124 replies to this topic

[mitch_de]

All OpenCL GPU FPS Benches:

NEW OpenCL Raytraycing Benchmark (updated 1. Posting)

smallluxGPU
Does raytraycing by GPU, GPU+CPU or CPU only
Very complex (real life) computing, so less advantage for weak GPU than running more low level OpenCL Demos.
Does much better hybrid (CPU+GPU) than Galaxies.
Uses ALL openCL GPUs (up to 4) parallel which it find.
Also works with ATI 48XX GPUs .  

Update to V170 (always same link )[/b]
http://www.macupdate...632/smallluxgpu
Major update with console tab (you can see informations the gui also shows but now even more + errors)
Happy benching (times type 0 no changes, type 1 maybe little faster)
 33632_scr.jpg   253.91K   291 downloads


************





Older stuff / mostly not much real world like smalllux !
  Galaxies32K V2 + Galaxies 8K V2  + Grass + Displacement
+AO (raytraycing CPU/GPU) + Transpose Bandwith [/size]
Snow Leopard + Intel Macs ONLY !
ATI OpenCL GPUs (4850&4870)  not really working! - i am in contact with ATI DEVs -problems with OpemCL Drivers/Framework - must+will be fixed  with 10.6.1 or an ATI Driver Update

HOT NEWS - always updated here
- New Galaxies OpenCL Bench V2 build
- Apple updated / fixed some OpenCL API usage (maybe help ATI)
- little speed up (10% on my GT 9600)
Now i build an 32K and an 8K Version - 32K use for fast/highend GPU/CPU and 8K for lowend CPU/GPUs.
If GPU limits, there will be no difference in GPU Gigaflops. But on very fast GPUs 32K may give much higher GPU Gigaflops - more GPU load=less waste of OpenCL overhead time.

DL Links on 1 post

- OS X 10.6.1 updated ATI/Intel+Nvidia OpenGL drivers, but the OpenCL Framework stays same. So ATI will fail (i am sure) also with 10.6.1 running Apple OpenCL Demos (here listed).

Apple does an rewrite of the GALAXIES OpenCL Demo/Bench - Nvidia GTX285 will rise from 280 GigaFlops to around 400 GigaFlops
The CPU GigaFlops stays same around 28 GigaFlops C2D /100 GigaFlops MacPro 2009 .
This new Version will compiled and shared here like the last one. Slower OpenCL GPUs, like mine 9600 GT should not expect such an big GigaFlop boost with that new Galaxies (N-Body) Apple Demo.

OPENCL - Good to know :
- OpenCL is an API for universal GPU(CPU) computing
- main difference to CUDA / ATI STEAM is: both only working with their "own" gpu.
an CUDA (NV) app like badaboom(h264 on GPU) cant work on an ATI gpu and vice versa
- OpenCL is universal for different gpu vendors means:
- Xcode / GCC compiles an code which includes the source (in C as an string)  for the gpu programm
that c source is , different to CUDA/ATI STEAM , is compiled later by OpenCL Framework at runtime !
So same App can run on complete different gpus and also , without/less codechange om CPU if no
OpemCL gpu (newer ones) is found
The source (example below) for the gpu programm will be really compiled at runtime, not only interpreted.
So little differences between run of my bench may happen because of that compile on the run

NEWS:
form An Information from ATI OS X OpenCL divison dev team:
Thank you for the quick response and I hope you extend the benchmark application since it’s a really good idea. Regarding the sample applications posted on the developer.apple.com website (eg. Galaxies, Qjulia, etc), we are aware that some fail (or even crash) on AMD hardware and working to track down all these issues. We suspect that most of these issues will be resolved for the next graphics driver update in Snow Leopard.
BTW, I ran the demos on a iMac with a Radeon 4850 and I get the following results:

OpenCL Benches:

G A L A X I E S - an CPU vs GPU GigaFlop Bench
UPDATE:
Due to an hint from ATI, i  increased the count of thing to compute from 4K to 16K
New 16K version is available , shows 16K at the legend.
Apple reduces that count for GPUs without discrete VRAM = 9400M / 8600M, so it is set to  4K, even legend shows 16K.
You cant compare 9400M / 8600M 1:1 with the other, run 16K count

Using of GALAXIES:
Start Galaxy
key s = switch compute  Modes
>CPU>Single/Multi, CPU-Vector/SSE Single/Multi>GPU> GPU+CPU> (bold=start Mode)
key  SPACE  = Pause/go on
key 6 = Reset Szene
key Q = QUIT

