Optimize your GPU. Find the value the easy way.

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

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4 · 5 · 6 · 7 · 8 . . . 13 · Next

AuthorMessage
Profile S@NL - eFMer - efmer.com/boinc
Volunteer tester
Avatar

Send message
Joined: 7 Jun 99
Posts: 512
Credit: 148,746,305
RAC: 0
United States
Message 1277102 - Posted: 29 Aug 2012, 15:09:05 UTC - in response to Message 1277087.  

Thanks Fred.

-->
</config>

BTW, is this test system specific or does it give a general return for any/all cards?

This comment end isn't a match with a start, but I don't think the parser will complain.

The program isn't specific to any cards. Even ATI cards should work, with the right exe supplied.

As there are plenty of users who got this to work, I'm not sure what' causing this.

Win 7 ?
TThrottle Control your temperatures. BoincTasks The best way to view BOINC. Anza Borrego Desert hiking.
ID: 1277102 · Report as offensive
Russ Sheldon
Avatar

Send message
Joined: 18 Jan 01
Posts: 382
Credit: 530,470,939
RAC: 0
United States
Message 1277118 - Posted: 29 Aug 2012, 15:47:25 UTC - in response to Message 1277102.  
Last modified: 29 Aug 2012, 15:50:20 UTC

Yeah, I'm on Win7 64. To reiterate, what's odd to me is that the 41g code works fine but the 41z can't find the work_unit.sah file, AND their are plenty of other people that can seem to get it to work properly.
What I meant by card specific is will my GTXxxx card perform the same in MY system as your exact same GTXxxx card will perform in your system, or is performance system specific?
Proud member of the half bill club!
ID: 1277118 · Report as offensive
Profile S@NL - eFMer - efmer.com/boinc
Volunteer tester
Avatar

Send message
Joined: 7 Jun 99
Posts: 512
Credit: 148,746,305
RAC: 0
United States
Message 1277140 - Posted: 29 Aug 2012, 16:11:51 UTC - in response to Message 1277118.  

Yeah, I'm on Win7 64. To reiterate, what's odd to me is that the 41g code works fine but the 41z can't find the work_unit.sah file, AND their are plenty of other people that can seem to get it to work properly.
What I meant by card specific is will my GTXxxx card perform the same in MY system as your exact same GTXxxx card will perform in your system, or is performance system specific?

My test systems are the same.
The folder from the example should be x41z_64_cuda42 all lower case!
You could also try the 32 bit version included, but they should be identical.
But in any case don't expect any speed improvements in the 64 bit cuda version.
TThrottle Control your temperatures. BoincTasks The best way to view BOINC. Anza Borrego Desert hiking.
ID: 1277140 · Report as offensive
Russ Sheldon
Avatar

Send message
Joined: 18 Jan 01
Posts: 382
Credit: 530,470,939
RAC: 0
United States
Message 1277158 - Posted: 29 Aug 2012, 16:27:37 UTC - in response to Message 1277140.  

AAAAAAAAAAAAAAAAAAAAARRRRRRRRRRRRRRRRRRRGGGGGGGGGGGGGGGGGGGGGGHHHHHHHHHHHHHH!!!!

Yep, right before I saw your reply, I studied the config code and realized that it was calling <folder>x41z_64_cuda42</folder>. What I had done was to match the x41g folder naming my folder x41z instead of x41z_64_cuda42 so obviously (NOW) it was calling a folder that did not exist and could not find the .sah file. DDDDDUUUUUUUUHHHHHHHHHHHHHHH!!!

For others having the same problem as this dumbass:

<config>
<data>
<name>x41g</name>
<folder>x41g</folder>
<exe>Lunatics_x41g_win32_cuda32_perf.exe</exe>
<parameter>%20--device%20%device</parameter>
</data>

<!-- Some more settings: You can add additional data folders as shown below -->
<!-- Additional gui options are also available for experts -->


<data>
<name>x41z%20cuda%2032</name>
<folder>x41z</folder>
<exe>Lunatics_x41z_win32_cuda32_perf.exe</exe>
<parameter>%20--device%20%device</parameter>
</data>
<data>
<name>x41z%20cuda%2042</name>
<folder>x41z_cuda42</folder>
<exe>Lunatics_x41z_win32_cuda42.exe</exe>
<parameter>%20--device%20%device</parameter>
</data>
<data>
<name>x41z%2064%20bit%20cuda%2042</name>
<folder>x41z_64_cuda42</folder>
<exe>Lunatics_x41z_win32_cuda42.exe</exe>
<parameter>%20--device%20%device</parameter>
</data>

<gui>
<showParameters>1</showParameters>
<showReadXmlButton>1</showReadXmlButton>
<maxAutoCount>3</maxAutoCount>
<showExpert>1</showexpert>
</gui>

</config>


<!--

Make sure you open this file in a simple text editor, like notepad.exe only.
This file has to be stored in ascii format.

