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: Tony Viola who wrote (61628)6/13/1999 10:55:00 AM
From: Steve Porter  Respond to of 1574043
 
Tony,

While I wasn't there (the dinner) I do believe that AMD has kept it short and sweet (as you put it) between the latches.

Steve

PS. It looks liek June 14th has surfaced as a date again.. (www.aceshardware.com). I'm not sure what AMD is doing right now.



To: Tony Viola who wrote (61628)6/13/1999 12:42:00 PM
From: Scumbria  Respond to of 1574043
 
Tony,

I thought you said the K7 was all by-hand laid out. Wasn't that at least partially for speed optimization purposes?

I think it was done exclusively for speed optimization. My comments about speed work were based on the fact that it is a very immature design, and that Dirk said they hadn't pushed the clock speed very hard. Once they get on the e-beam, they will be able to start knocking the real speedpaths down quickly.

Did Meyer talk about design rules for the K7, like rule of thumb maximum number of logic levels between latches, or clocks? You have always been impressed by the deepness of the pipeline in the K7, did they also keep it "short and sweet" between the latches?

Good question. Dirk said that they went to great lengths to make sure the pipeline was evenly partitioned. Without the 3-cycle L1, even a 20 stage pipeline would be useless for improving MHz.

Scumbria



To: Tony Viola who wrote (61628)6/13/1999 1:02:00 PM
From: Ali Chen  Read Replies (1) | Respond to of 1574043
 
Tony Viola, <You have always been impressed by the
deepness of the pipeline in the K7, did they also keep it "short and sweet" between the latches?>
Such a remark from a seasoned "professional" like
you should not be tolerated, and I am very sad
that all my efforts to educate you (and mostly Ph.D Yousef)
have failed to significant extent. The answer
to your question is:

"The sole reason for deep pipelines is to keep
it "short and sweet" between the latches!"

But you are doing a good progress though.
"short and sweet between latches" is a good
concise description of the issue.
Congratulations.