AMD and Intel General Discussion (not for getting help)

Page 27 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Status
Not open for further replies.
All Ive heard is that AMD has working 32nm parts, and has had them, and their last mention was Liano was being pulled in for an earlier launch.
This is nothing new here, is what Im saying, its to be expected, and praising Intel by doing this isnt really such a good thing, as they did it on a larger node, making it easier to begin with, and cutting their teeth at the lower nodes may make GF wiser quicker too, tho it wouldnt fit the Intel is better, theres gotta be exclusive problems scenario
 
Oh OK - I guess any implication that just maybe Intel did its homework and made the right choice intially, is "being stupid" or "isn't really such a good thing" nowadays :).

The gist of the article, if read carefully, is that the gate first approach runs into more serious problems at the smaller nodes.

Methinks the IBM/AMD mindset "If Intel does it one way, we'll distinguish ourselves by doing it the opposite way, even if stupid from an engineering viewpoint" is in full play here.
 
Ive read this (not this article) before, with those claims. There were earlier claims that HKMG was needed at 45nm as well. Like I said, lets wait and see, maybe Intel did the pooch here and IBMs way is better, we simply dont know, but I for one wont pay alot of attention to the same ones who said wed need it at 45nm.
Until its here, and running with yields and on si, we wont know.
Im not about to priase either side for eithers approach until both approaches are mature and here and comparable.
 

I have been working my Arse off. I have been logging in and reading but have had no time to post. Its been a hell of a ride lately.
 
^^ Why would we wait until the actual facts are known, when there is a history of rumbling about IBM/AMD's 32nm being less than stellar?? After all, you didn't wait until the facts were known before slamming Larrabee 😀, despite myself and others saying let's see more hard info vs. the rumors. Turns out the wait for confirmation was a matter of a few weeks, but I guess even that was too long for the AMD fanbois 😀.

And yes, you were correct in that instance, since Intel has canceled the first gen chip, but the final chapter is not written yet for that story either. Larrabee may never pan out as a commercial product, but the pioneering research will pay off in other venues, according to InfoWorld ...

Correct me if I'm mistaken, but wasn't it you and Jennyh who were touting AMD's 32nm with HKMG as bringing all sorts of transistor improvements to equal or surpass Intel's, some 6 months ago? Of course, that was even more preliminary guesswork on your part, but now that the pattern is becoming a bit clearer, all of a sudden we need to stop & wait for the actual product to appear?? 😀

And seriously - Intel "did the pooch" on gate-last? From other articles I've read, Intel has brought the P-channel transistor performance (historically the weak link in CMOS) up to the level of the N-channel devices on their 2nd gen HKMG. They have 2 fabs cranking out the 32nm at the moment, with 2 more by next summer. I'd say that, far from doing the pooch, Intel has demonstrated their continuing dominance as the world's leading fab process technology company, bar none! 😀

Personally I don't think the gate-first approach is going to be a show-stopper, but it is shaping up to be the 2nd-class choice, both yield and performance-wise, and possibly abandoned before 22nm....
 
“By January 2006, this process was yielding fully functional test chips. Now, products at Intel’s Fab 32 in Arizona are matching the yields with the same defect densities as our fab in Oregon,” he said.
http://www.semiconductor.net/article/207942-Intel_Takes_45_nm_HKMG_Process_to_IEDM.php

This article was written a year apart at least between Intels first working part, and seeing their yields do well in the Oregon fab.
IBMs 32nm was on par/slightly better than Intels, as per D Kantor, argue with him.
What I just showed, and what Ive been saying is, given time, more than a year, Intel too made decent yields on gate last.
There werent all these so called experts coming around and decrying :itll never work this way" etc at the time when Intel was in dev, which is what we have nowwith IBM.
All Im saying is, give them the same time, ignore the peanut gallery for now, see what happens, and then we will know.
This by itself isnt being treated the same, too many "experts"
Even without all those "experts", funny how Intel managed over time too?
So screw the "experts", especially a VP, like theyd really know

My whole point about doing the pooch would be not attributable to Intel, which is what theo "experts" decided to throw in in a pointless way anyways.
RDRs, you follow them, regardless, so the "experts" are full of it.
Gate first?
Its actually more conducive for many approaches over gate last, but again, you have to follow your RDRs, but all these "experts" feel the "need" to chime in, where the hell were these "experts" when Intel was first doing this?
The "experts" need to shut the %&&* up, as usual, too many "experts", and we need to see what comes.
 