%20 = space

<data> Is the selection data that shows up in the combi box. (more than 1 item)

<parameter> %device = device number %device replaced by 0, 1, 2, 3, 4
<showParameters> 1 = show field where you can put in additional parameters. This value is added to <parameter> in <data>. Make sure to add an additional space.
e.g. <parameter>%20--device%20%device and parameter field --test send to the application -> --device 0 --test
<showexpert> shows additional information in the screen log.
<maxAutoCount> 2 = 0.5, 3 = 0.33, 4 = 0.25 So fill in 1 or 2 or 3 or 4

-->



Depending upon which flavor _z you are using, choose the corresponding config block and name your folder the same so config can find the .sah file. Seems pretty simple now. DAMN!
Proud member of the half bill club!
ID: 1277158 · Report as offensive
Russ Sheldon
Avatar

Send message
Joined: 18 Jan 01
Posts: 382
Credit: 530,470,939
RAC: 0
United States
Message 1277201 - Posted: 29 Aug 2012, 17:24:38 UTC - in response to Message 1277158.  

Finally, now that I got this jewel to run, am I reading the figures correctly?
Best bet would be 0.5 or 0.33?



Starting automatic test: (x41z 64 bit cuda 42)
29 August 2012 - 12:11:00 Start, devices: 2, device count: 1 (1.00)
29 August 2012 - 12:11:00 Set: x41z 64 bit cuda 42 , Exe: lunatics_x41z_win32_cuda42.exe , Parameters: --device %devicedevice 0
29 August 2012 - 12:12:50 Runtime: Device: 1, count: 0, 108 seconds
29 August 2012 - 12:12:50 Device: 1, Count: 0, finished.
29 August 2012 - 12:12:51 Runtime: Device: 0, count: 0, 109 seconds
29 August 2012 - 12:12:51 Device: 0, Count: 0, finished.
Ready ---------------------------------------------------------------------
Results:
Device: 0, device count: 1, average time / count: 109, average time on device: 109 Seconds (1 Minutes, 49 Seconds)
Device: 1, device count: 1, average time / count: 108, average time on device: 108 Seconds (1 Minutes, 48 Seconds)
Next ---------------------------------------------------------------------
29 August 2012 - 12:12:53 Start, devices: 2, device count: 2 (0.50)
29 August 2012 - 12:12:53 Set: x41z 64 bit cuda 42 , Exe: lunatics_x41z_win32_cuda42.exe , Parameters: --device %devicedevice 0
29 August 2012 - 12:16:04 Runtime: Device: 1, count: 0, 190 seconds
29 August 2012 - 12:16:04 Device: 1, Count: 0, finished.
29 August 2012 - 12:16:05 Runtime: Device: 1, count: 1, 191 seconds
29 August 2012 - 12:16:05 Device: 1, Count: 1, finished.
29 August 2012 - 12:16:08 Runtime: Device: 0, count: 1, 194 seconds
29 August 2012 - 12:16:08 Runtime: Device: 0, count: 0, 192 seconds
29 August 2012 - 12:16:08 Device: 0, Count: 1, finished.
29 August 2012 - 12:16:08 Device: 0, Count: 0, finished.
Ready ---------------------------------------------------------------------
Results:
Device: 0, device count: 2, average time / count: 193, average time on device: 96 Seconds (1 Minutes, 36 Seconds)
Device: 1, device count: 2, average time / count: 190, average time on device: 95 Seconds (1 Minutes, 35 Seconds)
Next ---------------------------------------------------------------------
29 August 2012 - 12:16:10 Start, devices: 2, device count: 3 (0.33)
