Folding Forum

Art, did you miss the stated 233 days to complete the WU? That's 0 ppd, even on UNI work units.

7im wrote:Art, did you miss the stated 233 days to complete the WU? That's 0 ppd, even on UNI work units.

No, I did not miss that, but that was on a different system:

BIG_RED wrote:I also got v6 to fold on an emulated x86 linux client. If i remember correctly it was would have finished in 233 days. The things to remember are that it was going through an emulator, running on top of android and for some reason it was only using have the cpu, but still it was neat to see it run.

I predict that Stanford will NOT provide a client that uses any kind of emulated x86. Even if there were a "jillion" IOS devices and an equal number of Android devices, everything is currently being aimed at compiler-generated parallel code using native instructions (e.g.-SSE2). Nobody is going to be interested in folding at a fraction of the speed of a Pentium M for 1/3 at a rate of a couple dozen PPD -- earning maybe 8- 9 points in 8 hours with no chance of a bonus -- even if WUs are completed before the final deadline.

Feel free to correct my estimates if you have more definitive data.

The emulated x86 linux client was on the same nexus 10. The difference in bogo mips was about 6x faster on the ubuntu arm client(997 ubuntu to 176 x86). I have run hardinfo 0.5.1 benchmarks here are the results for ubuntu arm.

Cpu blowfish 19.359, cpu crytohash 85.519, cpu fibonacci 4.721, cpu n-queens 12.642, fpu fft 11.015 fpu raytracing 16.837

Not sure what benchmark would be more comparable to fah. We are still about 2 years from doing anything on arm, but the gpu is were it's at but no drivers yet to benchmark the gpu. Once the drivers are out i will try to get the folding benchmark to run.

Edit: Looks like they just released the mali OpenCL SDK today. So I will see what happens and post in the Anything goes section if there is any big news.

bruce wrote:I predict that Stanford will NOT provide a client that uses any kind of emulated x86. Even if there were a "jillion" IOS devices and an equal number of Android devices, everything is currently being aimed at compiler-generated parallel code using native instructions (e.g.-SSE2). Nobody is going to be interested in folding at a fraction of the speed of a Pentium M for 1/3 at a rate of a couple dozen PPD -- earning maybe 8- 9 points in 8 hours with no chance of a bonus -- even if WUs are completed before the final deadline.

Feel free to correct my estimates if you have more definitive data.

Agreed. From the title of the topic and the posts here, especially this one:

7im wrote:No work is being done on those platforms. Stanford has limited development resources, and they are currently focused on the most powerful processors. Quad core CPUs, and GPUs, and the PS3. If they ever have the spare resources to work on clients for less powerful processors, I'm sure they would consider all possible options.

it seemed to be just another Thought experiment, not intended to be taken as a serious suggestion for FAH/Stanford/PG.