I'm sick and tired of people saying...

I agree, partially.

It is always fashionable to look for the end of CMOS, but at the end of the day it is cheap and a well understood material.

On the other hand, I think you'll find that AMD/Intel etc are the few companies to continue shrinking processes. I can see a whole lot of companies staying at 0.13u and above for a very long time - to use my usual example, look how long it has taken for any of the GPU companies to move to 0.13u.

0.13u and 90nm has some very horrible signal integrity and process variation issues that are making life quite difficult for designers at the moment.

Quote:

---

Originally posted by Arcanum
I agree, partially.

It is always fashionable to look for the end of CMOS, but at the end of the day it is cheap and a well understood material.

On the other hand, I think you'll find that AMD/Intel etc are the few companies to continue shrinking processes. I can see a whole lot of companies staying at 0.13u and above for a very long time - to use my usual example, look how long it has taken for any of the GPU companies to move to 0.13u.

0.13u and 90nm has some very horrible signal integrity and process variation issues that are making life quite difficult for designers at the moment.

---

IBM will as well. I’m not sure what the foundries like TSMC and UMC will do, but they are not noted for cutting edge performance. IBM, Intel and AMD may be the only three companies to develop process technology suitable for a high-end microprocessor.

That’s why I have doubts about AMD’s ability to keep up in the longer term. IBM and Intel have the resources to throw at the problem, AMD doesn’t, and as you say the resources required go up significantly below .13.

AMD could continue to exist as a design only company, but those types of companies haven’t had a great history in the X86 world. Transmetta needed their ASP to be over $150 just to break even after paying IBM to fab their processors. I could be wrong I’m not convinced that business model can work for X86 processors.

(BTW, just so there is no confusion, I don’t think these things necessarily apply to embedded processors like you work on since the competitive issues are significantly different. In particular, in embedded processors (IMO :) ) complexity is punished instead of rewarded, which should help keep design costs more reasonable in that space. I.E. brute force isn’t effective.)

Quote:

---

Originally posted by Chronus


There are other vectors to be followed, vectors that will be taking over the whole industry in the next 5 to 10 years,
Chronus

---

I think you will find that it's nearly impossible, with out support from intel new standards in the desktop market don't amount to much, you have to remember marketshare.

Looks like strategic partnerships, rather than self contained production, will be the answer to the massive cost of fabs and manufacturing.

Quote:

---

Germany's Infineon Technologies said on Tuesday it had teamed up with AMD Inc. and Taiwanese foundry UMC to develop technology to produce the small structures in next generation chips. Infineon said the development program on 65/45 nanometer manufacturing platform technologies -- would take place at United Microelectronic Corp.'s facility in Hsinchu, Taiwan, where later the chips will be produced. The first chips using structures made from the technology are likely to be produced in 2005, an Infineon spokesman said. He added that as the number of elements on a chip doubles every year, semiconductor makers are under pressure to develop new microelements to fit on chips, which are also shrinking. At the moment the size of the smallest element on a chip is 130 nm.

---

As of today, the market value of these companies: (http://moneycentral.msn.com)

AMD
Market Cap. USD$2.885 Bil

Intel
Market Cap. USD$119.1 Bil

Another interesting piece of financial information, the projected fiscal year 2002 estimates for growth rate:

AMD
Growth Rate FY(12/02) -1,330.56%

Intel
Growth Rate FY(12/02) 12.85%

Simple numbers that don't reveal the entire picture, but give a good idea of where the companies stand.

Quote:

---

Originally posted by Un4given
Looks like strategic partnerships, rather than self contained production, will be the answer to the massive cost of fabs and manufacturing.

---

Certainly an interesting move, and something AMD likely must do to stay in business long term. I can’t help feeling IBM wouldn’t be a better fit though.

Each of these three companies is likely to have its own requirements from a process. AMD produces needs high performance, with reasonable power output. UMC is a foundry wants low cost above all else. Infinion produces DRAM, which is very different again. They also produce some other types of chips, but I’m not quite sure where they fit in.

