Intel FUD Versus AMD Fact

H

HardOCP News

[H] News
Joined
Dec 31, 1969
Messages
0
InfoWorld’s Tom Yeager has a thought provoking editorial posted today titled “Intel FUD versus AMD fact.” The key word here might be “provoking” as I am sure Intel fans might not be too happy with what Mr. Yeager has to say.

Fear, uncertainty, and doubt (FUD) marketing hits harder now than ever, when investors and analysts fancy themselves armchair experts in semiconductor process technology, and it’s not in AMD’s character to fight Intel’s FUD with fire. It’s trying to fight it with facts instead, hoping -- perhaps unrealistically -- that someone will listen.
Click to expand...
 
I enjoyed the article. Obviously I have to trust what he says about AMD, but it's renewed my hope for them. I like his angle that AMD goes at their own pace, not in reaction to Intel. They change their methods only when it becomes necessary in order to satisfy newer technology demands.

Hopefully everything he said is true. AMD may find some air to breathe afterall.
 
Interesting article.. but it's weird then that AMD went with hypertransport at a point when the FSB still hadn't outlived its usefulness. Everybody knows that the 2ghz HTT in single-core A64's on DDR ram is very underutilized as far as bandwidth goes. Now HTT is great when we're talking quad-core, and at that point FSB technology becomes a limiter... but the fact that AMD was using HTT beofre it was "necessary" seem to run couter to their arguments about not going 65nm and 45nm until it's "necessary."
 
AMD is clearly the leader in Hot Air Marks 2007. Yeager is certainly entitled to his opinions, which is what that article is.
 
What I care about is that there is always competition. They will trade off the top spot to each other repeatedly as the years go on but the fact that they compete against each other makes us consumers winners.

Honestly, I like AMD better than Intel. But I don't want to see Intel fail either. I will usually buy the speed winner of the year. 2007's winner is Intel. I'm hoping however that 2008 is AMD's year again as I think their new procs have more potential than Intels.
 
interesting article for sure. i try to remain neutral on it, but i really liked how he talked about intel boasting bigger wafers where AMD using better yields instead.

it reminds me of people talking about energy efficientcy and thats all they look at. a former roomate had a hard time grasping that in a room where we had both electric baseboard heat and a small gas fireplace, the gas fireplace was a better deal cost wise. very true that (believe it or not) the electric baseboard is much more efficient and he couldn't look past that. i also took cost of fuel (gas vs. electricity) into account and then the heating costs change.

i like to say to people this: do you want 40% of $10 or 10% of $100?
 
From the article...

As a result, Opteron and Athlon FX, AMD’s two leading chip lines, have years-long stability in power, cooling, and socket specifications, enabling competition among third-party chip-set and motherboard makers.
Click to expand...

Socket lasts for years? Tell that to my 939 opty :\ Not much to take from that opinion piece IMO
 
hegulator said:
Interesting article.. but it's weird then that AMD went with hypertransport at a point when the FSB still hadn't outlived its usefulness. Everybody knows that the 2ghz HTT in single-core A64's on DDR ram is very underutilized as far as bandwidth goes. Now HTT is great when we're talking quad-core, and at that point FSB technology becomes a limiter... but the fact that AMD was using HTT beofre it was "necessary" seem to run couter to their arguments about not going 65nm and 45nm until it's "necessary."
Click to expand...

I think AMD used HTT with their A64 line because it was already developed for the Opteron (servers benefit greatly from HTT). And why not go ahead and implement it now, get familiar with it while the tech is there and ready to use, so you don't run into problems later when the FSB does become obsolete? That makes a lot of sense when you think where AMD is trying to come from.
 
TruthInRuin said:
I think AMD used HTT with their A64 line because it was already developed for the Opteron (servers benefit greatly from HTT). And why not go ahead and implement it now, get familiar with it while the tech is there and ready to use, so you don't run into problems later when the FSB does become obsolete? That makes a lot of sense when you think where AMD is trying to come from.
Click to expand...

Bingo. AMD does not design desktop parts and has not since K8. They adapt their server parts for desktop with a memory controller and that is about it. This is a very simplified statement, but it does strike to the heart of the conversation.
 
He is full of shit,totally full of shit,its that or Anand and Kyle are as they are saying vastly different things from this idiots article about the CTO 'summit'

Read up on Anand's CTO summary,and 'final thoughts' and then tell me this is the real deal.Someone feels pity for AMD it seems.Is this the business world or a love affair ? :p
 
One of the rare AMD/Intel discussions that hasn't devolved into rabid fanboyism. Yet.

I agree most strongly with the poster who said that continued competition is the key for consumers, not the performance of one particular company's chips. In light of the fact that Intel still owns nearly three quarters of the x86 processor market, I really have to hope that AMD continues to expand.
 
am i the only one who thought of the PS3's Cell processor as soon as it said "partnership with IBM"?

AMD and IBM on the leading edge?

this i'll have to see
 
True.

Their MO for the past couple of years has been to introduce their new architecture first in the server market, and migrate variants of that technology to the desktop market a quarter or two later.

It makes sense from a profit perspective, as that is where the highest margins are. It makes sense from a publicity perspective, because the top dog in the server world garners considerable press. The technical perspective is more subtle, but just as important as the preceding two, as it allows them to leverage to great advantage, investments already made in the server segment, to compete effectively in the desktop/mobile markets. From business perspective, what could be better than the cost savings incurred from NOT having to maintain design and engineering teams for each separate platform.

I have marveled at Intel's achievement with the Core2, but I have always pulled for the underdog, and remain a believer that while AMD's development cycles are not exactly in synch with Intel's, they are still competitive, and just waiting for the timing to come together to return them to the top. Expecting AMD to match the pure manufacturing capability of a monster like Intel is unrealistic, at least presently.

Perhaps in the future, their innovation and continued competition in the market will allow them to expand their manufacturing capacity to the extent that changing processes (200 to 300 millimeter wafers, or .65 to .45) won't be such a strenuous venture. But I rather like the fact that they are able to squeeze more out of each process through innovation, and engineering. They have had great success in wringing every bit of performance out of their FABs and their 200mm wafers. I am sure at the present time, they do this because they have to. In the future, when they are a better match for Intel's capacity, let us hope they continue to do it because they want to.

I hope in the next couple of months, we will be hearing about how the Phenom family of processors will vault AMD back into the performance lead, where they will hopefully remain until Intel takes their next technological leap. That is competition, and it is a necessary part of this, or any other business in order for the consumer to remain a factor. Without it, the outrageous price of FX and Extreme Edition processors could become the norm for much more mid-line models.
 
Hopefully AMD has something for Intel by the time Intel release their new chips because I've been looking for an upgrade for a while, and so far, this is going to be my very first upgrade with an Intel CPU.
 
Haha, wow who has been passing out the weed at AMD? They are clueless. Honestly the truth is they can't compete in terms of money or resources. Intel screwed up with p4, they admit it. They left a window of opportunity for amd to excel. However, with penryn then nehalem don't think they are sitting on their ass now. They got a damn good wakeup call. AMD's problem is simple, their current execution sux. Delay after delay. Just cause the problem is simple doesn't mean the solution is though. Think they might be laying off some more people soon.
 
I sincerely hope AMD do get their act together; for my part, Intel's planned flood of the market for Q3 looks like a tasty upgrade. Honestly, CPUs are ahead of GPUs in terms of what we need (GPUs, while evolving quicker than CPUs, just arent enough for current games/apps whereas few games outside of SUpCom are CPU limited) so if AMD releases a better mouse trap after I upgrade to Intel *shrug* so what? I hope they do to keep competition going but I feel Intel has just owned this round with tis C2D and C2Q.
 
Before we start getting to full ourselves at AMDs expense, let us not forget that the Core2 is basically a re-engineered variation of the old Pentium III. A damned good re-engineering, true enough. But it is proof positive that Intel is not infallible. They are a huge company, so they can afford to bleed cash over mistakes like the Pentium 4. And lest we not forget a little thing called RAMBUS, where Intel took it upon themselves to attempt to dictate the memory standard to the rest of the industry?

AMD has made their bones by getting the most bang for the buck out of their design and engineering, and yes, capitalizing on the mistakes of a competitor. Sounds like how business is done in a capitalistic environment to me. They have had their missteps, no doubt. But it hurts them a great deal more than the same situation at Intel.

I can certainly understand that there are those that would use nothing but "Intel Inside", but the lack of respect for AMD is kind of surprising. The fact that they did not have an immediate answer for Core2, and the are scorned and scolded. They have been doing more with less, for such a long time, and have remained in the game, that alone deserves some kind of recognition.

I have a lot of respect for the fact that AMD scrapped 6 months worth of R&D on a new processor design that was heavily invested towards optimizing multi-threaded applications, to the detriment of single threaded applications. They read the handwriting on the wall, and figured out they were headed in the wrong direction (at least for the present and foreseeable future), and started over again. When Intel was faced with a situation like that with the Pentium 4 and their seemingly endless pipelines to accommodate speed grades that they were never able to reach, they went ahead with it anyway. This instead of sticking with a......re-engineered and improved version of the Pentium III architecture.......which is where they have come back to years later with the Core2, and struck gold.
 
Axiomatic said:
What I care about is that there is always competition. They will trade off the top spot to each other repeatedly as the years go on but the fact that they compete against each other makes us consumers winners.

Honestly, I like AMD better than Intel. But I don't want to see Intel fail either. I will usually buy the speed winner of the year. 2007's winner is Intel. I'm hoping however that 2008 is AMD's year again as I think their new procs have more potential than Intels.
Click to expand...

Hear hear! I just wish that OEMs would follow the same path that you take: speed winner of the year gets the contract.

In addition I welcome other CPU manufacturers to join the fray (VIA?), we as the consumers can only benefit.
 
People either don't know, forget or like to ignore the fact that Intel has FAB capacity that dwarfs AMD's. It's the same for the engineering budget and just about everything else. Yet AMD took on Intel and for a while kicked it around. This is partially due to AMD's engineering ideas and Intel's engineering ideas. AMD decided to go with a more efficient approach while Intel went with the brute force approach. AMD ended up winning that round.

We've already seen the results of Intel's rushing of process shrinks. Anyone else remember Intel's original foray into the 90 nanometer arena? It was supposed to usher in lower power and lower heat for the P4 line. What happened instead? The damn Preshott. Part of the problem with heat was the new P4 stepping but the process change didn't help the power consumption and heat output like it was supposed to. AMD waited until they had the 90 nanometer process to a point where it would actually make a difference. I think I remember them stating they had no need for 90 nanometer at the time Intel introduced it because AMD chips were still running cooler than Intel at 90. When AMD did debut 90 nanometer, it had none of the problems that Intel had.

Since I have no clue what the author of the article was shown (and even if I did I wouldn't be able to comment on it) I can't back his statements and opinions 100%. However, the article does follow AMD's track records and simply states that what he's seen indicates to him that AMD is continuing with the track record it has had over the last few years.

AMD has always played it's cards close to it's chest and is continuing to do so. To do any long lasting damage, I think Barcelona would have to be one stinker of a chip. I base this on the fact that trends in the CPU world move slowly. If they didn't, would Intel have survived the P4 era once the A64 and Opteron came on the scene? Intel and AMD had been going back and forth on performance since the PIII/Athlon days. I don't see any indication this trend will end anytime soon. AMD has gained more attention, respect and marketshare of the x86 market than any previous competitor to Intel.

Intel is pushing technology at a fast pace right now. This has definitely been good as it has birthed the Core architecture among other things. However, pushing the bleeding edge can have it's pitfalls. We saw this with their switch to 90 nanometer as I mentioned before. AMD has generally preferred to to wait it out a little until they have everything in place for release instead of getting into a pissing contest with Intel which could cause them to make a ruinous mistake. The author of the article is saying nothing more than this is what AMD is currently doing.

You can praise someone for being the first to market with something all you want. That normally does not matter in the long run in a case such as process technology. It's the implementation, the execution and the end result success which will matter.

 
So basically, everyone, like Tom Yeager, has their opinion about AMD and is convinced of its truth. ;)
 
The points cited are not fiction. They are just a quick recital of recent history. How you interpret them, and their impact on the debate, greatly influences each of our definitions of "truth" on the subject.

My main issue is that AMD was the hot child when they were beating the crap out of Intel. Hardly anyone had anything bad to say about them. They were "David" taking down the giant! Now that Intel has come roaring back, as Intel is apt to do when backed into a corner, they are practically writing AMDs epitaph, and wondering how the hell they are even still around.

I think they have earned a certain degree of respect for the past endeavors, and perhaps a bit of trust that they could perhaps pull it off again. I believe that any success that Intel has with its Core2, happened as a direct result of their competition with AMD, and would not have happened without it. Remember, if you will, the days when Intel was all alone in the field. When new architecture was YEARS apart, and incremental speed grades were HUNDREDS of dollars higher than their predecessors. Without AMD in the picture, and in a competitive way, we could easily see those days again.

Bottom line, when Intel succeeds, it is no biggie because that is what people expect of them. They are Intel for gods sake! When they screw up, folks scratch their heads and wonder how it could have happened. Put AMD in those situations, and the expectations are exactly reversed. If I have a bias towards AMD, it is mainly because of the way they compete. Leaner, more focused, and refined. They don't have the cash reserves, or the FABs to spare. They are the underdog, expected to fail. But when they manage something that gives Intel a swift kick in the ass, it is all the more satisfying to see them succeed, and the competition continues in earnest. Which is the way it should be.
 
hegulator said:
Interesting article.. but it's weird then that AMD went with hypertransport at a point when the FSB still hadn't outlived its usefulness. Everybody knows that the 2ghz HTT in single-core A64's on DDR ram is very underutilized as far as bandwidth goes. Now HTT is great when we're talking quad-core, and at that point FSB technology becomes a limiter... but the fact that AMD was using HTT beofre it was "necessary" seem to run couter to their arguments about not going 65nm and 45nm until it's "necessary."
Click to expand...
The front side bus isn't so great for memory access, and AMD has had a key advantage in accessing ram with lower latency through the onboard memory controllers. Once you give up the front side bus, and HT was already developed (to connect multiple CPUs together), they simply added that. I doubt they lost much sleep in shipping a bus that is "too fast".

