Re: AMD Bulldozer Criticism

They'd also need a minimum of three to five years to produce a product that could even think of taking on anything Intel has to offer. The reason I say AMD's loss in the market place wouldn't change pricing too much is due to the fact that the market simply won't tolerate Intel hiking the prices on all it's CPUs. We might see a short price hike while they claim supply and demand issues with the vacuum left by AMD but after a short while prices would normalize again. The market likes cheap consumer electronics. AMD's vacancy wouldn't change that. But instead of a two year tick-tock cycle we'd be looking at much longer 4 or 5 year product cycles and generations of rehashed CPUs, chipsets etc. Honestly as much as I love progress this wouldn't be the end of the world either given how software lags behind the hardware these days.

Competition is good but in a market like this, it isn't the only factor when it comes to setting prices. Where competition really makes all the differences is in the products themselves. We need AMD or someone around always to at the very list keep Intel moving forward. Occasionally leapfrogging them like AMD did with the Athlon 64 really shakes things up leading to even better products like the Core 2 Duo.
Does anyone remember the fire sale prices on the 'Cedar Mill' P4's? God, that was awesome!

You know what is amazing? That the x86 PC market is one of the few areas that a small competitor can somewhat keep the market giant in line.
 
Does anyone remember the fire sale prices on the 'Cedar Mill' P4's? God, that was awesome!

You know what is amazing? That the x86 PC market is one of the few areas that a small competitor can somewhat keep the market giant in line.

Can people stop with this nonsense arguments ? If AMD would rollover and disappear tomorrow it wouldn't make a difference.

Intel would still be forced to innovate continuously and maintain the current price structure ( from really cheap to high end ). Why ?

1. Intel's biggest competitor are the millions of CPUs it sold yesterday. For the company to maintain and accelerate the replacement period of PCs/servers it needs to offer compelling performance, power and perf/$ improvements so total ROI improves. Otherwise people would extend the lifespan of their PCs and server, delay their replacement and Intel's sales and growth will plummet. New buyers aren't capable of carrying the burden and since most of the new growth is in developing countries they are very cost conscious.

2. Regarding price structure, it acts as an entry barrier. The moment Intel would raise price above a certain threshold it makes the computing market attractive for all kinds of alternative uarchs : ARM, MIPS, etc. Hell, if the entry price is good enough you'd soon see PowerPC resurrected. You can see this already today. The rock bottom of the PC market is being eroded by ARM products which are moving up the food chain.

In summary, all the " we need AMD to have competition and to keep prices under control" are BS. The market would balance itself straight away if Intel would do anything stupid. And Intel's managers are fully aware of this, rest assured.
 
Can people stop with this nonsense arguments ? If AMD would rollover and disappear tomorrow it wouldn't make a difference.

Intel would still be forced to innovate continuously and maintain the current price structure ( from really cheap to high end ). Why ?

1. Intel's biggest competitor are the millions of CPUs it sold yesterday. For the company to maintain and accelerate the replacement period of PCs/servers it needs to offer compelling performance, power and perf/$ improvements so total ROI improves. Otherwise people would extend the lifespan of their PCs and server, delay their replacement and Intel's sales and growth will plummet. New buyers aren't capable of carrying the burden and since most of the new growth is in developing countries they are very cost conscious.

2. Regarding price structure, it acts as an entry barrier. The moment Intel would raise price above a certain threshold it makes the computing market attractive for all kinds of alternative uarchs : ARM, MIPS, etc. Hell, if the entry price is good enough you'd soon see PowerPC resurrected. You can see this already today. The rock bottom of the PC market is being eroded by ARM products which are moving up the food chain.

In summary, all the " we need AMD to have competition and to keep prices under control" are BS. The market would balance itself straight away if Intel would do anything stupid. And Intel's managers are fully aware of this, rest assured.

None of that is true in the least. First order of business is supply. No one not the billions of Arm licensee's, not IBM, not AMD all together can produce enough chips to supply half of what Intel is. For that reason government agencies would have a difficult time managing the Intel Monopoly. It would take 10 years minimum for people to band together design a product and develop the relationships and capacity that AMD has, without limitless money and government support.

Second we have seen what Intel will do both in a competitive and non competitive enviroment. Look back at the Pentium II, it was almost two years inbetween the launch of that and deschutes, and 3 years for an increase in 20% performance. At a time when every ounce of CPU performance was needed for every task. They would do it again. IA64 was an attempt to create and redo the entire corporate and home market to a architecture they didn't have to license out. If it wasn't for the Competitiveness of the Athlon, and later the launch of AMD64, Xeon's would be dead by now, and bit by bit people would be moving their server and professional software to support IA64. By 2015 Intel would have made desktop IA64 chips to go with Microsoft IA64 OS's. All so Intel could once again completely control the release schedule and pricing of chips.

Its the increase in ARM sales that pushed the Atom out. Its AMD that caused Intel to hurry up their release schedules and put their billions of dollars of profit to use. Its AMD's fault that we have a complete pricing and performance window that goes from $50-$1k for some serious performance. You only need to look at the amount of products between the $50-$300 price range and the number between 300-$1k to see what happens when they don't have competition.

If AMD died, Intel isn't going to kill off half their products overnight and then all of a sudden raise prices. But year after year less and less new releases will come out. They will be marginally better then the last, knowing that the replacement cycle is 3 years and that won't change much. They will keep low cost alternatives out there, but features will be slow to propagate to the lower end, performance lines won't blur to much with the economy lines, performance line CPU's will EOL instead of prices being dropped. They might shift most of their development efforts into ARM competition more then they already have because development spending on X86 standard lines will have already been dropped by quite a bit. It would happen over a 5-7 year period and people would probably not notice, mostly just accepting the slightly increased costs on inflation. As for performance, who's to say that Intel isn't trying their best, so while the geek sees less new things overall, most people will accept the increases they have knowing nothing about Moore's law, or what happen behind the scenes in 2000-2006.
 
None of that is true in the least. First order of business is supply. No one not the billions of Arm licensee's, not IBM, not AMD all together can produce enough chips to supply half of what Intel is. For that reason government agencies would have a difficult time managing the Intel Monopoly. It would take 10 years minimum for people to band together design a product and develop the relationships and capacity that AMD has, without limitless money and government support.

Second we have seen what Intel will do both in a competitive and non competitive enviroment. Look back at the Pentium II, it was almost two years inbetween the launch of that and deschutes, and 3 years for an increase in 20% performance. At a time when every ounce of CPU performance was needed for every task. They would do it again. IA64 was an attempt to create and redo the entire corporate and home market to a architecture they didn't have to license out. If it wasn't for the Competitiveness of the Athlon, and later the launch of AMD64, Xeon's would be dead by now, and bit by bit people would be moving their server and professional software to support IA64. By 2015 Intel would have made desktop IA64 chips to go with Microsoft IA64 OS's. All so Intel could once again completely control the release schedule and pricing of chips.

Its the increase in ARM sales that pushed the Atom out. Its AMD that caused Intel to hurry up their release schedules and put their billions of dollars of profit to use. Its AMD's fault that we have a complete pricing and performance window that goes from $50-$1k for some serious performance. You only need to look at the amount of products between the $50-$300 price range and the number between 300-$1k to see what happens when they don't have competition.

If AMD died, Intel isn't going to kill off half their products overnight and then all of a sudden raise prices. But year after year less and less new releases will come out. They will be marginally better then the last, knowing that the replacement cycle is 3 years and that won't change much. They will keep low cost alternatives out there, but features will be slow to propagate to the lower end, performance lines won't blur to much with the economy lines, performance line CPU's will EOL instead of prices being dropped. They might shift most of their development efforts into ARM competition more then they already have because development spending on X86 standard lines will have already been dropped by quite a bit. It would happen over a 5-7 year period and people would probably not notice, mostly just accepting the slightly increased costs on inflation. As for performance, who's to say that Intel isn't trying their best, so while the geek sees less new things overall, most people will accept the increases they have knowing nothing about Moore's law, or what happen behind the scenes in 2000-2006.
And, really, the way Intel could compete with ARM is by building a better ARM-based chip (Intel still has a license for ARM). Think of Cortex-A9/A15 with Intel's process technology... Just on sheer output alone, Intel could dominate the entire Smartphone/Tablet market. Intel could cripple the market by just buying Imagination Technologies.
 
And, really, the way Intel could compete with ARM is by building a better ARM-based chip (Intel still has a license for ARM). Think of Cortex-A9/A15 with Intel's process technology... Just on sheer output alone, Intel could dominate the entire Smartphone/Tablet market. Intel could cripple the market by just buying Imagination Technologies.

Exactly the reason Intel hasn't been competing against full steam is due to the investment they made in making the X86 line stronger when they sold off the Arm development lines in 2005ish.

Think of it this way, they are designing 5 major lines right now. Atom, the desktop/mobile lines, the standard server chip line, the advanced server chip line, and IA64. With AMD competing with Intel in all of those lines with the exception of the Itanium products. Think of how strained Intel is, and considering that AMD is doing that with half of the chip variations its a minor miracle.

Its going to be another 2 years or more before we see it in performance but now AMD is unshackled as well. They can now maintain as many if not more product variations as Intel does. Selling GloFo was the best business decision ever made. Worst case scenario if they struggle in a couple of years AMD makes a joint venture with IBM for a Fab plant or two, and then buys out IBM later. But till then just to take off both the production of the plants, the development of the process, and all the major issues of shifting production off their hands is a huge boon.

Personally I am happy that Smartphone and Handheld market is able to grow without Intel's input right now. Sure their competition would push everyone, but the turnaround everybody else has on design licensing to production would be halved with Intel and their supply chain could flood the market. Literally killing them all off in 6 months.
 
Re: Topic title...

Some of the criticism of Bulldozer is coming from a couple of ex-AMD engineers. If you don't know who they are, look them up. It's from last year, but still appropriate:

Mitch Alsup
http://groups.google.com/group/comp.arch/msg/a7ce9a93e7ec42eb?hl=en
(Ahem) K10 is BullDozer, K8 is Opteron and follow-ons.
...

When I left, BD was supposed to be 20-25% faster frequency wise, and
loose a little architectural figure (5%-ish) of merit due to the
microarchitecture. The surprising thing was the lack of mention of
frequency in the market-droid-ing.
That confirms an old rumor/wish people here were hoping for: POWER7-like 5GHz clock speeds on "K10". How's that game "broken telephone" work? ;) Look up some of his other posts if you want to know what happened to K9 and how much work was done on it.

Cliff Maier
http://forums.macrumors.com/showthread.php?p=9683616#post9683616 (starts here, goes on for couple dozen responses... may be hard to read for two reasons, only one is technical :p)

choice quotes, if you want an insider's view of what happened at AMD after K8:

The team that did all the great design work for Athlon 64/Opteron is gone. Forced out or quit in disgust. They now work at Apple, Oracle, misc. startups, and, in my case, changed careers. http://forums.macrumors.com/showpost.php?p=9683707&postcount=48

The team that designed the K6-2 was the CMD team, which was formed by the acquisition of a company called Nexgen. That team also designed Athlon 64 and Opteron (Athlon was designed by the TMD team). By 2007, all the key CMD folks were gone. The team that was left sucks, and has accomplished little since then other than shrinks to smaller technology and bolting more of the same cores on.http://forums.macrumors.com/showpost.php?p=9683789&postcount=68

Yeah. Like replacing the 40-man team that designed A64 with a 250 man team for minor design changes. And like throwing away all the EDA tools because the vapor tools from the texas design team would be better if they ever were finished.http://forums.macrumors.com/showpost.php?p=9685563&postcount=242

The new CEO [Dirk Meyer] is a guy that once told the design team (when he was a manager in Texas) "if you don't like it, quit" - and 60 out of 100 people in Sunnyvale quit within a month. They used to hand place and hand instantiate each cell in the design for maximum efficiency and speed - now they rely on tools which perform 20% worse than humans (in order to save money). http://forums.macrumors.com/showpost.php?p=9706695&postcount=559

Guys - I have nothing to be disgruntled about. I left on my own accord. I had great times at AMD making a64 and creating amd64. AMD paid me enough to pay for a house in silicon valley and a porsche. I quit on my own and was not pressured to do so. They were sad to see me go, and i was sad to leave