As for LRB, what you may have considered rumors, and what many considered a superb showing for RT, well, I simply knew better.
I know far more about gfx than I do about cpus, and I was trying to point out those differences in a market sort of way alone to show how LRB is failing, let alone the HW and SW and Drivers etc. whic requires someone more familiar with gfx.
People continued to be hardheaded, not my fault, I tried telling them, showing them, reasoning it out for them, but hardheadedness I was still confronted with.
I didnt jump early, I gathered info as time went along, and there simply wasnt enough, and what there was wasnt convincing, actually the opposite, and thats when I really started to come down harder on LRB.
When Intel made their early claims about perf, they showed nothing to back it up, and never have doing raster.
The RT showings were weak and incomplete, and instead of people checking out what I was saying, they attacked what I said, because it wasnt what they wanted to hear, again, hardheadedness.
Maybe I wasnt early enough, and less people wouldve been surprised
 
I think an early sign that LRB was not going to be the "next great thing" was that parallel to the LRB hype was Intel beating the drum for game developers to dumb down their games to run on basically crap graphics solutions. That didn't strike me as someone with a Nvidia/ATI slayer in wait.
 
nvidiacomic.jpg
 
The use of Tesla S1070 by the similarity-defining services has significantly boosted the rate of identification of unknown files, thus making for a quicker response to new threats and providing users with even faster and more complete protection. During internal testing, the Tesla S1070 demonstrated a 360-fold increase in the speed of the similarity-defining algorithm when compared to the popular Intel Core 2 Duo central processor running at a clock speed of 2.6 GHz.

http://www.kaspersky.com/news?id=207575979
Looks like cpus are losing their place in alot of areas, with more coming
 
Yes definately with simple massively parallel tasks.

I doubt a Tesla card could best a Pentium III at booting up a system and loading windows and typing a simple letter though.

It would take many many more clock cycles than an X86 CPU.

Even if both were clocked at the same speed ... hence the P3 reference (bout the same as the GPU on a Tesla).

I could be completely wrong though ... frequently in fact !
 
How much of a lead does Intel have at 32nm?


Intel will soon be releasing their latest processor, code named Westmere, built on 32nm technology. Architecturally, the processor will be similar to the Nehalem design introduced last year and marketed as Core i7. The big change, however, will be Intel's new 32nm fabrication technology. Semiconductor Insights was fortunate to obtain the desktop version of the processor early and perform an independent detailed analysis of the Westmere/Clarksdale's process technology. At 32nm, Intel is introducing their second generation of high-k metal gate technology as well as other interesting process innovations. That being said, Semiconductor Insights has yet to observe a metal gate technology in a commercial device from any other semiconductor manufacturer. So, where exactly are Intel's competitors in the 32nm race?


To this end, let's compare the two main processor companies, Intel and AMD, and the leading pure-play semiconductor foundry, TSMC. All data presented here was independently gathered by Semiconductor Insights. Since TSMC does not have a product per se, we analyzed devices from a fabless FPGA manufacturer using TSMC technology. FPGA makers are usually among the first adopters of new technology from foundries like TSMC, since smaller geometries allow them to include more logic elements in their product and increase the overall performance and flexibility of their design. Also, note that we did not include the IBM/Common Platform technology due to the limited amount of independent data at this time. However, it will be interesting to see how the relatively new technology alliance will fare in the future against the reigning champion of foundries, TSMC.


