GPU FLOPS: Theory vs Reality

Message boards : Number crunching : GPU FLOPS: Theory vs Reality
Message board moderation

To post messages, you must log in.

1 · 2 · 3 · 4 . . . 20 · Next

AuthorMessage
Profile Shaggie76
Avatar

Send message
Joined: 9 Oct 09
Posts: 282
Credit: 271,858,118
RAC: 196
Canada
Message 1891052 - Posted: 21 Sep 2017, 0:37:25 UTC

Evidently I haven't updated my old thread in so long that it had to be locked so here's a fresh thread.

I ran another scan today and a few things are new: the RX 570 and RX 580s are on the charts now -- surprisingly they aren't running quite as fast the RX 480s on the chart -- it might be luck but there are over a dozen hosts and over 2000 tasks counted for each so I'm not sure that this is just sampling error (in contrast the RX 480 stats in this scan covered almost 150 hosts and over 19000 tasks so I'm pretty confident).

There aren't enough Vega parts in circulation for them to qualify for stats - I'll run another scan in a month or so and see if there are enough then.

ID: 1891052 · Report as offensive
EdwardPF
Volunteer tester

Send message
Joined: 26 Jul 99
Posts: 389
Credit: 236,772,605
RAC: 374
United States
Message 1891055 - Posted: 21 Sep 2017, 0:47:44 UTC

I love this page!!

Thanks!

See 'ya again in a month or so.

Ed F
ID: 1891055 · Report as offensive
Profile betreger Project Donor
Avatar

Send message
Joined: 29 Jun 99
Posts: 11416
Credit: 29,581,041
RAC: 66
United States
Message 1891056 - Posted: 21 Sep 2017, 0:59:36 UTC - in response to Message 1891052.  

Thanx
ID: 1891056 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13855
Credit: 208,696,464
RAC: 304
Australia
Message 1891093 - Posted: 21 Sep 2017, 5:21:22 UTC - in response to Message 1891055.  

I love this page!!

Yep.
Greatly appreciated.
Grant
Darwin NT
ID: 1891093 · Report as offensive
Profile Shaggie76
Avatar

Send message
Joined: 9 Oct 09
Posts: 282
Credit: 271,858,118
RAC: 196
Canada
Message 1891126 - Posted: 21 Sep 2017, 12:10:19 UTC

One possible explanation occurred to me for the new RX 580 scores being lower than expected: if these cards recently replaced a older and slower card my scripts might misconstrue older results from the old card as from the RX 580 -- to ease pressure on the SETI servers I get host information and only the summary of the task stats (for me to dig into each task to handle this would be 20x more server queries).

I might also dig through and see if I can find some Vega parts for some preliminary results since I'm curious.
ID: 1891126 · Report as offensive
Profile petri33
Volunteer tester

Send message
Joined: 6 Jun 02
Posts: 1668
Credit: 623,086,772
RAC: 156
Finland
Message 1892054 - Posted: 27 Sep 2017, 21:36:22 UTC
Last modified: 27 Sep 2017, 21:39:06 UTC

Thank you Shaggie76,

My 3x1080 + 1x1080Ti do a 350 000+ per day: 350 000/24/4 = 3640+ cr/h per card.

They use, let me take a look, (3 x 145W + 224W) / 4 = 165W per card yieliding a 22 credits per Wh,

Can you draw a red X on your charts for other top 20 hosts.....

--
Petri.
To overcome Heisenbergs:
"You can't always get what you want / but if you try sometimes you just might find / you get what you need." -- Rolling Stones
ID: 1892054 · Report as offensive
Profile Wiggo
Avatar

Send message
Joined: 24 Jan 00
Posts: 36843
Credit: 261,360,520
RAC: 489
Australia
Message 1892066 - Posted: 27 Sep 2017, 22:10:34 UTC

The 3GB 1060's still do very well there and IMHO are still the best "bang for buck" going (now that the bit mining craze is over), but I may be a little prejudiced there. :-)

Cheers.
ID: 1892066 · Report as offensive
Profile RueiKe Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 14 Feb 16
Posts: 492
Credit: 378,512,430
RAC: 785
Taiwan
Message 1893331 - Posted: 5 Oct 2017, 7:13:57 UTC - in response to Message 1891126.  

Hi Shaggie76, I recently installed a waterblock on my Vega FE. There should be about 1 day of results since then. Tasks ran prior to that would be thermally throttled. The hosted is: 8341269

One possible explanation occurred to me for the new RX 580 scores being lower than expected: if these cards recently replaced a older and slower card my scripts might misconstrue older results from the old card as from the RX 580 -- to ease pressure on the SETI servers I get host information and only the summary of the task stats (for me to dig into each task to handle this would be 20x more server queries).

I might also dig through and see if I can find some Vega parts for some preliminary results since I'm curious.

