New applications for GPU released


Message boards : News : New applications for GPU released

Message board moderation

To post messages, you must log in.
Previous · 1 · 2 · 3 · 4 · 5 . . . 8 · Next
AuthorMessage
Profile Sabroe_SMC

Send message
Joined: 20 Jun 12
Posts: 10
Credit: 54,138,618
RAC: 0
Message 2261 - Posted: 30 Dec 2013, 9:29:37 UTC

Last modified: 30 Dec 2013, 9:35:38 UTC
GTX 780Ti ~2250 s http://asteroidsathome.net/boinc/result.php?resultid=28194421
GTX Titan ~2300 s (recommended CUDA double precision disabled)
GTX 680 ~5300 s
GTX 650Ti ~8400 s
GTX 580 ~4200 s
GTX 570 ~3900-5000 s
GTX 480 ~4300-5300 s
GTX 560Ti ~5800 s
.
GTX 660M ~15000 s
ID: 2261 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Alessandro Freda

Send message
Joined: 13 Jan 13
Posts: 14
Credit: 148,362,928
RAC: 12,049
Message 2263 - Posted: 30 Dec 2013, 10:02:17 UTC - in response to Message 2227.  
It works also on Quadro series ?
I have this one:

NVIDIA GPU 0: Quadro FX 1800 (driver version 331.82, CUDA version 6.0, compute capability 1.1, 768MB, 8381137MB available, 264 GFLOPS peak)


but cannot get work:

30/12/2013 10:57:55 | Asteroids@home | Requesting new tasks for NVIDIA
30/12/2013 10:58:00 | Asteroids@home | Scheduler request completed: got 0 new tasks

ID: 2263 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile HA-SOFT, s.r.o.
Project developer
Project tester

Send message
Joined: 21 Dec 12
Posts: 176
Credit: 134,881,433
RAC: 2,368
Message 2266 - Posted: 30 Dec 2013, 11:17:58 UTC - in response to Message 2263.  
It works also on Quadro series ?
I have this one:

NVIDIA GPU 0: Quadro FX 1800 (driver version 331.82, CUDA version 6.0, compute capability 1.1, 768MB, 8381137MB available, 264 GFLOPS peak)


but cannot get work:

30/12/2013 10:57:55 | Asteroids@home | Requesting new tasks for NVIDIA
30/12/2013 10:58:00 | Asteroids@home | Scheduler request completed: got 0 new tasks



No. Your card must be Compute capability 2.0 or better.
ID: 2266 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile HA-SOFT, s.r.o.
Project developer
Project tester

Send message
Joined: 21 Dec 12
Posts: 176
Credit: 134,881,433
RAC: 2,368
Message 2267 - Posted: 30 Dec 2013, 11:18:35 UTC - in response to Message 2261.  
GTX 780Ti ~2250 s


Thanks
ID: 2267 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
cyrusNGC_224@P3D

Send message
Joined: 1 Apr 13
Posts: 37
Credit: 153,496,537
RAC: 0
Message 2268 - Posted: 30 Dec 2013, 11:27:08 UTC - in response to Message 2261.  
GTX 780Ti ~2250 s http://asteroidsathome.net/boinc/result.php?resultid=28194421
GTX Titan ~2300 s (recommended CUDA double precision disabled)
GTX 680 ~5300 s
GTX 650Ti ~8400 s
GTX 580 ~4200 s
GTX 570 ~3900-5000 s
GTX 480 ~4300-5300 s
GTX 560Ti ~5800 s
.
GTX 660M ~15000 s
GT 635M ~35000 s (Linux) http://asteroidsathome.net/boinc/result.php?resultid=27529265

Too slow than it is worth.
ID: 2268 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
skgiven
Avatar

Send message
Joined: 28 Jul 12
Posts: 13
Credit: 1,616,360
RAC: 0
Message 2269 - Posted: 30 Dec 2013, 11:40:25 UTC - in response to Message 2228.  