29 August 2012 - 12:16:10 Set: x41z 64 bit cuda 42 , Exe: lunatics_x41z_win32_cuda42.exe , Parameters: --device %devicedevice 0
29 August 2012 - 12:20:48 Runtime: Device: 1, count: 2, 274 seconds
29 August 2012 - 12:20:48 Device: 1, Count: 2, finished.
29 August 2012 - 12:20:50 Runtime: Device: 1, count: 1, 276 seconds
29 August 2012 - 12:20:50 Device: 1, Count: 1, finished.
29 August 2012 - 12:20:51 Runtime: Device: 1, count: 0, 277 seconds
29 August 2012 - 12:20:51 Device: 1, Count: 0, finished.
29 August 2012 - 12:20:54 Runtime: Device: 0, count: 0, 280 seconds
29 August 2012 - 12:20:54 Device: 0, Count: 0, finished.
29 August 2012 - 12:20:56 Runtime: Device: 0, count: 2, 282 seconds
29 August 2012 - 12:20:56 Device: 0, Count: 2, finished.
29 August 2012 - 12:20:56 Runtime: Device: 0, count: 1, 282 seconds
29 August 2012 - 12:20:56 Device: 0, Count: 1, finished.
Ready ---------------------------------------------------------------------
Results:
Device: 0, device count: 3, average time / count: 281, average time on device: 93 Seconds (1 Minutes, 33 Seconds)
Device: 1, device count: 3, average time / count: 275, average time on device: 91 Seconds (1 Minutes, 31 Seconds)
Next ---------------------------------------------------------------------
29 August 2012 - 12:20:58 Start, devices: 2, device count: 4 (0.25)
29 August 2012 - 12:20:58 Set: x41z 64 bit cuda 42 , Exe: lunatics_x41z_win32_cuda42.exe , Parameters: --device %devicedevice 0
29 August 2012 - 12:27:06 Runtime: Device: 1, count: 3, 366 seconds
29 August 2012 - 12:27:06 Device: 1, Count: 3, finished.
29 August 2012 - 12:27:09 Runtime: Device: 1, count: 1, 369 seconds
29 August 2012 - 12:27:09 Device: 1, Count: 1, finished.
29 August 2012 - 12:27:10 Runtime: Device: 1, count: 2, 370 seconds
29 August 2012 - 12:27:10 Runtime: Device: 1, count: 0, 370 seconds
29 August 2012 - 12:27:10 Device: 1, Count: 0, finished.
29 August 2012 - 12:27:10 Device: 1, Count: 2, finished.
29 August 2012 - 12:27:15 Runtime: Device: 0, count: 1, 375 seconds
29 August 2012 - 12:27:15 Device: 0, Count: 1, finished.
29 August 2012 - 12:27:16 Runtime: Device: 0, count: 2, 376 seconds
29 August 2012 - 12:27:16 Runtime: Device: 0, count: 0, 376 seconds
29 August 2012 - 12:27:16 Device: 0, Count: 2, finished.
29 August 2012 - 12:27:16 Device: 0, Count: 0, finished.
29 August 2012 - 12:27:16 Runtime: Device: 0, count: 3, 376 seconds
29 August 2012 - 12:27:16 Device: 0, Count: 3, finished.
Ready ---------------------------------------------------------------------
Results:
Device: 0, device count: 4, average time / count: 375, average time on device: 93 Seconds (1 Minutes, 33 Seconds)
Device: 1, device count: 4, average time / count: 368, average time on device: 92 Seconds (1 Minutes, 32 Seconds)
The best average time found: 92 Seconds (1 Minutes, 32 Seconds), with count: 0.33 (3





Proud member of the half bill club!
ID: 1277201 · Report as offensive
Profile Snowmain
Avatar

Send message
Joined: 17 Nov 05
Posts: 75
Credit: 30,681,449
RAC: 83
United States
Message 1277388 - Posted: 30 Aug 2012, 3:07:30 UTC
Last modified: 30 Aug 2012, 4:05:08 UTC

Another data point. An oldy.