GitHub: Ricks-Lab
Instagram: ricks_labs
ID: 1893331 · Report as offensive
Profile Brent Norman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester

Send message
Joined: 1 Dec 99
Posts: 2786
Credit: 685,657,289
RAC: 835
Canada
Message 1893332 - Posted: 5 Oct 2017, 7:18:22 UTC - in response to Message 1893331.  

How many GPU tasks are you running at once?
ID: 1893332 · Report as offensive
Profile RueiKe Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 14 Feb 16
Posts: 492
Credit: 378,512,430
RAC: 785
Taiwan
Message 1893333 - Posted: 5 Oct 2017, 7:36:40 UTC - in response to Message 1893332.  

How many GPU tasks are you running at once?


I’m running 1 at a time.
GitHub: Ricks-Lab
Instagram: ricks_labs
ID: 1893333 · Report as offensive
Profile Shaggie76
Avatar

Send message
Joined: 9 Oct 09
Posts: 282
Credit: 271,858,118
RAC: 196
Canada
Message 1893499 - Posted: 6 Oct 2017, 0:15:56 UTC

Host: 8341269

gfx901 (Anonymous)
    1052 Credit / Hour
     37% Core / Task
     403 Tasks
I'd have expected it to be a bit higher but maybe it's too soon to tell?
ID: 1893499 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13164
Credit: 1,160,866,277
RAC: 1,873
United States
Message 1893511 - Posted: 6 Oct 2017, 0:32:59 UTC - in response to Message 1893333.  

How many GPU tasks are you running at once?


I’m running 1 at a time.

Rick, looked at your stderr output and the only thing I might offer is increasing -tt to 1500. I don't know anything about the ATI cards but it might respond to that parameter like the Nvidia cards. Gives the GPU more time to process data before switching out from the kernel. From the docs ...
-tt F: Sets desired target time for kernel sequence. That is, how long kernel/kernel sequence can executes w/o interruption and w/o switching
	to another tasks like GUI update. F is floating point number in milliseconds. Default is 15ms. App will try to adapt kernels (currently
	implemented for PulseFind kernels) to run designated amount of time. To increase performance try to increase this value. ?High values
	could result in GUI lags. If use_sleep active try to use target times divisible on sleeping time quantum for your particular system.
	For example at least some AMD-based systems have 15ms sleep quantum. That is, Sleep(1) will actually sleep 15ms instead of 1ms.
	Has no effect in iGPU build (USE_OPENCL_INTEL path).

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 1893511 · Report as offensive
Profile RueiKe Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 14 Feb 16
Posts: 492
Credit: 378,512,430
RAC: 785
Taiwan
Message 1893516 - Posted: 6 Oct 2017, 0:39:58 UTC - in response to Message 1893499.  

Host: 8341269

gfx901 (Anonymous)
    1052 Credit / Hour
     37% Core / Task
     403 Tasks
I'd have expected it to be a bit higher but maybe it's too soon to tell?


Me too! With 60% increase in clock speed, I expected it would be significantly better than my Fiji based cards. I ran a bench test on a WU between Fiji and Vega and only saw a 10% improvement. I have not tried tweak command line arguments at all, so there may be some unrealized potential.
GitHub: Ricks-Lab
Instagram: ricks_labs
ID: 1893516 · Report as offensive
Profile RueiKe Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 14 Feb 16
Posts: 492
Credit: 378,512,430
RAC: 785
Taiwan
Message 1893518 - Posted: 6 Oct 2017, 0:42:08 UTC - in response to Message 1893511.  

How many GPU tasks are you running at once?


I’m running 1 at a time.

Rick, looked at your stderr output and the only thing I might offer is increasing -tt to 1500. I don't know anything about the ATI cards but it might respond to that parameter like the Nvidia cards. Gives the GPU more time to process data before switching out from the kernel. From the docs ...
-tt F: Sets desired target time for kernel sequence. That is, how long kernel/kernel sequence can executes w/o interruption and w/o switching
	to another tasks like GUI update. F is floating point number in milliseconds. Default is 15ms. App will try to adapt kernels (currently
	implemented for PulseFind kernels) to run designated amount of time. To increase performance try to increase this value. ?High values
	could result in GUI lags. If use_sleep active try to use target times divisible on sleeping time quantum for your particular system.
	For example at least some AMD-based systems have 15ms sleep quantum. That is, Sleep(1) will actually sleep 15ms instead of 1ms.
	Has no effect in iGPU build (USE_OPENCL_INTEL path).


