EVGA Nvidia 470, attempting GPU work. All fail. Please read my results and figure out the issue and correct the code. It is exceptionally irriatating. I really do not care for attempting to determine on my own why your project will not work on this system. Tyan system board, Win7 64 bit, 2 AMD opteron 275, 1 EVGA 470, running this, Seti, and GPU grid. Only working Nvidia driver 258.

honestly I don't know, but I have an i7 quad under the hood. It is working, I have 4 different lines of status bar on my screen.

You are running cpu units, not the gpu units.
Here is an example of what an Nvidia card can do and what a unit looks like:
463.14 222.04 106.88 MilkyWay@Home v1.02 (opencl_nvidia)

Here is a couple of yours:
31,615.23 30,655.79 118.25 MilkyWay@Home N-Body Simulation v1.08
21,062.19 20,428.95 213.76 MilkyWay@Home v1.00

You can see the time to completion is MUCH different as well as the unit name itself. At the end of my unit name is (opencl nvidia), that is missing from your units. The completion times WILL be different between my Nvidia gpu and your gpu, I have a 560Ti and you have a 260M, but the unit name at the end should still be the same when you get the gpu units working.

Have you gone to the web page setting and told MilkyWay to send you gpu units? It is under Your Account and then MilkyWay @home preferences. These are my settings as I use my cpu's elsewhere:
Use CPU
Enforced by version 6.10+ no

Use AMD (ATI) GPU
Enforced by version 6.10+ yes

Use NVIDIA GPU
Enforced by version 6.10+ yes

I hope the new "Updated GPU Requirements" thread is sufficient. Please post there with comments, issues or workarounds.

ABSOLUTELY, thank you VERY much!! It was getting hard to have to scroll thru the list to figure out what to buy to make sure it worked here.

---

EVGA Nvidia 470, attempting GPU work. All fail. Please read my results and figure out the issue and correct the code. It is exceptionally irriatating. I really do not care for attempting to determine on my own why your project will not work on this system. Tyan system board, Win7 64 bit, 2 AMD opteron 275, 1 EVGA 470, running this, Seti, and GPU grid. Only working Nvidia driver 258.

One easy answer is I am using driver version: 314.7 and you are using Driver version: 258.96, meaning yours is VERY old in gpu driver history. Also if you look here:
http://milkyway.cs.rpi.edu/milkyway/workunit.php?wuid=334090369

You can see that only one pc had trouble with the unit, your pc, that means it is unlikely to b a unit or MilkyWay problem, but instead a problem with your pc. Upgrade the Nvidia driver and see if that fixes the problem on its own, sometimes it does. Another issue could be if you are overclocking or playing games or otherwise doing pc intensive things while crunching, if so this too could be causing the pc to 'lose focus' and the units could error out. IF you are doing something else, stop for the next two hours and let Boinc just crunch, you CAN leave the Boinc Manager box open to watch if you'd like, but you are looking to see if the units still error out when the pc is otherwise idle. If you are overclocking back it off back to stock settings for the next two hours and see if the errors go away, if so then the problem is the pc is overclocked too far for MilkyWay to work properly.

honestly I don't know, but I have an i7 quad under the hood. It is working, I have 4 different lines of status bar on my screen.

You are running cpu units, not the gpu units.
Here is an example of what an Nvidia card can do and what a unit looks like:
463.14 222.04 106.88 MilkyWay@Home v1.02 (opencl_nvidia)

Here is a couple of yours:
31,615.23 30,655.79 118.25 MilkyWay@Home N-Body Simulation v1.08
21,062.19 20,428.95 213.76 MilkyWay@Home v1.00

You can see the time to completion is MUCH different as well as the unit name itself. At the end of my unit name is (opencl nvidia), that is missing from your units. The completion times WILL be different between my Nvidia gpu and your gpu, I have a 560Ti and you have a 260M, but the unit name at the end should still be the same when you get the gpu units working.

Have you gone to the web page setting and told MilkyWay to send you gpu units? It is under Your Account and then MilkyWay @home preferences. These are my settings as I use my cpu's elsewhere:
Use CPU
Enforced by version 6.10+ no

Use AMD (ATI) GPU
Enforced by version 6.10+ yes

Use NVIDIA GPU
Enforced by version 6.10+ yes

I went to my settings and unchecked GPU Box. I didn't see the versions you are talking about(NVIDA/AMD)listed. This is what I saw " Suspend GPU work while computer is in use?
Enforced by version 6.6.21+ no" I changed it from yes to NO

Ill see if this changes anything. Ill request more work in the AM. Thanks

honestly I don't know, but I have an i7 quad under the hood. It is working, I have 4 different lines of status bar on my screen.

