HCC GPU Launched.

[hendermd]

I am running a HD 7770 and getting between 21 and 24 wus/hr on 30 watts, my HD 5670 does about 8/hr.

[werdwerdus]

I picked up a 7850, will add it in my sig rig #4 for now. Is it possible to have an nvidia and amd card in the same machine? Ok let me rephrase that I know it's POSSIBLE (semantics ) but can you have both drivers running at the same time?

[kjeldoran]

I think we can do some more tweaking to get more WUs in one day. Unlike other projects, knocking off seconds is a huge percentage of overall time.

For example, I saw no noticeable decrease in time increasing core clocks of my 7850 from 900mhz to 950mhz. This most likely has to deal with the fact my GPU usage is pegged at a max of 87%. Inefficiencies in coding or maybe shader use?

However, changing the percentage of CPU usage in boinc to 87% (freeing one thread) decreased my time between 7 and 18 seconds. Yes, you lose one thread for other CPU projects, but you increase the amount of WUs being done by the GPU.

Previous time per WU 2:06, with 87% cpu usage being used it dropped it by an average 10 seconds. A drop in 10 seconds produces and extra 59 WUs. One thread of missing CPU HCC is only about 24WUs on a 4.5ghz 2600k. Losing a CPU thread provides an overall increase in WU completion .

Of course, I only watched this for like 10 minutes. With some daily stats and more tweaking we could maybe drop the overall time of GPU WUs with the right settings.

[Mydog]

I am running a HD 7770 and getting between 21 and 24 wus/hr on 30 watts, my HD 5670 does about 8/hr.

How do you count the WU's??
I tried but cant seem to get it right, my 680L uses 0.05 hour run time on each WU

[Martijn]

I picked up a 7850, will add it in my sig rig #4 for now. Is it possible to have an nvidia and amd card in the same machine? Ok let me rephrase that I know it's POSSIBLE (semantics ) but can you have both drivers running at the same time?

IIRC this should work when one of the cards uses Windows drivers and the other card nVidia or ATI. That was a few years ago though, when 7 just came out.

[Moo]

Looked around but couldn't find an answer. How long before we can gpu crunch on linux?

[bluestang]

7870 doing 28 WU/HR @ 33.2 pts/WU = 22310 BOINC PPD = 156170 PPD WCG

Oh!...and that's with 54 watts more than what I was running.

I had a HD 3650 in my WorkPC (see sig) and the system pulls 168W full load not doing with GPU.

When I swapped GPUs it went to 222W full load running HCC GPU

[haschioz]

if someone is doing a spreadsheet please list cpu type and speed too as it is said (wcg forum) to have quite an impact. and if 7770 also does around 30wu/h its gonna be KILLA BANG FOR BUCK.

[OldChap]

Lifted straight from WCG forums http://www.worldcommunitygrid.org/fo...3962_offset,90

Back to an earlier question in this thread. Has anyone successfully used an app_info.xml to run more than one HCC GPU task at a time?

Yup, on my GTX560Ti 448 setting the thing to run 2 units at the same time will slowly converge to the situation you describe after around 3-4 units and then stays there. It alternates each unit between CPU and GPU so that both are always fully utilized. It essentially doubles the crunching speed - pretty awesome - but runs hotter.

The following is provided as-is, no guarantees that it will not make your GPU explode, burn your cat, kill your wife etc.:
HCC GPU-only x2 appinfo: http://pastebin.com/DJ0kijf3
(You need to manually fetch the files)

