Apologies if I've missed a previous relevant post, but after upgrades that have taken me from GT430's through a GT640 to (currently) a GTx750 coexisting with a GTX640, I was expecting to see some cuda50 tasks drifting that PC's way (my other GTX750 box mostly runs cuda50).

I know the scheduler may have some preconceived notions about the machine because I kept the ID on upgrading. So SETI may think its still the older hardware. Or it may be looking at the 640 and being cautious, if SETI does cautious.
Is there any way bar endless patience to get the scheduler to trial some of the 'newer' CUDA tasks on an upgraded machine?

---

When running two different GPUs from the same family SETI will report you as having two of the more powerful GPU. In your case the GTX750 is not only the newer (in terms of technology) but also the more powerful, thus it will report as having two GTX750.
Using the stock applications thee is no way to, initially, select the version of CUDA that is to be used. Until the server has learnt what your GPUs are capable of it will send you a few dozen or more of each of the CUDA types.

One way, and not the easiest to do is to install "optimised" applications, and only install the CUDA 50 application. If you are feeling bold, then download the correct installer from a site like [url=http://mikesworldnet.de/download.html[Mike's World[/url] then run it to install your desired versions of the CPU and GPU applications.

Please, take note: Make sure you have turned the Windows drivers update service OFF, and are running a version of the Nvidia driver from Nvidia NOT Microsoft - there have been no end of problems with MS sending out drivers that cripple GPUs from the point of view of using them for doing calculations.

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?