To better understand that future, it's sometimes helpful to look at the past. Table 1 shows the dates of introduction for the 90nm, 65nm, 45nm technology nodes as well as the upcoming 32nm node from our three manufacturers. We can see that Intel has consistently introduced a new process technology at the beginning of the year, every two years. This pace is dictated by Moore's law, a trend Intel has diligently followed since first observed by Intel's co-founder, Gordon E. Moore. Intel has been so regular in the launch of new technologies, we can almost safely predict that the next 22nm process from Intel will be ready around the beginning of 2012.


AMD has also stayed on track with Moore's law for the past tech nodes, introducing new processes during the same year but usually later than Intel. However, a period of more than two years is now expected between the introduction of AMD's 32nm technology and the previous 45nm node first seen in late 2008. This slowdown may be partly explained by the transfer of AMD's manufacturing operations to GlobalFoundries but is most likely a sign of how difficult it will be to keep up with Moore's law beyond the 32nm node.
Click to expand...

Not exactly in-depth as the above IEDM article, but a bit interesting anyway...
 
A couple things stick out
"However, it will be interesting to see how the relatively new technology alliance will fare in the future against the reigning champion of foundries, TSMC. "
Yea, that 40nm is raining on that parade
Also, I find it interesting they didnt mention Intels cheating, almost breaking AMD, and then the delays because of lack of funds, and the delay of the NY fab, but if you wanna lick Intel, just keep being sweet.
Poor writing in including all the facts.
The NY fab, if AMD had had the funds, but couldnt since NO OEM COULD BUY FROM THEM...ahem, wouldve been doing 32nm by now, but was delayed.
Now, theyre making TSMC look better than they are, and so too Intel.
Those advantages for Intel are forever gone now, and just like they did for abu dhabi, they can drop another 10 billion into GF and just leave Intel behind, so to me, this entire article was short sighted, boastful of other underserving companies, and just what some people want to hear
 



That is absolute BS and you know it.
 
^ +1. I guess we'll hafta listen to the already tired old tune "but Intel cheated" to explain everything from process to design advantages for Intel...

@Jay - please explain how "but Intel cheated" made IBM & AMD choose a gate-first HKMG approach, after Intel demoed the advantages of gate last (esp. noting that 2 GigaPascals strain advantage for PMOS that the article Badtrip linked to shows)??
 
The statements "The NY fab, if AMD had had the funds, but couldnt since NO OEM COULD BUY FROM THEM...ahem, wouldve been doing 32nm by now, but was delayed." and "just like they did for abu dhabi, they can drop another 10 billion into GF and just leave Intel behind" don't really agree with each other. Wasn't AMD planning on spinning off its fabs in 2007?? Why would they invest in a new fab in Luther Forest, when they had another fab in Germany they could have retooled at half the cost anyway? If they had enough $$ to buy ATI, they had enough to retool the fab, or else sell more junk bonds to raise the cash.

But if they were going to spin off the fabs anyway, why not let the UAE spend the $$...
 
The new fab was planned way before 07, and would be on line now, with 32 here.
Tell me, who makes IBMs parts and AMDs parts?
There has to be a need to drive it, but again, the need was delayed for lack of funds, and just as the fabs in Germany have to be upgraded, the new fab wouldnt have to have been, thus putting the needs/abilities/product in better time than we had.
You can argue, hell, you argued Intel was innocoent until the end, go ahead, it doesnt matter.
I have said over and over again, I dont care, and its only being a fan where people get blinded, and I see it all for what it is, and not some BS story licking Intels boots as well as TSMCs.
No mention of GF, isnt that funny?
Havnt you questioned this at all?
Since when did AMD come close to having 65nm?
I dont recall, but a mere 10 months difference isnt that bad, with maybe 12.
They didnt mentioned except in praise of course, Intels HKMGs process, and how that took a lil more time for Intel as well.
They act like 22 will be as hard as 32 is inacting HKMG for the first time, all the while, praising Intel.
They praise TSMC, yet have basically nothing to say, you dont find this tid bit interesting?