Overclocked 8800 ultra 612/666Mhz 1100/1152Mhz mem
Starting test: (x41g)
29 August 2012 - 22:49:26 Start, devices: 1, device count: 2 (0.50)
29 August 2012 - 22:59:29 Runtime: Device: 0, count: 1, 601 seconds
29 August 2012 - 22:59:29 Device: 0, Count: 1, finished.
29 August 2012 - 22:59:30 Runtime: Device: 0, count: 0, 602 seconds
29 August 2012 - 22:59:30 Device: 0, Count: 0, finished.
Ready ---------------------------------------------------------------------
Results:
Device: 0, device count: 2, average time / count: 601, average time on device: 300 Seconds (5 Minutes, 0 Seconds)

Starting automatic test: (x41g)
29 August 2012 - 23:01:08 Start, devices: 1, device count: 1 (1.00)
29 August 2012 - 23:06:06 Runtime: Device: 0, count: 0, 297 seconds
29 August 2012 - 23:06:06 Device: 0, Count: 0, finished.
Ready ---------------------------------------------------------------------
Results:
Device: 0, device count: 1, average time / count: 297, average time on device: 297 Seconds (4 Minutes, 57 Seconds)
Next ---------------------------------------------------------------------
29 August 2012 - 23:06:06 Start, devices: 1, device count: 2 (0.50)
ID: 1277388 · Report as offensive
w1hue Project Donor
Volunteer tester

Send message
Joined: 4 Aug 00
Posts: 69
Credit: 5,492,898
RAC: 7
United States
Message 1277401 - Posted: 30 Aug 2012, 4:50:39 UTC - in response to Message 1275682.  


So ... what am I not doing correctly? (Yes, I stopped BOINC Manager before executing the performance utility).

V 1.1 will have a lot more diagnostics.

The files are copied
C:\Users\username\AppData\Roaming\eFMer\SetiPerformance\slotx

In that location the exe is started. Is it there? (You have to press start first to copy the files.)


There are 255 "slot" folders there, but none contain any files (exe or otherwise). Press "start" WHERE?? I don't see any "start"...

Since I'm not using Lunatics apps, I might as well just forget the whole thing...
ID: 1277401 · Report as offensive
Profile S@NL - eFMer - efmer.com/boinc
Volunteer tester
Avatar

Send message
Joined: 7 Jun 99
Posts: 512
Credit: 148,746,305
RAC: 0
United States
Message 1277408 - Posted: 30 Aug 2012, 5:34:28 UTC - in response to Message 1277401.  


There are 255 "slot" folders there, but none contain any files (exe or otherwise). Press "start" WHERE?? I don't see any "start"...

Since I'm not using Lunatics apps, I might as well just forget the whole thing...

Make sure Devices is set to 1, unless you have more than one card.
Use V 1.1!
Press "Test automatically"

TThrottle Control your temperatures. BoincTasks The best way to view BOINC. Anza Borrego Desert hiking.
ID: 1277408 · Report as offensive
juan BFP Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 16 Mar 07
Posts: 9786
Credit: 572,710,851
RAC: 3,799
Panama
Message 1277555 - Posted: 30 Aug 2012, 14:06:04 UTC

Fred

I just put the 1.1 to work on my 590, this is the test report, i belive something still wrong, 1 WU is the best? on a 590?


Starting automatic test: (x41z cuda 42)
30 August 2012 - 10:56:38 Start, devices: 2, device count: 1 (1.00)
30 August 2012 - 10:58:30 Runtime: Device: 1, count: 0, 109 seconds
30 August 2012 - 10:58:30 Runtime: Device: 0, count: 0, 109 seconds
30 August 2012 - 10:58:30 Device: 1, Count: 0, finished.
30 August 2012 - 10:58:30 Device: 0, Count: 0, finished.
Ready ---------------------------------------------------------------------
Results:
Device: 0, device count: 1, average time / count: 109, average time on device: 109 Seconds (1 Minutes, 49 Seconds)
Device: 1, device count: 1, average time / count: 109, average time on device: 109 Seconds (1 Minutes, 49 Seconds)
Next ---------------------------------------------------------------------
30 August 2012 - 10:58:31 Start, devices: 2, device count: 2 (0.50)
30 August 2012 - 11:02:16 Runtime: Device: 1, count: 1, 220 seconds
30 August 2012 - 11:02:16 Runtime: Device: 1, count: 0, 220 seconds
30 August 2012 - 11:02:16 Runtime: Device: 0, count: 0, 220 seconds
30 August 2012 - 11:02:16 Runtime: Device: 0, count: 1, 220 seconds
30 August 2012 - 11:02:16 Device: 1, Count: 0, finished.
30 August 2012 - 11:02:16 Device: 1, Count: 1, finished.
30 August 2012 - 11:02:16 Device: 0, Count: 1, finished.
30 August 2012 - 11:02:16 Device: 0, Count: 0, finished.
Ready ---------------------------------------------------------------------
Results:
Device: 0, device count: 2, average time / count: 220, average time on device: 110 Seconds (1 Minutes, 50 Seconds)
Device: 1, device count: 2, average time / count: 220, average time on device: 110 Seconds (1 Minutes, 50 Seconds)
The best average time found: 109 Seconds (1 Minutes, 49 Seconds), with count: 1.00 (1)

