Optimize your GPU. Find the value the easy way.


log in

Advanced search

Message boards : Number crunching : Optimize your GPU. Find the value the easy way.

Previous · 1 . . . 6 · 7 · 8 · 9 · 10 · 11 · 12 . . . 13 · Next
Author Message
Profile Sunny129
Avatar
Send message
Joined: 7 Nov 00
Posts: 190
Credit: 3,163,755
RAC: 0
United States
Message 1280880 - Posted: 7 Sep 2012, 5:20:56 UTC - in response to Message 1280878.

the <count>n</count> statement is the one that controls the number of tasks running in parallel, where n=1 corresponds to 1 task, n=0.5 corresponds to 2 tasks, n=0.33 corresponds to 3 tasks, and so on and so forth...

Well, I know about <count> and I finally found some info on <flops>, but there is stuff in there that I don't entirely understand (for example, what's <avg_ncpus> mean?). It would be nice if the parameters in the app_info file were documented someplace...

I am currently running two SETI tasks in the 520 -- they appear to complete in somewhat less than twice the time for a single task, so It looks like I am coming out ahead. GPU-Z shows 99% GPU Load, 74% Memory Controller Load, 466 MB Memory Used and GPU Temp of 79 deg C when running two SETI enhanced tasks in parallel.

But it would be nice if I could get Fred's test program to run on my machine...

hmm...it appears you have some room to push. i could have sworn that 3 simultaneous tasks would require more than 1GB of VRAM, but 2 simultaneous tasks on your GT 520 appear to be using only 466MB...i must be mixing up SETI w/ some other project as far as VRAM consumption is concerned. so i suppose you could try 3 simultaneous tasks to see if they finish in less than three times the run time of a single task by itself.
____________

Grant (SSSF)
Send message
Joined: 19 Aug 99
Posts: 5683
Credit: 56,026,531
RAC: 49,951
Australia
Message 1280922 - Posted: 7 Sep 2012, 8:19:32 UTC - in response to Message 1280880.
Last modified: 7 Sep 2012, 8:19:56 UTC

so i suppose you could try 3 simultaneous tasks to see if they finish in less than three times the run time of a single task by itself.

If you look at the numbers posted in this thread, you'll see that for most video cards 2 is the number.
Only highend cards get any benefit from 3, on most cards you end up doing less work.
____________
Grant
Darwin NT.

Profile Sunny129
Avatar
Send message
Joined: 7 Nov 00
Posts: 190
Credit: 3,163,755
RAC: 0
United States
Message 1280972 - Posted: 7 Sep 2012, 12:07:37 UTC - in response to Message 1280922.

so i suppose you could try 3 simultaneous tasks to see if they finish in less than three times the run time of a single task by itself.

If you look at the numbers posted in this thread, you'll see that for most video cards 2 is the number.
Only highend cards get any benefit from 3, on most cards you end up doing less work.

yeah, that's what i originally thought and seemed to remember...then i saw that the SETI tasks weren't consuming as much VRAM as i thought they were. then i started to think about all the other projects i'm involved in, and figured i might be confusing SETI's VRAM requirements w/ that of another project...and so i started to second guess myself. but as i mentioned in a previous post, i was only able to run 2 Multibeam tasks at a time on either of my GTX 560 Ti's.
____________

jthon
Volunteer tester
Avatar
Send message
Joined: 12 May 08
Posts: 18
Credit: 11,023,334
RAC: 1,837
United States
Message 1281000 - Posted: 7 Sep 2012, 13:29:02 UTC

For those who use nVIDIA cards that have high temp readings you can adjust the cooling fan speed using the following software from nVIDIA:

http://www.nvidia.com/object/nvidia-system-tools-6.08-driver-th.html

This will not negate the use of 306.02 beta driver for you cards. Works great
My cards went from 70c to 53c in a matter of min. Fan speed max is 70 %.

John

Profile Snowmain
Avatar
Send message
Joined: 17 Nov 05
Posts: 74
Credit: 10,029,796
RAC: 16,543
United States
Message 1281009 - Posted: 7 Sep 2012, 14:08:50 UTC - in response to Message 1281000.

Or you could just download MSI Afterburner and design you own custom fanspeed vs. Gpu temp Curve.

Max fan speed 100%.
____________

jthon
Volunteer tester
Avatar
Send message
Joined: 12 May 08
Posts: 18
Credit: 11,023,334
RAC: 1,837
United States
Message 1281015 - Posted: 7 Sep 2012, 14:30:37 UTC - in response to Message 1281009.
Last modified: 7 Sep 2012, 14:32:29 UTC

Or you could just download MSI Afterburner and design you own custom fanspeed vs. Gpu temp Curve.

Max fan speed 100%.


I do not have an MSI motherboard or MSI graphics card. Will it work on any system?
Besides you need a graphics cards with CHL8318+uP6262 voltage regulators.

John

Profile Sunny129
Avatar
Send message
Joined: 7 Nov 00
Posts: 190
Credit: 3,163,755
RAC: 0
United States
Message 1281018 - Posted: 7 Sep 2012, 14:36:13 UTC - in response to Message 1281015.
Last modified: 7 Sep 2012, 14:39:18 UTC

Or you could just download MSI Afterburner and design you own custom fanspeed vs. Gpu temp Curve.

Max fan speed 100%.


I do not have an MSI motherboard or MSI graphics card. Will it work on any system?

yes, it'll work on just about any system. nevertheless, some people occasionally have trouble with it, and have to resort to an alternative method of GPU monitoring. so thank you for posting a link to the nVidia System Tools w/ SEA Support, as i'm sure some folks will find it to be a viable alternative to MSI Afterburner.
____________

Profile Cliff Harding
Volunteer tester
Avatar
Send message
Joined: 18 Aug 99
Posts: 881
Credit: 49,091,927
RAC: 30,200
United States
Message 1281569 - Posted: 8 Sep 2012, 17:58:06 UTC

Attempted to run the app against a EVGA GTX460SE and got the following message:

Starting automatic test: (x41g)
08 September 2012 - 13:55:21 Start, devices: 1, device count: 1 (1.00)
ERROR: Unable to copy from: D:\My Documents\Downloads\BOINC\BOINC Tasks\x41g\ to: C:\Users\Cliff Harding\AppData\Roaming\eFMer\SetiPerformance, slot: 0
Aborted

____________


I don't buy computers, I build them!!

Profile Alex Storey
Volunteer tester
Avatar
Send message
Joined: 14 Jun 04
Posts: 534
Credit: 1,627,204
RAC: 590
Greece
Message 1281613 - Posted: 8 Sep 2012, 19:41:42 UTC - in response to Message 1280834.

@Snowmain

Another awesome graph, thanx! One thing I noticed is it's around a dollar a Watt per year. Crude, but easy to remember:)

w1hue
Volunteer tester
Send message
Joined: 4 Aug 00
Posts: 48
Credit: 1,675,875
RAC: 1,412
United States
Message 1281731 - Posted: 9 Sep 2012, 0:18:42 UTC - in response to Message 1280858.

This brings up another question: Since there is no Lunatics NVIDIA GPU app for Astropulse (as of yet...), will my machine run GPU Astropulse WU's using the standard SETI app, or do I need to add something the the app_info file? And if so, what? The answer may be out there somewhere, but a search here and on the Lunatics site hasn't turned it up...

yes, once you enable AP tasks via your web preferences, your host should eventually download the stock nVidia OpenCL Astropulse binaries, and tasks will of course follow when they become available. you really only need an entry for it in the app_info.xml if you want to run multiple tasks in parallel/increase GPU utilization/decrease CPU utilization/mitigate GUI lag/etc.

Well, that does not appear to be the case as I received the following message:

SETI@home: notice from server
your app_info.xml file doesn't have a usable entry of AstroPulse V6

So obviously, I need to add entry(s) the the app_info file -- but what?? I think that I will start a new thread on this and see if anyone can help...
____________

Profile Snowmain
Avatar
Send message
Joined: 17 Nov 05
Posts: 74
Credit: 10,029,796
RAC: 16,543
United States
Message 1281756 - Posted: 9 Sep 2012, 1:44:15 UTC - in response to Message 1281613.
Last modified: 9 Sep 2012, 2:40:24 UTC

@Snowmain

Another awesome graph, thanx! One thing I noticed is it's around a dollar a Watt per year. Crude, but easy to remember:)


Yup 1 to 1 ruffly when your power cost 12 cents a killo watt hour.
Which I think is pritty average...right?(shrug)

Anyhow your welcome :) Glad to do it.