<app_info> 
<app>
    <name>hcc1</name>
    <user_friendly_name>Help Conquer Cancer</user_friendly_name>
</app>
<file_info>
    <name>wcg_hcc1_img_6.56_windows_intelx86__nvidia_hcc1</name>
    <nbytes>3499008.000000</nbytes>
    <max_nbytes>0.000000</max_nbytes>
    <status>1</status>
    <executable/>
    <signature_required/>
    <file_signature>
792397d3ec34c91119da45507eb1f8e732452b9d40c1d4adc313cb4d0b6db671
0382de7f034a2dea10b75e2ab2e012b1ce9c3effa738a48f50daa615fa3e6b82
3f4e28d9986968d347552f0298571f3755a601f8834ab9f554d457ea2f0a71c4
843ef8c8b5e4b04633591321627f9a403f12f380bcb5702593f2e42ca0ea5514
.
</file_signature>
</file_info>
<file_info>
    <name>hcckernel.cl.6.56</name>
    <nbytes>63144.000000</nbytes>
    <max_nbytes>0.000000</max_nbytes>
    <status>1</status>
    <signature_required/>
    <file_signature>
a2633e96b0df14406470c5c98e49533b903776e06356cf2dfad1ee853eb15c42
190a29639bed90e5a02c426669ff46441f90f1dad3fe303f05314befeffc28e5
7433da8e2545665f9900ec714a6f9f2d83246de5c32c6c060ac017d1392e0ed6
34839df4b734c410bb5007fe6108d793be40665a4c538654240dc28055bc01fe
.
</file_signature>
</file_info>
<file_info>
    <name>hcc1_image01_6.56.tga</name>
    <nbytes>262188.000000</nbytes>
    <max_nbytes>0.000000</max_nbytes>
    <status>1</status>
    <signature_required/>
    <file_signature>
7117813faaff1bceb99d172432ff67417397b809172fc09c47a0db414106c68d
de9bf88ed64f518e78b5e7ab79637335c9b166003bd3f32a29915101c1758746
d1db7a90bc201c35b1c65fcfb2a9b091253b1979bc6f9db748feb5a6e9d5c804
cd875399b2ee4016fc210f5b76f15aa0902477ce1c0a945fb56c178a0d65de4c
.
</file_signature>
</file_info>
<file_info>
    <name>hcc1_image01_6.56.tga</name>
    <nbytes>262188.000000</nbytes>
    <max_nbytes>0.000000</max_nbytes>
    <status>1</status>
    <signature_required/>
    <file_signature>
7117813faaff1bceb99d172432ff67417397b809172fc09c47a0db414106c68d
de9bf88ed64f518e78b5e7ab79637335c9b166003bd3f32a29915101c1758746
d1db7a90bc201c35b1c65fcfb2a9b091253b1979bc6f9db748feb5a6e9d5c804
cd875399b2ee4016fc210f5b76f15aa0902477ce1c0a945fb56c178a0d65de4c
.
</file_signature>
</file_info>
<file_info>
    <name>hcc1_image02_6.56.tga</name>
    <nbytes>32812.000000</nbytes>
    <max_nbytes>0.000000</max_nbytes>
    <status>1</status>
    <signature_required/>
    <file_signature>
0d3314db1591dcf067aa6551e3582f5274ff10287fb384c11934a98245fd1631
be5249db0ae6d0138b191618ae9502e853a681f879e7fa3c94fec8effdd6685b
a325ec7538ec8abb1fd719629c691bf721da031a87cf69d20b0f501a38e16a92
dbdd7442d90b64f30deff58bc46c5096fb017f4d6410eb8e9bfacb09b2c6a8ae
.
</file_signature>
</file_info>
<file_info>
    <name>hcc1_image03_6.56.tga</name>
    <nbytes>131116.000000</nbytes>
    <max_nbytes>0.000000</max_nbytes>
    <status>1</status>
    <signature_required/>
    <file_signature>
58915b00c9fae53d18ef72f179c6ff1222c134b0c0e696c3bbfddb91c650b26f
c5856311eea2fe2128d9dbdd558b47963332cf9f9ae910689a133b03e7113715
39babe2ce7303b73e9bfc69b1f01cc4199c7dfab13322d5b40adcb8ee167832d
81df23835ba4e12430a50aeb534b8bfed9a82ce54fb0a4018104f1adb36778d9
.
</file_signature>
</file_info>
<file_info>
    <name>hcc1_image04_6.56.tga</name>
    <nbytes>262188.000000</nbytes>
    <max_nbytes>0.000000</max_nbytes>
    <status>1</status>
    <signature_required/>
    <file_signature>
59882941f8d7b844b3ac20301fba70c98ab9cedd5a41855d24b0721b6b6275e6
f1de708d8b43ae7f0198f21924f528c1c710b6d36776d881fee6ebb539cf2f4e
4c4f63997be3eb6157b9a3a5922bdbdebb26fc1cfeae4d13980682800c697ffa
54a7a64cc14694ec457132320f0c2bcc9c3fed675f71cbe388d9ec55332c1bae
.
</file_signature>
</file_info>
<app_version>
    <app_name>hcc1</app_name>
    <version_num>656</version_num>
    <platform>windows_intelx86</platform>
    <avg_ncpus>1.000000</avg_ncpus>
    <max_ncpus>0.783972</max_ncpus>
    <flops>56530729222.498611</flops>
    <plan_class>nvidia_hcc1</plan_class>
    <api_version>7.1.0</api_version>
    <file_ref>
        <file_name>wcg_hcc1_img_6.56_windows_intelx86__nvidia_hcc1</file_name>
        <main_program/>
    </file_ref>
    <file_ref>
        <file_name>hcckernel.cl.6.56</file_name>
        <open_name>hcckernel.cl</open_name>
    </file_ref>
    <file_ref>
        <file_name>hcc1_image01_6.56.tga</file_name>
        <open_name>boinc_wcg_skin_w-logo.tga</open_name>
    </file_ref>
    <file_ref>
        <file_name>hcc1_image02_6.56.tga</file_name>
        <open_name>pbibmablk.tga</open_name>
    </file_ref>
    <file_ref>
        <file_name>hcc1_image03_6.56.tga</file_name>
        <open_name>UHN_stacked2.tga</open_name>
    </file_ref>
    <file_ref>
        <file_name>hcc1_image04_6.56.tga</file_name>
        <open_name>HCC_LOGO.tga</open_name>
    </file_ref>
    <coproc>
        <type>CUDA</type>
        <count>0.5000000</count>
    </coproc>
    <gpu_ram>266338304.000000</gpu_ram>
</app_version>
</app_info>