I am not disgruntled. I am sad. Sad that AMD squandered their lead because new management decided that the way to compete was design CPUs the way that chips in toasters are designed. And worse, to do it with giant teams of unskilled designers instead of a small team of highly experienced engineers who know how to design transistor by transistor. http://forums.macrumors.com/showpost.php?p=9736602&postcount=598

And the biggie on how Hector's Fusion strategy stabbed the remaining carcass:
What did happen is that management decided there SHOULD BE such cross-engineering ,which meant we had to stop hand-crafting our CPU designs and switch to an SoC design style. This results in giving up a lot of performance, chip area, and efficiency. The reason DEC Alphas were always much faster than anything else is they designed each transistor by hand. Intel and AMD had always done so at least for the critical parts of the chip. That changed before I left - they started to rely on synthesis tools, automatic place and route tools, etc. I had been in charge of our design flow in the years before I left, and I had tested these tools by asking the companies who sold them to design blocks (adders, multipliers, etc.) using their tools. I let them take as long as they wanted. They always came back to me with designs that were 20% bigger, and 20% slower than our hand-crafted designs, and which suffered from electromigration and other problems.

That is now how AMD designs chips. I'm sure it will turn out well for them [/sarcasm]http://forums.macrumors.com/showpost.php?p=9746191&postcount=620
 
Sorry, I don't buy it, these guys are sad that the world of processor design changed on them and they were not ready. I am not handing over a 1.5 Billion transistor CPU design over to 40 guys to hand craft the transistors. I don't care if they could get that down 1.25 Billion. That's a task of 31 million transistors per person. Even for a 5 year design, 20 days a month, 8 hours a day a, person would have to design 3 thousand transistors a day. There probably is some trade offs and software design isn't great, but the times of CPU's that do one thing great is gone, and technical innovation has exponentially grown since the shift. Deviation has grown. Now AMD has 7 different CPU's instead of 2.

This is a world has changed and people remembering the good old days.
 
Sorry, I don't buy it, these guys are sad that the world of processor design changed on them and they were not ready. I am not handing over a 1.5 Billion transistor CPU design over to 40 guys to hand craft the transistors. I don't care if they could get that down 1.25 Billion. That's a task of 31 million transistors per person. Even for a 5 year design, 20 days a month, 8 hours a day a, person would have to design 3 thousand transistors a day. There probably is some trade offs and software design isn't great, but the times of CPU's that do one thing great is gone, and technical innovation has exponentially grown since the shift. Deviation has grown. Now AMD has 7 different CPU's instead of 2.

This is a world has changed and people remembering the good old days.
while I agree that pxc is just trolling, as usual, the AMD dude may have a point. We all have to admit that its quite clear that AMD has been way too content to iterate instead of innovate. Phenom II isn't much more than a straight-line evolution of A64, and Llano isn't all that impressive when you consider that they just bolted a 5550 into an Athlon II quad. Its cool but the process shrink bought them what, 15w off the TDP? not too great
 
Re: Topic title...

Some of the criticism of Bulldozer is coming from a couple of ex-AMD engineers. If you don't know who they are, look them up. It's from last year, but still appropriate:

Mitch Alsup
http://groups.google.com/group/comp.arch/msg/a7ce9a93e7ec42eb?hl=en
That confirms an old rumor/wish people here were hoping for: POWER7-like 5GHz clock speeds on "K10". How's that game "broken telephone" work? ;) Look up some of his other posts if you want to know what happened to K9 and how much work was done on it.

Cliff Maier
http://forums.macrumors.com/showthread.php?p=9683616#post9683616 (starts here, goes on for couple dozen responses... may be hard to read for two reasons, only one is technical :p)

choice quotes, if you want an insider's view of what happened at AMD after K8:











And the biggie on how Hector's Fusion strategy stabbed the remaining carcass:

wow i'm so glad I read this great post thanks!
 
Hello, namecallers! ilu2

Sorry, I don't buy it, these guys are sad that the world of processor design changed on them and they were not ready. I am not handing over a 1.5 Billion transistor CPU design over to 40 guys to hand craft the transistors. I don't care if they could get that down 1.25 Billion. That's a task of 31 million transistors per person.
That's not how design works. Re-read this quote:
"Intel and AMD had always done so at least for the critical parts of the chip."

I really don't think anyone who has unwavering faith in a comeback understands how much damage Hector and Dirk caused to AMD. You can ignore what changed, and things have really changed since K8, but it's ignoring the reality of the situation.

BTW, I left out some really damning posts by both. You can use the links above to read more if you want.
 
Can people stop with this nonsense arguments ? If AMD would rollover and disappear tomorrow it wouldn't make a difference.

this thread is stupid, but that is not true. when amd is competing better, intel prices are lower to compete. when amd is competing worse, intel prices are higher because they can be. this would just make sense on its own even if I hadn't seen it in practice for 15 years.
 
Sorry, I don't buy it, these guys are sad that the world of processor design changed on them and they were not ready. I am not handing over a 1.5 Billion transistor CPU design over to 40 guys to hand craft the transistors. I don't care if they could get that down 1.25 Billion. That's a task of 31 million transistors per person. Even for a 5 year design, 20 days a month, 8 hours a day a, person would have to design 3 thousand transistors a day. There probably is some trade offs and software design isn't great, but the times of CPU's that do one thing great is gone, and technical innovation has exponentially grown since the shift. Deviation has grown. Now AMD has 7 different CPU's instead of 2.

This is a world has changed and people remembering the good old days.

The bulk of transistors is in the cache. Once you have a single piece setup of each cache, the rest can be cloned.

That is probably going to drop at least 70% of the work load in modern processor design.

I'm also not too sure about how they are saying that they "designed" each transitor as there is going to be a whole lot of using the same exact transistor in different circuits.

I can see how they would specifically design the different circuits, especially the critical ones.. but not every single transistor.

Sounds like people blowing things way out of proportion to me.
 
The bulk of transistors is in the cache. Once you have a single piece setup of each cache, the rest can be cloned.

That is probably going to drop at least 70% of the work load in modern processor design.

I'm also not too sure about how they are saying that they "designed" each transitor as there is going to be a whole lot of using the same exact transistor in different circuits.

I can see how they would specifically design the different circuits, especially the critical ones.. but not every single transistor.

Sounds like people blowing things way out of proportion to me.

That's not how it read. At least two of those guys said every transistor. I could buy it when CPU's were 120 million with cache. Its the same thing with lines of code in software and OS's as they get bigger, the work everyone is given is simpler and they exponentially higher more people. It sounds like to me that they like sitting there toying away with a design and taken there sweet ass "when its done" time like game developers like to. While Dirk and Hector and everyone else wanted to turn AMD into a much more scheduled and steady product manufacturer. This is probably when development on the K9 died. Handful of people pouring unrepeatable time into it left and AMD had to restart the development.
 
The bulk of transistors is in the cache. Once you have a single piece setup of each cache, the rest can be cloned.
Maier mentions a couple of key things above: the ability the design with transistors and the benefits of manually laying out "cells", which can be AMD's own designed circuits. There are many circuits repeated throughout the chip, and a cell consisting of many transistors can be used as a unit. The cache is a good example of cell usage.
 
Maier mentions a couple of key things above: the ability the design with transistors and the benefits of manually laying out "cells", which can be AMD's own designed circuits. There are many circuits repeated throughout the chip, and a cell consisting of many transistors can be used as a unit. The cache is a good example of cell usage.

I sense concern for your AMD short position/ IntEl long. Better get that FUD rolling, not much time left!
 
That's not how it read. At least two of those guys said every transistor.
You misread. "who know how to design transistor by transistor" and "used to hand place and hand instantiate each cell in the design for maximum efficiency and speed" is not the same as your interpretation.

It sounds like to me that they like sitting there toying away with a design and taken there sweet ass "when its done" time like game developers like to. While Dirk and Hector and everyone else wanted to turn AMD into a much more scheduled and steady product manufacturer. This is probably when development on the K9 died. Handful of people pouring unrepeatable time into it left and AMD had to restart the development.
Using automated tools was supposed to speed up development, but it didn't. Maier mentions problems he encountered with the tools he evaluated, but that's probably not the whole story.

If you don't want to read Alsup's history of K9, it was cancelled back in 2004, after 6 months of work. The problem wasn't the designers, but the design. It was an 8 wide execution chip. ;) From another thread I read, the BD design has been around for a very long time. Think of an era where a chip was designed with a DDR2 memory controller. :p
 
Re: Topic title...

Some of the criticism of Bulldozer is coming from a couple of ex-AMD engineers. If you don't know who they are, look them up. It's from last year, but still appropriate:

Mitch Alsup
http://groups.google.com/group/comp.arch/msg/a7ce9a93e7ec42eb?hl=en
That confirms an old rumor/wish people here were hoping for: POWER7-like 5GHz clock speeds on "K10". How's that game "broken telephone" work? ;) Look up some of his other posts if you want to know what happened to K9 and how much work was done on it.

Cliff Maier
http://forums.macrumors.com/showthread.php?p=9683616#post9683616 (starts here, goes on for couple dozen responses... may be hard to read for two reasons, only one is technical :p)

choice quotes, if you want an insider's view of what happened at AMD after K8:

And the biggie on how Hector's Fusion strategy stabbed the remaining carcass:

Wow I can't believe all this was being posted on the MR forums and not on a forum like this. It's pretty disappointing to read about all that though :(
 
You misread. "who know how to design transistor by transistor" and "used to hand place and hand instantiate each cell in the design for maximum efficiency and speed" is not the same as your interpretation.

Using automated tools was supposed to speed up development, but it didn't. Maier mentions problems he encountered with the tools he evaluated, but that's probably not the whole story.

If you don't want to read Alsup's history of K9, it was cancelled back in 2004, after 6 months of work. The problem wasn't the designers, but the design. It was an 8 wide execution chip. ;) From another thread I read, the BD design has been around for a very long time. Think of an era where a chip was designed with a DDR2 memory controller. :p

Still ignores the point (though can you link the K9 history never got a full feeling of what happened there). Its hard to dig out the highly technical differences peoples writing like that because he sets the stage that every guy who watches his company get big does. We used to be a close nit group, we were small and everybody did everything. Now half the stuff is done for us, not as well, and there are 30 people doing 1 persons job.

That stuff happens all of the time. So the software that he analyzed wasn't as good as his hands. As long as its faster and they can come up with new ways to make up the difference then so be it. Its easy to say things where great the old ways and look at them now. But fact is chances are if they went the old route they might be even more screwed.
 
That stuff happens all of the time. So the software that he analyzed wasn't as good as his hands. As long as its faster and they can come up with new ways to make up the difference then so be it. Its easy to say things where great the old ways and look at them now. But fact is chances are if they went the old route they might be even more screwed.
You seem to be stuck on thinking he means automation is bad. Some parts of the chip have been and continue to be made with automated tools, tweaked as necessary. He's complaining about the reliance on those and the detrimental effect it has on products. (One post I didn't link was Maier's observation that AMD no longer has any real CPU architects left. That would go along with his complaint about the huge number of low level designers.)

The problem is that the inefficiencies are NOT being worked around or made up for. AMD's strategy is to make a chip with 2x as many cores, often with a much larger die size and sell the product for less than the competitions mid-range chips. Penny-wise, pound-foolish comes to mind.

I'm not exactly sure anything could have been worse than taking 7 years to make the cancelled K9's replacement, using automated tools or not.

Maybe I'm just not getting it. Hector and Dirk were not good CEOs/CTOs, and in fact they were worse than mediocre. Do I sense a disagreement? Their reign has left AMD a hollow shell of its former self.
 
AMD is no longer the AMD I knew 5 years ago. I used to be their fan until sandy bridge comes out. Let's see how the BD retaliates, hopefully in a month from now.
 