I say it was a crap article, incomplete as to whats actually gone down and whats actually going on, as they havnt a clue from reading this article, and give no clues to other that do or will read this tripe.
Theyve totally missed the mark, praised Intel at every turn, praised TSMC, not sure why, and yet maintain, they know whats going on, by saying no one will keep up, whoever they are, whther its the "reigning champion" TSMC or Intel, and of course, will poor GF be able to keep up? Stay tuned next article for a possible conclusion!!!!
Tripe I tell yas
 


IIRC AMD, if they had not spun off their fabs, would have waited until NY coughed up the $1.4B for Luther Forest. I don't think that happened before 2007. AMD don't do nuthin' without a handout from somebody, be it the poor NY taxpayers or Intel stockholders...

You can argue, hell, you argued Intel was innocoent until the end, go ahead, it doesnt matter.
Click to expand...

No, I didn't. What I said was that Intel deserved its day in court, since all the facts weren't known at the time. All we *still* know for a fact is that some governmental bodies (not courts of law) have self-servedly fined Intel anywhere from chump change to some serious $$.

Now if the situation were reversed, and it was AMD getting the shaft from a gov't agency, you'd be crying foul all over the place 😀.

I have said over and over again, I dont care, and its only being a fan where people get blinded, and I see it all for what it is, and not some BS story licking Intels boots as well as TSMCs. No mention of GF, isnt that funny?
Click to expand...

I'd hardly accuse IEEE (parent org. of EE Times) of boot-licking...

Here's another EETimes article that you probably won't like 😀 :

GlobalFoundries takes on the world

Advanced Micro Devices Inc. spun off its manufacturing unit this year into an independent foundry company, rendering itself fabless. It was a seismic shift for the chip company, which once considered its fabs a competitive weapon against Intel Corp. But the move also altered the foundry business landscape.

AMD's move amounts to an admission that it could not keep up in the capital spending race with Intel. The company believed it needed to cut loose its fabs to ensure its survival. Time will tell whether the strategy works for AMD and what it will mean for the foundry business at large.


The spinoff, GlobalFoundries Inc., opened its doors with backing from both AMD and the Abu Dhabi government. It has fab lines running in Dresden, Germany, and a facility under construction in Malta, N.Y.

But the upstart company has seemed ill prepared to take up its desired role as a leading-edge wafer foundry. Early on, GlobalFoundries boasted, for example, that it had a high-yield 45-nanometer process running, whereas foundry leader Taiwan Semiconductor Manufacturing Co. (TSMC) did not.

But that process was essentially tuned for a single customer--AMD--and is based on silicon-on-insulator (SOI) technology, which is decidedly not mainstream.

By mid-2009, the new foundry provider acknowledged that it had only two customers: AMD and STMicroelectronics Inc. Even AMD's graphics chip unit, ATI, was still going with TSMC.

GlobalFoundries expects to have a mainstream, non-SOI 45-/40-nm process out next year, but by then it will be late to the game.

Realizing that the foundry needed to play catch-up, Abu Dhabi's Advanced Technology Investment Co. in September acquired Singapore-based Chartered Semiconductor Manufacturing Co. Ltd. for $3.9 billion. Over time, ATIC will fold Chartered into GlobalFoundries, bringing the new company a ready-made customer base and established processes and fabs--along with some much-needed credibility.

But nobody expects the merger to be easy. It will require the blending of corporate cultures and of geographically divergent teams. Some industry watcher foresee a brain drain, predicting that Chartered management and engineering talent will look to leave once the two foundry companies are integrated.

The tricky merger work will proceed without the guidance of former GlobalFoundries chairman Hector Ruiz, who stepped down in November. Ruiz is alleged to have given confidential information about AMD's GlobalFoundries spinoff plans to a defendant in the Galleon Group insider-trading case.

GlobalFoundries may one day give TSMC a run for its money. But that day isn't in sight quite yet.
Click to expand...
 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom