29 Oct
2002
29 Oct
'02
1:49 a.m.
At 08:47 AM 10/29/02 -0500, you wrote:
Hmmm, are you sure about this?
A 'real' DX-50 will clean the clock of a DX2-66, due to the true 50 Mhz bus.
If your seeing a DX2-66 beat a DX-50, then the DX-50 is not being run at
full (local bus) speed.
I second this opinion, at least if running Linux. I never made that
speed test under Winblows. This was in a machine with a 512K L2 cache;
the local bus speed could be configured to 33MHz, 40MHz and 50MHz.
Right now my daughter uses that machine, but it doesn't have that
486DX50 anymore; now it has one of those 3X/4X AMD 486 replacements
sold by PowerLeap; I think it is running at 3X with a bus speed of 40MHz,
for a 120MHz internal clock [1]. Quite speedy for a 486.
[1] Another test revealed that the combination 40MHz local/120MHz internal
was substantially faster than 33MHz local/133MHz internal. I would
have tried 50MHz local/100MHz internal, but I think it wasn't supported.
I suspect it might have been even faster. Another case of raw
internal clock speed not being an appropriate measure of system performance.
carlos.
--------------------------------------------------------------
Carlos E. Murillo-Sanchez carlos_murillo(a)nospammers.ieee.org