ID: 1277555 · Report as offensive
Profile S@NL - eFMer - efmer.com/boinc
Volunteer tester
Avatar

Send message
Joined: 7 Jun 99
Posts: 512
Credit: 148,746,305
RAC: 0
United States
Message 1277571 - Posted: 30 Aug 2012, 15:03:31 UTC - in response to Message 1277555.  

Fred
I just put the 1.1 to work on my 590, this is the test report, i belive something still wrong, 1 WU is the best? on a 590?

The 590 is here: http://www.efmer.eu/forum_tt/index.php?topic=981.0

My card:

Device: 0, device count: 3, average time / count: 321, average time on device: 107 Seconds (1 Minutes, 47 Seconds)
Device: 2, device count: 3, average time / count: 322, average time on device: 107 Seconds (1 Minutes, 47 Seconds)
= 53


Device: 0, device count: 4, average time / count: 396, average time on device: 99 Seconds (1 Minutes, 39 Seconds)
Device: 2, device count: 4, average time / count: 401, average time on device: 100 Seconds (1 Minutes, 40 Seconds)
= 50

It should run 3 easily.
Try a manual run with count 0.25 and see what you get.

The problem with the card is that it tends to get too hot.
If it does it clocks down to prevent overheating.

The Z version is faster on the card, but it down clocks more, so the result may even be worse.
My fan is running close to 100%. Temperature 83 C.

TThrottle Control your temperatures. BoincTasks The best way to view BOINC. Anza Borrego Desert hiking.
ID: 1277571 · Report as offensive
juan BFP Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 16 Mar 07
Posts: 9786
Credit: 572,710,851
RAC: 3,799
Panama
Message 1277637 - Posted: 30 Aug 2012, 17:10:44 UTC - in response to Message 1277571.  

Thanks

Will try later, the card is in another location.

I just think why the wierd results, the same time for all the test, in all the other hosts i run, with diferent GPU combinations, the build works OK.

Actualy i run 2 WU at a time due the XP32 memory limitation, but expect to change soon to windows 7 64 as our final arrangementes with MS for our corporate licence renewal ends.

My card normaly runs at 80C with 80% fan but it has an extra fan atache at the back of the GPU to help cooling.
ID: 1277637 · Report as offensive
Horacio

Send message
Joined: 14 Jan 00
Posts: 536
Credit: 75,967,266
RAC: 0
Argentina
Message 1277675 - Posted: 30 Aug 2012, 17:50:49 UTC - in response to Message 1277637.  

I just think why the wierd results, the same time for all the test, in all the other hosts i run, with diferent GPU combinations, the build works OK.

It may be due to the short running time of the WU used to do the tests.
Ive been testing yesterday and Im still doing some more tests, so I have no definately conclusions yet, but until now, the best count number changes if I use a longer WU...
In my gtx680 using the WU provided it gives me the same average time on board for counts of 3 to 6, while using a WU that takes around a bit more than twice the time, count at 3 becomes the third faster option pretty close to the times with count 6 and several seconds behind the times of count at 5.

I guess that with shorter WU most of the time its used to set the slot, load the data and return the result, while the neat crunching time is so small that any change on it cant be accurately measured using an integer number of seconds... (and this may be worst if you have a really fast GPU on a extremelly slow system)
ID: 1277675 · Report as offensive
juan BFP Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 16 Mar 07
Posts: 9786
Credit: 572,710,851
RAC: 3,799
Panama
Message 1277832 - Posted: 31 Aug 2012, 0:07:03 UTC - in response to Message 1277675.  
Last modified: 31 Aug 2012, 0:08:12 UTC

I just think why the wierd results, the same time for all the test, in all the other hosts i run, with diferent GPU combinations, the build works OK.

It may be due to the short running time of the WU used to do the tests.
Ive been testing yesterday and Im still doing some more tests, so I have no definately conclusions yet, but until now, the best count number changes if I use a longer WU...
In my gtx680 using the WU provided it gives me the same average time on board for counts of 3 to 6, while using a WU that takes around a bit more than twice the time, count at 3 becomes the third faster option pretty close to the times with count 6 and several seconds behind the times of count at 5.