I pm'ed about 30 other members with unique video cards with links to this thread.

edit: My power cost .1596 per killowatt hour.

____________

Profile Sunny129
Avatar
Send message
Joined: 7 Nov 00
Posts: 190
Credit: 3,163,755
RAC: 0
United States
Message 1281757 - Posted: 9 Sep 2012, 2:02:47 UTC - in response to Message 1281731.

This brings up another question: Since there is no Lunatics NVIDIA GPU app for Astropulse (as of yet...), will my machine run GPU Astropulse WU's using the standard SETI app, or do I need to add something the the app_info file? And if so, what? The answer may be out there somewhere, but a search here and on the Lunatics site hasn't turned it up...

yes, once you enable AP tasks via your web preferences, your host should eventually download the stock nVidia OpenCL Astropulse binaries, and tasks will of course follow when they become available. you really only need an entry for it in the app_info.xml if you want to run multiple tasks in parallel/increase GPU utilization/decrease CPU utilization/mitigate GUI lag/etc.

Well, that does not appear to be the case as I received the following message:

SETI@home: notice from server
your app_info.xml file doesn't have a usable entry of AstroPulse V6

So obviously, I need to add entry(s) the the app_info file -- but what?? I think that I will start a new thread on this and see if anyone can help...

yeah, i wasn't thinking about the fact that you were already running an app_info.xml to pass parameters to the nVidia Multibeam app. b/c you are employing an app_info.xml file, your host will only download work for the apps specified in the app_info.xml. so i stand corrected - you must have an entry for even the stock nVidia OpenCL Astropulse app since you're already using an app_info.xml. at any rate, i agree that its best to start a new thread since that's not the topic of this thread. that being said, if you just cut & paste the section of code i posted above into your app_info.xml, you should be good to go. i wouldn't worry about the <avg_ncpus>0.04</avg_ncpus> and <max_ncpus>0.2</max_ncpus> parameter values, as these have been pretty well established.. the only thing you need to manipulate is <count>. the executable of the app itself is the same despite my running Windows7 x64 and your running WinXP x32, so that doesn't need to be changed either.
____________

w1hue
Volunteer tester
Send message
Joined: 4 Aug 00
Posts: 48
Credit: 1,675,875
RAC: 1,412
United States
Message 1281772 - Posted: 9 Sep 2012, 4:55:02 UTC - in response to Message 1281757.

that being said, if you just cut & paste the section of code i posted above into your app_info.xml, you should be good to go. i wouldn't worry about the <avg_ncpus>0.04</avg_ncpus> and <max_ncpus>0.2</max_ncpus> parameter values, as these have been pretty well established.. the only thing you need to manipulate is <count>. the executable of the app itself is the same despite my running Windows7 x64 and your running WinXP x32, so that doesn't need to be changed either.

OK - thanks! The executable name is not the same as in my oldApp_backup folder, but I'll leave it as copied from your post and see what happens. I set <count> to 1.0 -- still not clear if 0.5 will work with the stock app, but I can experiment with that later. I added <flops>34000000000</flops> after <max_ncpus>0.2</max_ncpus> as that resulted in a reasonable time estimate for the setiathome_enhanced WUs.

____________

pester500
Send message
Joined: 6 Jul 04
Posts: 2
Credit: 36,302,684
RAC: 6,745
United States
Message 1282147 - Posted: 10 Sep 2012, 5:41:42 UTC - in response to Message 1281756.

This is a great app! I definitely noticed about a 6K bump in my RAC after adjusting my app_info.xml.