Perhaps one of you clever b'tards would like to explain what this is and where to put it ???

I am thinking that this will mean that for me the 1m 20s that the GPU is idle out of each 3m 30 wu would be recovered and worked....It may even take the GPU to 100% utilisation

Double points???? 140% at least.....surely!

[Movieman]

Dooh!

Looks like I forgot that is BOINC points, not WCG. So.....22310 x 7 = 156170 PPD WCG

Now I feel better!

WOOT

No, it's

[OldChap]

7870 doing 28 WU/HR @ 33.2 pts/WU = 22310 BOINC PPD = 156170 PPD WCG

Surely that is 33.2 WCG points per wu that is found on WCG device results page????

I reckon on My Super 2600K doing 60pts per hour per thread taken from those statistics....That is ..11520 pts per day...... OK I will shut up now (thinks: Must check first then post later)

THE RESULTS STATISTICS PAGE ON WCG IS BOINC POINTS

[Movieman]

Surely that is 33.2 WCG points per wu that is found on WCG device results page????

Nope, the points you see on WU on your boinc manager web page are BOINC points, look closely..

[D_A]

Perhaps one of you clever b'tards would like to explain what this is and where to put it ???

I COULD make a suggestion but ...

http://boinc.berkeley.edu/wiki/Anonymous_platform

Create a file app_info.xml in the project directory.

I take that to mean the directory /projects/www.worldcommunitygrid.org in your data folder, wherever it is on your machine. It's in /home/michael/BOINC on this machine (Linux, Berkley install)
Hope that helps.

[Movieman]

Application data folder/BOINC I would guess on a Windows machine, same place we put the config file to "return results immediately"

[Otis11]

Lifted straight from WCG forums http://www.worldcommunitygrid.org/fo...3962_offset,90

Perhaps one of you clever b'tards would like to explain what this is and where to put it ???

I am thinking that this will mean that for me the 1m 20s that the GPU is idle out of each 3m 30 wu would be recovered and worked....It may even take the GPU to 100% utilisation

Double points???? 140% at least.....surely!

If I'm reading that right the key part is this:

<coproc>
   <type>CUDA</type>
   <count>0.5000000</count>
</coproc>

That's telling it that any of the WUs that fall into the categories listed above only need half of the CUDA coprocessor - aka 0.5 GPUs. That allows BOINC to overschedule the GPU and just let's the card take care of it - which it actually does a pretty good job of.

I might be wrong though, just skimmed over it in 20 seconds cuz I gotta run to lab now. I'll dig into more depth asap.

Until then someone get to putting a chart together for the GPUs - I need to start planning my christmas list!!!

