on your PPC.
In the end of the day its real speed thats counts and thats where today
Ultra-DMA-ide crashes your 5 year old
SCSI even if its driven through a emulator.
> --
>
> -Steve.
Bjørnar Bolsøy
Here I can play 5 or 6, simoultaneously, of the OS3.9 CD avi's
before it starts to skip.
Regards...
Bjørnar Bolsøy
> Benoit is also highly parameter and gfx system dependent, these
> configs aren't even close to matching. I can easily get an 060
> startup render under .1 secs by playing with the app/OS config.
Just do what everyone else does, fire it up, press "calc"
and post the result.
Regards...
Troy Parker
> Interestingly, UAE-JIT on my laptop is 60% faster per MHz than
> your Celeron. I wonder why a K6 with crappy laptop-gfx is so
> much faster?
The K6 is better at integer math than the celeron, and the K6 was
traditionally bundled with motherboards providing usually between 512-
1Mb of L2 cache.
And, lets just say the early celerons, well, sucked.
The K6, however, had production problems relating to it's L1 cache, many
early units had to be returned or run without L1.
Troy Parker
> Thats might be a more interesting point if (1) the emulation was
> seriously usable--its not, and (2) it was true.
Unusable? Says who?
There's a good number of Amiga developers drooling over the opportunity
to port their Amiga apps to Amithlon/x86 native. Newscoaster might be
ported soon, from what I hear.
It would be interesting to see how an Amithlon system with x86
recompiled apps would perform.
Tom and Lisa Peters
> Again, the benchmarks are completely screwed up. With a 50MIPS
> or so, average (its wildly different everytime you hit the test
> button), SysSpeed performance, I'd peg this emulation at stock
> 020 speed overall, tops. Somethings are indeed faster than that,
> somethings are indeed slower.
So let me get this straight. You would rather base an emulation speed
on a meaningless indicator as MIPS (and a highly variable one, according
to you), as opposed to a timed fractal benchmark ?
Are you this dumb every day ?
> -Steve.
Mad Dog
SG
> SG wrote:
> > Again, the benchmarks are completely screwed up. With a 50MIPS
> > or so, average (its wildly different everytime you hit the test
> > button), SysSpeed performance, I'd peg this emulation at stock
> > 020 speed overall, tops. Somethings are indeed faster than that,
> > somethings are indeed slower.
>
> So let me get this straight. You would rather base an emulation speed
> on a meaningless indicator as MIPS (and a highly variable one, according
> to you),
Its 100% perfectly controlled when running the same instruction
set using the same code on two sets of actual hardware. You are
confusing buzz words that you once overheard with actual thought.
Hardware vs hardware emulated in software with different levels
of test caching, half the machine not emulated properly, running
a purely gfx rendering test using completely different gfx driver
subsystems and gfx hardware, and using completely different
processor instruction sets is not a valid test. Even though in
this case the Amiga hardware wins by a massive margin.
> as opposed to a timed fractal benchmark ?
>
> Are you this dumb every day ?
Do you understand what I just taught you? If no, ask me again,
I'll be glad to explain further.
> > -Steve.
>
> Mad Dog
--
-Steve.
SG
Agian, how do you know its not dropping frames? And Bjørnar, an
040 Amiga can play more than one instance smoothly--this isn't
Windows.
I can tell you for a fact that a WinUAE on a 266 (faster version
than his) is completely unusable for smooth playback of a single
Action stream--thats with no sound.
--
-Steve.
SG
You can't, the times are way different based on OS params.
--
-Steve.
Thomas Tavoly
In article <3C4D3AC1.M...@erols.com> "SG" <s...@erols.com> writes:
> Stefan Kleinheinrich <SKleinh...@t-online.de> wrote:
>
> > Oh Steve
> > my good old d K6/3 with its disfunctional FPU running WinUAE0.6??? without
> > JIT had about the same speed as my
> > overclocked A2630 at 32Mhz. That on everyday compiler runs and believe me i
> > would have noticed even the smallest
> > difference. Hey but who cares as you can buy a brand new AthlonXP-maschine
> > for the price of a used lowend PPC-
>
> Thats might be a more interesting point if (1) the emulation was
> seriously usable--its not, and (2) it was true.
Truth hurts. Amithlon for example is seriously usable, I haven't switched
on my 4000/060/CV64/3D in something like two months now because it's on
average 10-15 times slower compared to my Athlon.
> New PPC cards cost around $650 with an 060, PPC, and UWSCSI,
Sorry, $925 at softhut.com, I'm sure they would like to know where you get
yours from. $200 for nearly complete 1 GHz Athlon system, see old message.
> and are amazingly fast,
For 1997, and for the few things that run on it (nothing, back then).
> in general use by direct comparision
> (reality, not just UAE-goofy benchmarks. CS-PPC cards don't
> provide 1.1GB/s sustained disk performance from IDE drives, for
> example).
Blablabla. Try DTA100.lha in disk/misc off Aminet on a large file (for
example 1 GB) instead. Use a stopwatch. Use your brain.. Use any recent
UDMA or SCSI controller/mobo and it will beat your CSPPC's UWSCSI because
of its crippled memory bus, even in emulation.
...
_ . Thomas Tavoly
. _ // . aTm...@amiga.xs4all.nl
. \X/ http://www.tavoly.net
... 5.4
SG
> In article <3C4D3AC1.M...@erols.com>, s...@erols.com says...
>
>
> > Thats might be a more interesting point if (1) the emulation was
> > seriously usable--its not, and (2) it was true.
>
> Unusable? Says who?
>
> There's a good number of Amiga developers drooling over the opportunity
> to port their Amiga apps to Amithlon/x86 native. Newscoaster might be
> ported soon, from what I hear.
Wow.
> It would be interesting to see how an Amithlon system with x86
> recompiled apps would perform.
That has nothing to do with anything written above.
--
-Steve.
Bjørnar Bolsøy
What OS parameters? It's the same here on a fresh OS3.9
install, OS3.9 upgrade and a souped up OS3.1 install.
Regards...
Bjørnar Bolsøy
> Bjørnar Bolsøy <bbo...@hotmail.nospam> wrote:
>> Here I can play 5 or 6, simoultaneously, of the OS3.9 CD avi's
>> before it starts to skip.
>
> Agian, how do you know its not dropping frames?
It's smooth as a baby's butt.
Right up to when I fire up the 5th or 6th avi.
> And Bjørnar, an
> 040 Amiga can play more than one instance smoothly--this isn't
> Windows.
An 040 doesn't even manage high quality mp3s, forget the
OS3.9 avi's.
> I can tell you for a fact that a WinUAE on a 266 (faster version
> than his) is completely unusable for smooth playback of a single
> Action stream--thats with no sound.
How do you know it's dropping frames?
Regards...
al
"Casper" <Cas...@netnet.net> wrote in message
news:sE.dE.jH292rf...@netnet.net...
> "SK" == "Stefan Kleinheinrich" writes:
>
>
> SK> news:SB.GB.iH279mW...@netnet.net...
> SK> > "SK" == "Stefan Kleinheinrich" writes:
>
> SK> > SK> Duron or Celeron running at the same clockspeed. Same goes for
> SK> > SK> the mobile versions
> SK> > SK> who have smaller and slower caches, and it won't even run on my
> SK> > SK> old
> SK> < K6/400.
>
> SK> > What? You mean I'm not going to be able to give this stuff a try on
> SK> > the K6 2/500 that someone gave me yesterday! :0
>
> SK> My board wasn't really designed for the K6/3 (Intel TX Chips) so you
> SK> may have more luck. 060/50 using WinUAE or 100mhz 060 using Amithlon
> SK> is my guess. A 500mhz Athlon would be at ~200mhz (Amithlon).
>
> Does anyone know if it would even worth the effort to try either
> WinUAE JIT or Amithlon on a pretty well loaded out K6 2/500?
> It only has 60MB's of RAM btw...but that could easily be remedied.
>
> regards,
> --
> ------------------------------------------------------------------------
> Posted with Amiga NewsRog
> ------------------------------------------------------------------------
I'd say definetly go with amithlon, that way you don't have such a HUGE
processor overhead by having to run windows first and then the emulation.
Amithlon is a lot quicker, but less compatible than WinUAE, plus no AGA
support (WinUAE has).
Al.
Graham Briggs
>It also benches 1120MB/s sustained from a 12MB/s IDE disk (maybe
>1MB/s actual Amiga side performance). I'm going on Amiga
>experience when the numbers are obviously screwed up, you have
>none so you can't.
Ignore benchmarks that benchmark hardware, because obviously these are not
going to be accurate on non-Amiga hardware (i.e., emulation).
>Benoit is also highly parameter and gfx system dependent, these
>configs aren't even close to matching. I can easily get an 060
>startup render under .1 secs by playing with the app/OS config.
Looks like an application that does a lot of calculations over time to me.
Perfect as a benchmark because it tests the emulated 68k processor, which
in the previously given example rated at around fast '030 to '040 speed.
Speed of WritePixel is irrelevant here, you ain't doing no fancy polygon
shifting.
What about doing a complex operation in Photogenics? Say a gaussian blur
with a radius of 10 on a predefined image?
Any other "real world" benchmarks that we can set up? I know SG will just
ignore them and keep on wittering about 1GB/s disk throughput benchmarks
everytime he is losing an argument, but we can but try...
Graham
Graham Briggs
>Hardware vs hardware emulated in software with different levels
>of test caching, half the machine not emulated properly, running
>a purely gfx rendering test using completely different gfx driver
>subsystems and gfx hardware, and using completely different
>processor instruction sets is not a valid test. Even though in
>this case the Amiga hardware wins by a massive margin.
It measures getting a task done in an application running under AmigaOS.
What better benchmark could you get?
>Do you understand what I just taught you? If no, ask me again,
>I'll be glad to explain further.
A classic sign of someone who has lost an argument and doesn't know it is
to attack the person they are arguing with.
Graham
SG
> In article <3C4D3AC1.M...@erols.com> "SG" <s...@erols.com> writes:
>
> > Stefan Kleinheinrich <SKleinh...@t-online.de> wrote:
> >
> > > Oh Steve
> > > my good old d K6/3 with its disfunctional FPU running WinUAE0.6??? without
> > > JIT had about the same speed as my
> > > overclocked A2630 at 32Mhz. That on everyday compiler runs and believe me i
> > > would have noticed even the smallest
> > > difference. Hey but who cares as you can buy a brand new AthlonXP-maschine
> > > for the price of a used lowend PPC-
> >
> > Thats might be a more interesting point if (1) the emulation was
> > seriously usable--its not, and (2) it was true.
>
> Truth hurts. Amithlon for example is seriously usable, I haven't switched
> on my 4000/060/CV64/3D in something like two months now because it's on
> average 10-15 times slower compared to my Athlon.
>
> > New PPC cards cost around $650 with an 060, PPC, and UWSCSI,
>
> Sorry, $925 at softhut.com, I'm sure they would like to know where you get
> yours from.
Vesalia has 'em for $607, + $30 for insured 4 day shipping
anywhere in the world--060 and PPC hardware included. Actually
Amiga compatible too.
> $200 for nearly complete 1 GHz Athlon system, see old message.
A comparable UWSCSI card alone costs that much. And it can't
even run AOS at all (as of 02/02). And I've been testing the
Amiga JIT emulation extensively (with a much more powerful than
Amithlon, 256MB JIT engine), they are entire unusable for
instability alone, in addition to downright slow, quite
unfortunately.
I would have enjoyed converting my laptop to AOS, but it simply
doesn't run it for more than a few minutes at a time (at 000 to
020 speed most of the time. The JIT cached benchmarking is a
complete joke to anyone with a clue, sorry, my 266MHz benches at
180MIPS (most recent test) while taking as much as 50 to 100
times longer than the 060 side alone to do most common tasks.
> Blablabla. Try DTA100.lha in disk/misc off Aminet on a large
> file (for example 1 GB) instead. Use a stopwatch. Use your brain.. Use
> any recent
Done. UAE-JIT is a joke speed-wise (and not just in its horribly
slow (actual) disk performance), all the while benching
hilariously... 1120MB/s sustained from my 12MB/s under Windows
2.5" IDE.
--
-Steve.
SG
> "SG" <s...@erols.com> wrote in <3C4D4E9E.M...@erols.com>:
>
>
> >It also benches 1120MB/s sustained from a 12MB/s IDE disk (maybe
> >1MB/s actual Amiga side performance). I'm going on Amiga
> >experience when the numbers are obviously screwed up, you have
> >none so you can't.
>
> Ignore benchmarks that benchmark hardware, because obviously these are not
> going to be accurate on non-Amiga hardware (i.e., emulation).
>
> >Benoit is also highly parameter and gfx system dependent, these
> >configs aren't even close to matching. I can easily get an 060
> >startup render under .1 secs by playing with the app/OS config.
>
> Looks like an application that does a lot of calculations over time to me.
> Perfect as a benchmark because it tests the emulated 68k processor, which
> in the previously given example rated at around fast '030 to '040 speed.
Even that falls short given Amigas tight hardware integration.
You can't turn off everything but the 68K chip in fully
iterative, nearly 100% instruction cached isolation, and call it
"an Amiga" or even representative of simple 68K performance.
Fact is, UAE-JIT posts SysSpeed scores that are utterly absurd
given its actual observed performance. Anyone with experience
using fast Amigas will quickly agree.
> Speed of WritePixel is irrelevant here, you ain't doing no fancy polygon
> shifting.
>
> What about doing a complex operation in Photogenics? Say a gaussian blur
> with a radius of 10 on a predefined image?
All you need to do is open your first Workbench window to know
you are dealing with 000-020 speed, max (talking UAE-JIT @ "only"
266MHz).
> Any other "real world" benchmarks that we can set up? I know SG will just
> ignore them and keep on wittering about 1GB/s disk throughput benchmarks
> everytime he is losing an argument, but we can but try...
1120MB/s is hardly "losing."
And I've posted myraids of real tests, UAE-JIT is often 50 to
100x slower than my 060 alone doing real world stuff, that it can
actually run without crashing (which is not much, I'm afraid).
--
-Steve.
SG
> "SG" <s...@erols.com> wrote in <3C4DE2E4.M...@erols.com>:
>
> >Hardware vs hardware emulated in software with different levels
> >of test caching, half the machine not emulated properly, running
> >a purely gfx rendering test using completely different gfx driver
> >subsystems and gfx hardware, and using completely different
> >processor instruction sets is not a valid test. Even though in
> >this case the Amiga hardware wins by a massive margin.
>
> It measures getting a task done in an application running under AmigaOS.
> What better benchmark could you get?
How about just opening standard WB window and timing at a full 20
times longer to render? How about launching an app from the RAM
Disk that takes 100 times longer? This is real world
performance.
Its really not that complicated.
> >Do you understand what I just taught you? If no, ask me again,
> >I'll be glad to explain further.
>
> A classic sign of someone who has lost an argument and doesn't know it is
> to attack the person they are arguing with.
You obviously didn't understand it, would you like me to explain
it again.
--
-Steve.
Bjørnar Bolsøy
news:a2nkjq$s9b$1...@paris.btinternet.com:
> Any other "real world" benchmarks that we can set up? I know SG
> will just ignore them and keep on wittering about 1GB/s disk
> throughput benchmarks everytime he is losing an argument, but we
> can but try...
We've posted a whole bunch. Here's one testing how fast
Visage with and without the WarpJpeg datatype displays
this large 2.9MB 6000x894 jpeg:
http://www.eso.org/outreach/press-rel/pr-2001/phot-11-01-hires.jpg
WinUAE Cel640mhz = 7.76 (nojpeg)
WinUAE Cel640mhz = 6.50 (nodatatype)
WinUAE Athlon900mhz = 5.49 (nojpeg)
AMIthlon Athlon1300mhz = 2.23 (nojpeg)
AMIthlon Athlon1300mhz = 1.69 (nodatatype)
Steve never responded.
Regards...
Casper
a> I'd say definetly go with amithlon, that way you don't have such a HUGE
a> processor overhead by having to run windows first and then the emulation.
a> Amithlon is a lot quicker, but less compatible than WinUAE, plus no AGA
a> support (WinUAE has).
a> Al.
I appreciate the reply and advice. I'll order it and give it a whirl!
I don't really care about AGA support anyway. :)
Thomas Tavoly
In article <Xns919EA9EA...@193.213.112.19> "Bjørnar Bolsøy" <bbo...@hotmail.nospam> writes:
> "SG" <s...@erols.com> wrote in news:3C4CDAAD.M...@erols.com:
> > Bjørnar Bolsøy <bbo...@hotmail.nospam> wrote:
>
> >> Fullscreen CybGfx 640*480 - 19fps
> >> Fullscreen P96 640*480 - 8fps
> >> WB window 640*480 - 7fps
> >
> > You guys should really check what's been on the web for years...
> >
> > http://amigapro.com/cvppc.html
> >
> > That screenshot was before overclocking all three CS-PPC clocks,
> > its around 130fps now, my CS-PPC is a mere 7x faster than UAE-JIT
> > at 640MHz.
>
> Seems you are testing 320x240, and I'm testing 640x480.
>
> But my figures were a bit off. All my tests were run in
> 32bit, windowed on the 32bit Win desktop.
>
> These are more correct:
>
> Fullscreen CybGfx 640*480@8 - 26fps
> Fullscreen P96 640*480@8 - 31fps
>
> Fullscreen CybGfx 800*600@8 - 11fps
> Fullscreen P96 800*600@8 - 8fps
>
> WB window 320*240@8bit - 85fps (my max freq)
> WB window 320*240@16bit - 35fps
> WB window 800*600@8bit - 20fps
>
> So my Celly is 3-4x your K6, and right up there with
> your CSPPC before overclocking.
If I may butt in with some Amithlon advocacy again :)
Fullscreen CybGfx 640*480@8 - 154 fps
Fullscreen P96 640*480@8 - 180 fps
Fullscreen CybGfx 800*600@8 - 112 fps
Fullscreen P96 800*600@8 - 136 fps
WB window 320*240@8bit - 529 fps
WB window 320*240@16bit - 280 fps
WB window 800*600@8bit - 135 fps
The WB tests ran on a 1024x768 WB screen with depth set to 8 and 16 bit as
noted behind the window resolutions.
Test system had Athlon 1400@1533, GeForce 2 GTS (200/333, too bad there is
no tool yet to overclock it like for Windows :).
Steve wrote previously:
> That screenshot was before overclocking all three CS-PPC clocks,
> its around 130fps now, my CS-PPC is a mere 7x faster than UAE-JIT
> at 640MHz. I'll have to send WarpUp on over to the laptop,
> standby for more...
I have made a screenshot too: http://www.xs4all.nl/~ttavoly/Amithlon_Voxel.gif
Looks like Steve's PPC is utterly annihilated, in 68k emulation. If we
(very crudely) factor in the difference in speed (2.4x based only on MHz)
between my system and Bjørnar's Celeron 640, WinUAE is also between 2.4-7.1
times slower than Amithlon for this particular test.
Thomas Tavoly
In article <3C4F1EFB.M...@erols.com> "SG" <s...@erols.com> writes:
> Thomas Tavoly <aTm...@amiga.xs4all.nl> wrote:
>
> > In article <3C4D3AC1.M...@erols.com> "SG" <s...@erols.com> writes:
> >
> > > New PPC cards cost around $650 with an 060, PPC, and UWSCSI,
> >
> > Sorry, $925 at softhut.com, I'm sure they would like to know where you get
> > yours from.
>
> Vesalia has 'em for $607, + $30 for insured 4 day shipping
> anywhere in the world--060 and PPC hardware included. Actually
> Amiga compatible too.
Strange, where do you see that? www.vesalia.de has them for $725 (without
shipping) as an Internet action price until 31-Jan-2002. I wonder how much
it will be after that. Still more than a fully decked killer Athlon system.
> > $200 for nearly complete 1 GHz Athlon system, see old message.
>
> A comparable UWSCSI card alone costs that much.
Adaptec AHA-2940UW $55 (www.star-components.com, seen on www.pricewatch.com).
Of course you don't need one, because unlike on Amiga the on board IDE is
damned fast, not to mention doesn't gobble up CPU. I'm getting 11 MB/s
sustained in emulation with my old 9 GB drive. That's not a lot less than
what it gets native. And with a new 10K rpm drive like the Fujitsu MAN
series (55 MB/s sustained read speed) no real Amiga can keep up because its
main memory isn't that fast, PPC equipped or not. SCSI is nice if you have
a system that can't bear the few % CPU overhead of UDMA IDE, but on a 1+
GHz machine, emulated or not, it's irrelevant.
Just for fun, I also ran DTA on a ~390 MB file about halfway across my
30 GB UDMA/100 drive (IBM 75GXP) with Windows on it, mounted through fat95,
and got 27 MB/s. On a smaller file of 'only' 66 MB nearer the beginning
of the disk I got over 28 MB/s. And that with the overhead of fat95 reading
from a FAT32 partition, not a native Amiga partition/filesystem, that would
probably be a bit faster still.
For comparison: this drive does 35 MB/s average sustained read speed over
the *first 8 GB* in Win 98, according to HDTach 2.61. That's pretty damned
near what Amithlon does.
For further comparison I also ran WinUAE (I think two or three revisions
behind the latest) on the same file copied to my DCFFS hardfile, so not
exactly optimal for throughput, but it still managed 17 MB/s. Note that
while Amithlon's timing is correct (14s), DTA reports 9 seconds and 42 MB/s
for this 390 MB file while it did actually take 22 seconds (17 MB/s), so
it's safe to assume you need to use a stopwatch for this too. WinUAE
settings were every frame refresh, 1024x768x16, fastest CPU emulation, full
sound emulation with slider set to 1, 8 MB JIT cache.
> And it can't even run AOS at all (as of 02/02).
Huh? What do you mean, the SCSI card, or the mobo? Amithlon supports
many SCSI controllers (including the mentioned Adaptec) and should work
with the mobo I listed. And what's with the date? Unless you mean a future
PPC only AmigaOS? "Two more weeks".. By which time of course Amithlon just
might support PPC code too, perhaps even on a PCI PPC card, if we are
speculating.
> And I've been testing the
> Amiga JIT emulation extensively (with a much more powerful than
> Amithlon, 256MB JIT engine),
Oh goody. What's a 256 MB JIT engine? Do you mean 256 MB JIT cache? I can
up the JIT cache in Amithlon with a simple line in startup-sequence up to
about 450 MB right now, although that would be silly (or up to over 950 MB
if I put in another 512 MB worth of DDR). Currently I just use 64 MB, the
default being 16 MB, and with the usual stuff loaded have 400 MB or so main
memory left free.
Just for fun, I upped the JIT cache to 350 MB, and now have 100 MB Fast and
7 MB Chip free. No noticable speed increase that I can tell right off the
bat. Perhaps after hours of doing different things it might make a small
difference if I went back to the first things I did?
Also, care to explain what JIT engine this is, where to find more info, and
how it should be much more powerful than Amithlon's? Unless you mean UAE
set to use 256 MB JIT cache.
UAE's JIT engine is less powerful than that of Amithlon's, simply because
Amithlon's is a newer, rewritten one, faster in itself, optimized for the
new task, and without the overhead of chipset emulation (additionally not
held up by chipset timing restrictions), or some bloated OS abstraction
underneath.
> they are entire unusable for
> instability alone, in addition to downright slow, quite
> unfortunately.
I think that is a matter of selecting the right hardware and setting that
and the emulation up correctly. Although WinUAE is known to be buggy it
seems reasonably stable here, and Amithlon is very stable. I don't have
more crashes than on my 4000 (which is, when still used, also very stable
BTW).
> I would have enjoyed converting my laptop to AOS, but it simply
> doesn't run it for more than a few minutes at a time (at 000 to
> 020 speed most of the time.
WinUAE can be very unstable/slow depending on system or setup, but doesn't
need to be. Amithlon of course should be better in any case.
> The JIT cached benchmarking is a complete joke to anyone with a clue,
JIT cached is simply a way of how code is executed, it doesn't impact
benchmarking if done properly, i.e. do real world tasks, use a stopwatch
and not some silly buggy benchmark program that was never intended for
more than 040 Amigas.
Oh, and even if the whole benchmark fits in JIT cache, if the real world
application that works along the same principles is equally sped up, then
what's the problem? That's the whole idea behind having a cache.
You should only have an issue with caching if it creates a false
image of speed that does not reflect the real world performance after
that (like your 1100 MB/s disk read speed), but this is a matter of
improper testing. Good benchmarks don't suffer from this, and that's why
Sysspeed, SysInfo etc. are partially or completely useless on anything
above a real 040 Amiga.
> sorry, my 266MHz benches at
> 180MIPS (most recent test) while taking as much as 50 to 100
> times longer than the 060 side alone to do most common tasks.
MIPS stands for Meaningless Information on Processor Speed. Also Millions
of Instructions Per Second. It doesn't say anything about how fast
something is executed or not. A CPU could do six bazillion MIPS and thrash
around in useless code 99.999% of the time, or the program you used to get
this number does not reflect real world application speed, or is simply
buggy/never meant for the system you are forcing it on.
> > Blablabla. Try DTA100.lha in disk/misc off Aminet on a large
> > file (for example 1 GB) instead. Use a stopwatch. Use your brain.. Use
> > any recent
>
> Done. UAE-JIT is a joke speed-wise (and not just in its horribly
> slow (actual) disk performance), all the while benching
> hilariously... 1120MB/s sustained from my 12MB/s under Windows
> 2.5" IDE.
Instead of hammering on whatever you think this signifies, why not post the
DTA number (with stopwatch in hand) from a properly set up UAE? As I said,
for comparison my old 5400 rpm 9 GB IDE drive gets 11 MB/s sustained and
about 28 MB/s with a newer, but not top of the line 7200 rpm drive under
Amithlon and 17 MB/s from the same newer drive under WinUAE.
Thomas Tavoly
AMIthlon Athlon1400@1533mhz = 1.74 (nojpeg) RAM: 1.72
AMIthlon Athlon1400@1533mhz = 1.57 (nodatatype) RAM: 1.34
Note how the nojpeg number is lower than expected when corrected for MHz on
my machine (probably due to FSB/DDR) but almost the same when testing from
RAM: and the nodt number is higher than the expected correction except when
done from RAM:.
This image is already approaching sensible benchmark limits, close to the
one second boundary the disk load time becomes a limiting factor, you
should take a much larger picture to test :)
Another thing to take into account is possible timing incorrectness on
the part of WinUAE, you should always verify with stopwatch in hand.
> Steve never responded.
Not much of a surprise.
Gunnar Wretlind
>
> In article <3C4F1EFB.M...@erols.com> "SG" <s...@erols.com> writes:
> > Thomas Tavoly <aTm...@amiga.xs4all.nl> wrote:
> >
> > > In article <3C4D3AC1.M...@erols.com> "SG" <s...@erols.com> writes:
> > >
> > > > New PPC cards cost around $650 with an 060, PPC, and UWSCSI,
> > >
> > > Sorry, $925 at softhut.com, I'm sure they would like to know where you get
> > > yours from.
> >
> > Vesalia has 'em for $607, + $30 for insured 4 day shipping
> > anywhere in the world--060 and PPC hardware included. Actually
> > Amiga compatible too.
>
> Strange, where do you see that? www.vesalia.de has them for $725 (without
> shipping) as an Internet action price until 31-Jan-2002. I wonder how much
> it will be after that. Still more than a fully decked killer Athlon system.
The prices on Vesalia's site include German sales taxes and since Non EU
residents don't have to pay German sales taxes Steve has removed the tax
from Vesalia's prices (-16%).
Anyway I think it is still to much for such slow CPU's.
//Gunnar
Thomas Tavoly
> The prices on Vesalia's site include German sales taxes and since Non EU
> residents don't have to pay German sales taxes Steve has removed the tax
> from Vesalia's prices (-16%).
Hah, good one. But you still have to add local/import tax then, which in
some cases could be higher. Typical.
> Anyway I think it is still to much for such slow CPU's.
Definitely. I paid the equivalent of $115 for a replacement PSU and $320
for a replacement 4000D mobo not too long ago. Amiga gear is expensive..
Marcel DeVoe
> "a" == "al" writes:
> a> I'd say definetly go with amithlon, that way you don't have such a HUGE
> a> processor overhead by having to run windows first and then the emulation.
> a> Amithlon is a lot quicker, but less compatible than WinUAE, plus no AGA
> a> support (WinUAE has).
> a> Al.
> I appreciate the reply and advice. I'll order it and give it a whirl!
> I don't really care about AGA support anyway. :)
It's not the AGA that WinUAE supports but the custom chips support that
it's famous for. Same for AmigaXL. So if it's AmigaXL that you are mainly
interested in for classic support, then you might as well use WinUAE.
--
Marcel J. DeVoe - mde...@shore.net - Team *AMIGA*
A4091scsi CV64 96 megs CDRW M1764-17" Catweasel FUSION/Emplant
A4000/060 CyberStorm MKII overclocked 66mhz - see "How to Overclock!"
and "DIY A4000 Tower for $45" @ http://www.shore.net/~mdevoe
Tom and Lisa Peters
SG wrote:
>
> Thomas Tavoly <aTm...@amiga.xs4all.nl> wrote:
>
> > In article <3C4D3AC1.M...@erols.com> "SG" <s...@erols.com> writes:
> >
> > > Stefan Kleinheinrich <SKleinh...@t-online.de> wrote:
> > >
> > > > Oh Steve
> > > > my good old d K6/3 with its disfunctional FPU running WinUAE0.6??? without
> > > > JIT had about the same speed as my
> > > > overclocked A2630 at 32Mhz. That on everyday compiler runs and believe me i
> > > > would have noticed even the smallest
> > > > difference. Hey but who cares as you can buy a brand new AthlonXP-maschine
> > > > for the price of a used lowend PPC-
> > >
> > > Thats might be a more interesting point if (1) the emulation was
> > > seriously usable--its not, and (2) it was true.
> >
> > Truth hurts. Amithlon for example is seriously usable, I haven't switched
> > on my 4000/060/CV64/3D in something like two months now because it's on
> > average 10-15 times slower compared to my Athlon.
> >
> > > New PPC cards cost around $650 with an 060, PPC, and UWSCSI,
> >
> > Sorry, $925 at softhut.com, I'm sure they would like to know where you get
> > yours from.
>
> Vesalia has 'em for $607, + $30 for insured 4 day shipping
> anywhere in the world--060 and PPC hardware included. Actually
> Amiga compatible too.
Both Stephan and Thomas pay VAT as they are in Europe.
= $718 (USD)
Even if you were in the US, $630 for a 233 MHz 604e (no L2 cache), a 060,
and a UW SCSI controller ?
THe equivalet equipment can be bought for a mac for ~$65 (no need for the
hot and slow 060).
> > $200 for nearly complete 1 GHz Athlon system, see old message.
>
> A comparable UWSCSI card alone costs that much. And it can't
AHAHAHAHAHAHAHAHAHAHAHAHAHAHA
$200 for an adaptec 2940UW ? (<-- the closest comparable card)
AHAHAHAHAHAHAHAHAHAHAHAHAHAHA
I can pick those up for $50. Hell, I bought a 2940U2B (LVD)
for $90 about a year ago.
> > Blablabla. Try DTA100.lha in disk/misc off Aminet on a large
> > file (for example 1 GB) instead. Use a stopwatch. Use your brain.. Use
> > any recent
>
> Done. UAE-JIT is a joke speed-wise (and not just in its horribly
> slow (actual) disk performance), all the while benching
> hilariously... 1120MB/s sustained from my 12MB/s under Windows
> 2.5" IDE.
Note that Steve once again failed to do the test required of him.
> -Steve.
Mad Dog
PLZI
"Tom and Lisa Peters" <gionp...@home.NOSPAM.com> wrote in message
news:3C51ACBF...@home.NOSPAM.com...
>
>
> SG wrote:
> > A comparable UWSCSI card alone costs that much. And it can't
>
> AHAHAHAHAHAHAHAHAHAHAHAHAHAHA
>
> $200 for an adaptec 2940UW ? (<-- the closest comparable card)
>
And who would want sloooow ultrawide, when you can get Ultra160
scsi-controllers under 200$? Now let me think who. Oh, ok, now I know one
person.
- PLZI
SG
>
> In article <MPG.16ba67ec...@news1.telia.com> Gunnar Wretlind <gunnar.wret...@telia.com> writes:
>
> > The prices on Vesalia's site include German sales taxes and since Non EU
> > residents don't have to pay German sales taxes Steve has removed the tax
> > from Vesalia's prices (-16%).
>
> Hah, good one. But you still have to add local/import tax then, which in
> some cases could be higher. Typical.
That was total.
> > Anyway I think it is still to much for such slow CPU's.
>
> Definitely. I paid the equivalent of $115 for a replacement PSU
$10 tops for a $150W AT, if you have a clue. $30 or so for 300W.
You should've asked me first.
> and $320
> for a replacement 4000D mobo not too long ago. Amiga gear is
> expensive..
Right, $320 every $12 years is a lot more than the $20,000 or so
you'd have thrown away on 6 molasses-speed Windows pcs.
--
-Steve.
Bjørnar Bolsøy
> Thomas Tavoly <aTm...@amiga.xs4all.nl> wrote:
>> Definitely. I paid the equivalent of $115 for a replacement PSU
>
> $10 tops for a $150W AT, if you have a clue. $30 or so for 300W.
>
> You should've asked me first.
>
>> and $320
>> for a replacement 4000D mobo not too long ago. Amiga gear is
>> expensive..
>
> Right, $320 every $12 years is a lot more than the $20,000 or so
> you'd have thrown away on 6 molasses-speed Windows pcs.
http://www.connect-computers.com/root/store/sf_proddetail.asp?
sf_p_ID=SW62RUH1CE68I9BSEPHC85CME8
kit Duron with 700Mhz Processor
Case with Duron CPU with fan& heat sink, 810LMR
MotherBoard with Video,56k Fax modem, Sound & Lan Card
Sale Price: $119.00
Regards...
SG
> In article <aTmos...@amiga.xs4all.nl>, aTm...@amiga.xs4all.nl says...
> > Strange, where do you see that? www.vesalia.de has them for $725 (without
> > shipping) as an Internet action price until 31-Jan-2002. I wonder how much
> > it will be after that. Still more than a fully decked killer Athlon system.
You mean Windows system--nobody cares.
> The prices on Vesalia's site include German sales taxes and since Non EU
> residents don't have to pay German sales taxes Steve has removed the tax
> from Vesalia's prices (-16%).
>
> Anyway I think it is still to much for such slow CPU's.
That runs an OS environment thats 100s of times faster.
--
-Steve.
Stefan Kleinheinrich
"SG" <s...@erols.com> schrieb im Newsbeitrag
news:3C524BAC.M...@erols.com...
> Gunnar Wretlind <gunnar.wret...@telia.com> wrote:
>
> > In article <aTmos...@amiga.xs4all.nl>, aTm...@amiga.xs4all.nl says...
> > > Strange, where do you see that? www.vesalia.de has them for $725
(without
> > > shipping) as an Internet action price until 31-Jan-2002. I wonder how
much
> > > it will be after that. Still more than a fully decked killer Athlon
system.
>
> You mean Windows system--nobody cares.
Hello STEVE !!! Is there somebody inhere ???
This whole discussion is about WinUAE (and sometimes Amithlon). WinUAE runs
on
x86-based systems though they are on topic. As you started this thread you
should know.
Your assumption: Athlonsytem = Win-System. Bullshit !!!
I know a lot of people who use primarily for Linux or Amithlon, some of them
don't even
have Windows installed.
I for myself see my PC as a Amithlon/Linux-system that can run
DirectX-Games.
(They ONLY thing Windows is really usable for)
>
> > The prices on Vesalia's site include German sales taxes and since Non EU
> > residents don't have to pay German sales taxes Steve has removed the tax
> > from Vesalia's prices (-16%).
> >
> > Anyway I think it is still to much for such slow CPU's.
>
> That runs an OS environment thats 100s of times faster.
????
Do you mean AmigaOS is faster on the 060 (there is no 4.0 yet) then AmigaOS
on a
Athlon-maschine ? NO WAY
Or do you compare it to Windows itself ? Everybody knows WinSlows sucks on
every
given HW ,
If the 604 would still be in production a realistic price would be <$10
(just the CPU).
If a resonable.priced G4-sytem+AOS4 hits the market some of us happy
Amithlon-user may
reconsider our options. This won't happen anytime before this summer. (and
has been delayed
ever and ever again for the last 4 or 5 years)
> --
>
> -Steve.
Stefan
Lasse Bodilsen
<bbo...@hotmail.nospam> wrote:
>> Right, $320 every $12 years is a lot more than the $20,000 or so
>> you'd have thrown away on 6 molasses-speed Windows pcs.
>
> http://www.connect-computers.com/root/store/sf_proddetail.asp?
>sf_p_ID=SW62RUH1CE68I9BSEPHC85CME8
>
> kit Duron with 700Mhz Processor
> Case with Duron CPU with fan& heat sink, 810LMR
> MotherBoard with Video,56k Fax modem, Sound & Lan Card
>
> Sale Price: $119.00
Neat price tag. But do stay away from the PCChips 810MLR mainboard.
its total crap.
We ordered 100 mainboard from ECS (makers of PCchips mainboard, and
had to return the whole lot, because it was 50/50 chance they would
work :-)
Lasse Bodilsen <-> Sca 'at' BodyCount.dk
AMItlhon@1300/MatroxG100/256mb/13gig/Os3.9
SG
>
> "Tom and Lisa Peters" <gionp...@home.NOSPAM.com> wrote in message
> news:3C51ACBF...@home.NOSPAM.com...
> >
> >
> > SG wrote:
> > > A comparable UWSCSI card alone costs that much. And it can't
> >
> > AHAHAHAHAHAHAHAHAHAHAHAHAHAHA
> >
> > $200 for an adaptec 2940UW ? (<-- the closest comparable card)
> >
>
> And who would want sloooow ultrawide, when you can get Ultra160
> scsi-controllers under 200$?
No difference. Again you are confusing interface speed with
drive speed--thats what they want you to do. Standard Windows
novice.
--
-Steve.
Gunnar Wretlind
> Gunnar Wretlind <gunnar.wret...@telia.com> wrote:
>
> > In article <aTmos...@amiga.xs4all.nl>, aTm...@amiga.xs4all.nl says...
> > > Strange, where do you see that? www.vesalia.de has them for $725 (without
> > > shipping) as an Internet action price until 31-Jan-2002. I wonder how much
> > > it will be after that. Still more than a fully decked killer Athlon system.
>
> You mean Windows system--nobody cares.
I prefer Windows, because of the applications, but that does not mean
that one has to run Windows on x86 computers. Ever heard of Linux or
Amithlon?
> > The prices on Vesalia's site include German sales taxes and since Non EU
> > residents don't have to pay German sales taxes Steve has removed the tax
> > from Vesalia's prices (-16%).
> >
> > Anyway I think it is still to much for such slow CPU's.
>
> That runs an OS environment thats 100s of times faster.
Too bad most people use their computers to run applications.
//Gunnar
Stefan Kleinheinrich
> No difference. Again you are confusing interface speed with
> drive speed--thats what they want you to do. Standard Windows
> novice.
> --
>
> -Steve.
Steve just a few basic facts:
ULTRA-DMA100 on todays mobos is faster than your 5 year old UW-SCSI.
Thats about interface-speed.
For the HD itself:
My 60gb IBM-HD had cost me less than $160 including taxes.
A UW-SCSI IBM-HD with the same speed costs nearly $260, the only problem
is its only 36gb as IBM don't produce bigger UW-HDs.
The next option would be a 73gb U160 drive which is actually ~20% faster
then
the 60gb IDE, only problem >$1000.
Conclusion:
Today nobody sensible uses SCSI-HDs in normal workstation-type computer.
Todays IDE-drives are even faster than your memory-bus.
SCSI only makes sence in a server.
PLZI
"SG" <s...@erols.com> wrote in message news:3C527FBF.M...@erols.com...
> PLZI <janne....@eisikanautaa.sonera.com> wrote:
>
> > And who would want sloooow ultrawide, when you can get Ultra160
> > scsi-controllers under 200$?
>
> No difference.
Really? Wonder why they make those things, then.
> Again you are confusing interface speed with
> drive speed--thats what they want you to do. Standard Windows
> novice.
So, you are using only one drive on your controller, so you've never managed
to saturate the scsi bus? Thought as much. You spend all your money on
overpriced hardware, no wonder you can not afford to buy more drives. Poor
you.
- PLZI
PLZI
> drive speed--thats what they want you to do. Standard Windows
> novice.
And while we're at it, would you please explain this:
(IBM Ultrastar 15k scsi hard drive specs)
http://www.storage.ibm.com/hdd/ultra/36z15data.htm
Media transfer rate 453-647 Mbits/sec
Interface transfer rate 160 MB/sec
453 Mbits/sec is around 56Mbytes / sec. And this is the LOWER limit of the
media speed. Want to demonstrate me how this thing ALONE would not saturate
your UW-scsi bus?
- PLZI
Keith Blakemore-Noble
something about "Re: Hey, UAE-JIT is pretty neat."...
ROTFL!!!!
Nice to see Stevieboy is still talking out of his @r$e as usual.
Hey, Stevieboy, if the interface is unimportant, why are you so
insistant that it has to be UW SCSI as opposed to normal SCSI, eh?
Ho hum.
SG
> > Vesalia has 'em for $607, + $30 for insured 4 day shipping
> > anywhere in the world--060 and PPC hardware included. Actually
> > Amiga compatible too.
>
> Both Stephan and Thomas pay VAT as they are in Europe.
> = $718 (USD)
>
> Even if you were in the US, $630 for a 233 MHz 604e (no L2 cache), a 060,
> and a UW SCSI controller ?
That runs the fastest popular OS (100s of times faster than
Windows) the world has ever know. You bet.
> THe equivalet equipment can be bought for a mac
Too bad MacOS is the slowest popular OS the world has ever known.
--
-Steve.
SG
> Hi there SG, on or around Sat, 26 Jan 2002 16:12:16 GMT you wrote
> something about "Re: Hey, UAE-JIT is pretty neat."...
> > PLZI <janne....@eisikanautaa.sonera.com> wrote:
> >
> > >
> > > "Tom and Lisa Peters" <gionp...@home.NOSPAM.com> wrote in message
> > > news:3C51ACBF...@home.NOSPAM.com...
> > > >
> > > >
> > > > SG wrote:
> > > > > A comparable UWSCSI card alone costs that much. And it can't
> > > >
> > > > AHAHAHAHAHAHAHAHAHAHAHAHAHAHA
> > > >
> > > > $200 for an adaptec 2940UW ? (<-- the closest comparable card)
> > > >
> > >
> > > And who would want sloooow ultrawide, when you can get Ultra160
> > > scsi-controllers under 200$?
> >
> > No difference. Again you are confusing interface speed with drive
> > speed--thats what they want you to do. Standard Windows novice.
>
> ROTFL!!!!
YOu actually think using a higher bandwidth interface speeds up
you drives. No wonder you believed UAE was actually getting
1.1GB/s sustained from a 2.5" IDE drive.
--
-Steve.
Stefan Kleinheinrich
HaHaHaHaHa
Been waiting for two days to see the 1,1GB/s bullshit.
Steve
Once and for all NOBODY (exept you of course) cares about this stupid
benchmark.
I don't know how much you can get out of your outdated laptop-hd in a even
more
outdated laptop and i don't care. (or do you own one of the rare
A4000-laptops ?)
If you compare the drivespeed of your A4000 to Amithlon running on a
ASUS-AV7-
266mb + IBM IC35L060 (or similar) on a reallife test we can start talking
seriously.
Seems hell has to freeze over before we can see something like that.
Stefan
Stefan Kleinheinrich
> Tom and Lisa Peters <gionp...@home.NOSPAM.com> wrote:
>
> > > Vesalia has 'em for $607, + $30 for insured 4 day shipping
> > > anywhere in the world--060 and PPC hardware included. Actually
> > > Amiga compatible too.
> >
> > Both Stephan and Thomas pay VAT as they are in Europe.
> > = $718 (USD)
> >
> > Even if you were in the US, $630 for a 233 MHz 604e (no L2 cache), a
060,
> > and a UW SCSI controller ?
>
> That runs the fastest popular OS (100s of times faster than
> Windows) the world has ever know. You bet.
So does any Athlon or P3/4 cpu (i'm talking Amithlon again)
>
> > THe equivalet equipment can be bought for a mac
>
> Too bad MacOS is the slowest popular OS the world has ever known.
So what ?
Tom and Lisa just stated that the price-tag on pup-system is high enough
to have been in Woodstock.
> --
>
> -Steve.
Keith Blakemore-Noble
something about "Re: Hey, UAE-JIT is pretty neat."...
> Keith Blakemore-Noble <Ke...@Blakemore-Noble.net> wrote:
>
> > Hi there SG, on or around Sat, 26 Jan 2002 16:12:16 GMT you wrote
> > something about "Re: Hey, UAE-JIT is pretty neat."...
> > > PLZI <janne....@eisikanautaa.sonera.com> wrote:
> > >
> > > >
> > > > "Tom and Lisa Peters" <gionp...@home.NOSPAM.com> wrote in
> > > > message news:3C51ACBF...@home.NOSPAM.com...
> > > > >
> > > > >
> > > > > SG wrote:
> > > > > > A comparable UWSCSI card alone costs that much. And it
> > > > > > can't
> > > > >
> > > > > AHAHAHAHAHAHAHAHAHAHAHAHAHAHA
> > > > >
> > > > > $200 for an adaptec 2940UW ? (<-- the closest comparable
> > > > > card)
> > > > >
> > > >
> > > > And who would want sloooow ultrawide, when you can get Ultra160
> > > > scsi-controllers under 200$?
> > >
> > > No difference. Again you are confusing interface speed with
> > > drive speed--thats what they want you to do. Standard Windows
> > > novice.
> >
> > ROTFL!!!!
>
> YOu actually think using a higher bandwidth interface speeds up you
> drives.
Do you actually read what people write, Stevieboy, or do you simply
make things up to please those voices in your head?
'cos you sure as hell ain't answering what I wrote (as usual).
> No wonder you believed UAE was actually getting
> 1.1GB/s sustained from a 2.5" IDE drive.
LOL!!!
Quote where I said that Stevieboy.
Quote NOW or shut the fsck up you dimwitted troll.
Stefan Kleinheinrich
"Keith Blakemore-Noble" <Ke...@Blakemore-Noble.net> schrieb im Newsbeitrag
news:65.15.nH2lJf...@Blakemore-Noble.net...
schnippeldieschnapp
> Quote where I said that Stevieboy.
>
> Quote NOW or shut the fsck up you dimwitted troll
Oh please not. Steve is so much fun.
Stefan
Tony Mulvihill
S> > The prices on Vesalia's site include German sales taxes and
S> > since Non EU residents don't have to pay German sales taxes
S> > Steve has removed the tax from Vesalia's prices (-16%).
S> >
S> > Anyway I think it is still to much for such slow CPU's.
S>
S> That runs an OS environment thats 100s of times faster.
heh :-) that's true.
I demo'ed my towered 50 mhz 060 A1200 to Windows users (at Comdex)
and then asked them (based on what they'd seen) what speed processor
they thought was powering my A1200. Almost all answers were between
700~900 mhz.
I never realized Windows was such a hog till that day.
Cheers :-)
--
Tony Mulvihill
Happy Amiga User. A1200 060 scsi ide ram rom blah blah
Member & Secretary of the AUG <http://www.aug.org.au>
"Workbench" Editor. The magazine of the AUG
ICQ # 20578798
Bjørnar Bolsøy
news:e1.V1.oH2R7jk...@net2000.com.au:
> "S" == "SG" writes:
> S> That runs an OS environment thats 100s of times faster.
>
>
> heh :-) that's true.
>
> I demo'ed my towered 50 mhz 060 A1200 to Windows users (at
> Comdex) and then asked them (based on what they'd seen) what
> speed processor they thought was powering my A1200. Almost all
> answers were between 700~900 mhz.
>
> I never realized Windows was such a hog till that day.
Well, what did you show them that made them conclude
the way they did?
Regards...
Brett O'Callaghan
>"Tony Mulvihill" <to...@net2000.com.au> wrote in
>> I demo'ed my towered 50 mhz 060 A1200 to Windows users (at
>> Comdex) and then asked them (based on what they'd seen) what
>> speed processor they thought was powering my A1200. Almost all
>> answers were between 700~900 mhz.
> Well, what did you show them that made them conclude
> the way they did?
Not running Unreal Tournament, I'd hazard to guess.
Byeeeee.
