qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-devel] [v2 1/2] cutils: add avx2 instruction optimization


From: Juan Quintela
Subject: Re: [Qemu-devel] [v2 1/2] cutils: add avx2 instruction optimization
Date: Thu, 12 Nov 2015 12:30:26 +0100
User-agent: Gnus/5.13 (Gnus v5.13) Emacs/24.5 (gnu/linux)

Paolo Bonzini <address@hidden> wrote:

>
> The main issue here is that you are not testing whether the compiler 
> supports gnu_indirect_function.
>
> I suggest that you start by moving the functions to util/buffer-zero.c
>
> Then the structure should be something like
>
> #ifdef CONFIG_HAVE_AVX2
> #include <immintrin.h>
> #endif
>
> ... define buffer_find_nonzero_offset_inner ...
> ... define can_use_buffer_find_nonzero_offset_inner ...
>
> #if defined CONFIG_HAVE_GNU_IFUNC && defined CONFIG_HAVE_AVX2
> ... define buffer_find_nonzero_offset_avx2 ...
> ... define can_use_buffer_find_nonzero_offset_avx2 ...
> ... define the indirect functions ...
> #else
> ... define buffer_find_nonzero_offset that just calls
> buffer_find_nonzero_offset_inner ...
> ... define can_use_buffer_find_nonzero_offset that just calls
> can_use_buffer_find_nonzero_offset_inner ...
> #endif

My understanding for this was that glibc is better than hand made asm,
and paolo4_memzero (or whatever was it called) was the best approach.
And just remove SSE.  Have I missed something?


Later, Juan.



reply via email to

[Prev in Thread] Current Thread [Next in Thread]