Re: [perfmon2] AMD Fam10h PREFETCH_INSTRUCTIONS_DISPATCHED

2009-03-10 Thread frederick . ryckbosch
Hi, I've done some more testing. Using an example from the block prefetching paper from AMD, I wrote a test that contains a fixed number of prefetches. I've varied the number of prefetches and measured them. The resulting graph is attached. The counter seems to work, as the number of inse

Re: [perfmon2] AMD Fam10h PREFETCH_INSTRUCTIONS_DISPATCHED

2009-03-09 Thread Frederick Ryckbosch
Op 9-mrt-09, om 16:58 heeft Robert Richter het volgende geschreven: > On 09.03.09 15:37:38, frederick.ryckbo...@ugent.be wrote: >> I'm working on a program that dynamically inserts prefetches in >> program code. >> Now I'm trying to measure the number of prefetches executed by the >> program, wh

Re: [perfmon2] AMD Fam10h PREFETCH_INSTRUCTIONS_DISPATCHED

2009-03-09 Thread Robert Richter
On 09.03.09 15:37:38, frederick.ryckbo...@ugent.be wrote: > I'm working on a program that dynamically inserts prefetches in program code. > Now I'm trying to measure the number of prefetches executed by the > program, when I use: > > pfmon -u -e PREFETCH_INSTRUCTIONS_DISPATCHED:ALL ./myprog Wha

[perfmon2] AMD Fam10h PREFETCH_INSTRUCTIONS_DISPATCHED

2009-03-09 Thread frederick . ryckbosch
Hi, I'm working on a program that dynamically inserts prefetches in program code. Now I'm trying to measure the number of prefetches executed by the program, when I use: pfmon -u -e PREFETCH_INSTRUCTIONS_DISPATCHED:ALL ./myprog I get: 0 PREFETCH_INSTRUCTIONS_DISPATCHED:ALL I'm sure the prefe