Last modified: 30 Dec 2013, 12:00:06 UTC
GTX 780Ti ~2250 s
GTX Titan ~2300 s (recommended CUDA double precision disabled)
GTX 680 ~5300 s
GTX 650Ti ~8400 s
GTX 580 ~4200 s
GTX 570 ~4200 s
GTX 560Ti ~5500 s
.
GTX 660M ~15000 s

.
.
.
in progress.


GTX770 3,616s
GTX670 4,980s

Win7 x64, 8GB DDR2133, Intel(R) Xeon(R) CPU E3-1265L V2 @ 2.50GHz (8threads)

28124806 11741759 5194 29 Dec 2013, 22:10:28 UTC 30 Dec 2013, 1:16:06 UTC Completed, waiting for validation 3,615.51 35.62 pending Period Search Application v100.00 (cuda55)

28038972 11728673 5194 29 Dec 2013, 18:46:25 UTC 29 Dec 2013, 21:48:12 UTC Completed and validated 4,980.47 41.03 480.00 Period Search Application v100.00 (cuda55)

MSI Afterburner (RivaTuner):
GPU power, 56% and 62%
GPU temperature, 63C and 47C
GPU usage, 98% and 97%
Core clock, 1163MHz and 1110MHz
GDDR clock, 3506MHz and 3005MHz
GDDR usage, 621MHz and 457MB


Regarding AVX vs SSE3, on an i73770K there seems to be from 3% to about 15% difference in run times. I expect you would need an ix-4xxx CPU to benefit fully from AVX code developments:

28194832 11760270 53841 30 Dec 2013, 0:47:58 UTC 30 Dec 2013, 4:06:53 UTC Completed and validated 8,747.55 8,665.78 480.00 Period Search Application v102.10 (sse3)
28194788 11755863 53841 30 Dec 2013, 0:47:58 UTC 30 Dec 2013, 4:06:53 UTC Completed, waiting for validation 7,855.23 7,774.79 pending Period Search Application v102.10 (sse3)
28136775 11746774 53841 29 Dec 2013, 22:36:14 UTC 30 Dec 2013, 2:58:03 UTC Completed and validated 7,588.37 7,517.23 480.00 Period Search Application v102.10 (avx)

If the GPU and CPU WU's are the same size, just different apps, then a GTX770 may be twice as fast as a single i7 CPU thread, but only has 1/4 the performance of the entire CPU. This seems to be the case because they get the same credit (480).
.
ID: 2269 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile HA-SOFT, s.r.o.
Project developer
Project tester

Send message
Joined: 21 Dec 12
Posts: 176
Credit: 134,881,433
RAC: 2,368
Message 2270 - Posted: 30 Dec 2013, 11:57:37 UTC - in response to Message 2269.  

Last modified: 30 Dec 2013, 11:59:27 UTC

Regarding AVX vs SSE3, on an i73770K there seems to be from 3% to about 15% difference in run times. I expect you would need an ix-4xxx CPU to benefit fully from AVX code developments


Yes. AVX on 3xxx stuck on memory loading (CPU does two SSE3 op). AVX is for 4xxx CPUs.

If the WU's for the GPU and the CPU are the same, just different apps, then a GTX770 is twice as fast as a single i7 CPU thread, but only has 1/4 the performance of the entire CPU.


Yes. It is.
ID: 2270 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile [AF>Amis des Lapins] Phil1966

Send message
Joined: 10 Jul 13
Posts: 21
Credit: 10,363,676
RAC: 0
Message 2273 - Posted: 30 Dec 2013, 12:44:14 UTC - in response to Message 2270.  

Regarding AVX vs SSE3, on an i73770K there seems to be from 3% to about 15% difference in run times. I expect you would need an ix-4xxx CPU to benefit fully from AVX code developments


Yes. AVX on 3xxx stuck on memory loading (CPU does two SSE3 op). AVX is for 4xxx CPUs.

