I was talking to pm and he said he thought I probably wouldn't be able to write one... he's right of course. All my attempts result in lower CPU temps than prime 95 
.
Anyway, maybe some people have input as to what I'm doing wrong.
I'm trying to do an FPU & ALU operation every iteration, an L2 access every ~24 iterations, and a main memory access every 75-150 cycles. I figure that should provide some nice heating. everything would actually probably be about half that, since each "iteration" would probably be longer than 1 clock cycle... especially if you test for validity of results
Two problems with actually implementing that:
1. to actually produce useful results, you need to TEST everything to make sure it comes out right. how do you do that without reducing heating? (since I assume that the processor can make "mistakes" before completely locking)
2. I dont know how to force an L2 or main memory access without creating huge arrays and jumping around. thoughts?
.Anyway, maybe some people have input as to what I'm doing wrong.
I'm trying to do an FPU & ALU operation every iteration, an L2 access every ~24 iterations, and a main memory access every 75-150 cycles. I figure that should provide some nice heating. everything would actually probably be about half that, since each "iteration" would probably be longer than 1 clock cycle... especially if you test for validity of results
Two problems with actually implementing that:
1. to actually produce useful results, you need to TEST everything to make sure it comes out right. how do you do that without reducing heating? (since I assume that the processor can make "mistakes" before completely locking)
2. I dont know how to force an L2 or main memory access without creating huge arrays and jumping around. thoughts?
Facebook
Twitter