I guess that with shorter WU most of the time its used to set the slot, load the data and return the result, while the neat crunching time is so small that any change on it cant be accurately measured using an integer number of seconds... (and this may be worst if you have a really fast GPU on a extremelly slow system)


Could be that but this 590 is not my fast GPU, and is feeded by a Core(TM)2 Quad CPU Q8300 @ 2.50GHz with one core separated just to feed the GPU so feeding speed must not be the problem i think, and the build runs fine in all my other hosts, just in this host and with this GPU the problem apears. I try to use another wu (a bigger one with processing time of 4 minutes) for test and the results are very similar.

Tomorrow i will switch the 590 on this host with another with a 690 to see if the problem is related to the GPU itself or the Host. Will return with the results ASAP.
ID: 1277832 · Report as offensive
Profile Snowmain
Avatar

Send message
Joined: 17 Nov 05
Posts: 75
Credit: 30,681,449
RAC: 83
United States
Message 1277855 - Posted: 31 Aug 2012, 1:37:36 UTC
Last modified: 31 Aug 2012, 2:20:47 UTC

hmmm
ID: 1277855 · Report as offensive
Profile Snowmain
Avatar

Send message
Joined: 17 Nov 05
Posts: 75
Credit: 30,681,449
RAC: 83
United States
Message 1277866 - Posted: 31 Aug 2012, 2:24:02 UTC
Last modified: 31 Aug 2012, 3:09:51 UTC

$PerKWH0.12 X41g.........watts....$ per year...WU per year.....WU per $

............sec. per WU
8800 ultra OC......603....210........220.75....52298.51...........236.91*
GT630................406....65...........68.33....77674.88..........1136.79
GTX460 oc..........158....190........199.73...199594.94.........999.33*
GTX660Ti...........158....150........157.68...199594.94.........1265.82
GTX560.............150....150........157.68...210240.00.........1333.33
GTX560Ti oc.......133....234........303.80...242584.62.........963.95
GTX295.............130....289........303.80...242584.62...........798.51
GTX670.............129....170........178.70...244465.12.........1367.99
GTX680.............107....195........204.98...294728.97.........1437.81
GTX480.............100....250........262.80...315360.00.........1200.00
GTX580 oc..........80.....300........315.36...394200.00.........1250.00*
GTX690..............58.....300........315.36...543724.14.........1724.14Holy
GTX590..............53.....365........383.69...595018.87.........1550.79

I looked up some overclocked cards for the power consupmtion number however they range alot. + - 20%
$ per year assumes 100% load 24/7/365
Interesting data. Nice to finally have something empirical.
The GT630 is a bit of a suprise. GTX 560 as well.

I am in the market myself for a new main display card. I have found the GTX 480 new for $200....everything else in that area of perfromance seems to be closer to $500.
GTX 560..$150
GTX 480..$200...Kinda hard not to go that direction..
ID: 1277866 · Report as offensive
Profile arkayn
Volunteer tester
Avatar

Send message
Joined: 14 May 99
Posts: 4438
Credit: 55,006,323
RAC: 0
United States
Message 1277871 - Posted: 31 Aug 2012, 2:50:23 UTC - in response to Message 1277866.  

$PerKWH0.12 X41g.........watts....$ per year...WU per year.....WU per $

............sec. per WU
8800 ultra OC......603....195........204.98....52298.51...........255.13*
GT630................406....65...........68.33....77674.88..........1136.79
GTX460 oc..........158....180........189.22...199594.94.........1054.85*
GTX660Ti...........158....150........157.68...199594.94.........1265.82
GTX560.............150....150........157.68...210240.00.........1333.33
GTX560Ti oc.......133....190........199.73...237112.78.........1187.18
GTX295.............130....289........303.80...242584.62...........798.51*
GTX670.............129....170........178.70...244465.12.........1367.99
GTX680.............107....195........204.98...294728.97.........1437.81
GTX480.............100....250........262.80...315360.00.........1200.00
GTX580 oc..........80.....300........315.36...394200.00.........1250.00*
GTX690..............58.....300........315.36...543724.14.........1724.14Holy
GTX590..............53.....365........383.69...595018.87.........1550.79

I guessed on the power consumption on the OC cards buts it with in 10-15%.
$ per year assumes 100% load 24/7/365
Interesting data. Nice to finally have something empirical.
The GT630 is a bit of a suprise. GTX 560 as well.