If the WU's for the GPU and the CPU are the same, just different apps, then a GTX770 is twice as fast as a single i7 CPU thread, but only has 1/4 the performance of the entire CPU.


Yes. It is.


Then I don't understand why you launched this GPU app ?
Is it a test for a futur "different / new type of" WU's ?

Have tryed AVX WU's on an i7-4770K ... but I don't know why, very very slow +
only increasing the CPU temp. I guess one should run AVX on 4 cores only => no real gain <-> SSE3 on 8 cores ...

Wish you a Happy New Year !
ID: 2273 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile HA-SOFT, s.r.o.
Project developer
Project tester

Send message
Joined: 21 Dec 12
Posts: 176
Credit: 134,881,433
RAC: 2,368
Message 2274 - Posted: 30 Dec 2013, 12:58:10 UTC - in response to Message 2273.  

Last modified: 30 Dec 2013, 13:21:54 UTC
Then I don't understand why you launched this GPU app ?


To give an option for crunchers to select "right" app and of course to explore usage of GPUs for next development.

Is it a test for a futur "different / new type of" WU's ?


Not for now, but we will see in future.

Have tryed AVX WU's on an i7-4770K ... but I don't know why, very very slow +
only increasing the CPU temp. I guess one should run AVX on 4 cores only => no real gain <-> SSE3 on 8 cores ...


HT cores share FPU units so no profit from HT.

For example:
cpus without HT like i5-4670 can make the same work with AVX as 4770 (SSE3,AVX) and has better power/price ratio.
ID: 2274 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile HA-SOFT, s.r.o.
Project developer
Project tester

Send message
Joined: 21 Dec 12
Posts: 176
Credit: 134,881,433
RAC: 2,368
Message 2276 - Posted: 30 Dec 2013, 13:19:27 UTC - in response to Message 2258.  
Could we get an OpenCL-App for AMD GPUs too ?


Open-Cl is on our todo list. One of our volunteers decided to develop opencl app as shool project so we are waiting for him.

What about AVX-App on current AMD CPUs ?


You can try it with app_info, but AMD is very poor on AVX. AMD CPUs simulate AVX with SSE3.
ID: 2276 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile ritterm
Avatar

Send message
Joined: 22 Jun 12
Posts: 21
Credit: 6,288,600
RAC: 0
Message 2277 - Posted: 30 Dec 2013, 13:50:13 UTC

Last modified: 30 Dec 2013, 13:52:06 UTC
Ack! 11,000-12,000 sec on my FX-8150 host with a 550Ti... :-( Can that be optimized?
ID: 2277 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Freddykrug [Astronomy.Ru Forum...

Send message
Joined: 22 Jun 12
Posts: 7
Credit: 1,383,253
RAC: 63
Message 2278 - Posted: 30 Dec 2013, 14:00:48 UTC

Last modified: 30 Dec 2013, 14:08:05 UTC
It is a good news. You have done a wonderful Christmas gift! :) But - when will the release of the application for ATI (Radeon)?

Sorry, I do not read written above.
ID: 2278 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile WinterGuard1944
Avatar

Send message
Joined: 26 Apr 13
Posts: 2
Credit: 5,334,240
RAC: 0
Message 2280 - Posted: 30 Dec 2013, 14:35:06 UTC - in response to Message 2227.  
GT 540M ~32 000 s :-)
ID: 2280 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile HA-SOFT, s.r.o.
Project developer
Project tester

Send message
Joined: 21 Dec 12
Posts: 176
Credit: 134,881,433
RAC: 2,368
Message 2281 - Posted: 30 Dec 2013, 14:40:17 UTC - in response to Message 2277.  
Ack! 11,000-12,000 sec on my FX-8150 host with a 550Ti... :-( Can that be optimized?