Why should we give AMD the benefit of the doubt? It's amazing how many people you run into on the net that have these ill-contrived allegiances to big companies that could give two craps about them. I loved AMD for quite some time. I thought during the time where they popped Intel in the mouth with their Athlon 64 chips at competitive pricing would've made them a much more successful company than they are today. Since that time period, they have made one bone headed mistake after another. In fact in AMD's history, only once in the early- to mid-2000s did they ever reach or exceed Intel's performance level, which in turn brought out the fury of the stagnant, monopolistic, greedy corporation. AMD has been riding on the coattails of that successful architecture for over half a decade, which is closer to a century in the computer and consumer electronics industry. They could've used that liver blow, back then, to form a credible lead or at least remain competitive with Intel, but they never did. Then came the promises of Fusion and Bulldozer for year, after year, to the point that they are no longer nearly as relevant now as they would have been as recently as 18 months ago. Hell, even Intel beat them to the punch with low-end, discrete-class graphics on die. That was the very thing AMD was touting being the forefront of innovation of for YEARS! Don't get me wrong, AMD now seems to realize it's position in the processing world and is focusing on the more mainstream-level consumer. Llano is amazing in that they have continued prolonging an ancient archeticture but have been able to market it towards users who do not require extensive processing process. Genius. However, what happens when Core i3 desktops and laptops with low-mid end graphics start popping out at cheaper prices (and in some areas they already have. You can get a thin-n-light Acer Timeline X (yes i know acer blows) with an i5 2410m, 540m graphics, and 4gb of ram for $750....or you could get an A8 system with mid-level graphics from HP's Dv6 lineup for around the same price. The Acer should shred it. Brazos is the only thing, I can see AMD, holding onto long term. It's a serious threat to Atom, if not it's killer. Additionally their focus on tablets will renew some market share in other areas of the computing industry. All that being said, they are still riding technology that was novel at it's inception but dated in it's facilitation.

I don't play favorites and will side with whatever company treats the consumer fairly, offers the most bang for the buck, and continues enticing me with eloquently robust innovation. But if AMD's history is anything it is predictable and proven. With that being said, I find it highly dubious that anyone should give them the benefit of the doubt.
 
Why should we give AMD the benefit of the doubt? It's amazing how many people you run into on the net that have these ill-contrived allegiances to big companies that could give two craps about them. I loved AMD for quite some time. I thought during the time where they popped Intel in the mouth with their Athlon 64 chips at competitive pricing would've made them a much more successful company than they are today. Since that time period, they have made one bone headed mistake after another. In fact in AMD's history, only once in the early- to mid-2000s did they ever reach or exceed Intel's performance level, which in turn brought out the fury of the stagnant, monopolistic, greedy corporation. AMD has been riding on the coattails of that successful architecture for over half a decade, which is closer to a century in the computer and consumer electronics industry. They could've used that liver blow, back then, to form a credible lead or at least remain competitive with Intel, but they never did. Then came the promises of Fusion and Bulldozer for year, after year, to the point that they are no longer nearly as relevant now as they would have been as recently as 18 months ago. Hell, even Intel beat them to the punch with low-end, discrete-class graphics on die. That was the very thing AMD was touting being the forefront of innovation of for YEARS! Don't get me wrong, AMD now seems to realize it's position in the processing world and is focusing on the more mainstream-level consumer. Llano is amazing in that they have continued prolonging an ancient archeticture but have been able to market it towards users who do not require extensive processing process. Genius. However, what happens when Core i3 desktops and laptops with low-mid end graphics start popping out at cheaper prices (and in some areas they already have. You can get a thin-n-light Acer Timeline X (yes i know acer blows) with an i5 2410m, 540m graphics, and 4gb of ram for $750....or you could get an A8 system with mid-level graphics from HP's Dv6 lineup for around the same price. The Acer should shred it. Brazos is the only thing, I can see AMD, holding onto long term. It's a serious threat to Atom, if not it's killer. Additionally their focus on tablets will renew some market share in other areas of the computing industry. All that being said, they are still riding technology that was novel at it's inception but dated in it's facilitation.

I don't play favorites and will side with whatever company treats the consumer fairly, offers the most bang for the buck, and continues enticing me with eloquently robust innovation. But if AMD's history is anything it is predictable and proven. With that being said, I find it highly dubious that anyone should give them the benefit of the doubt.

:rolleyes:


You understand the only reason we have Sandy Bridge in 2011 is the competition AMD brings to the market?

Seriously, the best today might be Core Duos priced $500 per chip without competition in the market.

Supporting AMD brings better products, faster and cheaper to the market FROM INTEL.
 
AMD is no longer the AMD I knew 5 years ago. I used to be their fan until sandy bridge comes out. Let's see how the BD retaliates, hopefully in a month from now.

I totally agree bro.

I've owned more AMD systems than intel but I switched at Nehalem and I haven't seen anything yet that will make me switch back.

My money goes to towards performance, AMD hasn't held that title since 939 which happends to be my last AMD build.
 
Last edited:
:rolleyes:


You understand the only reason we have Sandy Bridge in 2011 is the competition AMD brings to the market?

Seriously, the best today might be Core Duos priced $500 per chip without competition in the market.

Supporting AMD brings better products, faster and cheaper to the market FROM INTEL.


Tell me how AMD has pressed Intel to progress? In all honesty some of the higher end Core 2 Quads still give some of the newer Phenoms a run for their money....Core 2 Quads from 5 years ago! If AMD had actually BEEN competitive, we may not have seen a delay on Ivy, Sandy Bridge in 2010, and a 6-core Core i7 in the mainstream. Intel is only competing with itself on all fronts other than the extremely low end. I built a Core i3 desktop from scratch for $250+tax. With i3 processors being under a 100 and competing with most of AMD's offerings, even at the high end, I fail to see the 'pressure' AMD is putting on Intel to be innovative at the majority of market segments. The only segment they have them beat now is at the netbook/cheap laptop market...which Intel has steadily held for many years.

I never said I don't want AMD to succeed, I AM saying that based on their history it's unlikely that they will make a competitive, long-term move this generation. I, with all honesty, want BD to be successful. I'm with you, competition is great for the industry. Unfortunately for the mid-end to high-end segment, I just don't see it happening.
 
Last edited:
Why should we give AMD the benefit of the doubt? It's amazing how many people you run into on the net that have these ill-contrived allegiances to big companies that could give two craps about them. I loved AMD for quite some time. I thought during the time where they popped Intel in the mouth with their Athlon 64 chips at competitive pricing would've made them a much more successful company than they are today.

Not when your main problem lies on the supply side... AMD can oddly enough supply more chips now, than they could during the K7/K8 days.

