SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Politics : Formerly About Advanced Micro Devices -- Ignore unavailable to you. Want to Upgrade?


To: Dan3 who wrote (103085)4/8/2000 3:39:00 PM
From: pgerassi  Read Replies (1) | Respond to of 1574883
 
Dear Dan:

A point: Athlon decodes into 3 uop PAIRS. Thus each Athlon uop does one ALU/FP and/or load/store. Thus it may be that it takes 3.5 Willi uops average to do a x86 instruction but maybe 2 Athlon uops. Since you pointed out that most code does one (or more) load and/or stores per x86 instruction (reason why CISC uses less instructions), it may show that Willi may not have the appropriate balance with typical applications.

Also, any time a CPU requires more than 1.5 to 2 times the Mhz to do the same task in the same time, it will become obvious to all users, even neophytes, that Mhz is no longer a good metric. The reason it is displaying a tenacity now is that Intel and AMD perform within 10 percent of each other at the same clock. So if Willi at 1.5 Ghz runs like a Mustang at 1 Ghz, people will buy a Mustang at 1.25 Ghz over a 1.5 Ghz Willi.

Pete



To: Dan3 who wrote (103085)4/10/2000 12:17:00 AM
From: tejek  Respond to of 1574883
 
Remember, the guys in this discussion who are most convinced that Willamette will be faster than Athlon were certain that Athlon would slaughter coppermine clock for clock.

Dan,

Don't you mean just the opposite: "that cumine would slaughter the Athlon clock for clock"?

ted