Daniel Lemire
Daniel Lemire
> So I don't really see the point of disabling BMI2 It is not enabled by default. So we are not disabling it. However, it is interesting to check whether...
@Validark > This now compiles. Is it faster? Thanks. I will review and assess. We might introduce a kernel between icelake and haswell if the performance gains are sufficiently clear....
> They aren't used much, but they do appear in simdjson. It is an empirical question. By how much does enabling BMI2 in LLVM or GCC improves performance. As I...
@Validark At least under some versions of GCC, and some hardware, enabling BMI2 appears to lower the performance. It does increase slightly the instruction count. See https://github.com/simdjson/simdjson/pull/2243 For LLVM, it...
@Validark Even here, in this PR, you can see that our CI reports a performance degradation: However, I am not giving up. I think that if we are sufficiently careful,...
@pauldreik Interesting. I am going to check it out!
I have traced it and I cannot understand what the issue might be. It looks like a false positive to me. https://github.com/simdutf/simdutf/pull/636
@pauldreik @clausecker Is this a confirmed issue?
@madscientist Can you help review?
I will leave this open. I am not at all pushing to get this merged.