Re: MMX based IP-checksumming patch, 2.1.105, RFC

Tkil (tkil@scrye.com)
19 Jun 1998 13:28:18 -0600


[sent to linux-kernel and cc'd to cited author]

>>>>> "mingo" == mingo <mingo@hal.cobaltmicro.com> writes:

mingo> [...] I'm posting it here because maybe someone out there has ideas
mingo> how to prevent the quite expensive FPU_SAVE / RESTORE
mingo> operations somehow ... the routine itself basically clobbers
mingo> only 2 MMX registers.

mingo> [...] but for most RL MTU's the FPU save/restore operation eats
mingo> a considerable amount of the saved cycles ...

has this been tried on an AMD K6 yet? i thought they had a fast
MMX/FPU switch, an order of magnitude faster than the P5MMX. or is
this switch independent of the FPU save/restore times?

just a thought,
t.

p.s. if you would like me to try this myself, i'll try to do so in the
next few days. i'm getting a k6-2 whenever i can convince the
UPS people to give it to me...

-- 
Tkil * <URL: http://www.scrye.com/~tkil> * hopelessly hopeless romantic.
  "So amplify this little one 	|   She hears as much as she can see
   She's a volume freak       	|   And what she sees, she can't believe."
        -- Catherine Wheel, _Happy Days_, "Judy Staring At The Sun"

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu