Just been looking through the forum at - /Application Code Discussions/CUDA for Milkyway@Home/ thread, and see many possible answers to my question. However, I thought, with your patience, to ask the questions all together here.

BACKGROUND

I bought a couple of low riser nVidia GeForce 9,500GT 1024MB PCI (yes legacy PCI) graphics cards last year. The aim was to use them in an old dual P3 Win2K pro server for CUDA crunching.

I placed one of them in the primary graphics slot, and it would not boot as it interfered with the 3 PCI card Promise Mirror-1 RAID this rig sports.

I reinstalled the ATI Radeon 7,000 series card and placed the 9,500GT in the last empty PCI expansion slot with the same result.

PROPOSAL

I have another old dual Xeon server which currently crunches Milkyway using an ATI HD3850 in the AGP graphics slot.

The OS is 32bit WinXP Pro SP3 and I run BOINC Manager 6.10.13 (have for many months).

This desktop server has 4 empty full width PCI-X slots (not PCI or PCI-E but PCI-X), and I would like to place both nVidia cards in 2 of these slots (well spaced for cooling). The nVidia cards would initially run with attached screens, using VGA plugs, then later dummy video plugs (if they work).

QUESTIONS

1. Can GeForce 9,500GT cards crunch double precision for MW?

2. Are these 9,500GT cards compatible with CUDA 1.3 for the 0.24 CUDA application to crunch MW?

3. Can a mixture of ATI 3850 AGP (crunching MW now) and 1 or 2 CUDA capable nVidia cards coexist together for MW crunching?

If the ATI and nVidia cards can coexist, would it be better to crunch MW with the HD3850 and use the 9,500GTs for another CUDA capable project at the same time?

4. I understand that I would need to install a CC_info file to allow all cards to be recognised and then run (assuming I can get the right nVidia CUDA drivers installed)?

If so, can anyone point me to a CC_info file source, please?

5. I am sure there are questions I have missed, so thanks for the guidance (in anticipation).

John

---

Go away, I was asleep

Hi John
I can answer one of your questions at least. The 9500 gt cant do double precision for MW. I have one it gets pretty good credits ay Collatz though. Hope that helps a little anyway.

---

Yes it does, Bruce.

I will have to let Milkyway and Collatz run on the same PC. Collatz, hopefully on the nVidia cards and Milkyway continuing on the HD3850.

Unfortunately, previous experience says that Collatz takes over the HD3850 in preference to Milkyway, which may be an issue?

Clearly I will need to think this one through, and try and obtain a relevant copy of a cc_info file.

Long term, after getting my DNETC given credit to the same levels as I have for Collatz and Milkyway, I can split the cards to - HD3850/2 x 9,500GT to Collatz; the HD4850 to Milkyway; and, the HD5850 to DNETC?

---

Go away, I was asleep

Assign a different project prefs for that host and set Collatz to just run nvidia and set MW to only run ati. That way, MW will have the sole use of the ati card and Collatz will only use the nvidia cards.

Assign a different project prefs for that host and set Collatz to just run nvidia and set MW to only run ati. That way, MW will have the sole use of the ati card and Collatz will only use the nvidia cards.

I will remember to try that point, Haris Dublas.

i can set the MW preferences as at home, and the Collatz preferences for work. Hopefully I will get around it that way.

I would say I plan to spend a month or so with DNETC on the main ATI cards, then try this our when the total credit has grown to a decent amount.

---

Go away, I was asleep

Hi,
this is a copy of postings in the GPUGRID Forum and gives answers about compute capabilities:

The following are mostly compute capability 1.1:
(check versions for obsolete G80 GPU versions)

GeForce 9600 GT 512MB, est. 34GFLOPS to 37GFLOPS
Geforce 9600 GSO, 768MB (DDR2) est. 46GFLOPS
Geforce 9600 GSO, 384MB (DDR3) est. 48GFLOPS
GeForce 8800 GT 512MB, est. 60GFLOPS
GeForce 9800 GT 512MB, est. 60GFLOPS
GeForce 9800 GT 1024MB, est. 60GFLOPS
GeForce 8800 GTS 512MB, est. 77GFLOPS
GeForce 9800 GTX 512MB, est. 77GFLOPS
GeForce 9800 GTX+ 512MB, est. 84GFLOPS
GeForce GTS 250 1024MB, est. 84GFLOPS
GeForce 9800 GX2 512MB, est. 138 GFLOPS

COMPUTE CAPABILITY 1.3:

GeForce GTX 260(192sp) est. 85GFLOPS (120)
Tesla C1060 est. 93GFLOPS (131)
GeForce GTX 260 est. 96GFLOPS to 111GFLOPS (135 to 156)
GeForce GTX 275 est. 123GFLOPS (173)
GeForce GTX 285 est. 127GFLOPS (179)
GeForce GTX 280 est. 130GFLOPS (183)
GeForce GTX 295 est. 212GFLOPS (299)


The following GeForce cards are the present mainstream choice, with reference clocks,

GT 220 GT216 40nm Compute Capable 1.2 128 BoincGFlops peak
GT 240 GT215 40nm Compute Capable 1.2 257 BoincGFlops peak
GTX 260 GT200b 55nm Compute Capable 1.3 596 BoincGFlops peak (sp216)
GTX 275 GT200b 55nm Compute Capable 1.3 674 BoincGFlops peak
GTX 285 GT200b 55nm Compute Capable 1.3 695 BoincGFlops peak
GTX 295 GT200b 55nm Compute Capable 1.3 1192 BoincGFlops peak
GTX 480 GF100 40nm Compute Capable 2.0 1345 BoincGFlops peak
GTX 470 GF100 40nm Compute Capable 2.0 1089 BoincGFlops peak
GTX 465 GF100 40nm Compute Capable 2.0 855 BoincGFlops peak

Two new cards will work here soon,

GTX 460 GF104 40nm Compute Capable 2.1 907 BoincGFlops peak (768MB)
GTX 460 GF104 40nm Compute Capable 2.1 907 BoincGFlops peak (1GB)

Values are approximate, and unconfirmed here. There will be a large performcne variety of GTX 460 cards, as many do not follow reference design!

It is expected that a GTX 475 following the above GF104 architecture will be released in the Autumn - it should have a full complement of 384shaders and use all 8 GPU cores.

Hope that helps a little bit.

Regards
Alexander