My config:
Win7 64-bit, Core i7-2600K @ 4.2 GHz
2x GeForce GTX 570 (no OC)

Output from x41g:

Starting automatic test: (x41g)
10 September 2012 - 00:22:06 Start, devices: 1, device count: 1 (1.00)
--------------------------------------------------------------------------- Results: Device: 0, device count: 1, average time / count: 127, average time on device: 127 Seconds (2 Minutes, 7 Seconds)
Next :---------------------------------------------------------------------------
10 September 2012 - 00:24:15 Start, devices: 1, device count: 2 (0.50)
--------------------------------------------------------------------------- Results: Device: 0, device count: 2, average time / count: 218, average time on device: 109 Seconds (1 Minutes, 49 Seconds)
Next :---------------------------------------------------------------------------
10 September 2012 - 00:27:56 Start, devices: 1, device count: 3 (0.33)
--------------------------------------------------------------------------- Results: Device: 0, device count: 3, average time / count: 349, average time on device: 116 Seconds (1 Minutes, 56 Seconds)
>> The best average time found: 109 Seconds (1 Minutes, 49 Seconds), with count: 0.50 (2)
____________

Doc's Brains
Send message
Joined: 11 Jun 12
Posts: 26
Credit: 752,217
RAC: 0
Canada
Message 1282440 - Posted: 10 Sep 2012, 23:02:57 UTC

Just a quick question, I have just updated my comps to crunch with Lunatics programs, and I am wondering if the program that is being talked about in this thread would be a good thing for me to use as well. I have an I5 laptop with the 525M Nvidia GPU?

Is there a similar program that I can use on my older P4's( I have 2) that have Nvidia's 6200 GPU's installed in them?

Any response will be greatly appreciated.

Cheers!
____________

Doc's Brains
Send message
Joined: 11 Jun 12
Posts: 26
Credit: 752,217
RAC: 0
Canada
Message 1282473 - Posted: 11 Sep 2012, 2:13:17 UTC - in response to Message 1282440.

Please disregard the previous post \i made with regards to using the program, I have ran the program after reading further, and here are my results;

Toshiba Satellite L770
Intel I5 2450m @ 2.5Ghz Sandy Bridge
NVidia GeForce GT 525M
Driver version nvlddmkm 8.17.13.142(Force Ware 301.42
Windows7 64 bit GPU clock; 475Mhz, Memory; 900Mhz, Shader 950Mhz



Starting automatic test: (x41g)
10 September 2012 - 18:17:04 Start, devices: 1, device count: 1 (1.00)
--------------------------------------------------------------------------- Results: Device: 0, device count: 1, average time / count: 981, average time on device: 981 Seconds (16 Minutes, 21 Seconds)
Next :---------------------------------------------------------------------------
10 September 2012 - 18:33:28 Start, devices: 1, device count: 2 (0.50)
--------------------------------------------------------------------------- Results: Device: 0, device count: 2, average time / count: 1560, average time on device: 780 Seconds (13 Minutes, 0 Seconds)
Next :---------------------------------------------------------------------------
10 September 2012 - 18:59:34 Start, devices: 1, device count: 3 (0.33)
--------------------------------------------------------------------------- Results: Device: 0, device count: 3, average time / count: 2338, average time on device: 779 Seconds (12 Minutes, 59 Seconds)
>> The best average time found: 780 Seconds (13 Minutes, 0 Seconds), with count: 0.50 (2)

With the testing now done, I know it's in my best interest to allow a device count of 2. My question now is how do I take this information and utilize it, in other words, how do I change the GPU to allow 2 instances to run at the same time?

Is there a link or 2 that I can read, that will help me learn how to do this? Reading prior posts in this thread, some coding is probably involved, and while I don't know how to do this on my own, I can learn real fast, cut and paste with the best, and can follow directory instructions without getting lost.

I know I have been needy in the last week or so, asking for help to accomplish these tasks, but I feel we're almost at an end of what I can do in to improve my computers behaviour when completing S@H tasks.

Once again, Thanks for your kind words of knowledge and your willingness to share them with me, it really is appreciated.

Cheers!
____________

Profile BilBg
Volunteer tester
Avatar
Send message
Joined: 27 May 07
Posts: 2560
Credit: 5,854,462
RAC: 2,470
Bulgaria
Message 1282512 - Posted: 11 Sep 2012, 4:46:35 UTC - in response to Message 1282473.


Go again at the link/page where you get Seti Performance

There is short explanation, read after:
"Now that I found out that a count of 2 (0.5) is the best setting, what now? ..."


