Slashdot Mirror
AMD: What Went Wrong?
Barence writes "In 2006, AMD could seemingly do no wrong. Its processors were the fastest in the PC market, annual revenue was up a record 91%, expansion into the graphics game had begun with the high-profile acquisition of ATI, and it was making exciting plans for a future where it looked like it could 'smash Intel's chip monopoly' for good. Now the company is fighting for its very survival. How did AMD end up surrendering such a advantageous position – and was it given an unfair shove on the way down? This article has plotted AMD's decline, including the botched processor launches, the anti-competitive attacks from Intel and years of boardroom unrest."
Intel has had its share of buggy and bad designs, and that's even without going into discussion of the HMSS Itanic. Some AMD chips do great job of bang for the buck, my laptop has a nice dual core one that made the cost much less than comparable Intel chip would.
Still, AMD needs to get more risky with heavy investment into more advanced design and fab. mediocrity just isn't tolerated in processor design.
AMD made good products too, they just made them for the wrong market. This is why commercial semiconductor manufacturing is so difficult. You give your engineers a set of constraints and then about 5 years later you have a product. Intel, back around 2003, bet heavily on laptops and power-conscious servers. AMD bet on desktops. Intel's market predictions were better and so the products that they brought to market turned out to be the ones customers wanted by the time the related products made it to market. AMD's were not. Of course, Intel only made this bet after seeing how badly they'd underestimated the importance of power consumption with the Pentium 4 so, if anything, their later products were an overreaction to the poor performance of the P4.
So, in summary, AMDs problem was that they didn't screw up the previous generation, so assumed that the next one could be more of the same and missed the industry shift to mobile devices.
I am TheRaven on Soylent News
Intel just succeeded HUGE. They few years of AMD dominance were more a result of intels missteps.
Bingo. The P4 was a dead end, Intel were betting on Itanium for the 64-bit market, and AMD just kept on building better x86 chips.
Once Intel realised they were falling behind, they dropped their brain-dead policies and pushed out better chips than AMD's.
Let's not forget the underhanded tactics that Intel used. They were forced to pay a minimal $1bn to AMD for it. I always thought its too small an amount for losing their position as leaders in the CPU market. And now look how things turned out...
AMD will never fail because Intel won't let it fail since it is their DOJ defence against being a monopoly. The couple of times AMD got ahead of Intel on technology, like x86-64, Intel started a money losing price war to put AMD back in its place. When AMD is struggling, Intel raises profit margins on its products to help them out. There are also less advertised ways Intel helps keep AMD afloat: Patent sharing, employee no-stealing, joint tools development like OpenAccess, etc. Having worked in that industry I was always surprised that the DOJ never came down on them for those agreements. The patent sharing and joint tools ones are official even though Intel puts like 10X more into them as AMD does. I left that industry after 5 years since I saw it as a dead end since you only have a few companies competing for your skills. As my manager at Intel told me, "I won't give you a raise since you only have one other place that would even care about the skills you picked up here, AMD and we really control them too."
Potentially, AMD is still favored by many people who don't mind tinkering. For instance, for under $100 you can get an AMD x4 with a top end of 3.8ghz or more. My development box as an AMD x6 that was $130 running all 6 cores at 4.2ghz solid. To buy anything comparable from Intel would be well over $400. It's the same story in server land. AMD vs Intel really depends on application. AMD has true physical core superiority. Intel bet on hyperthreading, and it works well for many projects, until you actually need 12 physical cores for number crunching and not just thread spawning. Then it's AMD by a mile.
I use Intel Xenons in my mid to low web/caching servers and I use AMD 12 cores+ in my data servers/larger VM hosts. It just seems to be the recipe that gets me the best bang for my buck, but to each their own.
Also Intel bribed big OEMs to use their processors instead of AMD's. Dell was an especial example of this: in the K7/K8 days they'd make noise every year or two about how they were considering selling AMD-based systems rather than being exclusively Intel, and those of us in IT who wanted /better/ computers would get very excited, but then Intel reliably came along and gave Dell an even better sweetheart deal on their CPUs, which was probably Dell's objective the whole time.
It wasn't AMD's fault for choosing the wrong market; they'd made a far better desktop and mobile processor than the P4, it was just that Intel was abusing its market position.
Hail Eris, full of mischief...
E pluribus sanguinem
The article is pretty explicit about how AMD dug its own grave. I don't think blaming an Intel monopoly is all that convincing.
Really? The article mentions how Intel managed to get Sony money to cancel ALL AMD shipments, and how they paid Dell roughly 3/4 of a billion dollars in a single quarter to not use AMD chips. But I'm sure you're right, I'm sure keeping AMD out of all of the major OEMs(except to some extent HP) had nothing to do with it.
If i had one dollar for every brain you dont have, i would have $1.
That campaign really had a lot of success. The only people who buy AMD are geeks who only do it when it gives a good price performance ratio. It does for me, going AMD simply means you can spend your budget on a fast SSD which will do a hell of a lot more for your performance then a faster more expensive intel CPU with a regular HD.
But people like me are the exception and AMD never really managed to remove "a computer has an intel inside" from the consumers mind. Just try your local electronic store.
Netbooks were a chance, AMD didn't put restrictions on its netbooks but they failed to push high end netbooks before Intel again stole their thunder with smart books. My netbook has got 8gb in it, it makes it a very smooth machine, just light and cheap enough to lug around and not worry about it getting dented or worse, stolen. Netbooks partially failed because they sold with slow HD's and tiny amounts of memory, hurting their performance no end.
AMD just never had the clout to sell its chips on even terms. And it is sad because Intel dropped the ball completely when they believed they had no competition. There is a reason that 64 bit linux is report as AMD64. Intel failed and AMD delivered but for AMD to have truly broken through they need a long string of victories and no losses like Bulldozer.
If AMD wants to succeed, they might consider something that Intel is also thinking of doing. Intel is having trouble gettings its chips into tablets and phones especially, so they have considered making their own... AMD could do a lot better getting their CPU's in PC's if they started selling them. Control the whole supply line and pass the savings on to the consumer and beat Intel and Intel Inside PC makers on price. Intel can't do that for fear of pissing of all its customers but AMD doesn't have many bridges to burn.
Yes, making PC's is a very low margin industry but that is partly because you are buying all the parts from third parties. AMD wouldn't be doing that. The profit on the CPU inside the PC would be part of the profit of their PC. The profit on the graphics card would be part of the profits on the PC.
Risky and unconventional but unless THEY build the PC, they are always going to have a hard time getting their CPU into the PC.
MMO Quests are like orgasms:
You may solo them, I prefer them in a group.
Dell – then the world’s biggest PC maker – received billions of dollars to “remain monogamous” with Intel. At their peak in the first quarter of 2007, payments from Intel made up 76% of Dell’s quarterly operating income: $723 million against a total of $949 million.
And I really wonder why Intel hasn't been gutted and salted for monopoly abuse, with its CEO and main backers arrested. How can it not be MORE clear than that ?!?
Non-Linux Penguins ?
Man, has Slashdot gone downhill or what? You guys can't tell the difference between an Apple Shill and a Microsoft Shill?
This is getting scary.
Faster! Faster! Faster would be better!
In essences what AMD was evolution vs Intels attempted revolution. They evolved x86 with a 64-bit extension rather than attempt to revolutionize like Intel went for.
Now however the roles have switched. Intel goes for a evolution, while AMD tries for revolution with their APU concept of shifting floating point onto the GPGPU.
comment first, facts later. http://chem.tufts.edu/AnswersInScience/RelativityofWrong.htm
>The problem was AMD didn't bet hard enough on ramping up clock speeds.
Clock speeds have barely moved lately.
>Hertz to Hertz, AMD makes a better processor.
Not since the Core2, and even less competitive in the "i" era
Climate Progress - Hell and High Water
No, it's not Slashdot. These days, it has become hard to tell the difference between Apple and Microsoft, except maybe Apple sues more people.
Excuse me, wtf r u doin?
If you want to buy 5000 computers every year how many companies can you buy from?
If you want to buy one computer a year you can build your own for all it matters. If you want 1 computer every 5 years you probably don't have the desire or skills to build your own, nor is saving that small amount of money worth it for a lot of people.
When apply either of those two constraints Dell IBM and HP were the big dogs for a long time, and they were basically in bed with intel. People who don't have the skills to build their own want to buy from someone with a name brand who will stay in business long enough to honour a warranty, and people who want to buy 5000 computers this year are only going to buy from a big outfit, for basically the same reasons, and because there aren't a lot of places that can supply you will 1000 computers by the end of the week. If you're a really big outfit you're looking at buying something like 20-100k computers a year, and when you start talking numbers like that even your acer, asus and toshiba guys will have trouble keeping up.
AMD has the same problem in two different sectors. They had one really good product, and then someone released a better one. In the GPU business AMD will have the best parts for a couple of month then nvidia will come along and take the crown, and neither of them are competing in the high volume business desktop market that intel has (and has gone so far as to put it into the CPU package). For the CPU business Intel has been toying with them for at least 6 years. How do you know that? Because you can overclock an i7 (or a core 2 series) by 30% on air easily. Everytime AMD gets close to matching the performance/watt, performance/dollar or whatever, intel just ups the clocks a bit and boom, they're back in first place. They're basically a full process (die size) ahead of AMD, and they always have been, which gives them a huge advantage. In the GPU business AMD is doing as well as they can, if you look at the steam numbers they're up around 40% of the market. The problem is that the gaming market, which is where the money is on a per unit basis, isn't all that big. nVidia has a revenue of about 3.7 Billion USD, AMD 6.4, and Intel 54. The money is in volume, and AMD can't get volume because their price per unit, per performance, per watt are all just not up to match Intel, yes, Intel was anti-competitive for a while, but they only need to do that for about 4 years to get themselves back out into the lead by a wide margin.
Then when you look at the server market, they do not compete at all anymore. Performance per watt and per dollar both lag badly behind the Xeon.
That's untrue in my experience and has been for 5+ years. In the higher-end 2-4 CPU server range, AMD has had the best performance/price ratings for a long time because competitive Xeons are much too expensive. For example, the highest-performing 4P Linux servers (e.g. SPECint2006 rate) are currently Xeon E7-4870 based, followed closely by Opteron 6282 SE, but for the Xeons you'll pay 4 times as much as for the Opterons. A typical server configuration with 256GB memory will cost you ~8000 EUR ($10500) if you go with the Opterons and ~20000 EUR ($26000) if you go with the E7-4870 (and if you can actually find one on the market). More affordable Intel-based servers are not competitive performance-wise with the 6282 SE. If you don't need much parallelism and a lot of RAM, you might be able to get a more affordable offer using Xeons (with 2 of their 4C CPUs e.g.), but even there C32 based Opterons will offer much better performance per Dollar at comparable or lower TDP even (e.g. 2 x X7542 vs 2 x Opteron 4238). We've always been comparing closely when purchasing beefy 1U/2U servers over the past 10 years and Intel has not had competitive offers since their socket 604/Clarksboro Xeons when they allowed decent amounts of RAM (24 FB-DIMM sockets) in 1U compared with socket 940 Opterons (at somewhat sane prices). YMMV if your CPU needs are different, although I'd like to know how ...
"I love my job, but I hate talking to people like you" (Freddie Mercury)
One big thing you leave out: Intel stepped up their anti-competitive behavior, buying off the big computer makers to get them to cancel AMD-based computer lines.
Dell had the Optiplex 740 line. It was a damn good line, very effective, came in $150 under an "equivalent" Intel computer. What was Intel's response? They stepped up their monopolist subsidization and got Dell to back down. Repeat for a number of other manufacturers, and AMD's stuck in a bind again.