Since that time period, they have made one bone headed mistake after another. In fact in AMD's history, only once in the early- to mid-2000s did they ever reach or exceed Intel's performance level, which in turn brought out the fury of the stagnant, monopolistic, greedy corporation. AMD has been riding on the coattails of that successful architecture for over half a decade, which is closer to a century in the computer and consumer electronics industry. They could've used that liver blow, back then, to form a credible lead or at least remain competitive with Intel, but they never did.

There were two things AMD needed then to get even close to Intel: A stable platform to offer with their processors and more fabs. When you're dependent on 3rd parties to provide an accompanying chipset/platform for your processors, you will not make a lot of inroads. That problem existed across 3 generations of AMD processors: K5/K6, K7 and K8.

Then came the promises of Fusion and Bulldozer for year, after year, to the point that they are no longer nearly as relevant now as they would have been as recently as 18 months ago. Hell, even Intel beat them to the punch with low-end, discrete-class graphics on die. That was the very thing AMD was touting being the forefront of innovation of for YEARS! Don't get me wrong, AMD now seems to realize it's position in the processing world and is focusing on the more mainstream-level consumer. Llano is amazing in that they have continued prolonging an ancient archeticture but have been able to market it towards users who do not require extensive processing process. Genius. However, what happens when Core i3 desktops and laptops with low-mid end graphics start popping out at cheaper prices (and in some areas they already have. You can get a thin-n-light Acer Timeline X (yes i know acer blows) with an i5 2410m, 540m graphics, and 4gb of ram for $750....or you could get an A8 system with mid-level graphics from HP's Dv6 lineup for around the same price. The Acer should shred it. Brazos is the only thing, I can see AMD, holding onto long term. It's a serious threat to Atom, if not it's killer. Additionally their focus on tablets will renew some market share in other areas of the computing industry. All that being said, they are still riding technology that was novel at it's inception but dated in it's facilitation.

This still comes down to the portion of AMD's budget that is for R&D. AMD was bleeding just from running their fabs, SOI is nice, but I still think it wasn't worth the investment. IBM needed a fool to throw money at SOI, and AMD took the bait. This is my opinion, though.

I don't play favorites and will side with whatever company treats the consumer fairly, offers the most bang for the buck, and continues enticing me with eloquently robust innovation. But if AMD's history is anything it is predictable and proven. With that being said, I find it highly dubious that anyone should give them the benefit of the doubt.

I've been with AMD for the last 4 years, starting from a X2 3600 up to my 555, before then, I was using a S478 Celerons and Pentium 4s, before then I was on Socket A, and before then Slot 1 P2/P3s, Super7, all the way back to both Intel and AMD 486s.

Sticking with AMD now, because all I have to do now is buy a 990 board, then later upgrade to Zambezi, though after that, it is all up in the air.
 
I totally agree bro.

I've owned more AMD systems than intel but I switched at Nehalem and I haven't seen anything yet that will make me switch back.

My money goes to towards performance AMD hasn't held that title since 939 which happends to be my last AMD build.

My last AMD build was a dual Opteron 254 system on a Tyan K8WE (S2895). Great system. Before that I had an Athlon 64 3800+ "Newcastle" on an ASUS A8N-SLI Deluxe. Though way back in the day I did have Cyrix, AMD, IBM Blue Lightning and Intel 486's. After the 486 era, I had a Cyrix 6x86 PR200+. I later moved to an Intel Pentium 233MHz MMX, dual Pentium Pros, then PII, dual Celery's, PIIIs, etc. I didn't have an AMD based system until the late Pentium III or early P4 days when I had an Athlon 1.33GHz Thunderbird. I believe I still have one of those CPUs floating around here somewhere. Aside from those, I've had mostly Intel based systems over the years.

In days past I went for what I could afford. That's why I had Cyrix and AMD processors. It wasn't because I wanted to I assure you. Then moving on to when I could afford Intel processor based systems I pretty much always went with the most powerful of the two. Which ever that was at the time.
 
My last AMD build was a dual Opteron 254 system on a Tyan K8WE (S2895). Great system. Before that I had an Athlon 64 3800+ "Newcastle" on an ASUS A8N-SLI Deluxe. Though way back in the day I did have Cyrix, AMD, IBM Blue Lightning and Intel 486's. After the 486 era, I had a Cyrix 6x86 PR200+. I later moved to an Intel Pentium 233MHz MMX, dual Pentium Pros, then PII, dual Celery's, PIIIs, etc. I didn't have an AMD based system until the late Pentium III or early P4 days when I had an Athlon 1.33GHz Thunderbird. I believe I still have one of those CPUs floating around here somewhere. Aside from those, I've had mostly Intel based systems over the years.

In days past I went for what I could afford. That's why I had Cyrix and AMD processors. It wasn't because I wanted to I assure you. Then moving on to when I could afford Intel processor based systems I pretty much always went with the most powerful of the two. Which ever that was at the time.

I've owned similar systems (minus the Tyan board) to yours listed. I think we are probably in the same age range just looking at it =).


At the Thunderbird stage I believe I first got my hands on a 700 mhz duron which overclocked pretty nicely. Socket A was fun back in the day I remember the 1.4 Thunderbird being the hottest CPU out. Now that I look back at all those builds over the years I miss each one of those machines...

I really only did intel on a p2 350mhz much earlier when top speeds for cpu's were around 500-600mhz. Prior to this I would have been on the Cryix chip but running windows 95/98 she was good enough.

The last AMD build is also using ASUS A8N32-SLI Deluxe + Opteron 170 and a 4890 which is currently my HTPC in a silverstone case hooked up with HDMI to my 46' Samsung LCD power hungry yes but machine sleeps most of the time and I wanted to reuse it after having her sit so long when I upgraded to Nehalem.

And my story is the same having the funds to buy top end was lacking in those days, now that i'm older and have a career that is no longer an issue.
 
:rolleyes:


You understand the only reason we have Sandy Bridge in 2011 is the competition AMD brings to the market?

I disagree . AMD has been lagging behind Intel for 5 years. The phenom 2 architecture (x4 cpus anyway) is on par with/maybe slightly better than the Core 2 Quads, and certainly not enough of a threat to Intel to cause them to have to release Sandy Bridge to get past the performance of AMD. Intel was already beating AMD without Sandy, so I do not believe that the "competition" from AMD put much pressure on Intel to have to hit back, but of course, this is my opinion.
 
