More timings

From: Alwyn Jones (alwyn@xray.bmc.uu.se)
Date: Tue Dec 10 1996 - 08:41:23 CST


spin tests in seconds, 100 updates
test		O2		Impact 		500/50T	/40T 	250/40T
		R5000/ 		R4400/		400Mhz		233MHz
		180MHz		250MHz
11000 vecs	4/5		2/3		5/5	6/7	7
36000 vecs	9/10		6/7		7/7	10	17
sketch_auto	8/9		4/4		4/5

now we have the powerstorm 50T in the alphastation500/400MHz. it has almost
the same speed as the impact in the standard test. if you zoom in/clip, the
36,000 vector times drop to 4 secs for the 500/50T, stay the same for the
impact.

for those that may be interested, here are some cpu tests on windows NT
systems, running a molecular dynamics code from johan aqvist. the number is
proportional to the speed of the machine, so big is best.

Alphastation 500/400 Mhz, DEC Unix, DEC Fortran: 100
Alpha XL 433MHz, no sec. cache, WindowsNT 3.51, DEC FORTRAN:123
PWS Pentium Pro 200 Mhz, Windows NT 3.51, Microsoft FORTRAN: 62

in swden, we can get 3 alpha XLs for the cost of a 500/400 workstation



alwyn



This archive was generated by hypermail 2b29 : Tue Mar 20 2007 - 03:13:19 CST