The Intel compiler, widely regarded as the best compiler available for x86, still produces code designed to make Intel chips look better than any others.
http://www.agner.org/optimize/blog/read.php?i=49
That page was posted three years ago. Scroll to the bottom, and read the latest additions to the discussion there: "New Intel compiler version - still the same!"
http://www.agner.org/optimize/blog/read.php?i=49#179
This makes it difficult to be sure how much better Intel chips really are than AMD chips. When the Intel chip scores higher on a benchmark, and the benchmark includes Photoshop, was the Intel chip actually better or was Photoshop compiled with the Intel compiler?
Sadly, I think Intel chips really are better now; given that Intel is leapfrogging past AMD on process technology, they have major advantages so their chips ought to be better.
But I still buy AMD. Yeah, I'm giving up some increment of performance... but the chips these days are so fast, I can survive on only 90% performance or whatever. And I prefer to avoid doing business with a company that continues to sell a compiler that sabotages performance on competitor's chips.
Personally, I would love to see AMD sell a line of processors that return "GenuineIntel" for the CPU ID, and thus run Intel compiler code at full speed. When Intel sues them, they can argue that this is necessary for full compatibility with the code produced by Intel's own C compiler. (Yeah, I know. It will never happen. It's a fun daydream but that's all.)
Even if AMD doesn't have the top performing chips, they continue to score very well on price/performance, and the performance is good enough for me. And they are less evil than Intel. So I remain an AMD customer.
steveha
lf(1): it's like ls(1) but sorts filenames by extension, tersely
If I were a major computer manufacturer these days, I'd spec in AMD CPUs (Black Editions, etc.), then attach a self-contained coolant system to it, and crank it until it reached the temperatures that the i7 normally operates at. The $500 in cost savings would appeal to my customers, and I'd be able to price my competitors out of the market.
A core i7 2700k (unlocked for overclocking) only costs $369 on Newegg ( http://www.newegg.com/Product/Product.aspx?Item=N82E16819115095 ). Pair it with the most expensive LGA 1155 motherboard they have at $339 ( http://www.newegg.com/Product/Product.aspx?Item=N82E16813131760 ), and you're paying $708. Do you mean to tell me that you can get an equivalently powerful AMD processor, with a motherboard with similar features, for less than $210?
Now what if I were to tell you that you can get a motherboard that ticks all the same boxes as the other one, for $129? ( http://www.newegg.com/Product/Product.aspx?Item=N82E16813157271 ). That brings the total cost down to $498. Could you please enlighten me on how it is possible to *save* $500 by building an AMD solution instead of Intel, when the Intel option is less than $500, at retail ?
The only consumers spending over $1000 on a CPU are the folks with too much money and not enough brains. And while you can spend that much on an extreme edition 6-core Intel processor, you're forgetting that it's also overclockable, by about 30%, and that you'd really be pushing things if you tried to get an FX6 running stably at 4.5GHz. You'd also be forgetting that unless something is massively parallel, the i7 still retains a performance edge over the bulldozer architecture. Chiefly, though, you'd be forgetting that for 99% of what you do, you'll never see the difference between the i7 2700 and the FX6, except perhaps that the ability to use a small SSD as a cache drive to improve spinny platter drive performance, something that's built into the Intel Z68 motherboard chipset and, at least last I checked, didn't exist on the AMD platform, would actually give the i7 a boost in real world usage, for significantly less price (pair a 32GB SSD with a 3TB spinny platter drive, and you get the write speed of the SSD, coupled with the capacity of the spinny platter drive). You may see a performance increase in things like video encoding, depending on the software you're using, but it's not going to get you any more frames per second in Skyrim. And truthfully? When I rip a DVD, I queue up the transcode in Handbrake before I go to bed, set it to turn the computer off when it's done, and don't really care if it finishes 2 minutes earlier.