Wumpus
 
I was giving him the benefit of the doubt until I read
and safe overclocking are uncommon among Intel-branded chip sets
Click to expand...
. Can someone explain that comment? I've used Intel chipsets such as 965, 865, 440BX (and others I can't recall) in my overclocked systems through the years and I can't ever recall the chipset being an issue for me.

And regarding this
AMD claims that its yields lead the industry
Click to expand...
, can anyone point me to actual numbers on this? I am curious as to what they are.
 
I searched with Google for "AMD yield data" and "AMD manufacturing yield data" as well as "Intel manufacturing yield data" and found no hard numbers for you.

It is most likely because they do not want to allow yield data to affect market prices for processors. For instance, if we knew the yield data for Intel's processors, and it turned out that their yields give them a manufacturing cost per processor of, say, $25, how would the market react? Everyone would hold out for the price to drop.

AMD is likely the same way. Would they be able to sell those FX chips for $800 if we all knew it only costs them $50 to make them? Well... yeah, some people would be that gullible. But not as many as when we don't know.

Obviously, this means we can't just take them at their word, but given AMD's track record compared to Intel's, I would be more inclined to believe them over Intel. That may be just my opinion, but logic led me to that opinion, not emotion or fanboyism... contrary to what my sig suggests. :D
 
[H]ocusPocus;1031075384 said:
And regarding this , can anyone point me to actual numbers on this? I am curious as to what they are.
Click to expand...
Yield numbers are closely held secrets. AMD is making an unverifiable claim and some suckers, i mean journalists, are buying it as fact
 
pxc said:
Yield numbers are closely held secrets. AMD is making an unverifiable claim and some suckers, i mean journalists, are buying it as fact
Click to expand...

Figure out how many FABs Intel is using for churning out CPU cores and compare that to how many FABs AMD is using for churning out CPU cores then match that to the numbers of CPUs they put out (and don't forget to take into account 200mm wafer vs 300mm wafer) and you'll probably get a rough idea.

Personally, I don't doubt the statement that AMD has higher yields. I remember it being stated before by AMD and Intel's response wasn't exactly contradictory. Not to mention that just because AMD has few FABs in comparison to Intel, they have to get the absolute best yields possible to put out volume to stay competitive.

We have no hard numbers but I've never seen any proof to dispute the claim.

 
Well thanks for looking anyway. I did some Google searches too and got the same kind of results as you. To be honest, I didn't expect to find such numbers released officially, but thought maybe someone had some really good estimates. But given multiple product lines, wafer size, process size, inventory levels, fab maintenance schedules, etc. it might be difficult to estimate within any meaningful margin of error.
 
one thing that affects yeilds is what happens to the chips with defects.

with amd, you very often see chips in semprons that were originally dual core dies, and had half the cache disabled. while the defects are still there, amd simply turned the defect off and sold the otherwise functioning core

to be honest, i don't really see intel doing this all that often, if ever. for that reason alone, i believe amd is right in saying that they have better yeilds

who has the lower defect density? very good question ;)
 
(cf)Eclipse said:
one thing that affects yeilds is what happens to the chips with defects.

with amd, you very often see chips in semprons that were originally dual core dies, and had half the cache disabled. while the defects are still there, amd simply turned the defect off and sold the otherwise functioning core

to be honest, i don't really see intel doing this all that often, if ever. for that reason alone, i believe amd is right in saying that they have better yeilds

who has the lower defect density? very good question ;)
Click to expand...

For one thing, Intel was the one that started disabling cache to make Celerons. They're still doing it across the lineup. If they can disable cache, they can disable cores. The question is not if they can, of course.

You're right about not seeing Intel disabling bad dual-core dies to make single cores. Their yields are so high, they don't need to. But, they eventually do which is why we've seen single-core Celeron-M roll-outs from both Core Duo and Core 2 Duo lines along with Core Solo which was very low volume. But, while the CD Celeron-M 4xx came early, the C2D Celeron-M 520 and 530 are just coming out. It just shows you how high their yields are.

In the past, Intel underbinned and crippled Celerons just to have processors at lower price points to compete against Cyrix and AMD. It's the reason why early Celerys hit 50% OCs while PIIs maxed out at near the same frequency. It's been discussed to death back in the day.
 
You must log in or register to reply here.
Back
Top