Latest drivers (NVidia and ATI)

[Jason G]

... NVIDIA GPU 0: GeForce 8600 GTS (driver version unknown, CUDA version 4000, compute capability 1.1, 240MB, 96 GFLOPS peak)

is that an Optimus notebook/platform ? I've provided a fix for this to Berkeley via messenger pigeon.  It's cosmetic only really, but very annoying. That should be fixed in some future Boinc (It's not a driver problem).

Jason

[efmer (fred)]

Fred, you're getting me really confused here.

Me too  .

I tried 280.26 on 2 computers and had to back to 275.33. 275.33 works just fine.
But this is a driver pack and drivers are probably specific to certain cards.

[corsair]

is there any improvement with last ATI 11.8??

[TouchuvGrey]

Hello Fred:

            i notice that you are only linking to Throttle and BoincTasks.
Do you no longer support/reccomend your rescheduler ?

      i recall recently seeing some discussion of whether it was neccessary or
desireable any longer with some of the recent changes in TaskManager.

      Not looking to open a can of worms here but i would like to hear what you have
to say and what settings you would suggest. intel corei7 920 CPU, 12 gigs tri channel DDR3.
GTS 250 and GTX460 GPU's

[benool]

... NVIDIA GPU 0: GeForce 8600 GTS (driver version unknown, CUDA version 4000, compute capability 1.1, 240MB, 96 GFLOPS peak)

is that an Optimus notebook/platform ? I've provided a fix for this to Berkeley via messenger pigeon.  It's cosmetic only really, but very annoying. That should be fixed in some future Boinc (It's not a driver problem).

Jason

No it is a desktop PC. I'm still running Boinc client version 6.10.60 if it matters.

So far it is running fine. It seems to be slightly faster too although for some task it ends in < 30s (using X38g)

[Jason G]

OK, you must have just found some other weird configuration that Boinc driver version detection (cosmetic) fails under.  Just checking 

[Richard Haselgrove]

Out of curiosity, how many entries are there in the 'display adapters' section of device manager - and what else is listed, apart from your 8600 GTS? (i'm thinking motherboard video here).

[Mike]

is there any improvement with last ATI 11.8??

Have tried 11.8. SDK 2.5 yesterday.

CPU times was nearly the same as GPU time on my host.
Since i see same behaviour on new nvidia drivers (OpenCL) i get the impression its by design.

[benool]

Out of curiosity, how many entries are there in the 'display adapters' section of device manager - and what else is listed, apart from your 8600 GTS? (i'm thinking motherboard video here).

It is a mixed host having an ATI 4550 (soon to be replaced by a 6670) along with the 8600 GTS.

For this BOINC client AMD card is disabled by use of cc_config.xml  (I run a boinc deamon for each Nvidia/AMD/CPU)

[Richard Haselgrove]

OK, you must have just found some other weird configuration that Boinc driver version detection (cosmetic) fails under.  Just checking 

Yup, I reckon that qualifies as 'one David didn't think of' 

[Jason G]

Yup, I reckon that qualifies as 'one David didn't think of' 

LoL, to be fair (this time), the particular nvApi call definition in question does loosley imply iterating through active displays is needed.  Too bad if you interpret it that way that it will miss systems with no active display attached to an nVidia GPU, such as idling Optimus & crunching only cards like Tesla with no display driven by an nVidia GPU.

[efmer (fred)]

i notice that you are only linking to Throttle and BoincTasks.
Do you no longer support/reccomend your rescheduler ?

A bit off topic.
Not super active, as the need for a rescheduler is low. But I released an updated V 2.6 no too long ago.
http://www.efmer.eu/forum_tt/index.php?topic=428.0

[Pepi]

"driver version unknown, CUDA version 4000,"
I get this "error" all the time, and it is Boinc related.
When I got it: under windows 7 I use GPU on motherboard for monitor, and Nvidia 560 TI for crunching. So nothing is connected to Nvidia, and then I get tis message. If I disable on-board GPU, and user Nvidia for picture and crunching, then all is ok, driver version is known...

[benool]

Yup so that is exactly the same as me with a 8600 with no display attached.

The "no display attached" seems to be the key to driver version detecion failing

[Richard Haselgrove]

I think we ought to ask Jason if he can supply you with either a modified copy of BOINC, or the test probe we used to track down the problem originally, to confirm that they fix the issues for both benool amd Pepi - David Anderson has accepted the patch for 'Optimus' laptop chipsets for the next version of BOINC, but I'd like to know whether it'll work for you too as well, or whether there is still something else which needs tracking down.