Interesting CUDALucas quirk

bcp19 · 2011-11-20, 01:46

Don't know if this will occur on other CPU's, but I have a Core 2 Duo E4500 @ 2.2GHz and I was doing some testing with CUDALucas. If I have nothing but CUDALucas running, it takes 1:18 to run 10,000 iterations, but if I run P95 with TF to low limits on the same core as CUDALucas, it drops to 1:07 for 10,000 iterations. Kind of a weird little quirk, but nice to use.

Dubslow · 2011-11-20, 04:44

Same core? You mean you set the affinity for CUDALucas? As far as I know, it should operate independently of the CPU.

LaurV · 2011-11-20, 06:35

For me it is no difference between "CL with P95" and "CL alone", except in the temperatures. What you describe is a bit weird, if CL is faster when both running. Maybe some odd settings? Please also note that the time per iteration is not constant, for example for me with a 56M expo, one iteration can take everything between 7.2 and 8.4 milliseconds, CL alone. This could be related to some FFT internal stuff (how big are the residues, how many zeros and ones do they have to multiply, etc, I have no idea).

bcp19 · 2011-11-20, 11:48

I have both cores running P9, with Cudalucas set to core 2, and task manager says CL takes up ~3% cpu.

It's been fairly consistent overnight, 6.71-6.73 milliseconds. Opening IE can cause it to jump up to 6.8.

2011-11-20, 01:46	#1
bcp19 Oct 2011 1010100111₂ Posts	Interesting CUDALucas quirk Don't know if this will occur on other CPU's, but I have a Core 2 Duo E4500 @ 2.2GHz and I was doing some testing with CUDALucas. If I have nothing but CUDALucas running, it takes 1:18 to run 10,000 iterations, but if I run P95 with TF to low limits on the same core as CUDALucas, it drops to 1:07 for 10,000 iterations. Kind of a weird little quirk, but nice to use.

2011-11-20, 06:35	#3
LaurV Romulan Interpreter "name field" Jun 2011 Thailand 24063₈ Posts	For me it is no difference between "CL with P95" and "CL alone", except in the temperatures. What you describe is a bit weird, if CL is faster when both running. Maybe some odd settings? Please also note that the time per iteration is not constant, for example for me with a 56M expo, one iteration can take everything between 7.2 and 8.4 milliseconds, CL alone. This could be related to some FFT internal stuff (how big are the residues, how many zeros and ones do they have to multiply, etc, I have no idea). Last fiddled with by LaurV on 2011-11-20 at 06:39

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
CudaLucas Residual	evoflash	GPU Computing	21	2017-11-13 12:04
Don't DC/LL them with CudaLucas	LaurV	Data	131	2017-05-02 18:41
CUDALucas gives all-zero residues	fivemack	GPU Computing	4	2016-07-21 15:49
CUDALucas: which binary to use?	Karl M Johnson	GPU Computing	15	2015-10-13 04:44
Trying to run CUDALucas on Windows 8 CP	Rodrigo	GPU Computing	12	2012-03-07 23:20

2011-11-20, 04:44	#2
Dubslow Basketry That Evening! "Bunslow the Bold" Jun 2011 40<A<43 -89<O<-88 1C35₁₆ Posts	Same core? You mean you set the affinity for CUDALucas? As far as I know, it should operate independently of the CPU.

2011-11-20, 11:48	#4
bcp19 Oct 2011 7·97 Posts	I have both cores running P9, with Cudalucas set to core 2, and task manager says CL takes up ~3% cpu. It's been fairly consistent overnight, 6.71-6.73 milliseconds. Opening IE can cause it to jump up to 6.8.