Latest drivers (NVidia and ATI)

[Fredericx51]

Not very lucky, getting mostly VLAR tasks, marked _vlar, so nothing to do for the FERMI's except GPUGrid and Einstein.
And still using the  258.96. drivers, giving a better response, no 'lag' or late response from the mouse or
keyboard.

Since a lot of Einstein tasks, which errorred out, after I made an mistake in the app_info.xml, were
resend, I can do them again.......

 

[M_M]

I have upgraded to 260.63 drivers, including nvcuda.dll (CUDA 3.2.1), supposed to be much better optimized for fermi then existing CUDA 3.0.

From app_info.xml I see that Lunatics_x32f_win32_cuda30_preview.exe is probably using it own (CUDA 3.0) dll's, cudart32_30_14.dll, cufft32_30_14.dll, libfftw3f-3-1-1a_upx.dll. Is there somewhere a CUDA 3.2.1 version of such dll's? Shall we soon expect 0.38 unified installer and optimized application based on cuda 3.2.1?

[Jason G]

DLL swapping is not advised as was practical in the past.  Nvidia has made them 'strongly versioned' as of recently, so jiggering things to use libraries other than what the application was built with is an 'at your own risk' affair... and is not guaranteed to work or be stable.

Cuda DLLs are, as always, found in the respective Cuda SDK.

There are a lot of improvements in Cuda 3.2, and I intend to make builds with it, though the benefits are mostly feature additions rather than performance of existing code, and use of these new features is yet to be implemented here.  We skipped Cuda 3.1 because of specific bugs, which apparently are resolved in 3.2.  That will allow development to continue, where it had stalled with Cuda 3.1 issues, forcing a return to Cuda 3.0.

Since the features that Cuda 3.2 adds require further development to make use of, it'll be 'a while' before any new release uses it.  Not a lot of point making Cuda 3.2 builds that are the same speed as teh Cuda 3.0 builds & don't use the new features    Using those will take time.

Jason

[Zeus Fab3r]

I have upgraded to 260.63 drivers, including nvcuda.dll (CUDA 3.2.1), supposed to be much better optimized for fermi then existing CUDA 3.0.

Pa mogao si i mene da pitaš  Jason ti je sve lepo odgovorio, nema više kao nekad skineš dll-ove i baciš u folder. Ja još uvek ne mogu da provalim ima li ikakve vajde da sa GTX260/191.07/cuda2.3 prelazim na 2xx.xx drajvere i x32f. Brzina je verujem ista a jedino se manje javljaju -12 greške. Nego kad pre nabavi Fermi ?ż 

pozz, ZF

[edit]sorry guys, it's my fellow countryman[/edit]

[MarkJ]

<snip>

Since the features that Cuda 3.2 adds require further development to make use of, it'll be 'a while' before any new release uses it.  Not a lot of point making Cuda 3.2 builds that are the same speed as teh Cuda 3.0 builds & don't use the new features    Using those will take time.

Jason

Would a cuda 3.2 version without any additional code changes make sense for the Fermi guys? Given the supposed speed increases on Fermi perhaps it might be worthwhile.

[Jason G]

Would a cuda 3.2 version without any additional code changes make sense for the Fermi guys? Given the supposed speed increases on Fermi perhaps it might be worthwhile.

Without additional code changes, probably not, or at least not much.  That's because the reported speed increases are primarily double precision and weird sized FFTs (which we don't use)

Having said that, after some downtime due to health, and issues migrating development enviroment to use some newer tools, I'm *almost* back up to continuing development.  There are new features in Cuda 3.2 to look at that will help on both Fermi and earlier gen cards,  considerable work on other areas of the code still to be done, and I beleive Raistmer has some other improvements from the OpenCL that may be applicable for incorporation.

So slow progress, but fairly solid nontheless.

Jason

[Richard Haselgrove]

Just been doing some long-overdue maintenance on my hosts, during the outage. Found an oddity on the NVidia driver download page.

NVidia UK (GBR) is offering 160.89 WHQL, released today, as recommended for GeForce 9, XP. But when you click through, it's still described as Beta, released last week. And the US site can't find it at all. Anybody got any experience with it yet, or should I stay one step behind?

[Jason G]

Hi Richard,
   160.89 is current WHQL according to the advanced search page: http://www.nvidia.com/Download/Find.aspx?lang=en-us

Given the mentioned release date is 18th October, My guess is they may even still be in the process of uploading it & editing the pages.

[Richard Haselgrove]

That's what I saw, but the search result doesn't match the download page.

They don't usually keep the same version number for Beta and WHQL, do they? Just wonder if there were any clues from your developer area access.

[Jason G]

That's what I saw, but the search result doesn't match the download page.

They don't usually keep the same version number for Beta and WHQL, do they? Just wonder if there were any clues from your developer area access.

  None that I saw, but checking again anyway...

[Edit:]  Nothing since Cuda 3.2 Release Candidate, which was issued ith 260.61 'development drivers'. 
My guess is 260.89 is primarily just a stopgap release for introduction of GTS 450 & GT 430 cards.  Haven't fully read the release notes yet though.

[Miep]

I think somebody might be hangover over there
I get this from the main DL search:

Verde Notebook Release 260 
Version: 260.89 WHQL
Release Date: 2010.10.14
Operating System: Windows Vista, Windows 7
Language: English (U.S.)
File Size: XX MB 
 
Note File size XX and release date 14th (beta/archive says 18th)
I'll let them sort that out before I grab it. Need to backup anyway before I mess with the driver...

Is Cuda 3.2 officially out then?

[Richard Haselgrove]

OK, I'll leave it until the US have woken up and the coffee has kicked in - they'll probably sort it out then. Plenty of M$ patches to be getting on with.

This is my debt graphing machine, which, once v6.12.3 is out, will be asked to run stable for the next six months or so. So I want to get it all done right in one go.

[Richard Haselgrove]

Is Cuda 3.2 officially out then?

Not, according to http://developer.nvidia.com/object/cuda_archive.html - still the RC from last month. But maybe they're planning to sort that out after coffee, too. 

[Jason G]

Is Cuda 3.2 officially out then?

Not, according to http://developer.nvidia.com/object/cuda_archive.html - still the RC from last month. But maybe they're planning to sort that out after coffee, too. 

I'd say so.  That RC was originally available through registered developer channels only, but seems publicly accessible now.  Probably it remains tagged with RC because 'we've' yet to find issue with it... So not a lot of point renaming everything & potentially breaking something that seems to work.

It supposedly fixes the problem in Cuda 3.1 that Ghost and I were able to isolate with Mixed Fermi/Pre-Fermi cards (That I had an open ticket on, yet for us to verify.).  So it'll have some tasty features to make use of (Fermi especially, but pre-Fermi also) over the last usable release for production builds (3.0).  It's quite a technobbery jump though, so it's going to take 'more than a quick rebuild' to use all that stuff.

[Edit:] handiest page in NVida's site is http://developer.nvidia.com/object/cuda_archive.html  ,  3.2RC is in the latest release section along with poor broken 3.1 (pretty obscure breakage, I know  ).

[Richard Haselgrove]

[Edit:] handiest page in NVida's site is ....

... the one I linked?