C-Ray FP/CPU Benchmark Test Results

SGI hardware problems, solutions, tips, hacks, etc.
Forum rules
Any posts concerning pirated software or offering to buy/sell/trade commercial software are subject to removal.
User avatar
theinonen
Posts: 380
Joined: Wed Feb 21, 2007 11:32 am
Location: Finland

Re: C-Ray FP/CPU Benchmark Test Results

Unread postby theinonen » Sat May 10, 2008 12:35 pm

Here is result for IBM 9112-165, 450 MHz POWER3-II

Operating system: AIX 5.1, compiler used: gcc-4.0


(Test 1): 'scene' at 800x600 --> Rendering took: 4 seconds (4106 milliseconds).
Test 2: 'sphract' at 800x600 --> Rendering took: 129 seconds (129587 milliseconds).
Test 3: 'sphract' at 1024x768 with 8X oversampling --> Rendering took: 1827 seconds (1827911 milliseconds).
Test 4: 'scene' at 7500x3500 --> Rendering took: 256 seconds (256194 milliseconds).

User avatar
mapesdhs
Posts: 2519
Joined: Mon Nov 10, 2003 4:17 pm
Location: Edinburgh, Scotland
Contact:

Re: C-Ray FP/CPU Benchmark Test Results

Unread postby mapesdhs » Fri May 23, 2008 1:49 am

theinonen writes:
> Here is result for IBM 9112-165, 450 MHz POWER3-II

Thanks for that! Sorry I didn't spot your post earlier.

Note that the C-Ray page has been redesigned slightly (header/footer) and is now also a part of the
larger SGI General Performance Comparisons page, just one of many tests available to view, each
with its own separate page for easy access.

Btw, anyone got a quad-core Opteron or a seriously overclocked Core2Extreme they can test? Just curious.

Still hoping Bishop2K can run the tests on his mega-Origin... 8)

Ian.

User avatar
nekonoko
Site Admin
Site Admin
Posts: 8145
Joined: Thu Jan 23, 2003 1:31 am
Location: Pleasanton, California
Contact:

Re: C-Ray FP/CPU Benchmark Test Results

Unread postby nekonoko » Fri May 23, 2008 2:05 am

mapesdhs wrote:Btw, anyone got a quad-core Opteron or a seriously overclocked Core2Extreme they can test? Just curious.


I run my Q6600 (quad core) at 3.2GHz if that helps.
Twitter: @neko_no_ko
IRIX Release 4.0.5 IP12 Version 06151813 System V
Copyright 1987-1992 Silicon Graphics, Inc.
All Rights Reserved.

User avatar
mapesdhs
Posts: 2519
Joined: Mon Nov 10, 2003 4:17 pm
Location: Edinburgh, Scotland
Contact:

Re: C-Ray FP/CPU Benchmark Test Results

Unread postby mapesdhs » Fri May 23, 2008 1:25 pm

nekonoko writes:
> I run my Q6600 (quad core) at 3.2GHz if that helps.

Yes, definitely!

Meanwhile, someone give Bishop2K a nudge to run C-Ray on his huge Origin. :D

Ian.

User avatar
Dr. Dave
Posts: 2311
Joined: Fri Feb 13, 2004 10:37 pm
Location: Ottawa, Canada >burp<

Re: C-Ray FP/CPU Benchmark Test Results

Unread postby Dr. Dave » Wed Jun 18, 2008 6:34 pm

New one for you... same Octane system as previous, except a dual-360 overclocked to 438.75, cache running at 292.5 Mhz - and it seems to be running totally clean as well, got lucky on this one!

Code: Select all

cat scene | ./c-ray-mt -t 32 > foo.ppm
c-ray-mt v1.1
Rendering took: 1 seconds (1166 milliseconds)

cat sphfract | ./c-ray-mt -t 32 > foo.ppm
c-ray-mt v1.1
Rendering took: 32 seconds (32095 milliseconds)

cat sphfract | ./c-ray-mt -t 32 -s 1024x768 -r 8 > foo.ppm
c-ray-mt v1.1
Rendering took: 421 seconds (421375 milliseconds)

cat scene | ./c-ray-mt -t 32 -s 7500x3500 > foo.ppm
c-ray-mt v1.1
Rendering took: 71 seconds (71459 milliseconds)

The PM20 has had it's multiplier jumped to 4.5 * 97.5 MHz, up from 4 * 90 MHz, the 97.5 oscillator module taken from an R10k-195 module.
:O3000: <> :O3000: :O2000: :Tezro: :Fuel: x2+ :Octane2: :Octane: x3 :1600SW: x2 :O2: x2+ :Indigo2IMP: :Indigo2: x2 :Indigo: x3 :Indy: x2+

Once you step up to the big iron, you learn all about physics, electrical standards, and first aid - usually all in the same day

User avatar
mapesdhs
Posts: 2519
Joined: Mon Nov 10, 2003 4:17 pm
Location: Edinburgh, Scotland
Contact:

Re: C-Ray FP/CPU Benchmark Test Results

Unread postby mapesdhs » Thu Jun 19, 2008 1:52 am

Dr. Dave writes:
> New one for you... same Octane system as previous, except a dual-360 overclocked to 438.75, cache running at 292.5 Mhz

Thanks for that!! Page updated.

Ian.

User avatar
Dr. Dave
Posts: 2311
Joined: Fri Feb 13, 2004 10:37 pm
Location: Ottawa, Canada >burp<

Re: C-Ray FP/CPU Benchmark Test Results

Unread postby Dr. Dave » Thu Jun 19, 2008 8:07 pm

Yet another one - this was originally a dual R10K-250, 1M cache per CPU - now it's a dual R12k-344, cache running at 172 Mhz (cache divisor of 2, 6ns cache chips), SysAD of 114.5476 MHz.

Code: Select all

cat scene | ./c-ray-mt -t 32 > foo.ppm
c-ray-mt v1.1
Rendering took: 1 seconds (1493 milliseconds)

cat sphfract | ./c-ray-mt -t 32 > foo.ppm
c-ray-mt v1.1
Rendering took: 40 seconds (40402 milliseconds)

cat sphfract | ./c-ray-mt -t 32 -s 1024x768 -r 8 > foo.ppm
c-ray-mt v1.1
Rendering took: 528 seconds (528140 milliseconds)

cat scene | ./c-ray-mt -t 32 -s 7500x3500 > foo.ppm
c-ray-mt v1.1
Rendering took: 89 seconds (89239 milliseconds)
:O3000: <> :O3000: :O2000: :Tezro: :Fuel: x2+ :Octane2: :Octane: x3 :1600SW: x2 :O2: x2+ :Indigo2IMP: :Indigo2: x2 :Indigo: x3 :Indy: x2+

Once you step up to the big iron, you learn all about physics, electrical standards, and first aid - usually all in the same day

User avatar
mapesdhs
Posts: 2519
Joined: Mon Nov 10, 2003 4:17 pm
Location: Edinburgh, Scotland
Contact:

Re: C-Ray FP/CPU Benchmark Test Results

Unread postby mapesdhs » Fri Jun 20, 2008 3:36 am

Dr. Dave writes:
> Yet another one - this was originally a dual R10K-250, 1M cache per CPU - now it's a dual R12k-344, cache
> running at 172 Mhz (cache divisor of 2, 6ns cache chips), SysAD of 114.5476 MHz.

So is it really an R12K though? I mean, are the CPUs still R10Ks but just oc'd? If it's an oc'd R10K, it wouldn't
be correct to call it an R12K since the latter does have internal design changes.

Ian.

User avatar
Dr. Dave
Posts: 2311
Joined: Fri Feb 13, 2004 10:37 pm
Location: Ottawa, Canada >burp<

Re: C-Ray FP/CPU Benchmark Test Results

Unread postby Dr. Dave » Fri Jun 20, 2008 6:39 am

It's an R12k, the CPU's have been swapped for R12k-300's.
:O3000: <> :O3000: :O2000: :Tezro: :Fuel: x2+ :Octane2: :Octane: x3 :1600SW: x2 :O2: x2+ :Indigo2IMP: :Indigo2: x2 :Indigo: x3 :Indy: x2+

Once you step up to the big iron, you learn all about physics, electrical standards, and first aid - usually all in the same day

User avatar
mapesdhs
Posts: 2519
Joined: Mon Nov 10, 2003 4:17 pm
Location: Edinburgh, Scotland
Contact:

Re: C-Ray FP/CPU Benchmark Test Results

Unread postby mapesdhs » Fri Jun 20, 2008 6:48 am

Dr. Dave wrote:It's an R12k, the CPU's have been swapped for R12k-300's.


Ah, I see! So you've taken the chips from two single-300s, put them into the dual-250 module
and then oc'd it? Neat!

Hey, how about a proper writeup on all these different mods you've done? I'd be happy to house
it on my site & mirrors.

Ian.