I am in the market myself for a new main display card. I have found the GTX 480 new for $200....everything else in that area of perfromance seems to be closer to $500.
GTX 560..$150
GTX 480..$200...Kinda hard not to go that direction..


The GTX-480 is an absolute power hog and runs very warm to just plain hot.

If you can wait a bit longer, see what the GTX-660(Non-Ti) comes in at. Should be out in a week or 2.

http://videocardz.com/nvidia/geforce-600/geforce-gtx-660

ID: 1277871 · Report as offensive
Profile Snowmain
Avatar

Send message
Joined: 17 Nov 05
Posts: 75
Credit: 30,681,449
RAC: 83
United States
Message 1277879 - Posted: 31 Aug 2012, 3:15:18 UTC - in response to Message 1277871.  
Last modified: 31 Aug 2012, 3:45:08 UTC

The GTX-480 is an absolute power hog and runs very warm to just plain hot.


Yeah I saw a video online of a gtx 480 over 103 C...

My system is in a different room form my monitor and keyboard so heat isn't as much of an issue as it is for others, but you still gotta move the heat out with the ac unit 4 months out of the year.

However the 660 will almost certianly be $300 plus when it comes out..which is out of my budget.

The WU per $ between the 660ti and the 480 is barely 5% different. Do you think the 660 will be significantly more efficient? Not to mention it will take 2 of them to touch a gtx 480....umm am I missing something?


So I would need to buy 2 in order to keep up with the 480 and the electrical cost to break even between 2 660 and 1 480 is around the 4 year point. Right?

Now 2 660 are much(1/3) faster than a 480 but cost 2.25 ro 2.5 time more.
ID: 1277879 · Report as offensive
Profile arkayn
Volunteer tester
Avatar

Send message
Joined: 14 May 99
Posts: 4438
Credit: 55,006,323
RAC: 0
United States
Message 1277885 - Posted: 31 Aug 2012, 3:24:45 UTC - in response to Message 1277879.  

The GTX-480 is an absolute power hog and runs very warm to just plain hot.


Yeah I saw a video online of a gtx 480 over 103 C...

My system is in a different room form my monitor and keyboard so heat isn't as much of an issue as it is for others, but you still gotta move the heat out with the ac unit 4 months out of the year.

However the 660 will almost certianly be $300 plus when it comes out..which is out of my budget.

The WU per $ between the 660ti and the 480 is barely 5% different. Do you think the 660 will be significantly more efficient? Not to mention it will take 2 of them to touch a gtx 480....umm am I missing something?


Actually they are figuring around $229 to $249.


ID: 1277885 · Report as offensive
w1hue Project Donor
Volunteer tester

Send message
Joined: 4 Aug 00
Posts: 69
Credit: 5,492,898
RAC: 7
United States
Message 1277894 - Posted: 31 Aug 2012, 3:42:21 UTC - in response to Message 1277408.  

OK, I downloaded V 1.1 - Here's the results:

Starting automatic test: (x41g)
30 August 2012 - 20:25:21 Start, devices: 1, device count: 1 (1.00)
30 August 2012 - 20:25:30 ERROR: Device: 0, Count: 0, program failed to start.
Ready ---------------------------------------------------------------------
Results:
Device: 0, device count: 1, average time / count: 0, average time on device: 0 Seconds (0 Minutes, 0 Seconds)
Aborted, detected an error, results may be invalid.

Here's the contents of the log file:

08/30/2012, 20:22:16 -- Startup
08/30/2012, 20:22:16 -- Start reading config.xml
08/30/2012, 20:22:16 -- <data> <name>x41g<folder>x41g<exe>lunatics_x41g_win32_cuda32_perf.exe<parameter> --device %device
08/30/2012, 20:22:16 -- <gui> <showParameters>0<showReadXmlButton>0<maxAutoCount>3<showexpert>0
08/30/2012, 20:22:16 -- Finished reading config.xml

Here's the contents of stderr.txt:

Can't open init data file - running in standalone mode
Can't open init data file - running in standalone mode
setiathome_CUDA: Found 1 CUDA device(s):
Device 1: GeForce GT 520, 1023 MiB, regsPerBlock 32768
computeCap 2.1, multiProcs 1
clockRate = 1892000
setiathome_CUDA: No device specified, determined to use CUDA device 1: GeForce GT 520
SETI@home using CUDA accelerated device GeForce GT 520
Priority of process raised successfully
Priority of worker thread raised successfully
Cuda Active: Plenty of total Global VRAM (>300MiB).
All early cuFft plans postponed, to parallel with first chirp.

