http://www.tgdaily.com/content/view/32076/136/

AMD's CTO, Phil Hester, in an interview with THG's Wolfgang Gruener accuses Intel of copying AMD. Of course there is some merit to this statement, but the gauntlet is thrown, so to speak.

Among other things, he calls Intel old school. For weeks AMD says nothing and when they do speak up.. well...

Getting on to more pressing and less mudslinging, he outlines AMD's position on Fusion illustrated by this road map:

Going back to derision for a bit, he chides Intel on "pushing nanometers" as a metric.

I thought it was because Intel was on crappy Netburst...

Comments?

Great interview and some very interesting comments from Hester. Generally, I believe he is right. But the deciding factor (as I'm sure he is aware) isn't who comes up with what first. It's who is first to market. That's not always true (i.e., Itanium), but I think for the next couple years it will be.

If AMD can navigate it's debt and put product delays behind them, they'll be very successful in 5 years. But if the delays continue... forget it. Intel is too quick of a company. They've shown consistently they can adapt to AMD's business models and ideas... esp w/ x86-64.

Another yawn...AMD needs to shut up and get off the obsession with Intel and focus more on delivering products as opposed to paper launches.

Bold words coming from the company that invented the whole x86 architecture. [/sarcasm]

Another fine example of Tom Yager's 'non reactionary' publicity

What happend to the wonderful AMD of the XP days? Less BS talk and more asskicking products. Nowadays it more asstalk and less kickin' products.

I wouldn't say most of his accusations are correct.

Intel did adopt some of AMD's technologies, but AMD also adopted Intel's technologies. The most notable is the shared cache technology.

As for AMD64, I would disagree with him that Intel copied AMD. Intel and AMD have an agreement to share licenses. However, AMD refused to share data on AMD64. In order to maintain a unified 64bit extension, Intel have no choice but to use AMD64 extension. I have no doubt that Intel can develop a 64bit extension on its own, but a unified extension is a lot better than different extensions.

IMO, the laughable part of this interview is that most of the technologies CTO accused Intel of weren't even developed by AMD. Hypertransport is developed by Hypertransport Consortium, which AMD is only a co-developer. I have no doubt IBM might be the main player behind this. IMC is also developed by IBM, implemented by AMD.

For Fusion, I believe Intel already released some technical data on putting a graphic processor on the CPU itself, while for AMD, its just white papers.

AMD have some pretty good ideas of its own, but accusing its competitor of copying all of the technologies from them is a little too...low.

EDIT: If AMD doesn't change this thinking, and start doing some major innovation, expect them to sink lower in the next year or two.

Those licenses do not extend to new technology. The only reason SSE is allowed to be copied by AMD is because a judge ruled the original MMX and SSE extensions couldn't be patented. And of course, for MMX and SEE to be successful they have to be "open source." So AMD (naturally) uses them.

Furthermore, x86-64/AMD64 was reverse engineered by Intel, which US patent laws is completely legal.

Tha seems a very astute assessment Jesse, though a lot of Phils comments, such as. "we are in a very competative position" and "you have two healthy competitors here" seem a bit....optimistic.

After seeing how much Intel is Inside (sorry, couldnt resist ) K10 courtesy of Anands K10Uarch review, and knowing how much each company has adopted from the other, the copying comments seem a bit...contrived. What do you think about the whole copying thing Jesse? Bollucks, no big deal-each companies been doing it right along, or if it werent for AMD, Intel would still be pushing 8086s?

a n y b e n c h m a r k s y e t? :x

I don't see where in that interview that Phil said Intel copied the HyperTransport. He only talked about the x86-64 and Fusion that Intel have copied, which is true.

In general, avoiding the little bit of mudslinging, I believe this was a good interview. There are good points, Intel are really pushing the Nm race, like they pushed the GHz race... Intel isn't the performance leader at the moment because of 65Nm, it's because their new uArch is better then the older AMD uArch... Put C2D on 90Nm and it would still perform similar. The Nm race is just another Intel ploy to squeeze AMD harder (Hell, if my company was smited for 3 years by the competition, when I regained the lead, I would want to smite them as hard as possible too ).

This is an interesting war none the less. All the power to AMD, keep Intel true, and push them hard.

I don't see where in that interview that Phil said Intel copied the HyperTransport. He only talked about the x86-64 and Fusion that Intel have copied, which is true.

In general, avoiding the little bit of mudslinging, I believe this was a good interview. There are good points, Intel are really pushing the Nm race, like they pushed the GHz race... Intel isn't the performance leader at the moment because of 65Nm, it's because their new uArch is better then the older AMD uArch... Put C2D on 90Nm and it would still perform similar. The Nm race is just another Intel ploy to squeeze AMD harder (Hell, if my company was smited for 3 years by the competition, when I regained the lead, I would want to smite them as hard as possible too ).

This is an interesting war none the less. All the power to AMD, keep Intel true, and push them hard.

Are they though, really? Intel was at 65nm for a fair while. AMDs 90nm kicked the bejeezus out of Intels pre C2D 65nm. I could have missed it, but I havent seen Intel publically placing the same impetus on nm as they did Ghz, and I cant see as they would have wanted to draw attention to it back when they were being spanked. For the most part, from my perspective (which could easily be wrong) theyve been pushing to get to 45nm rather than pushing 45nm or 65nm WRT PR. There is an undetermined possibility that K10 @ 65nm will pull a K8 on Penryn and beat it like a drum. I think Intel knows this and realizes they could wind up with egg on their faces yet again. From what Ive seen, they have been worrking to get to the lower node to stay ahead of AMD, not use it as a sales piont, except WRT power usage. Internally, they are obviously pushing for the lower nodes due to increased yields, to be sure, but unlike clockspeed, I dont recall any " buy Itel because we're smaller" ads. This nm PR stuff Hestor was remarking on feels more like internet rumour mill/out of context Inquirer stuff, than the hard core blue man "Intel at 3.4GHz" ads

Well, even though *we* think of Intel "smited" for 2+ yrs by AMD, in the marketplace, it was perhaps 6 -9 months, because all that Intel advertising really works!

Tha seems a very astute assessment Jesse, though a lot of Phils comments, such as. "we are in a very competative position" and "you have two healthy competitors here" seem a bit....optimistic.

After seeing how much Intel is Inside (sorry, couldnt resist ) K10 courtesy of Anands K10Uarch review, and knowing how much each company has adopted from the other, the copying comments seem a bit...contrived. What do you think about the whole copying thing Jesse? Bollucks, no big deal-each companies been doing it right along, or if it werent for AMD, Intel would still be pushing 8086s?

Certainly there's been a long, long history of the companies copying one another. However, I'd have to say that since the original Athlon AMD hasn't really copied any significant technology from Intel. Having said that AMD hasn't copied because quite frankly, up until the C2D, Intel's technology sucked. I mean let's get real here, it wasn't terribly difficult to do better than the Pentium 4. One thing is for sure though: AMD's technology and influence has matured 10 fold since the original Athlon.

As we all know, these things happen in cycles. The question this time around is how long AMD's cycle will last. Some may say it's already ended, but I disagree. That judgement can only be made after Torrenza and Fusion have debuted. And regardless of Intel copying Fusion, Torrenza, etc, they've got some GREAT stuff on the horizon. Demostrating an 80 core 2 teraflop chip is astounding and certainly doesn't copy anything AMD's got going on.

The key thing here is this: both companies need to figure out a better and easier way for programmers to write parallel software. Specifically, they need to give the industry better tools. The number one thing programmers are bitching about these days is the work and time that goes into writing a program that uses SMP. Thus far I haven't seen either company make any significant steps to assist developers on this issue. I would liken the problem to 64 bit computing. No one wanted Itanium because it sucked in 32 bit and the developer tools were cumbersome. The in came AMD w/ x86-64, no hit in 32bit performance and the x86-64 tools were simple. Both companies have a HUGE opportunity here to gain wide spread support from developers if they do more in this area to ease the burden. Of course, the answer isn't that simple. Getting developers to use one set of tools over another for 2 different processor brands would be difficult... especially since the industry is almost exclusively on x86 now. No one wants back the days of compiling different OS's for PowerPC, Alpha, SPARC, AND x86.

Itanium was no where near the first VLIW processor
(http://en.wikipedia.org/wiki/Very_long_instruction_word - Transmeta, all DPS chips, MIPs, etc...), but it's the very "widely available/used" one. It's still complete garbage BTW.

AMD's tech probably isn't earth shattering... only the first ones to mass produce it for consumers.

Drooling over mass market technology is like drooling over Yugos and Civics. I'd rather drool at Bugatti Veyrons...

I don't see where in that interview that Phil said Intel copied the HyperTransport. He only talked about the x86-64 and Fusion that Intel have copied, which is true.

In general, avoiding the little bit of mudslinging, I believe this was a good interview. There are good points, Intel are really pushing the Nm race, like they pushed the GHz race... Intel isn't the performance leader at the moment because of 65Nm, it's because their new uArch is better then the older AMD uArch... Put C2D on 90Nm and it would still perform similar. The Nm race is just another Intel ploy to squeeze AMD harder (Hell, if my company was smited for 3 years by the competition, when I regained the lead, I would want to smite them as hard as possible too ).

This is an interesting war none the less. All the power to AMD, keep Intel true, and push them hard.



The nm race is very, very important. The way I see it, reducing the size of the cpu helps in all three factors
price, power consumption and speed
It drastically reduces heat and $ and can be very effective for raising clock speeds
The only thing Phil Hester doesn’t like about nm progress is that he is behind



AMD, PUT UP OR SHUT UP

They had one for the HD2900XTX. 8O

It burned a hole clean through the test bench.

I know, but my comments were strictly confined to AMD vs Intel products. 64 bit processors had been around a long, long time before Itanium. But Itanium (or IA64) was supposed to eventually replace 32 bit processors, something no one had attempted before. This was the significance of Itanium and thus why I said "first to market."

Right. And you make a good point about manufacturing; Intel is definitely the leader in that. I can't think of a single instance where AMD was ahead of Intel in this particular arena.