@Faksnima

Doesn't take 1 or 2 years to come out with a new architecture. It takes 4+ years.

What has AMD done in the last 5 years? How about taken a performance crown from nVidia while working on 3 new CPU architectures at the same time (2 of them completely from ground up), which we are starting to see come to fruition this year.

Intel is milking the monopoly which they illegally achieved to be able to run their "tick tock". Intel is literally 10 times the size of AMD (actual order of magnitude).

I am perfectly fine with people buying whatever performs best at the time for their individual needs. But talking about AMD failing without really understanding what AMD is up against is ignorant at best.
 
Last edited:
If you're reading this thread and want to type something about the fundamentals of capitalism, please don't. It would just add more fuel for the fire and we don't need everyone to be economists anyways -> less pay.
 
Sorry, I don't buy it, these guys are sad that the world of processor design changed on them and they were not ready. I am not handing over a 1.5 Billion transistor CPU design over to 40 guys to hand craft the transistors. I don't care if they could get that down 1.25 Billion. That's a task of 31 million transistors per person. Even for a 5 year design, 20 days a month, 8 hours a day a, person would have to design 3 thousand transistors a day. There probably is some trade offs and software design isn't great, but the times of CPU's that do one thing great is gone, and technical innovation has exponentially grown since the shift. Deviation has grown. Now AMD has 7 different CPU's instead of 2.

This is a world has changed and people remembering the good old days.

the 1.5billion transistors are not unique, if it's an 8 core chip, now you have 1.5/8, then you have cache, which also takes up a fairly large chunk of the die, the mentioned certain key areas needing to be hand designed, not the entire chip on top of that.
 
You seem to be stuck on thinking he means automation is bad. Some parts of the chip have been and continue to be made with automated tools, tweaked as necessary. He's complaining about the reliance on those and the detrimental effect it has on products. (One post I didn't link was Maier's observation that AMD no longer has any real CPU architects left. That would go along with his complaint about the huge number of low level designers.)

The problem is that the inefficiencies are NOT being worked around or made up for. AMD's strategy is to make a chip with 2x as many cores, often with a much larger die size and sell the product for less than the competitions mid-range chips. Penny-wise, pound-foolish comes to mind.

I'm not exactly sure anything could have been worse than taking 7 years to make the cancelled K9's replacement, using automated tools or not.

Maybe I'm just not getting it. Hector and Dirk were not good CEOs/CTOs, and in fact they were worse than mediocre. Do I sense a disagreement? Their reign has left AMD a hollow shell of its former self.
Their is a lot of conjecture in your post but maybe I am miss understanding what they are saying. He didn't say that the software was worse then an alternative software option, he said he compared to their hand designed transistors and cells. I am not trying to defend AMD in this. They have failed to design competitive CPU's and each day without BD they get farther and farther behind.

But I am not reading true technical information. I am reading AMD design crew got bigger and farther away from what he felt made them great and a lot of the people who where part of the small family design crew left and they feel that is why AMD is suffering now. This happens all the time when a company "falls from grace". The fact is these type of transitions have to happen, you hope the company can make a smooth transition and maybe AMD did not. It doesn't mean AMD can't make a good CPU in the future or that AMD lost its vision. It just means that AMD outgrew those designers comfort level and now AMD has to learn to live without them.
 
My last AMD build was a dual Opteron 254 system on a Tyan K8WE (S2895). Great system. Before that I had an Athlon 64 3800+ "Newcastle" on an ASUS A8N-SLI Deluxe. Though way back in the day I did have Cyrix, AMD, IBM Blue Lightning and Intel 486's. After the 486 era, I had a Cyrix 6x86 PR200+. I later moved to an Intel Pentium 233MHz MMX, dual Pentium Pros, then PII, dual Celery's, PIIIs, etc. I didn't have an AMD based system until the late Pentium III or early P4 days when I had an Athlon 1.33GHz Thunderbird. I believe I still have one of those CPUs floating around here somewhere. Aside from those, I've had mostly Intel based systems over the years.

In days past I went for what I could afford. That's why I had Cyrix and AMD processors. It wasn't because I wanted to I assure you. Then moving on to when I could afford Intel processor based systems I pretty much always went with the most powerful of the two. Which ever that was at the time.

I have had very similar experiences, the only difference is the lack of Pros as I couldn't afford it but my dad did work with one.

My first AMD was the Tbird 1.4, man that was fast and I still have the cpu here. Tts gone bad as a while ago I placed it in the board gave it juice and all I got was a burning smell.

Funny thing, my cousin was given a free Cyrix 200 (or 166 don't remember), we had a still running pentium100 and loaded up winamp under 98SE on both. The Cryix took some heavey thinking to load a song while the pentium was super responsive, we still abused the cyrix as our music box though.

I never actually owned a P4, I did own a PentiumD system because I bought the parts (minus the case I have 8 spare cases anyways) for about $100 at the time. Decent CPU but that was it for that era.

I ran the Tbird all the way till the 3000+ was released, holy cow that was an upgrade no idea why I waited that long ... oh yea, money :D. The 3000+ was my first true OC'er other then my 300A celeron.

This is funny thinking back, it might just be me remembering only fond memories but it seems I was having more fun back then with computer hardware. It seems I had a better experience in the "darker" times of PC parts. Now its ... boring.
 
Using automated tools was supposed to speed up development, but it didn't.
Using automated tools does speed up development. Both AMD and Intel used synthesis and P&R tools before the Athlon and P4 (and continue to use).

Cliff's comment about asking EDA vendors to design ALU style circuits using their tools is silly and perhaps out of context because that's not what they're supposed to be used for. Those blocks are still custom outside of really simple implementations where the logic can efficiently be reduced to its canonical form (or something close).

I think any major miss-steps have been either managerial, architectural, or financial.
 
Last edited:
None of that is true in the least. First order of business is supply. No one not the billions of Arm licensee's, not IBM, not AMD all together can produce enough chips to supply half of what Intel is. For that reason government agencies would have a difficult time managing the Intel Monopoly.

?? I'm pretty sure the ARM licensee like Samsung, Qualcom, NVIDIA have acccess to capacities far in excess of what Intel has.I believe Apple buys significantly more silicon than Intel produces in the form of SOCs and memory.Intel is only the tip of the iceberg in the supply equation.
It would take 10 years minimum for people to band together design a product and develop the relationships and capacity that AMD has, without limitless money and government support.