DOWNLOADs each 6 MB:
NEW: http://rapidshare.co...xies_32K_V2.zip

NEW: http://rapidshare.co...axies_8K_V2.zip

mitch (C2D 3GHZ, NV 9600 GT , 1280x1024)
V1 /V2 24 Gigaflops : CPU ( SIM: Vector Multi-Core CPU. Mode)
112 Gigaflops[/b] : NV 9600 GT 16K
V2 142 Gigaflops : NV 9600 GT 32K[/b]
Users results (new 32K V2 Version):
CPU 18 G Nvidia 9600gt gpu 149 GigaFlops 32k v2.0 at 1680x1050
Gigabyte Ga-eg45m-ud2r, Intel e6750,
Mac Pro Nehalem 8 core 2.93GHz:http://www.barefeats.com/index.html
All tests 2500x1600!
Nvidia GTX 285 ...... soon!!
Mac Pro Nehalem 8 core CPU = .....

New OpenCL Transpose Bandwith - measures Bandwith of Matrix-Transpose
DL Link at the end of posting (very small, run like all other terminal OpenCL Bench apps)
mitch:
Nvidia 9600GT: around 39 Gigabyte/Sec
Mac Pro (1,1) 2.66Ghz 4GB RAM, 4870 1GB sapphire
Performing Matrix Transpose [256 x 4096]...
Bandwidth Achieved = 3.160816 GB/sec
MacPro 2009, NVidia GTX 285 Mac
Bandwidth Achieved =  around 80 GB/sec

UPDATE:
New OpenCL AO Bench (512*512 insted of 256*256 barefeat = barefeat results / 2)
DL Link at the end (very small, SSE4 optimized, 512*512 Window )
mitch
NV 9600 GT : 8 FPS (512*512)
C2D 3 GHZ : 0.8 FPS


So ATI users may try the new compiled
Procedural Geometric Displacement FPS Bench
Download for ATI + Nvidia USERS:
OpenCL_Displacement_Bench.zip  - with step by step HOW TO RUN readme -
Update:
An new compiled Displacement (the app only) which was build with GCC 4.2 very less optimzed compiler settings seems to run on ATI 4870 more stable / reliable.

If you have such problems with displacement, I attatched the small dl at the first post as displacement_ATi for overwrite+usage with the whole (normal) 7 MB dl.

QJulia1024 Results (the qJulia with 1024x1024 window size)
9600 GT , around 13  FPS static  - please let the bench first (wait a few seconds) show static FPS before you switch to animate (SPACE)
8-16 FPS when animating
Rob GTX 285 Mac
1024x1024 = 44 fps
eVGA GTX-285 1024MB  46.70fps
8800 GT 22.46 FPS
qJulia Results (800x800 window)
9600GT : around 29 FPS static, 16 - 60 FPS when animating (key SPACE)
eVGA GTX-285 1024MB  98.79 fps
Rob GTX 285 Mac
800x800 = 93 fps
2.4GHz C2D, 4GB RAM, 8600M GT (256MB)
qJulia (800x800):
static shows 10 - 11 FPS animated shows 9 - 11 FPS
MacBook Pro 13", GPU GeForce 9400M
running at 6,25 fps (6-6,50)

OpenCL Displacement FPS results (ATI should work !)
mitch
9600 GT 80 FPS first (white background+shadow) / 102 FPS second (with texture in backround)
ATI 4870 both around 90 FPS - but only 1/3 of start the bench was successfull - so also that bench didnt work 100% well - wait for OS X 10.6.1
Geforce GTX285 Mac : around 220 FPS, second shader test
Radeon 4870 1GB (sapphire)
Mac Pro (1,1) 2.66Ghz quad core, 4GB RAM
both szenes near 90 FPS
Most 4870 are near together between 84 and 90 FPS - but some test fail and some get bad result window graphics

GRASS simulates an scene grass sticks moving in the wind
Grass Results 4 Meg triangles + 170.000 Sticks to compute - big szene! ( 1024x1024 window size)
9600 GT , around 53 FPS
2.4GHz C2D, 4GB RAM, 8600M GT (256MB) 27 - 29 FPS
8800 GT  56.97 fps
i920 Overclocked to 4.4Ghz, 1760Mhz DDR3, PCIE-100Mhz,
eVGA GTX-285 1024MB  95.50 fps
Rob ( barefeats! Test mule is Mac Pro Nehalem 2.93 Octo)
GeForce GTX 285 Mac = 88 - 91 fps
Quadro FX 4800 = 77 fps steady
GeForce 8800 GT = 54 fps steady
GeForce GT 120 = 35 fps steady
DL for qJulia + Grass (has GUI) (at the end)
Read the readme - you will ger an file not found error (loads the qjulia.cl OpemCL source, if you didnt changed terminal directory to the app folder before running the command line app.

