3 ms·

I've never found cachegrind inaccurate, but maybe I'm not doing hardcore enough performance work. You can also use perf and get you numbers straight from the ha

I've never found cachegrind inaccurate, but maybe I'm not doing hardcore enough performance work. You can also use perf and get you numbers straight from the hardware if that's what you need. Truth be told I mainly use cachegrind because I prefer kcachegrind's UI to hotspot. (I even prefer cachegrind's approach since the numbers will be less distorted by other random background activity on the machine, but that could just be idealism on my part, who knows.) If perf or the vendor-specific tools like vtune/uprof aren't sufficient for you then I'm curious what do you use?

nkurz 1y ago
I switched from emulator tools like valgrind to tools with hardware support like perf, pmu-tools, and VTune. I generally found them sufficient, but sometimes buggy and difficult to use. Cachegrind is occasionally inaccurate due to an inaccurate model, but the greater problem was that cache hit percentages only tell a fraction of the story. To be able to predict performance I often needed to be able to accurately measure things like the number of memory requests in flight. Searching now for an example, I hit on a comment I made here a few years ago where this new tool probably would have been helpful: https://news.ycombinator.com/item?id=18442131 https://news.ycombinator.com/item?id=18442131 In general I have much greater faith in the on chip performance registers. That said, other than glancing at news stories like this I haven't been keeping up with recent advances. I guess it's possible that cachegrind and friends have improved since I was using them.
- do_not_redeem 1y ago
  I've always reached for llvm-mca when I need stuff like that, but again it's all predicted/ideal numbers, not live off the hardware. And you need to start off with another tool first to pinpoint where to look. I've never come across pmu-tools, thanks for the tip. I'll try it out next time I'm in the trenches.

nkurz 1y ago

I switched from emulator tools like valgrind to tools with hardware support like perf, pmu-tools, and VTune. I generally found them sufficient, but sometimes buggy and difficult to use. Cachegrind is occasionally inaccurate due to an inaccurate model, but the greater problem was that cache hit percentages only tell a fraction of the story. To be able to predict performance I often needed to be able to accurately measure things like the number of memory requests in flight. Searching now for an example, I hit on a comment I made here a few years ago where this new tool probably would have been helpful: https://news.ycombinator.com/item?id=18442131 https://news.ycombinator.com/item?id=18442131 In general I have much greater faith in the on chip performance registers. That said, other than glancing at news stories like this I haven't been keeping up with recent advances. I guess it's possible that cachegrind and friends have improved since I was using them.

do_not_redeem 1y ago
I've always reached for llvm-mca when I need stuff like that, but again it's all predicted/ideal numbers, not live off the hardware. And you need to start off with another tool first to pinpoint where to look. I've never come across pmu-tools, thanks for the tip. I'll try it out next time I'm in the trenches.