User avatar
mapesdhs
Posts: 2519
Joined: Mon Nov 10, 2003 4:17 pm
Location: Edinburgh, Scotland
Contact:

Re: C-Ray FP/CPU Benchmark Test Results

Unread postby mapesdhs » Fri Jun 20, 2008 3:58 pm

Just finished testing my 24 x R10K/195 POWER Challenge 10000. It does rather well:

Code: Select all

scene: 271ms  (88 threads)
sphfract: 6530ms  (128 threads)
sphfract @ 1024x768 / 8: 81481ms  (128 threads)
scene @ 7500x3500: 13687  (128 threads)


Not bad! 8)

Ian.

User avatar
nekonoko
Site Admin
Site Admin
Posts: 8145
Joined: Thu Jan 23, 2003 1:31 am
Location: Pleasanton, California
Contact:

Re: C-Ray FP/CPU Benchmark Test Results

Unread postby nekonoko » Fri Jun 20, 2008 4:04 pm

mapesdhs wrote:Not bad! 8)


My O350 still beats it in only 4Us of rack space. Whoohoo! ;)
Twitter: @neko_no_ko
IRIX Release 4.0.5 IP12 Version 06151813 System V
Copyright 1987-1992 Silicon Graphics, Inc.
All Rights Reserved.

User avatar
mapesdhs
Posts: 2519
Joined: Mon Nov 10, 2003 4:17 pm
Location: Edinburgh, Scotland
Contact:

Re: C-Ray FP/CPU Benchmark Test Results

Unread postby mapesdhs » Sat Jun 21, 2008 2:25 am

nekonoko wrote:My O350 still beats it in only 4Us of rack space. Whoohoo! ;)


:D

Interesting to note how, at least for the short scene test, quite a few R10K/195s are needed to
beat your 6 x R14K/600 system (5% faster) - communication overhead a bit of a hit for such a
short test I guess. MHz-wise, 4680 in the Challenge vs. 3600 in your O300. It does better for
the longer tests as expected, eg. 17% faster for the sphfract test and 21% faster for the complex
sphfract test. I can't see any data for your 6xR14K/600 for the high-res scene test - did you run it?

Hmm, the 'scene' results do show the shortness of the test is becoming a major factor for
multi-core/CPU systems though. I reckon the faster internal links in O300 are helping there,
compared to how Challenge has to route data around internally over a shared bus. When the tests
are longer, the com links are less of a factor, so the Challenge does comparatively better.

Meanwhile, where's Bishop2K? I wanna see an SGI on the top spot! :D Be really interesting to see
whether the communication overhead means his big O3K can't run the short scene test as fast as
it theoretically could, though it should certainly do well for the longer tests.

Ian.
Last edited by mapesdhs on Sat Jun 21, 2008 2:46 am, edited 1 time in total.

User avatar
nekonoko
Site Admin
Site Admin
Posts: 8145
Joined: Thu Jan 23, 2003 1:31 am
Location: Pleasanton, California
Contact:

Re: C-Ray FP/CPU Benchmark Test Results

Unread postby nekonoko » Sat Jun 21, 2008 2:33 am

mapesdhs wrote:I can't see any data for your 6xR14K/600 for the high-res scene test - did you run it?


Don't have the O300 6xR14K/600 system anymore - half of it is in France and the other half is in the eastern US. The replacement O350 system is 8xR16K/700; if there are any results you need on that one let me know.
Twitter: @neko_no_ko
IRIX Release 4.0.5 IP12 Version 06151813 System V
Copyright 1987-1992 Silicon Graphics, Inc.
All Rights Reserved.

User avatar
mapesdhs
Posts: 2519
Joined: Mon Nov 10, 2003 4:17 pm
Location: Edinburgh, Scotland
Contact:

Re: C-Ray FP/CPU Benchmark Test Results

Unread postby mapesdhs » Sat Jun 21, 2008 2:52 am

nekonoko writes:
> Don't have the O300 6xR14K/600 system anymore - half of it is in France and the other half is in the eastern US. ...

Oh!

I noticed that 8 x 600 system on eBay. Would have bid/bought, but couldn't afford it atm. :\


> The replacement O350 system is 8xR16K/700; if there are any results you need on that one let me know.

You've done all the C-Ray tests, but how about doing the Blender test? It runs ok on a server remotely. I tested
the Challenge using a net link from my Octane2. A bit of overhead, but not much. Your O350 has the optimum
no. of CPUs for running the test since Blender can't issue more than 8 threads. Use V2.44; the archive and the
test scene file can be downloaded from my site, links on the URL above.

Ian.


Return to “SGI: Hardware”

Who is online

Users browsing this forum: No registered users and 2 guests