____________



- ALF - "Find out what you don't do well ..... then don't do it!" :)

Profile BilBg
Volunteer tester
Avatar
Send message
Joined: 27 May 07
Posts: 2560
Credit: 5,854,462
RAC: 2,470
Bulgaria
Message 1282687 - Posted: 11 Sep 2012, 22:43:32 UTC



ATI tests using SetiPerformance 1.3

I think this is the first report for ATI tests (I managed to make it run both MB and AP ATI tests)

If you want to get this combined experimental package (all files from the original + ATI tests added by me) use the link bellow:
(I hope Fred will not be angry that I supply this)

SetiPerformance_32_64_1_3__ATI-added-by-BilBg.7z - 17 MB:
http://www.datafilehost.com/download-04a5fb4f.html


---===---===---

Before you run the tests:

* A) Look in config.xml and change the 'parameter' for MB/AP if you wish.

As you can see I use:
-period_iterations_num 20 (which is the default in the app_info.xml created by Lunatics installer)
-unroll 3 (which I choose for my AMD Radeon HD 6570 which have 'Number of compute units: 6'; if no -unroll is specified the .exe by (internal) default uses -unroll 2)

<parameter>%20--device%20%device%20-instances_per_device%204%20-period_iterations_num%2020%20</parameter>
<parameter>%20--device%20%device%20-unroll%203%20</parameter>

(The additional 'Space' char %20 at the end is intentional - to avoid error if you add something in the 'Parameters' field (in the program window) and forget to add initial Space ' ' there)


If you want e.g.
-period_iterations_num 44
-unroll 11

1) You may edit to:
<parameter>%20--device%20%device%20-instances_per_device%204%20-period_iterations_num%2044%20</parameter>
<parameter>%20--device%20%device%20-unroll%2011%20</parameter>


2) Or you may delete to:
<parameter>%20--device%20%device%20-instances_per_device%204%20</parameter>
<parameter>%20--device%20%device%20</parameter>

... and paste the options/parameters/cmdline:

-period_iterations_num 44 -unroll 11
in the 'Parameters' field (additional spaces are good, no space is bad).
(I think the .exe will just ignore any cmdline parameter it do not understand, e.g. MB app will ignore -unroll)


* B) For AP ATI tests: to get better/real timings - use 'bin' files compiled for your GPU:
- you need to do the following only once for a given GPU:
Start the AP test
Wait ~1 minute (or until you see high GPU load)
run Open Slots.cmd
go to slot0
copy all (2) 'bin' files from slot0 to:
...\SetiPerformance_32_64_1_3__ATI-added-by-BilBg\zip\AP6_win_x86_SSE2_OpenCL_ATI_r1316\

[Cancel] the test


(Now you can start SetiPerformance and [Test ...] again many times - the AP app will now use the prepared 'bin' files.
You can have a mix of 'bin' files (in zip\AP6_win_x86_SSE2_OpenCL_ATI_r1316\) compiled for different GPUs.

The work_unit.sah file is there only to fool SetiPerformance to run ;)
The real AP WU is in.dat
)


! Don't forget to exit BOINC before tests or at least Snooze GPU !


---===---===---


AP6_win_x86_SSE2_OpenCL_ATI_r1316:




For my AMD Radeon HD 6570 there is no difference in heat generation for 1...3 AP ATI tasks:




MB6_win_x86_SSE3_OpenCL_ATi_HD5_r390:





Judging from the times and lag and possible errors I decided to try (in BOINC/SETI/app_info.xml):
For AP ATI: <count>0.48</count>
For MB ATI: <count>0.51</count>

This way BOINC will run either: 1 MB, MB+AP, AP+AP
But these variations in usage cause DCF to fluctuate 1.0 ... 2.8
(runs this way ~12 hours now in MB+AP 'variant')

***

Test logs (AMD Radeon HD 6570, Win XP32, Catalyst 11.12):

For AP ATI:

Starting automatic test: (ap6_win_x86_sse2_opencl_ati_r1316)

