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: kash johal who wrote (93565)2/16/2000 12:54:00 PM
From: Process Boy  Respond to of 1572336
 
kash -

1) We need some benchmarks. I hope we don't have to wait too long.

2) I'm pretty happy with the clockspeed on A-0 Si with absolutely no speedpath work yet (one of your conditions was if AMD can keep up MHz; we have a bet on this, so you know where I stand :-))

3) We need some benchmarks.

4) We need some benchmarks.

Oh, and we need some benchmarks.

PB



To: kash johal who wrote (93565)2/16/2000 1:14:00 PM
From: Tenchusatsu  Read Replies (2) | Respond to of 1572336
 
Kash, <Doesn't compare to Athlon. x87 Latency is terrible. Fxch(?) no longer free.>

I don't know what the implications are revolving around FXCH. But I can tell you that the Willamette guys are confident they'll smoke Athlon's floating-point performance.

<The implication is that the l2 cache runs at 1/2 speed - similar to cumine. Will be a disadvantage to AThlons full speed l2 cache.>

Do you know whether Athlon will have a full-speed L2 cache? Will Athlon also feature a full-sized back-side bus so that they can complete the transfer of an entire cacheline in one clock? Would it even matter, since latency is the bigger determiner of performance?

<DRDRAM all the way. I don't need to say any more do i??>

Intel's current stance is that RDRAM is excellent for workstations. Intel also believes that RDRAM will even be viable for the low-end market in 2001. (Case in point: Sony Playstation 2 uses RDRAM.) Therefore, Intel will continue to push Rambus for desktops and workstations.

For servers, however, it seems customers overwhelmingly want DDR over RDRAM. Therefore, it's likely that Foster-based servers will start out on DDR.

Tenchusatsu