You are running cpu units, not the gpu units.
Here is an example of what an Nvidia card can do and what a unit looks like:
463.14 222.04 106.88 MilkyWay@Home v1.02 (opencl_nvidia)

Here is a couple of yours:
31,615.23 30,655.79 118.25 MilkyWay@Home N-Body Simulation v1.08
21,062.19 20,428.95 213.76 MilkyWay@Home v1.00

You can see the time to completion is MUCH different as well as the unit name itself. At the end of my unit name is (opencl nvidia), that is missing from your units. The completion times WILL be different between my Nvidia gpu and your gpu, I have a 560Ti and you have a 260M, but the unit name at the end should still be the same when you get the gpu units working.

Have you gone to the web page setting and told MilkyWay to send you gpu units? It is under Your Account and then MilkyWay @home preferences. These are my settings as I use my cpu's elsewhere:
Use CPU
Enforced by version 6.10+ no

Use AMD (ATI) GPU
Enforced by version 6.10+ yes

Use NVIDIA GPU
Enforced by version 6.10+ yes

I went to my settings and unchecked GPU Box. I didn't see the versions you are talking about(NVIDA/AMD)listed. This is what I saw " Suspend GPU work while computer is in use?
Enforced by version 6.6.21+ no" I changed it from yes to NO

Ill see if this changes anything. Ill request more work in the AM. Thanks

If you UNCHECK the gpu box then you will NOT get any gpu units. The ati and nvidia settings should be right there too. I go, on the MW webpage, to Your Account, Preferences for this project and then I click on Milkyway@home preferences and then in the top section where the top line says resource share, mine is set to 100, the 3rd and 4th items down talk about ATI and Nvidia, edit the preferences so that they say yes, or in your case at least the Nvidia line, says yes, save the changes. Then after you do an update from your pc, you should start getting gpu units.

Instead of continually sending these notice about "lacking double precision" to BOINC clients, why doesn't this project simply set it's servers so as NOT to attempt those downloads that require dp to computers whose gpu's are not on your list?
Or better yet, simply tell the volunteer which project tasks to uncheck on his/her Milkyway preferences page.
Simply sending those one-line notices with no relative references to the specific task units to which they pertain is entirely confusing and meaningless to the volunteer.

Are you crunching a cpu project using those cpu's at the same time?

Yes.

If so change the setting so one cpu is left free, per gpu, then set MW to run two units at once again and see if it helps. When you also run a cpu project on the same cpu you use to fee a gpu it can slow down both projects. Since the gpu projects crank out much more work it might make sense to drop one cpu core from the cpu projects and give more to the gpu projects.

Since I saw very little advantage run-time wise (per WU) running two units in the gpu as opposed to one, I believe that I am better off in the overall scheme of things in just running one gpu WU and not slowing down the cpu while crunching other projects. I only have a dual core machine and my (single) gpu a is a GT 520 which I chose for its low power consumption over a speedier power hog.

---

Instead of continually sending these notice about "lacking double precision" to BOINC clients, why doesn't this project simply set it's servers so as NOT to attempt those downloads that require dp to computers whose gpu's are not on your list?
Or better yet, simply tell the volunteer which project tasks to uncheck on his/her Milkyway preferences page.
Simply sending those one-line notices with no relative references to the specific task units to which they pertain is entirely confusing and meaningless to the volunteer.

All the programmers went to the Microsoft school of error writing, obscure and intentionally vague, ie 'your pc has crashed'!

As far as what preferences to check or uncheck just ask us fellow crunchers, we will be happy to help.

Are you crunching a cpu project using those cpu's at the same time?

Yes.

If so change the setting so one cpu is left free, per gpu, then set MW to run two units at once again and see if it helps. When you also run a cpu project on the same cpu you use to fee a gpu it can slow down both projects. Since the gpu projects crank out much more work it might make sense to drop one cpu core from the cpu projects and give more to the gpu projects.

Since I saw very little advantage run-time wise (per WU) running two units in the gpu as opposed to one, I believe that I am better off in the overall scheme of things in just running one gpu WU and not slowing down the cpu while crunching other projects. I only have a dual core machine and my (single) gpu a is a GT 520 which I chose for its low power consumption over a speedier power hog.

If you go into the Boinc Manager, the icon down by the clock, you can look under the Tasks tab and see how much cpu and gpu percentage Boinc is using, the more cpu percentage Boinc is using the more advantageous it is to leave a cpu core free. MOST Nvidia cards use much larger percentages of the cpu then AMD cards do, for instance my AMD cards here are using 0.05% of a cpu while the gpu is crunching. BUT my Nvidia cards are up into the 80+ percent range of a cpu. This is the result of manufacturer programing and each project optimizing its own app for us to use. On DistRTgen my AMD cpu is using 0.985% of a cpu to work at full speed, freeing up a cpu core made sense and lets me get more work done faster. Most projects optimize one brand or the other, they just don't have the resources to do both. So one works optimally and the other just works.