10 September 2012 - 20:08:27 Start, devices: 1, device count: 1 (1.00)
10 September 2012 - 20:08:27 Set: ap6_win_x86_sse2_opencl_ati_r1316 , Exe: ap6_win_x86_sse2_opencl_ati_r1316.exe , Parameters: --device %device -unroll 3
10 September 2012 - 20:21:06 Runtime: Device: 0, count: 0, 758 seconds
10 September 2012 - 20:21:06 Device: 0, Count: 0, finished.
---------------------------------------------------------------------------
Results: Device: 0, device count: 1, average time / count: 758, average time on device: 758 Seconds (12 Minutes, 38 Seconds)
Next :---------------------------------------------------------------------------
10 September 2012 - 20:21:06 Start, devices: 1, device count: 2 (0.50)
10 September 2012 - 20:21:06 Set: ap6_win_x86_sse2_opencl_ati_r1316 , Exe: ap6_win_x86_sse2_opencl_ati_r1316.exe , Parameters: --device %device -unroll 3
10 September 2012 - 20:45:20 Runtime: Device: 0, count: 0, 1452 seconds
10 September 2012 - 20:45:20 Device: 0, Count: 0, finished.
10 September 2012 - 20:45:26 Runtime: Device: 0, count: 1, 1458 seconds
10 September 2012 - 20:45:26 Device: 0, Count: 1, finished.
---------------------------------------------------------------------------
Results: Device: 0, device count: 2, average time / count: 1455, average time on device: 727 Seconds (12 Minutes, 7 Seconds)
Next :---------------------------------------------------------------------------
10 September 2012 - 20:45:26 Start, devices: 1, device count: 3 (0.33)
10 September 2012 - 20:45:26 Set: ap6_win_x86_sse2_opencl_ati_r1316 , Exe: ap6_win_x86_sse2_opencl_ati_r1316.exe , Parameters: --device %device -unroll 3
10 September 2012 - 21:21:24 Runtime: Device: 0, count: 2, 2156 seconds
10 September 2012 - 21:21:24 Device: 0, Count: 2, finished.
10 September 2012 - 21:21:32 Runtime: Device: 0, count: 1, 2164 seconds
10 September 2012 - 21:21:32 Device: 0, Count: 1, finished.
10 September 2012 - 21:21:33 Runtime: Device: 0, count: 0, 2165 seconds
10 September 2012 - 21:21:33 Device: 0, Count: 0, finished.
---------------------------------------------------------------------------
Results: Device: 0, device count: 3, average time / count: 2161, average time on device: 720 Seconds (12 Minutes, 0 Seconds)
Next :---------------------------------------------------------------------------
10 September 2012 - 21:21:34 Start, devices: 1, device count: 4 (0.25)
10 September 2012 - 21:21:34 Set: ap6_win_x86_sse2_opencl_ati_r1316 , Exe: ap6_win_x86_sse2_opencl_ati_r1316.exe , Parameters: --device %device -unroll 3
10 September 2012 - 22:08:26 Runtime: Device: 0, count: 1, 2809 seconds
10 September 2012 - 22:08:26 Device: 0, Count: 1, finished.
10 September 2012 - 22:08:36 Runtime: Device: 0, count: 3, 2819 seconds
10 September 2012 - 22:08:36 Device: 0, Count: 3, finished.
10 September 2012 - 22:08:47 Runtime: Device: 0, count: 0, 2830 seconds
10 September 2012 - 22:08:47 Device: 0, Count: 0, finished.
10 September 2012 - 22:09:00 Runtime: Device: 0, count: 2, 2843 seconds
10 September 2012 - 22:09:00 Device: 0, Count: 2, finished.
---------------------------------------------------------------------------
Results: Device: 0, device count: 4, average time / count: 2825, average time on device: 706 Seconds (11 Minutes, 46 Seconds)

>> The best average time found: 720 Seconds (12 Minutes, 0 Seconds), with count: 0.33 (3)


For MB ATI:

Starting automatic test: (mb6_win_x86_sse3_opencl_ati_hd5_r390)

