Not trying to pick holes, so feel free to ignore me. The P6T7 actually has *** 2x *** nf200 chips as well as the X58 under that rather large heatsink. Paragraph continues ....... Hmm. That's more than a little misleading. How about .... The P6T7 has seven physical PCI-E 2.0 x16 slots, although how many actual PCI-E 2.0 lanes will be allocated to each depends on which slots and how many are populated. Slot 7 will always have 16 lanes regardless of how the other slots are configured. Slots 1,3 and 5 (the blue primary slots) will be allocated 16 lanes each if slots 2,4 and 6 (the black, secondary slots) are not used. In the case that slots 1 through 6 are occupied, each will be allocated 8 lanes. So if you use 4x double width cards and plug them into slots 1,3,5 and 7, each will be allocated 16 lanes. If you use 7 single slots cards, slots 1 through 6 will be allocated 8 lanes, and slot 7, 16 lanes. I'll shut up now.
Could I visit and bring some for you to try (assuming you aren't in some far-flung location) ... and a flexible riser 'card' so we can see what happens with more than 8 GPUs. Not to be picky either (but it does show how closely we read your articles), Asus describe the board as CEB (12 inch x 10.5 inch) not ATX (12 inch x 9.6 inch)
I've been having an offline conversation with an nVidia developer and it's still the case, even with with the latest 190.xx and CUDA 2.3, that the hardcoded limit of 8 GPU's is still there. So the drivers will not recognise more than 8 GPU's. Ha, ha. Yes, nearly an extra inch, just like the R2E. Although the 9 mounting holes are standard ATX layout.
The CPC/bit-tech office is in central London, not that means anything if you're writing from outer Mongolia. More seriously though, unfortunately as JackOfAll has already intimated CUDA is still hardcoded with a maximum limit of 8 GPUs at the moment.
43 The 8800 GT referred to earlier is causing the rig to overheat, so despite a bit of fiddling with the fans last night, I've had to disable it
Dude... Where are you getting all these from? Do you have an nVidia tree at the bottom of your garden?!
Damm been found out lol, I wish It would would grow me some 295's! Lets just say these 260's I didn't fund directly out of my own pocket - basically I had a little bit of on the side work / big favour to a mate which raised a bit of cash which should of been going towards my Evo's gearbox rebuild but I had a weak moment as I seen some reasonably cheap Asus factory overclocked 260's No more now though and I do mean that this time, I now have 16 GPU's and 11 CPU's folding which should be good for 100,000 ppd average and that was my target and im giong to stick and maintain that
Well credit will be given where credit is due. You've now built a farm similar in size to the one that I intend on building in a single rack later this year. Output wise you're spot on with me too, about 100K PPD would make me a happy chap! Fold on!
Back to: 42 GTS 250 Roll on 1 million points! GPU (GTS 250) and SMP (Athlon X2 4200) clients on Ubuntu 9.04 GPU (GTX 260) and SMP (Core 2 Duo E8400) clients on Windows 7 RC [folding]un=Dr_MikeS&t=35947[/folding]
Countdown to 360 GPUs well no gpu folding for now.....Ive deceided when i build a server NO GPU FOLDING so the board should be cheeper my ppd is going to take a hit so when the WU gets done ill probably reboot in vista till i get some help with gpu foldin on linux.......I tried, but the idea of folding bonus wu and 3 gpus folding was very tempting
Whassup with OCUK 39 - GTS250 (1GB) from OCUK who I have found to be excellent over the years ... seems to be some negative vibes towards OCUK in this forum? Hey I just hit 2 million points