For all GLUT (Terminal Apps, Transpose+qJluia+AO) check the app preferences of SYNC is OFF (screenshoot OpenCL AO preferences)

Attached Files

•  OpenCL_Transpose__Bandwidh.zip   6.24K   742 downloads
•  displacement_ATI.zip   22.07K   601 downloads
•  opencl_aobench.zip   145.75K   1001 downloads
•  Grass.zip   429.8K   1250 downloads
•  qJulia1024.zip   12.68K   696 downloads
•  OpenCL_Qjulia_GPU.zip   92.24K   1010 downloads
•  Bildschirmfoto_2009_09_01_um_09.02.58.jpg   102.32K   272 downloads
•  Bildschirmfoto_2009_09_01_um_19.39.37.jpg   388.74K   313 downloads
•  Bildschirmfoto_2009_09_04_um_14.46.23.jpg   88.81K   362 downloads
•  FPS_8.1_Nvidia_9600GT.jpg   79.43K   197 downloads
•  Bildschirmfoto_2009_09_06_um_11.20.42.jpg   69.81K   276 downloads
•  Bildschirmfoto_2009_09_28_um_23.18.48.jpg   82.5K   348 downloads
•  screen.jpg   170.61K   183 downloads

[GLXOZ]

mitch_de, on Aug 31 2009, 06:08 AM, said:

GALAXY OpenCL-Speed Bench-Demo   !
Snow Leopard + Intel Macs ONLY !

  LLMV GCC 4.2 Compiler compiliert = high optimized Code
Using:
Start Galaxy

your link is dead %(

[GLXOZ]

mitch_de, on Aug 31 2009, 06:30 AM, said:

??? What excat happens ???  I can download that RS Link - tested it right now. I added an second DL Link.

With your second link all OK, thank you - great work !

[mitch_de]

GLXOZ, on Aug 30 2009, 09:47 PM, said:

...- great work !

Was Apple Work - i only compiled it !

[GLXOZ]

mitch_de, on Aug 31 2009, 07:30 AM, said:

Was Apple Work - i only compiled it !

No matter, you compile it - you do work.

[reinstaller]

1600x1200
20 Gigaflops / around 59 U/sec : CPU ( SIM: Vector Multi-Core CPU. Mode) E6600 @ 3GHz
170  Gigaflops / around 505 U/sec : Nvidia 9800 GTX+

 Screen_shot_2009_08_30_at_22.34.24.png   148.13K   182 downloads  Screen_shot_2009_08_30_at_22.34.57.png   189.76K   161 downloads

[Sherry Haibara]

1280x800
20 Gigaflops - around 60 U/sec (CPU: Core2Duo 2,53 ghz SIM Vector Multi-core CPU)
10 Gigaflops - around 30 U/sec (GPU: SIM GeForce 9400M)
5 Gigaflops - around 60 U/sec (CPU Multi-core + GPU)

Testing on latest MacBook Pro 13".

Sherry Haibara

[blackosx]

Mitch, this is lovely - Thanks

blackosx (C2D 2.66GHZ, NV 8800 GT , 1680x1050)
19 Gigaflops / around 57 U/sec : CPU ( SIM: Vector Multi-Core CPU. Mode)
60 Gigaflops / around 178 U/sec : Nvidia 8800 GT

[Chaz_UK]

I just did a little trial on my 20 inch iMac and it seems that the lack of OpenCL support for the ATi 2x00 cards is disappointing.


Thanks for sharing it!

I'm not sure if I ran it right but here is a screengrab:

Attached Files

•  Picture_2.png   151.44K   128 downloads

[AppleIIGuy]

9800GT
Core i7 920 stock speed

290 Updates/sec
98 Gigaflops

1900x1200
Sim: GeForce 9800 GT

[radov4n]

Great tool

1600x1200, i7

http://i28.tinypic.com/ml2i6o.jpg

http://i31.tinypic.com/35hel9h.jpg

http://i30.tinypic.com/2m5av6w.jpg

http://i30.tinypic.com/nd68a1.jpg

[johan]

why do result go up and down. in same session?

i go up and down from 80 to 135 gflop

with 8800 gtx

i expected results to be stable within a session

[mitch_de]

johan, on Aug 31 2009, 04:52 PM, said:

why do result go up and down. in same session?
i go up and down from 80 to 135 gflop
with 8800 gtx
i expected results to be stable within a session

Because ist an simulation of stars.  They move by gravity so a lot changes (less stars/ more stars)  of things to compute over the time.
Thats make the changes in GigaFlops also - its a dynamic simulation, not very static.

