[Mono-dev] Enabling Support for SIMD Intrinsics

Jerry Maine crashfourit at gmail.com
Fri Jan 23 11:48:58 EST 2009


Justin, I'm planning on working on that. You're welcome to help.

Rodrigo Kumpera wrote:
> Well, basically all of it is left. We didn't start it, but it should
> be pretty simple once the register allocator is dealt with.
>
> 2009/1/23 Justin Holewinski <shawmishrak at gmail.com
> <mailto:shawmishrak at gmail.com>>
>
>     How much work is left in the port to amd64?  I have only skimmed
>     the code for the Mono JIT compiler, but if there are some smaller
>     tasks that need done, I could try to help.
>
>
>     Thanks,
>
>     Justin Holewinski
>
>     2009/1/22 Jerry Maine <crashfourit at gmail.com
>     <mailto:crashfourit at gmail.com>>
>
>         Justin, you could help remedy that yourself and finish the
>         simd port
>         for amd64 in the jit engine or at least help with it.
>
>         Rodrigo Kumpera wrote:
>         > Yes, mono 2.2 and 2.4 will only
>         feature simd support for x86 32
>
>         > bits.
>
>         >
>
>         > 2009/1/22 Justin Holewinski <shawmishrak at gmail.com
>         <mailto:shawmishrak at gmail.com>
>
>         > <mailto:shawmishrak at gmail.com>>
>
>
>         >
>
>         > Is there a special configure script option to enable the SIMD
>
>         > intrinsic optimizations?
>
>         >
>
>         > I just compiled Mono from the 2.2 sources and compiled/executed
>         the
>
>         > following C# code snippet:
>
>         >
>
>         > Console.WriteLine("SIMD Mode:  {0}",
>
>         > SimdRuntime.AccelMode.ToString());
>
>         >
>
>         > This is printing out  "SIMD Mode:  None" regardless of the -O
>
>         > option to mono.  I have tried -O=simd and -O=all.
>
>         >
>
>         > I also tried the SIMD examples available at
>
>         > http://tirania.org/blog/archive/2008/Nov-03.html (linked from
>         Mono
>
>         > 2.2 Release Notes) and noticed no difference between the -O=simd
>
>         > and -O=-simd comparisons, and both were significantly slower than
>
>         > the C version.  This leads me to believe the code snippet above
>         is
>
>         > not just lying to me.
>
>         >
>
>         > This is on Fedora 9 x86_64 with an Intel Core i7 920 processor
>
>         > (definitely has SSE support!).  Is SIMD support only available on
>
>         > x86 (32-bit)?
>
>         >
>
>         >
>
>         > Thanks,
>
>         >
>
>         > Justin Holewinski
>
>         >
>
>         > _______________________________________________ Mono-devel-list
>
>         > mailing list Mono-devel-list at lists.ximian.com
>         <mailto:Mono-devel-list at lists.ximian.com>
>
>         > <mailto:Mono-devel-list at lists.ximian.com>
>
>         > http://lists.ximian.com/mailman/listinfo/mono-devel-list
>
>         >
>
>         >
>
>         >
>
>         >
>         ----------------------------------------------------------------------
>
>         >
>
>         >
>
>         > _______________________________________________ Mono-devel-list
>
>
>         > mailing list Mono-devel-list at lists.ximian.com
>         <mailto:Mono-devel-list at lists.ximian.com>
>
>         > http://lists.ximian.com/mailman/listinfo/mono-devel-list
>
>
>
>
>         _______________________________________________
>
>         Mono-devel-list mailing list
>         Mono-devel-list at lists.ximian.com
>         <mailto:Mono-devel-list at lists.ximian.com>
>         http://lists.ximian.com/mailman/listinfo/mono-devel-list
>
>
>
>     _______________________________________________
>     Mono-devel-list mailing list
>     Mono-devel-list at lists.ximian.com
>     <mailto:Mono-devel-list at lists.ximian.com>
>     http://lists.ximian.com/mailman/listinfo/mono-devel-list
>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.ximian.com/pipermail/mono-devel-list/attachments/20090123/9210d693/attachment-0001.html 


More information about the Mono-devel-list mailing list