10 September 2012 - 22:47:32 Start, devices: 1, device count: 1 (1.00)
10 September 2012 - 22:47:32 Set: mb6_win_x86_sse3_opencl_ati_hd5_r390 , Exe: mb6_win_x86_sse3_opencl_ati_hd5_r390.exe , Parameters: --device %device -instances_per_device 4 -period_iterations_num 20
10 September 2012 - 23:04:00 Runtime: Device: 0, count: 0, 986 seconds
10 September 2012 - 23:04:00 Device: 0, Count: 0, finished.
---------------------------------------------------------------------------
Results: Device: 0, device count: 1, average time / count: 986, average time on device: 986 Seconds (16 Minutes, 26 Seconds)
Next :---------------------------------------------------------------------------
10 September 2012 - 23:04:01 Start, devices: 1, device count: 2 (0.50)
10 September 2012 - 23:04:01 Set: mb6_win_x86_sse3_opencl_ati_hd5_r390 , Exe: mb6_win_x86_sse3_opencl_ati_hd5_r390.exe , Parameters: --device %device -instances_per_device 4 -period_iterations_num 20
10 September 2012 - 23:36:29 Runtime: Device: 0, count: 0, 1947 seconds
10 September 2012 - 23:36:29 Device: 0, Count: 0, finished.
10 September 2012 - 23:36:30 Runtime: Device: 0, count: 1, 1948 seconds
10 September 2012 - 23:36:30 Device: 0, Count: 1, finished.
---------------------------------------------------------------------------
Results: Device: 0, device count: 2, average time / count: 1947, average time on device: 973 Seconds (16 Minutes, 13 Seconds)
Next :---------------------------------------------------------------------------
10 September 2012 - 23:36:32 Start, devices: 1, device count: 3 (0.33)
10 September 2012 - 23:36:32 Set: mb6_win_x86_sse3_opencl_ati_hd5_r390 , Exe: mb6_win_x86_sse3_opencl_ati_hd5_r390.exe , Parameters: --device %device -instances_per_device 4 -period_iterations_num 20
10 September 2012 - 23:39:06 Runtime: Device: 0, count: 1, 153 seconds
10 September 2012 - 23:39:06 ERROR: Device: 0, Count: 1, finished.

***

Starting test: (mb6_win_x86_sse3_opencl_ati_hd5_r390)

11 September 2012 - 02:02:31 Start, devices: 1, device count: 3 (0.33)
11 September 2012 - 02:02:31 Set: mb6_win_x86_sse3_opencl_ati_hd5_r390 , Exe: mb6_win_x86_sse3_opencl_ati_hd5_r390.exe , Parameters: --device %device -instances_per_device 4 -period_iterations_num 20
11 September 2012 - 02:05:13 Runtime: Device: 0, count: 2, 153 seconds
11 September 2012 - 02:05:13 ERROR: Device: 0, Count: 2, finished.
11 September 2012 - 02:05:13 Runtime: Device: 0, count: 0, 158 seconds
11 September 2012 - 02:05:13 ERROR: Device: 0, Count: 0, finished.


*** 3 MB tasks on my AMD Radeon HD 6570 cause ERROR

(1 or 2 of the r390 processes really exited (as shown in Process Explorer).

The remaining run very slow (monitor <prog>0.XXXX</prog> in state.sah), but I see high >90% GPU load (in SIV) and I feel lag.
GPU temp (in TThrottle) lower than usual for MB (44°C instead of ~50°C)
VRAM usage (shown by old version of ATI MemoryViewer) is too low (<50 MB, normal VRAM usage for one r390 task is 220-250 MB)

I aborted the test after ~1 hour by [Cancel] button.
)



____________



- ALF - "Find out what you don't do well ..... then don't do it!" :)

Profile Mike
Volunteer tester
Avatar
Send message
Joined: 17 Feb 01
Posts: 23296
Credit: 31,631,793
RAC: 24,107
Germany
Message 1282697 - Posted: 11 Sep 2012, 23:42:06 UTC

Will have a look as soon i got some time.

Thank you.

____________

Profile Sunny129
Avatar
Send message
Joined: 7 Nov 00
Posts: 190
Credit: 3,163,755
RAC: 0
United States
Message 1282702 - Posted: 12 Sep 2012, 0:36:10 UTC - in response to Message 1282687.


Results: Device: 0, device count: 3, average time / count: 2161, average time on device: 720 Seconds (12 Minutes, 0 Seconds)

Results: Device: 0, device count: 4, average time / count: 2825, average time on device: 706 Seconds (11 Minutes, 46 Seconds)

>> The best average time found: 720 Seconds (12 Minutes, 0 Seconds), with count: 0.33 (3)

[scratches head]hmmm....[/scratches head]
____________

Previous · 1 . . . 6 · 7 · 8 · 9 · 10 · 11 · 12 . . . 13 · Next

Message boards : Number crunching : Optimize your GPU. Find the value the easy way.

Copyright © 2014 University of California