Messages in this thread | | | Subject | Re: [PATCH -mm crypto] AES: x86_64 asm implementation optimization | From | Andi Kleen <> | Date | Wed, 16 Apr 2008 10:23:04 +0200 |
| |
"Huang, Ying" <ying.huang@intel.com> writes: > > En. I have no AMD machine. So I have not tested the patch on it. Maybe > there are some pipeline or load/store unit difference between Intel and > AMD CPUs. Tomorrow I can split the patch into a set of small patches, > with one patch for one small step. Can you help me to test these patches > to find out the reason for degradation on AMD CPU.
It would be also quite possible to use two different implementations, one for AMD another for Intel. crypto frame work should have no problems dealing with that.
-Andi
| |