Quote from: bigjme on April 16, 2014, 10:33:55 AM

Quote from: cbuchner1 on April 16, 2014, 10:28:46 AM

Quote from: DrAlco on April 16, 2014, 09:47:24 AM

Quote from: bigjme on April 16, 2014, 06:22:47 AM

What version of cudaminer is it i keep seeing? Being every screen shot lately reports the users temps?

It's the latest compile from source files on github. There was an update on github ~2 days ago. If you compiled from that you should see the new temp readings. That said, the latest compile is a little slower for me compared to the previous version (scrypt-N at least, haven't tested other algos).

The NVML api only works with the 64 bit cudaminer version on windows. nVidia doesn't provide a 32 bit NVML DLL.

Now unfortunately the 64 bit cudaminer mines a bit slower in scrypt and scrypt-jane... I wonder if it is possible to combine 32 bit CUDA device code with a 64 bit host code?

most likely, most of us will be running 64 bit windows anyway so the 32bit code should run the same as normal, and the 64bit run as its meant to. you would have to try it as i am unable to compile 64 bit due to visual c++ express Sad

usually I have better result with x86 (to the level I usually don't compile the x64 version)
I believe too some big hack is possible at this level to make the api work with the x86 version of cudaminer... however it won't work with native windows x86 (in case some are still using it...)

I saw an email saying that Cuda 6 was out, anything to gain using it, or will it be like the RC ?