Nvidia drivers, MBv8 and APv7 (post/pre 350)

Message boards : Number crunching : Nvidia drivers, MBv8 and APv7 (post/pre 350)
Message board moderation

To post messages, you must log in.

AuthorMessage
Iztok s52d (and friends)

Send message
Joined: 12 Jan 01
Posts: 136
Credit: 393,469,375
RAC: 116
Slovenia
Message 1758979 - Posted: 24 Jan 2016, 10:13:24 UTC

This is just to remind you, if you have similar problem,
when Nvidia driver is upgraded and you use very old AP GPU application.


I was running old APv7 Nvidia client for long time, using
driver 340.65. Worked fine with both MBv7 and APv7 on Nvidia GPU.

With multibeam V8, I had to upgrade driver to 352.63, so
setiathome_8.04_x86_64-pc-linux-gnu__opencl_nvidia_sah can do the job.
(in beta, of course)

Then AP workunits started to show, and they all failed.
Some searhing of this forum pointed to
http://setiathome.berkeley.edu/forum_thread.php?id=77133postid=1665104#1665104


So, I replaced all bool2 with bool_2 in AstroPulse_Kernels_r2696.cl file,
and all works fine: AP Wus are processed and validated.


BR
s52d
ID: 1758979 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14650
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1758986 - Posted: 24 Jan 2016, 12:16:33 UTC

If I may borrow this thread for a related matter?

I'm preparing to do a big NVidia driver test. Along the way, I found:

GeForce HotFix driver 361.60 for Windows, released 12 January 2016.

This may, or may not, assist with/resolve the problems people have been reporting with NV driver 361.43
ID: 1758986 · Report as offensive
Profile jason_gee
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 24 Nov 06
Posts: 7489
Credit: 91,093,184
RAC: 0
Australia
Message 1758997 - Posted: 24 Jan 2016, 12:57:47 UTC - in response to Message 1758986.  
Last modified: 24 Jan 2016, 12:59:20 UTC

Skimming the forum, there isn't a lot of detail, though looks like it addresses memory clock issues. These could indeed manifest in high-DPC-latency-like behaviour, potentially with transfer timeouts, application failures/device&driverResets, and potentially BSODs.

Fingers crossed the specific mention of Photoshop and Illustrator implies Cuda Acceleration (as I suspect). Downloaded and will give it a try (though I've been on 359.00 with no issues)

After reading the information at https://forums.geforce.com/default/topic/910177/geforce-drivers/announcing-geforce-hotfix-driver-361-60/, particularly the description of the reasons for hotfix drivers, I'll be keeping a closer eye on them.


...
More information about Hot fix drivers:

A GeForce driver is an incredibly complex piece of software, with more lines of code than the entire Windows OS. We have an army of software engineers constantly adding features and fixing bugs. These changes are checked into the main driver branches, which are eventually run through a massive QA process and released.

Since we have so many changes being checked in, we usually try to align driver releases with significant game or product releases. This process has served us pretty well over the years but it has one significant weakness. Sometimes a change that is important to many users might end up sitting and waiting until we are able to release the driver.

The GeForce Hotfix driver is our way to trying to get some of these fixes out to you more quickly.
These drivers are basically the same as the previous released version, with a small number of additional targeted fixes. The fixes that make it in are based in part on your feedback in the Driver Feedback threads and partly on how realistic it is for us to quickly address them. These fixes (and many more) will be incorporated into the next official driver release, at which time the Hotfix driver will be taken down.

To be sure, these Hotfix drivers are beta, optional and provided as-is. They are run through a much abbreviated QA process. The sole reason they exist is to get fixes out to you more quickly. The safest option is to wait for the next WHQL certified driver. But we know that many of you are willing to try these out.

These HotFix drivers represent a lot of additional work by our engineering teams, I hope they provide value for you. We’ll try it out and see if people like the idea and want us to continue.

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.
ID: 1758997 · Report as offensive
Profile jason_gee
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 24 Nov 06
Posts: 7489
Credit: 91,093,184
RAC: 0
Australia
Message 1758999 - Posted: 24 Jan 2016, 13:09:19 UTC - in response to Message 1758998.  
Last modified: 24 Jan 2016, 13:09:55 UTC

A GeForce driver is an incredibly complex piece of software, with more lines of code than the entire Windows OS.

Really?? Does M$ agree with that statement?


Who knows, though going by filesizes alone, yeah I'd say the Core Windows OS Kernel and basic services would have fewer lines of code. Naturally most of what we might consider Windows is poorly debugged device drivers (paraphrasing a documentary I saw years ago about Netscape and the browser wars)
"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.
ID: 1758999 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14650
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1759000 - Posted: 24 Jan 2016, 13:15:13 UTC
Last modified: 24 Jan 2016, 14:07:08 UTC

Please guys, keep the thread hijack to a minimum. The OP posted important information, and I don't want it to get buried.

Link to opening post
ID: 1759000 · Report as offensive

Message boards : Number crunching : Nvidia drivers, MBv8 and APv7 (post/pre 350)


 
©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.