I will let my signature come thru this time, it is normally off, so you can see the effect of where AMD cards work best. I have 2 Nvidia cards, but I have 12 AMD cards crunching!

---

If you go into the Boinc Manager, the icon down by the clock, you can look under the Tasks tab and see how much cpu and gpu percentage Boinc is using, the more cpu percentage Boinc is using the more advantageous it is to leave a cpu core free. ...

Gee, I didn't know that ... I must be stupid or something.

Could anyone tell me what the video ram requirements are for the different types of MW WU's? With E@H, for example, 300MB is the number when you're trying to decide how many GPU WU's your card can handle simultaneously.

Thanks for any information that you can provide.

Regards,
Steve

If this is off topic for this thread please point me in the correct direction.
I would like to not use windows for my AM3 and HD6990 cards,
I can not get the GPU drivers installed in Ubuntu properly.
is there a step by step guide, or a simple way to start using a linux os?
Thank you for your assistance.
Greg

If this is off topic for this thread please point me in the correct direction.
I would like to not use windows for my AM3 and HD6990 cards,
I can not get the GPU drivers installed in Ubuntu properly.
is there a step by step guide, or a simple way to start using a linux os?
Thank you for your assistance.
Greg

I know you have to install 'proprietary drivers' but don't remember how without my paperwork in front of me. Try this link for some guidelines though
http://ubuntuforums.org/showthread.php?t=1131430

If this is off topic for this thread please point me in the correct direction.
I would like to not use windows for my AM3 and HD6990 cards,
I can not get the GPU drivers installed in Ubuntu properly.
is there a step by step guide, or a simple way to start using a linux os?
Thank you for your assistance.
Greg

I'm running Linux Mint 14 and 15 which is based on Ubuntu so here goes.

As Mikey said, you do need the "Restricted Drivers". In the past, there was a separate menu option for this but recently Mint (and maybe Ubuntu) have moved it so that it is a tab under Synaptic.

The other thing is that for the GPU's to work try the following.

BEFORE YOU START BOINC MANAGER every time ......

Open terminal as Admin or as sudo

sudo service boinc-client restart

You should get a bunch of "ok's",

goa ahead and close terminal

and THEN start BOINC Manager

Hopefully, that will do it for you. If not, depending on what happens (or doesn't happen) we'll try again.

Also, if you don't see any GPU WU's, look under the event log up near the start to see if there are any references to the GPU.

Regards,
Steve

Hello,

My AMD HD4850 is not recognized by BOINC 7.0.27 on Debian Wheezy 64-bit. Boinc tells me that there is "no usable GPU found".

Do I need a proprietary driver or something like that? Debian tells me I'm using Gallium 0.4 driver.

Hello,

My AMD HD4850 is not recognized by BOINC 7.0.27 on Debian Wheezy 64-bit. Boinc tells me that there is "no usable GPU found".

Do I need a proprietary driver or something like that? Debian tells me I'm using Gallium 0.4 driver.

YES you do need a proprietary driver, see Europa's note for some guidelines, I know it isn't exactly the same but hopefully you can figure it out, I am NOT a Linux guy.

I struggled a lot to install the AMD proprietary driver (it was easier on Ubuntu!) but finally I succeeded :

ATI GPU 0: ATI Radeon HD 4700/4800 (RV740/RV770) (CAL version 1.4.1734, 512MB, 425MB available, 2000 GFLOPS peak)

However, BOINC still has issues to recognize my GPU :

Requesting new tasks for ATI
App version needs OpenCL but GPU doesn't support it
App version needs OpenCL but GPU doesn't support it
[error] App version uses non-existent ATI GPU
[error] App version uses non-existent ATI GPU
[error] Missing coprocessor for task de_separation_80_DR_8_rev_3_2_1372784654_14591661_0; aborting

What's wrong? Here's my xorg.conf :

Section "Device"
Identifier "Default Video Card"
Driver "fglrx"
EndSection

Section "Monitor"
Identifier "Default Monitor"
VertRefresh 60
HorizSync 30-81
EndSection

Section "Screen"
Identifier "Default Screen"
Device "Default Video Card"
Monitor "Default Monitor"
DefaultDepth 24
SubSection "Display"
Modes "1280x1024" "1024x768" "800x600" "640x480"
EndSubSection
EndSection

Milkyway runs OK on CPUs as well

---

Go away, I was asleep

Yes, Milkyway runs fine on the CPUs (Core 2 Duo E4300) but I would like the GPU to compute as well :)