AMD ? Who is AMD ? Compare AMD with Samsung or Qualcomm. The later 2 would love to be the "foundation of the digital world". ARM moved up the ladder in 2 years from non-existant to a real threat to Intel's position. THe new QC chips in the pipeline will touch Intel's low end and threaten the bulk of the sales. And ARM's growth is exponential.
Second we have seen what Intel will do both in a competitive and non competitive enviroment. Look back at the Pentium II, it was almost two years inbetween the launch of that and deschutes, and 3 years for an increase in 20% performance.
??

PII was introduced in mid 1997 with 233MHz. Two years later we had PIII Katmai with 600MHz. By mid 2000 ( your 3 year horizon ) we had 1GHz PIIIs. That's a 4x increase in performance. Your premise is simply wrong.
At a time when every ounce of CPU performance was needed for every task. They would do it again. IA64 was an attempt to create and redo the entire corporate and home market to a architecture they didn't have to license out. If it wasn't for the Competitiveness of the Athlon, and later the launch of AMD64, Xeon's would be dead by now, and bit by bit people would be moving their server and professional software to support IA64. By 2015 Intel would have made desktop IA64 chips to go with Microsoft IA64 OS's. All so Intel could once again completely control the release schedule and pricing of chips.

And AMD has a back up IA64 design in the works at the time. Who would have thought ?

Regardless of this, you fail to grasp my argument. Intel needs to innovate so people replace PCs. If Intel slows the release schedule to once a year 5% faster and 50% more expensive people would not replace their PCs. How hard is this to grasp ?

Intel needs Nehalem to rock so millions of perfectly fine C2Ds PCs are discarded. SB/IB has to out do Nehalem so millions of Nehalem based systems get replaced.
How else can Intel dream of shipping 1 million CPUs per day ?? Do you think they'll sell many $500 per piece CPUs in Bangladesh or Zimbabwe ?
Its the increase in ARM sales that pushed the Atom out. Its AMD that caused Intel to hurry up their release schedules and put their billions of dollars of profit to use. Its AMD's fault that we have a complete pricing and performance window that goes from $50-$1k for some serious performance. You only need to look at the amount of products between the $50-$300 price range and the number between 300-$1k to see what happens when they don't have competition.

Intel doesn't have competition above $300 and there are exactly 3 products in that space. Otteline can say prayers all night but the market would not move to the segment irrespective of AMD. People who can afford a $400 PC and replace it every 3 years would pay the same with or without AMD. And if Intel positions itself above that segment, others would be more than happy to jump in. Just ask Samsung, NVIDIA, Qualcom, etc...
If AMD died, Intel isn't going to kill off half their products overnight and then all of a sudden raise prices. But year after year less and less new releases will come out. They will be marginally better then the last, knowing that the replacement cycle is 3 years and that won't change much. They will keep low cost alternatives out there, but features will be slow to propagate to the lower end, performance lines won't blur to much with the economy lines, performance line CPU's will EOL instead of prices being dropped.

That's BS and your whole argument starts from a false premise. If the new offers do not offer compeling upgrade reasons the 3 years will suddenly become 5. Look around you, the whole world is spinning around accelerated innovation. Each new generation of products floods you with new features just so that you upgrade sooner.

:rolleyes:


You understand the only reason we have Sandy Bridge in 2011 is the competition AMD brings to the market?

Seriously, the best today might be Core Duos priced $500 per chip without competition in the market. Supporting AMD brings better products, faster and cheaper to the market FROM INTEL.


Yes and Intel would sell so many of those, enough to fill 1/4 of one of their 5 production FABs. A true ROI masterpiece.
 
I'm of the opinion that none of that will matter. They'll start planning obsolescence like the appliance makers etc. You won't have chips or motherboards that last as long anymore.

If they end up trying to make money competing against themselves they'll find a way to remove the competition. This happens in a lot of different markets, its much more likely than your scenario.
 
I'm of the opinion that none of that will matter. They'll start planning obsolescence like the appliance makers etc. You won't have chips or motherboards that last as long anymore.

If they end up trying to make money competing against themselves they'll find a way to remove the competition. This happens in a lot of different markets, its much more likely than your scenario.

Yeah, but I haven't seen toasters tripling in price even if they last a lot less than before, the very essence of his argument - that lack of competition will bring higher prices.

You're basically agreeing with my scenario without realizing it.
 
Yeah, but I haven't seen toasters tripling in price even if they last a lot less than before, the very essence of his argument - that lack of competition will bring higher prices.

You're basically agreeing with my scenario without realizing it.

No I'm not. I disagree with you. The price of the toaster doesn't really have to triple for the cost to the consumer to increase.

However, toasters are not in a monopoly situation. The larger appliances like refrigerators etc have planned obsolescence out of necessity because big old refrigerators and washing machines don't really die. The appliance makers eventually started decreasing product quality to make the items cheaper, disposable, and guarantee future sales. A new washing machine doesn't have to be faster than an old one. People buy for energy efficiency, ease of transport, convenience, cost, etc.

Planned obsolescence in a monopoly can be as simple as reducing product quality to a point where the user is forced to buy a new one every few years. So now mom and dad might buy a new computer every 5-6 years. In the future that could be 2-3 due to product failure, built-in incompatibility, or something similar. The cost of the processor may stay the same or go down but the cost to the consumer is still higher in the end.

Intel licenses its motherboard designs as well. In a monopoly situation you would be changing sockets every generation, paying more for each product, getting a lower quality part, and if you want to game be prepared to shell out because each non-typical use case will be cost stratified to make things more expensive. If they want even more control they could go back to something like RD-RAM and control both cost and each iteration of product development by its peripheral compatibility.

There are tons of insidious ways for a company to screw you over in a technology monopoly. All they have to do is continue building faster processors and as you continue upgrading overall quality goes down and platform cost goes up. What you're saying might have a bit of truth if you could get away with using a Pentium 3 today. You can't really. Intel doesn't have to compete with itself, they can just obsolete their old parts with new instruction sets, deals with MS to utilize new features, or just by making junk.

You would end up paying more in an x86 monopoly, but it wouldn't always be just in the cost of the base product.
 
Back
Top