AMD's 'Shanghai' CPU Enters Production

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.

hellwig

Distinguished
May 29, 2008
1,743
0
19,860
I'm not sure why people claim AMD is in a terrible position. In the 80's and most of the 90's, they were either making exact duplicates of Intel chips, or poorly designed compatibles. When AMD released the K7/Athlon in 1999 they jumped way over Intel in performance. They held that performance lead until Intel released Core2 in 2006. As such. I don't see why AMD's current situation should be dire since they've only been without the lead for a couple years. I think AMD can and will excel past Intel again, just don't keep holding your breath every time a new processor rolls off the line, that's just dumb.
 
G

Guest

Guest
"I don't see why AMD's current situation should be dire since they've only been without the lead for a couple years."

It's call DEBT and >$5Billion of it. They have limited ability to raise cash; which has only worsened with the general worldwide credit tightening (not that AMD would likely qualify for credit even if this were not the case). This restricts AMD's ability to fund normal operating expenditures - hence job cuts, cuts in capital spending (which limits future capacity plans), cuts/slowdown in R&D spending.

AMD now pays $95Mil every quarter (or $0.38Bil/year) in pure interest expense. To put this in perspective, effectively ~$5-$6 of every CPU AMD sells goes just to paying interest.

People don't seem to understand business health is not merely measured by product performance. Perhaps now you can see why AMD's position may be a little more dire than just considering product performance gaps?
 
G

Guest

Guest
"The fact that AMD does not have this process right now might mean that IBM has not transferred a robust process to AMD or they are not sharing it yet.." (highK/metal gate)

IBM is now on record as saying they will not move to high K until 32nm for logic products, thus it would appear unlikely AMD would do so earlier (unless 32nm was way late and they had no choice). Despite AMD's claim, this is not a simple "drop-in" - there is new tooling involved, other process steps are impacted and would need to be re-tuned, products would need to re-certified (the gate oxide failure modes are different for HighK vs conventional SiON), and in all likelihood the CPU's would need to be re-layed out - meaning new mask sets and potentially a couple of steppings to debug.

As for the ultra low K, the press on this is just as misleading. AMD conveniently claims a 15% wiring delay improvement, which is completely true! Butttttttt, this will not translate to an overall 15% speed improvement as they are still limited by transistor switching speed! Hence you conveniently see no actual claim of overall speed improvement and you hear "enable greater processor performance".... well it is only enabling it if you also speed up the transistor (though this little factoid is conveniently omitted). If nothing is done to the transistor at the the time the magic ultra low K process is implemented you will see very little actual overall improvement - what will happen is you will get electrons to and from the transistor faster but you'll still be spending the majority of the time waiting for the transistor to switch on and off.

Hence the clever choice of words on all these press releases - everything is technically accurate and many people are completely misinterpreting it. Not truly AMD's fault, but they are somewhat complicit as they create these vague press releases and appear to be attempting to capitalize/prey on people's lack of technical understanding. If they wanted to they make these press releases a lot more direct.
 
G

Guest

Guest
I am still waiting for the AMD statement mentioning they are using metal gates in shanghai....

Now...penryn is already beating barcelona with and old FSB and without integrated memory controller. (see the latest CPU chart) If you add the IMC and QPI the performance boost must be huge.

Netburst vs athlon was a competely different history. Now intel has a clear advantage and AMD just have buggy chips and incomplete marketing statements :)

 

alpha754293

Distinguished
Aug 16, 2006
6
0
18,510
point of comparison between "Shanghai" Opterons and Core i7 is not a very smart one as you're comparing processors for VERY different market segments.

Most people just simply don't have the mind to be able to load a 4-core system to its fullest extent. And the people that can and/do, are probably also the same group of people whose demand for computational power will ALWAYS outpace what Moore's law otherwise allows (namely, the HPC, scientific, and engineering groups.)

Only a few have been able to successfully transition into this new programming and application development paradigm.
 

enforcer22

Distinguished
Sep 10, 2006
1,692
0
19,790
To the guy talking about AMD's debt. I believe they paid off that ATI aquisition or will have once the fab sale goes through i swear i read it paid off the 5 bill they barrowed to buy ATI. Could be wrong but im pretty sure if they still have debt its not near as close to 5 bill anymore.
 
Status
Not open for further replies.