[Cyberdog !]

don't work on my iMac Core Duo 1.83 Ghz + ATI X1600
10.6 32 bits.

[proengin]

W3520 overclocked to 4.1Ghz Turbo, PCIE 102Mhz, 1280x1024x75hz, standard scene

GTX-285 - 469 updates/sec, 157 Gigaflops
Vector Multi core - 262 updates/sec, 87 Gigaflops

[mitch_de]

proengin, on Sep 1 2009, 07:46 AM, said:

W3520 overclocked to 4.1Ghz Turbo, PCIE 102Mhz, 1280x1024x75hz, standard scene

GTX-285 - 469 updates/sec, 157 Gigaflops
Vector Multi core - 262 updates/sec, 87 Gigaflops

Thanks.
What exact is an W3520 (Xeon 4 core, Xeon i7 4 core + 4 core) ?

157 Gigaflops looks a bit less, compard to an other posted 170 GigaFlops with his 9800GTX.
Can you run the App again and compare results ?
But maybe the 9800GTX values arent "real".

HINT:
Also try the new qJulia OpenCL FPS Bench - high gpu usage of complex OpenCL code.
http://www.insanelym...howtopic=183237
Maybe you see there more difference to other gpus.
My 9600GT getswith qJulia OpenCL FPS Bench around 30 FPS.
Galaxy with my 9600GT was around 70 Gigaflops at 1280x1024 (60 Gigafops 1600x1200).
So in Galaxy your  GTX285 is 2,2 times faster (GPU) than my 9600GT.
I am very ?? if qJuilia is scaling on your GTX 285 GPU.
Perhaps qJulia GTX 285 is more than 2,2 times faster : my 9600 GT get there around 30 FPS in the 800x800 window.

[rob-ART]

I will be posting some results on BareFeats.com today showing various GPUs running Galaxy and two other OpenCL benchmarks.

When I ran Galaxy, I noticed that the rates dropped after running for a while. So I'm posting the "ending" numbers on updates per second and gigaflops.

For some reason, with the Radeon HD 4870 installed, Galaxy hung on startup. I've sent the dump to mitch_de and to ATI engineering to see if there's an easy fix.

[JBeed]

I'm a bit worried here.
My MacBook Pro (2.4GHz C2D, 4GB RAM, 8600M GT) is getting only 10 Gflops (30 Updates/s) when using the GPU. If I'm using the Vector Multi-core setting, I get about 20 Gflops from the CPU.

I've tried to shut down all the windows, and I've tried with the machine newly restarted. I've also tried installing CUDA (and the CUDA-files are there, checked), but there is no difference at all.
It's stuck at exactly 10 Gflops and 30 Updates/s, no change at all. The weird thing, though, is that even if I'm running a game in the background (World of Warcraft), I get the exact same performance in this application; isn't the performance supposed to be lower if there's another application using the GPU?

Anyone else here with a MacBook Pro running 8600M GT that can post their results, so I have something to compare against?

[mitch_de]

Your results are normal - dont worry.
The 8600M GT isnt an good OpenCL gpu. OpenCL uses the programming features (OpenCL works a bit like shaders in games - but much more flexibel+universal) which are "undersized" in GPUs before 2009 - even 2008 build desktop gpus normally arent good OpenCL gpus.
So dont worry about that, because OpenCL usage will start in 2010 - not before, because most people wont have an gpu which can do things much faster than cpu. Only an less cpu load may the result of OpenCL usage on low end OpenCL gpus.
Mobile GPUs (MacBook, MacMini, iMac also!) maybe must wait longer than 2010, because they are not build for high end number crunching jobs.
Poweful (for OpenCL) & low power usage mobile gpus will bot be there until end of 2010 !
Its simple impossible to build an GTX285 / 9800GTX+  speed into an mobile mac or Imac - overheats (iMac) or mobile (akku) time less then 10 minutes ?!.

[reinstaller]

All my results are with cuda drivers.