Out of interest is any body else folding using W7RC? I have some peculiar anomalies which people may be able too shed some light on. I am running 2 dual core 1.8Ghz Opterons with a single 295 GTX 2 gpu clients running on the GTX 4 seperate 6.23 clients running on the CPU's in the configuration tab I have backed off the CPU Clients from 100% to allow some usage by the GPU cores. Currently running 5911 work units GPU0 reports 6524.93 GPU1 3089 under FAHmon any ideas? My other issue is that I changed to individual clients for the CPU's due to any work previously using the SMP client automatically restarted at 0% if the machine is rebooted unfortunately the same is true of the individual clients all at 50+% rebooted the machine. All clients started at the beginning
SMP client only just completed on time in the 1st instance so I dont think VM would help plus I have no experience of VM not really what Im looking for as a resolution
While I don't have dual GPU clients running on Win 7, I would suggest increasing the priority on the slower GPU rather than decreasing the CPU usage percentage. I had a similar issue with a 295 that had me uninstalling and installing various drivers until it resolved itself.
morning vince, well ive been running 2 smp clients with vmware and 2 gpu console clients on the same machine (i7 920 + 9800gx2) with win7 rc for about 3 weeks or so now with no real issues plus ive never used vmware before which was easy to setup.i had to use vmware as i could not get smp client working with win7. cheers spunky
Hi mate, As the other guys said, I'd run a SMP 6.23 client on the CPU - no need for VM and no different to running one standard client, just a few more parameters to mess about with (and an administrator account to install it!) As for the GTX 295, I had exactly the same issue with mine running in a Dell 670 dual Xeon box - one GPU at 7,000ppd and the second at 2,300ppd. It was doing my head in, and what I ended up doing was downloading RivaTuner and running it to set the fans to 100%, as I thought it was a heat/throttling issue, but no. However, there is a hardware monitoring icon which shows you loads of graphs, and it was whilst looking at that I noticed that the GPU0 was running at GPU: 574Mhz Shaders: 1242Mhz RAM: 999Mhz (i.e. stock speeds), but that the second GPU was way below this at like 200/800/800 or something (i.e. very below stock!). So, I went into RivaTuner and reset the GPU1 to match the GPU0 settings, expecting it to ramp up to 7,000ppd, which it duly did, and knackered the power supply, frying my RAID card in the process. Now I hadn't touched the Nvidia config on that card (mainly because I had no idea how to), but the second GPU had ramped down it's clocks for some reason of it's own accord, so it may be that if your PSU is up to the job, then using RivaTuner to turn the wick up on the second GPU would get you to the 14,000ppd I was hoping for from the card. Of course, you have been warned about the power requirements!!! Hope this helps, Craig...
Thanks for the response Craig I am positive I have no issues with the psu as I changed an OCT 600W recently for a much beefier Unit. Does Rivatuner work with W7RC now? SMP unit has been run before but only just completed in time hence running individual clients my big issue is that the units comletely start from the begining if the machine is rebooted for any reason losing all the work that has already been completed
...errr, don't know about that - I'm still messing about in XP32. Whether it is via RivaTuner or the Nvidia driver controls, the thing to look out for is an unspecified underclock on the second GPU, pulling the ppd down. I can only assume it did it on mine because of the underpowered PSU in the Dell, as I haven't the foggiest idea how to mess about with the overclock settings. Maybe try GPU-Z to see what it reports for the GPU1 clocks? They'll probably have found the cures for all these diseases before I get onto Vista, never mind trying W7RC
Nope Rivatuner does not currently work in W7RC have not found anything yet to monitor cards or look at card settings have been across in Wales so had no time.
Try EVGA precision. Its not as good as Rivatuner (which is out of date the mo anyway) but does the job and works in Win7 x64. I use to look after my 9800GX2
Thanks fpr that have downloaded 1.71 set both cores to standard setting mem shader etc. Downloaded GPUZ aswell. Still have anomalies between the 2 cores in FAH Mon 1 core running project 5911 6019 PPD the other core running 5905 3624 PPD What are ideal temps for the GPU's on the 295 I would like to apply a mild overclock to up the ppd per core any suggestions?
OUCH! Thats mega hot mate, if I see any of mine go much over 80 I start trying to think of ways to keep them cooler but @ 104 I would be just waiting for it to die! Have you tried taking the plastic case off the GX2? all 3 of mine ive removed their cases which drops the temps by a good 10-15 degrees if you got decent case cooling...
Sorry Vince, didn't realise you were still having 295 problems. Currently getting 7.8K ppd per GPU running a mixture of 384 and 768 pointers at 1450 shader clock on XP64 rigs. 511s get 6.2K and 353s 8.2K.
Crikey thats way better than mine are you using XP Pro? What are all the settings your using for your overclock?
Yes, using professional x64 edition. I'm using EVGA precision to set shader clocks to 1450 with core and memory set at stock. I've got fans on 100% because I've two rigs, in close proximity, with 3 x 295s in each and two AMD 940 CPUs running 4 vms between them. This typically keeps GPU temps between 60 and 80 degrees C depending on WU and ambient temperature. The strange thing I've found is that I get more ppd from the 295s when the CPUs are loaded with the smp clients than when not Edit: I've corrected the 353 pointer ppd above. Finger trouble.