--
http://www.geocities.com/brettocallaghan
Home of QDNStats V2 - Newsgroup Stats for Agent
SG
You are the one who said you'd get 160MB/s sustained form SCSI
drives using a 160MB interface, not me.
--
-Steve.
SG
> In article <cZB48.230$e_1....@read2.inet.fi>,
> "PLZI" <janne....@eisikanautaa.sonera.com> wrote:
>
> > And while we're at it, would you please explain this:
> >
> > (IBM Ultrastar 15k scsi hard drive specs)
> > http://www.storage.ibm.com/hdd/ultra/36z15data.htm
> >
> > Media transfer rate 453-647 Mbits/sec
> > Interface transfer rate 160 MB/sec
> >
> > 453 Mbits/sec is around 56Mbytes / sec. And this is the LOWER limit of the
> > media speed. Want to demonstrate me how this thing ALONE would not saturate
> > your UW-scsi bus?
>
> Be warned that the conversion factor between media transfer rate in bits
> per second and real world sustained throughput in bytes per second is
> usually around 11 or 12 instead of 8 like you'd expect. The figure
> they're listing is the raw data rate right off the head -- but there's a
> bunch of overhead, ranging from ECC encoding to sector gaps and other
> things.
>
> Moot point though, since with a factor of 12, the real data rate you'll
> get is 36 to 51 MBytes/s, which would still saturate UW SCSI...
Not through a file system. AOS is extrememly robust and advanced
in this way. Sure, it decreases throughput a little, but who
cares when you only use your hard drive for storage, not as the
primarly memory bus. Can you really tell when the an AOS 1MB
power app exe launches in 1/20th vs 1/30th of a second?
--
-Steve.
SG
> "Bjørnar Bolsøy" <bbo...@hotmail.nospam> wrote:
>
> >"Tony Mulvihill" <to...@net2000.com.au> wrote in
>
> >> I demo'ed my towered 50 mhz 060 A1200 to Windows users (at
> >> Comdex) and then asked them (based on what they'd seen) what
> >> speed processor they thought was powering my A1200. Almost all
> >> answers were between 700~900 mhz.
Its a real shame the A1200 can't use 64-bit RAM modes. Ever
re-jumper a jumperable big box 060 accel from 64 to 32-bit mode?
It makes a huge difference.
> > Well, what did you show them that made them conclude
> > the way they did?
>
> Not running Unreal Tournament, I'd hazard to guess.
>
> Byeeeee.
> --
> http://www.geocities.com/brettocallaghan
> Home of QDNStats V2 - Newsgroup Stats for Agent
--
-Steve.
SG
> Hi there SG, on or around Sat, 26 Jan 2002 19:30:20 GMT you wrote
> something about "Re: Hey, UAE-JIT is pretty neat."...
> > > > > And who would want sloooow ultrawide, when you can get Ultra160
> > > > > scsi-controllers under 200$?
> > > >
> > > > No difference. Again you are confusing interface speed with
> > > > drive speed--thats what they want you to do. Standard Windows
> > > > novice.
> > >
> > > ROTFL!!!!
> >
> > YOu actually think using a higher bandwidth interface speeds up you
> > drives.
>
> Do you actually read what people write, Stevieboy, or do you simply
> make things up to please those voices in your head?
Read it, its still quoted up there. He clearly thinks the
controller's bandwidth is drive transfer speed.
--
-Steve.
SG
> "S" == "SG" writes:
>
> S> > The prices on Vesalia's site include German sales taxes and
> S> > since Non EU residents don't have to pay German sales taxes
> S> > Steve has removed the tax from Vesalia's prices (-16%).
> S> >
> S> > Anyway I think it is still to much for such slow CPU's.
> S>
> S> That runs an OS environment thats 100s of times faster.
>
>
> heh :-) that's true.
>
> I demo'ed my towered 50 mhz 060 A1200 to Windows users (at Comdex)
> and then asked them (based on what they'd seen) what speed processor
> they thought was powering my A1200. Almost all answers were between
> 700~900 mhz.
>
> I never realized Windows was such a hog till that day.
Really? I think I first noticed it when my Z248 286/8 wouldn't
move at all under Windows, but was fine using DOS. Meanwhile, my
Amiga1000 was 32-bit pure lightning.
But anyway, I wonder why they guessed 700-900MHz? In normal day
to day operations, there is really no way to tell whats under the
hood of a Windows machine--I challenge anyone to guess
accurately.
My 1GHz/256MB Win2K GW2K at work is substantially slower (as are
all few-hundred of them) than my Win98 266MHz/160MB laptop at
home (sometimes I bring it in to go "faster"). CPU power is
entirely MIA when using Windows normally.
> Cheers :-)
> --
> Tony Mulvihill
>
> Happy Amiga User. A1200 060 scsi ide ram rom blah blah
> Member & Secretary of the AUG <http://www.aug.org.au>
> "Workbench" Editor. The magazine of the AUG
> ICQ # 20578798
>
>
--
-Steve.
SG
>
> "SG" <s...@erols.com> wrote in message news:3C527FBF.M...@erols.com...
> > PLZI <janne....@eisikanautaa.sonera.com> wrote:
> >
> > > And who would want sloooow ultrawide, when you can get Ultra160
> > > scsi-controllers under 200$?
> >
> > No difference.
>
> Really? Wonder why they make those things, then.
Proably for those using a properly multitasking OS with a well
programmed fully rentrant scsi driver (i.e. not Windows) with 15
drives attached, all running at the exact same speed as they'd
run without 160MB/s of interface bandwidth.
> > Again you are confusing interface speed with
> > drive speed--thats what they want you to do. Standard Windows
> > novice.
--
-Steve.
SG
>
> "SG" <s...@erols.com> schrieb im Newsbeitrag
> news:3C527FBF.M...@erols.com...
> > No difference. Again you are confusing interface speed with
> > drive speed--thats what they want you to do. Standard Windows
> > novice.
> > --
> >
> > -Steve.
>
> Steve just a few basic facts:
> ULTRA-DMA100 on todays mobos is faster than your 5 year old UW-SCSI.
> Thats about interface-speed.
We've been over this a zillion times, IDE drives horrid seek
speeds relegate them to 1-2MB/s tops during actually use (there
are a few cheap "scsi" interfaced spinoffs that use the same
slow drive mechanism, btw. Plus the interface can't do more
than one thing at a time (very bad for use as your primary memory
bus). Not that it matters under Windows poorly programmed
non-reentrant scsi implementation garbage anyway. You should
stick with cheap IDE, it makes total sense given you lack of real
alternatives.
--
-Steve.
SG
> In article <3C524BAC.M...@erols.com>, s...@erols.com says...
> > Gunnar Wretlind <gunnar.wret...@telia.com> wrote:
> >
> > > In article <aTmos...@amiga.xs4all.nl>, aTm...@amiga.xs4all.nl says...
> > > > Strange, where do you see that? www.vesalia.de has them for $725 (without
> > > > shipping) as an Internet action price until 31-Jan-2002. I wonder how much
> > > > it will be after that. Still more than a fully decked killer Athlon system.
> >
> > You mean Windows system--nobody cares.
>
> I prefer Windows,
Then leave.
--
-Steve.