Hi Keith, Thanks for the recommendation. I will give this a try today. I probably need to redo some of the DOE work I did on Fiji to optimize command line options.
GitHub: Ricks-Lab
Instagram: ricks_labs
ID: 1893518 · Report as offensive
Profile Shaggie76
Avatar

Send message
Joined: 9 Oct 09
Posts: 282
Credit: 271,858,118
RAC: 196
Canada
Message 1895859 - Posted: 17 Oct 2017, 23:50:32 UTC

I was hoping for more Vega parts in circulation by now but we aren't quite there yet (I require a certain number of completed work units per card to qualify and then enough separate computers for the card to show up on the charts. It's close, but not quite there yet:

C:\SETI>grep -i Vega GPUs.csv
7626762,Radeon RX Vega
7842719,Radeon RX Vega
7854642,Radeon RX Vega
8081803,Radeon RX Vega
8103729,Radeon RX Vega
8230810,Radeon RX Vega
8243334,Radeon RX Vega
8249242,Radeon RX Vega
8261851,Radeon RX Vega
8307472,Radeon RX Vega
8330537,Radeon RX Vega
8334662,Radeon RX Vega
8341269,Radeon Vega Frontier Edition
8344100,Radeon RX Vega
8344505,Radeon RX Vega

I also fixed a bug in my code that was mixing up the two types of Titan X cards but even now there aren't a lot of the Pascal parts crunching for SETI yet either:

C:\SETI>grep -i Pascal GPUs.csv
4693382,TITAN X (Pascal)
6987408,TITAN X (Pascal)
7978195,TITAN X (Pascal)
8008690,TITAN X (Pascal)
8076145,TITAN X (Pascal)
8107587,TITAN X (Pascal)
8163371,TITAN X (Pascal)
8184679,TITAN X (Pascal)
8286999,TITAN X (Pascal)
8312198,TITAN X (Pascal)
8333199,TITAN X (Pascal)
ID: 1895859 · Report as offensive
Dimitar Stoynev
Avatar

Send message
Joined: 7 Jan 09
Posts: 19
Credit: 336,531
RAC: 0
Bulgaria
Message 1898334 - Posted: 30 Oct 2017, 21:10:57 UTC

EPYC work!

I just wonder how a measly 770 can produce the same credit as a RX 580?
ID: 1898334 · Report as offensive
rob smith Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer moderator
Volunteer tester

Send message
Joined: 7 Mar 03
Posts: 22536
Credit: 416,307,556
RAC: 380
United Kingdom
Message 1898684 - Posted: 2 Nov 2017, 14:08:30 UTC - in response to Message 1898334.  

Simple - the nVidia offerings are far better at number crunching than the AMD offerings at (just about) every price point.
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 1898684 · Report as offensive
Dimitar Stoynev
Avatar

Send message
Joined: 7 Jan 09
Posts: 19
Credit: 336,531
RAC: 0
Bulgaria
Message 1898702 - Posted: 2 Nov 2017, 16:03:47 UTC - in response to Message 1898684.  

Simple - the nVidia offerings are far better at number crunching than the AMD offerings at (just about) every price point.


Something tells me the entire alt coin mining community begs to differ.
ID: 1898702 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13855
Credit: 208,696,464
RAC: 304
Australia
Message 1898852 - Posted: 3 Nov 2017, 9:50:31 UTC - in response to Message 1898702.  

Simple - the nVidia offerings are far better at number crunching than the AMD offerings at (just about) every price point.


Something tells me the entire alt coin mining community begs to differ.

This isn't mining.
If you want to mine, but AMD, if you want to crunch Seti, buy Nvidia.
Grant
Darwin NT
ID: 1898852 · Report as offensive
Dimitar Stoynev
Avatar

Send message
Joined: 7 Jan 09
Posts: 19
Credit: 336,531
RAC: 0
Bulgaria
Message 1899090 - Posted: 4 Nov 2017, 14:48:28 UTC - in response to Message 1898852.  

Fanboy-ism and trolling aside, we all know historically AMD cards have (almost) always had better raw computational power in the consumer market. Professional offerings are almost neck-and-neck (drivers and SW support not taken into account).

A GTX580 (stock clocks) has ~1.6 GFLOPS of compute power. RX 580 on the other side has ~6.2 GFLOPS. That is 5 times (!) more raw power.

Do you really believe the alt coin miners would go for 100s of AMD cards if they had a way to make GTX580 profitable?

No way nvidia's GTX580 crunches more numbers than a RX 580, all other factors aside. This means that the key is in the "other factors", e.g. CUDA vs OpenCL, or other cruncher optimizations. Maybe the workload is just too non-typical and AMD cards have no shortcuts crunching it? Maybe I'm simply misinterpreting the chart?

Anyhow, yesterday I got an email from S@H about how much more processing power is needed for the new telescopes and projects. Maybe, just maybe, if some skilled individual(s) spend some time optimizing the code for AMD cards more, we'll get some of that needed power for "free"? Bear in mind however, that I don't really know how many of the AMD owners are contributing to SETI@home, with all the mining craze currently raging. It might be not worth it to optimize further for just a few AMD cards, which would be really sad for me, since my RX is crunching for SETI most of the time.
ID: 1899090 · Report as offensive
1 · 2 · 3 · 4 . . . 20 · Next

Message boards : Number crunching : GPU FLOPS: Theory vs Reality


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.