) _ _ _)_ o _ _
(__ (_( ) ) (_( (_ ( (_ (
not bad for a human... _)

Multibeam x41g Preview, Cuda 3.20

Legacy setiathome_enhanced V6 mode.
Work Unit Info:
...............
WU true angle range is : 0.775000
VRAM: cudaMalloc((void**) &dev_cx_DataArray, 1048576x 8bytes = 8388608bytes, offs256=0, rtotal= 8388608bytes
VRAM: cudaMalloc((void**) &dev_cx_ChirpDataArray, 1179648x 8bytes = 9437184bytes, offs256=0, rtotal= 17825792bytes
VRAM: cudaMalloc((void**) &dev_flag, 1x 8bytes = 8bytes, offs256=0, rtotal= 17825800bytes
VRAM: cudaMalloc((void**) &dev_WorkData, 1179648x 8bytes = 9437184bytes, offs256=0, rtotal= 27262984bytes
VRAM: cudaMalloc((void**) &dev_PowerSpectrum, 1048576x 4bytes = 4194304bytes, offs256=0, rtotal= 31457288bytes
VRAM: cudaMalloc((void**) &dev_t_PowerSpectrum, 1048584x 4bytes = 1048608bytes, offs256=0, rtotal= 32505896bytes
VRAM: cudaMalloc((void**) &dev_GaussFitResults, 1048576x 16bytes = 16777216bytes, offs256=0, rtotal= 49283112bytes
VRAM: cudaMalloc((void**) &dev_PoT, 1572864x 4bytes = 6291456bytes, offs256=0, rtotal= 55574568bytes
VRAM: cudaMalloc((void**) &dev_PoTPrefixSum, 1572864x 4bytes = 6291456bytes, offs256=0, rtotal= 61866024bytes
VRAM: cudaMalloc((void**) &dev_NormMaxPower, 16384x 4bytes = 65536bytes, offs256=0, rtotal= 61931560bytes
VRAM: cudaMalloc((void**) &dev_flagged, 1048576x 4bytes = 4194304bytes, offs256=0, rtotal= 66125864bytes
VRAM: cudaMalloc((void**) &dev_outputposition, 1048576x 4bytes = 4194304bytes, offs256=0, rtotal= 70320168bytes
VRAM: cudaMalloc((void**) &dev_PowerSpectrumSumMax, 262144x 12bytes = 3145728bytes, offs256=0, rtotal= 73465896bytes
VRAM: cudaMallocArray( &dev_gauss_dof_lcgf_cache, 1x 8192bytes = 8192bytes, offs256=40, rtotal= 73474088bytes
VRAM: cudaMallocArray( &dev_null_dof_lcgf_cache, 1x 8192bytes = 8192bytes, offs256=240, rtotal= 73482280bytes
VRAM: cudaMalloc((void**) &dev_find_pulse_flag, 1x 8bytes = 8bytes, offs256=0, rtotal= 73482288bytes
VRAM: cudaMalloc((void**) &dev_t_funct_cache, 1966081x 4bytes = 7864324bytes, offs256=0, rtotal= 81346612bytes
Thread call stack limit is: 1k


The machine has one GT 520 board installed (overclocked), driver version is 285.58. As stated before, I am NOT running Lunitics app.

ID: 1277894 · Report as offensive
Russ Sheldon
Avatar

Send message
Joined: 18 Jan 01
Posts: 382
Credit: 530,470,939
RAC: 0
United States
Message 1277914 - Posted: 31 Aug 2012, 4:18:55 UTC - in response to Message 1277832.  
Last modified: 31 Aug 2012, 4:19:33 UTC

Juan,

Aviso! Your results may vary because you are in the southern hemisphere. I'm getting that information straight from tbret.
Proud member of the half bill club!
ID: 1277914 · Report as offensive
Previous · 1 · 2 · 3 · 4 · 5 · 6 · 7 · 8 . . . 13 · Next

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


 
©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.