[Bun-Bun]

I picked up a 7850, will add it in my sig rig #4 for now. Is it possible to have an nvidia and amd card in the same machine? Ok let me rephrase that I know it's POSSIBLE (semantics ) but can you have both drivers running at the same time?

I run Intel onboard with my GTX560 on my work computer. Actually works well that when the GTX560 is crunching the Intel is the one displaying so I see no slow down or choppyness.

Anyway my point is having two different brand cards and drivers running simultaneously works fine. I assume boinc will identify each card individually.

[Fulen]

Installed 2 7770 im my Q9650 3,6MHz

Making points like mad and 21 pages of pending validation

Random timed WU:
CPU ~15 sec
GPU ~65 sec
CPU ~54 sec

Tot 2min 14sec

[Movieman]

Installed 2 7770 im my Q9650 3,6MHz

Making points like mad and 21 pages of pending validation

Random timed WU:
CPU ~15 sec
GPU ~65 sec
CPU ~54 sec

Tot 2min 14sec

Yea, I saw you today.. It was tough at the speed you were moving!

[OldChap]

A bit rough but from comments seen here so far:

card	|	mins	|	secs	|	wu/hr	|	wu/day	|	BOINC	|	WCG

HD5870	|	3	|	30	|	17.1	|	411	|	13659	|	95616

hd7870	|	2	|	9	|	27.9	|	670	|	22236	|	155654

HD7970	|	1	|	30	|	40.0	|	960	|	31872	|	223104

HD7770	|	2	|	14	|	26.9	|	645	|	21407	|	149846

HD7850	|	2	|	6	|	28.6	|	686	|	22766	|	159360

[Mydog]

A bit rough but from comments seen here so far:

card	|	mins	|	secs	|	wu/hour	|	wu/day	|	BOINC per day	|	WCG per day
HD5870	|	3.00	|	30.00	|	17.14	|	411.43	|	13659.43	|	95616.00
hd7870	|	2.00	|	9.00	|	27.91	|	669.77	|	22236.28	|	155653.95
HD7970	|	1.00	|	30.00	|	40.00	|	960.00	|	31872.00	|	223104.00
HD7770	|	2.00	|	40.00	|	22.50	|	540.00	|	17928.00	|	125496.00
HD7850	|	2.00	|	6.00	|	28.57	|	685.71	|	22765.71	|	159360.00

Very nice OldChap now we need some NV figures too

[jspace]

Followed Gamekiller's advice above and switched the only PC I currently have that's capable of running this.
i7-920 @3.7GHz
HD 6950 900MHz core, 1300MHz mem in CCC
Win7 x64
BOINC 7.0.28
Catalyst 12.8

Looks like it's taking ~2.5 minutes/wu.

I've gotta pick up some GPUs!

[Snow Crash]

Application data folder/BOINC I would guess on a Windows machine, same place we put the config file to "return results immediately"

Not quite ...

Note 1: the app_info.xml file goes inside the WCG project folder ... \projects\www.worldcommunitygrid.org

Note 2: Once you drop this file in then you will need to stop and restart BOINC ... and when you do that it will dump all of the tasks you have already downloaded.

Note 3: You are now responsible for updating any of the info in this file regarding programs, files, etc that WCG requires to run the WUs as WCG can no longer automatically update this for you.

I WILL be doing this once I have a little time to set things up properly and I WILL start a thread to help everyone who tries this get up to, and stay up to, speed.

[OldChap]

Very nice OldChap now we need some NV figures too

Anybody with a card running who wants to post the typical runtime they get and on what card I will endeavor to update in a timely fashion. (remember I am an hour ahead of WCG time here so when it is midnght eastern I have been snoring for some time )

I would be interested to see folks post their rig stats too: Say maybe their RAC or typical ppd before and after specifying the card(s) running

[[Ion]]

I COULD make a suggestion but ...

http://boinc.berkeley.edu/wiki/Anonymous_platform



I take that to mean the directory /projects/www.worldcommunitygrid.org in your data folder, wherever it is on your machine. It's in /home/michael/BOINC on this machine (Linux, Berkley install)
Hope that helps.

Is there any way to tweak BOINC to run two tasks per GPU and also run tasks on the CPU? That's what I'd love to do if possible.

[Movieman]

Oh I think this is going to be an interesting end to the year..VERY interesting!