Sorry, no luck here. On CC2.X cards we are limited by block/SMX and registers/SMX so card occupancy is 67% and 550Ti has only 4 SMX. So 4*8 periods calculated in parallel.
ID: 2281 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile HA-SOFT, s.r.o.
Project developer
Project tester

Send message
Joined: 21 Dec 12
Posts: 176
Credit: 134,881,433
RAC: 2,368
Message 2284 - Posted: 30 Dec 2013, 15:07:19 UTC - in response to Message 2283.  
My poor little GT610, which is probably one of the slowest GPUs on this project, doesn't think that the new app is very fast.
It took nearly 17 hours to do one work unit, compared with my CPU, which can finish two work units in less than 2½ hours!


610 is very slow, I have three 610 with passive cooling at work.
ID: 2284 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile ritterm
Avatar

Send message
Joined: 22 Jun 12
Posts: 21
Credit: 6,288,600
RAC: 0
Message 2285 - Posted: 30 Dec 2013, 15:21:09 UTC - in response to Message 2281.  
Ack! 11,000-12,000 sec on my FX-8150 host with a 550Ti... :-( Can that be optimized?

Sorry, no luck here. On CC2.X cards we are limited by block/SMX and registers/SMX so card occupancy is 67% and 550Ti has only 4 SMX. So 4*8 periods calculated in parallel.

Thanks for the feedback. :-)

Oh, well...I guess I'll keep that GPU running Einstein.
ID: 2285 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile den777

Send message
Joined: 22 Jun 13
Posts: 15
Credit: 3,118,560
RAC: 0
Message 2293 - Posted: 30 Dec 2013, 16:55:54 UTC
30.12.2013 23:51:51 |  | CUDA: NVIDIA GPU 0: GeForce GTS 450 (driver version 327.23, CUDA version 5.5, compute capability 2.1, 1024MB, 925MB available, 637 GFLOPS peak)
30.12.2013 23:53:27 | Asteroids@home | Requesting new tasks for NVIDIA
30.12.2013 23:53:29 | Asteroids@home | Scheduler request completed: got 0 new tasks


What's wrong?
ID: 2293 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile HA-SOFT, s.r.o.
Project developer
Project tester

Send message
Joined: 21 Dec 12
Posts: 176
Credit: 134,881,433
RAC: 2,368
Message 2294 - Posted: 30 Dec 2013, 17:06:52 UTC - in response to Message 2293.  
30.12.2013 23:51:51 |  | CUDA: NVIDIA GPU 0: GeForce GTS 450 (driver version 327.23, CUDA version 5.5, compute capability 2.1, 1024MB, 925MB available, 637 GFLOPS peak)
30.12.2013 23:53:27 | Asteroids@home | Requesting new tasks for NVIDIA
30.12.2013 23:53:29 | Asteroids@home | Scheduler request completed: got 0 new tasks


What's wrong?


see if it helps

http://asteroidsathome.net/boinc/forum_thread.php?id=234
ID: 2294 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Jan Vaclavik

Send message
Joined: 26 Jan 13
Posts: 31
Credit: 1,501,198
RAC: 270
Message 2296 - Posted: 30 Dec 2013, 17:23:05 UTC
Great news for the project, but to be honest I am somewhat disappointed by the performance of mid-range card. I guess I will be sticking around with my CPU somewhat longer...
ID: 2296 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile HA-SOFT, s.r.o.
Project developer
Project tester

Send message
Joined: 21 Dec 12
Posts: 176
Credit: 134,881,433
RAC: 2,368
Message 2297 - Posted: 30 Dec 2013, 17:30:57 UTC - in response to Message 2296.  
Great news for the project, but to be honest I am somewhat disappointed by the performance of mid-range card. I guess I will be sticking around with my CPU somewhat longer...


These are expected results. GPUs are very specialized and it's hard to fit code to every hw architecture.
ID: 2297 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Previous · 1 · 2 · 3 · 4 · 5 . . . 8 · Next

Message boards : News : New applications for GPU released