Re: [CFT] faster athlon/duron memory copy implementation

[Daniel Egger <degger@fhm.edu>]

[-- Attachment #1: Type: text/plain, Size: 1582 bytes --]

Am Don, 2002-10-24 um 19.15 schrieb Manfred Spraul:

> Attached is a test app that compares several memory copy implementations.
> Could you run it and report the results to me, together with cpu, 
> chipset and memory type?

SiS 735, Duron 1200, 512 MB PC133 (running at 100Mhz).

Athlon test program $Id: fast.c,v 1.6 2000/09/23 09:05:45 arjan Exp $ 

copy_page() tests 
copy_page function 'warm up run'         took 19860 cycles per page
copy_page function '2.4 non MMX'         took 21205 cycles per page
copy_page function '2.4 MMX fallback'    took 21262 cycles per page
copy_page function '2.4 MMX version'     took 19893 cycles per page
copy_page function 'faster_copy'         took 12746 cycles per page
copy_page function 'even_faster'         took 13112 cycles per page
copy_page function 'no_prefetch'         took 10217 cycles per page

Being interested in seeing how the Via Ezra system here performs I also
ran it there but experienced three segfaults in the last three tests; 
two of which I can explain, but no_prefetch is a stranger right now.
Anyway:

PLE133, Via Ezra 667 Mhz, 128 MB PC100 (probably at 66Mhz)
egger@tanja:~$ ./athlon 
Athlon test program $Id: fast.c,v 1.6 2000/09/23 09:05:45 arjan Exp $ 

copy_page() tests 
copy_page function 'warm up run'         took 23213 cycles per page
copy_page function '2.4 non MMX'         took 34971 cycles per page
copy_page function '2.4 MMX fallback'    took 34958 cycles per page
copy_page function '2.4 MMX version'     took 22774 cycles per page

-- 
Servus,
       Daniel

[-- Attachment #2: Dies ist ein digital signierter Nachrichtenteil --]
[-- Type: application/pgp-signature, Size: 189 bytes --]