IBM would not acquire AMD for four reasons, IMHO:
- they are focusing more on services rather than products
- they would lose all Intel business
- they would be bringing in a company that is not profitable, in a sector that has had diminishing margins
- even with their might, they'd still have a difficult time against Intel

Quote:

---

Originally posted by isochar
IBM would not acquire AMD for four reasons, IMHO:
- they are focusing more on services rather than products
- they would lose all Intel business
- they would be bringing in a company that is not profitable, in a sector that has had diminishing margins
- even with their might, they'd still have a difficult time against Intel

---

He didn't mean IBM aquiring them, he meant IBM fabbing for them.

Quote:

---

Originally posted by Un4given


He didn't mean IBM aquiring them, he meant IBM fabbing for them.

---

Idle fabs are the devils playground.

(Sorry, couldn't help myself)

Quote:

---

Originally posted by Un4given


He didn't mean IBM aquiring them, he meant IBM fabbing for them.

---

I think the fit would be good either way, though admittedly being purchased by IBM works better for AMD then it does IBM.

Quote:

---

Originally posted by isochar
IBM would not acquire AMD for four reasons, IMHO:
- they are focusing more on services rather than products
- they would lose all Intel business
- they would be bringing in a company that is not profitable, in a sector that has had diminishing margins
- even with their might, they'd still have a difficult time against Intel

---

IBM is a VERY large company that has extensive involvement in both products and services. IBM is currently doing 18.5 Billion dollars in business each quarter compared to only 6.7 Billion for Intel.

I’m not sure what Intel business you are referring to, but IBM and Intel look to be headed for a collision course in the mid-range server market. (I.E Unix servers). The only three players that haven’t decided to get out of designing processors for this market are Intel, IBM, and Sun.

Sun just keeps falling further and further behind and Intel is a newcomer. IMO that leaves IBM as the target to beat. But at the very least they are going head to head no matter what.

IBM also has one of the widest range of X86 server out there. Not only do they sell more traditional X86 servers, they also sell a line of X86 based NUMA servers that would mesh nicely with Hammer.

Additionally they have the best packaging technologies in the world and some of the best process technologies, and are more then capable of designing chipsets and motherboards, all of which cold be enormous benefit to AMD.

To top it off they are also a tier 1 vendor of workstations this could be an entry point for AMD into the corporate world.

The downside is that IBM no longer gives preference to its own business units. Even if they bought AMD, AMD would still get no preference for use in servers or corporate PCs. And as I said before, the benefits for IBM are much less then they would be for AMD.

The only real benefit to IBM would be increased size and capabilities in it’s processor design area, and increased utilization of it’s FAB capabilities. Consider, IBM’s Power 4 processor should have almost every conceivable advantage over Hammer, yet will likely end up with only a small performance advantage. (The Power 4 has a superior ISA, HUGE amount of cache, better compiler support, much more bandwidth, far more execution resources, a far larger transistor and Power budget, etc) Clearly AMD has some design talent that IBM could benefit from.

The economies of scale produced by having a popular consumer processor are not to be sneezed at either. AMD sells far more Athlons, then all of IBM server processors combined.


Still, I think you are probably right, and IBM has no particular desire to pick up AMD at this point, but I still think it would be a good fit with potential advantages for both companies. Besides, I just can't see today's kinder, gentler IBM wanting to take on Intel head to head.

this is a little off subject and sortof like comparing apples and oranges, but does anyone think apple is going to go under? i mean they suck for gaming and all they are living off of is the schools buying them, and the idea that an apple can do no wrong. i mean they are saying that apple doesnt have any problems and is comparing its new OS to the older windows OSs. Basically they are living off people who dont know jack about computers.:mad: i would never buy one and all the people i know think they suck. even our school (who has macs) say they suck. i just dont see them surviving very long.

Quote:

---

Originally posted by n24tg
this is a little off subject and sortof like comparing apples and oranges, but does anyone think apple is going to go under? i mean they suck for gaming and all they are living off of is the schools buying them, and the idea that an apple can do no wrong. i mean they are saying that apple doesnt have any problems and is comparing its new OS to the older windows OSs. Basically they are living off people who dont know jack about computers.:mad: i would never buy one and all the people i know think they suck. even our school (who has macs) say they suck. i just dont see them surviving very long.

---

People have been talking about that for years, but they seem to be holding onto their core market share of around 5%, and sell their product for a good profit. In the longer term they could have trouble, since it looks like Motorola is getting further and further away from the desktop with it’s processors.

At some point this could be a real problem, forcing them to switch processors yet again, and there just may not be many alternative processors out there for them. I’m not sure Apple (or sun for that matter) could survive if they switched to X86. Maybe Itanium would work for them though, or maybe Apple could get IBM to produce a Power 4 (or it’s descendants) with Altivec support.

Originally posted by Moridin

Quote:

---

I think the fit would be good either way, though admittedly being purchased by IBM works better for AMD then it does IBM.

---

No disagreement there. :)

Quote:

---

IBM is a VERY large company that has extensive involvement in both products and services. IBM is currently doing 18.5 Billion dollars in business each quarter compared to only 6.7 Billion for Intel.

---

Yes, but if you have been following their trends over the past decade, you would see that they have been gearing themselves for a Services-dominated model. (Much in the same way many technology companies are doing so they can stay afloat or maximize income)

Quote:

---

I’m not sure what Intel business you are referring to, but IBM and Intel look to be headed for a collision course in the mid-range server market. (I.E Unix servers). The only three players that haven’t decided to get out of designing processors for this market are Intel, IBM, and Sun.

---

IBM sells businesses a complete solution, which not only includes the servers but the desktops as well. Ironically, these are being outsourced to an outside company. *cough* Services-only *cough*

Quote:

---

Sun just keeps falling further and further behind and Intel is a newcomer. IMO that leaves IBM as the target to beat. But at the very least they are going head to head no matter what.

---

Definately. Unless there are some surprises, Intel and IBM will be duking it out for the server market.

Quote:

---

IBM also has one of the widest range of X86 server out there. Not only do they sell more traditional X86 servers, they also sell a line of X86 based NUMA servers that would mesh nicely with Hammer.

---

No doubt that all these x86 servers are Intel-only machines. IBM and other tier 1 vendors know exactly what would happen if they adopted AMD-only.

A little foresight into the future, but I really question the profitability of Hammer. These will be substantially more expensive to produce than Athlons, yet how much more will AMD be able to charge?
If Intel sees any threat from them, they just drop the Xeons' prices to a level where AMD will have no counter.

Quote:

---

Additionally they have the best packaging technologies in the world and some of the best process technologies, and are more then capable of designing chipsets and motherboards, all of which cold be enormous benefit to AMD.

---

Yes, but would it benefit IBM to dedicate all that production for AMD processors/chipsets/motherboards?

Quote:

---

To top it off they are also a tier 1 vendor of workstations this could be an entry point for AMD into the corporate world.

---

That could very well change as soon as they don't have Intel-inside.

Quote:

---

The downside is that IBM no longer gives preference to its own business units. Even if they bought AMD, AMD would still get no preference for use in servers or corporate PCs. And as I said before, the benefits for IBM are much less then they would be for AMD.

---

My brief glance at it seems to indicate that IBMs units work independently of each other and look for the best suppliers/customers when shopping around, even if they are not from IBM.

Quote:

---

The only real benefit to IBM would be increased size and capabilities in it’s processor design area, and increased utilization of it’s FAB capabilities. Consider, IBM’s Power 4 processor should have almost every conceivable advantage over Hammer, yet will likely end up with only a small performance advantage. (The Power 4 has a superior ISA, HUGE amount of cache, better compiler support, much more bandwidth, far more execution resources, a far larger transistor and Power budget, etc) Clearly AMD has some design talent that IBM could benefit from.

---

Why not let AMD bankrupt and then steal parts of their team for a much cheaper overall cost?

Quote:

---

The economies of scale produced by having a popular consumer processor are not to be sneezed at either. AMD sells far more Athlons, then all of IBM server processors combined.

---

True. But quantity doesn't always equal a profit. I guarantee you that IBM makes a substantially larger profit off their server processors than AMD does on their desktop processors.

Quote:

---

Still, I think you are probably right, and IBM has no particular desire to pick up AMD at this point, but I still think it would be a good fit with potential advantages for both companies. Besides, I just can't see today's kinder, gentler IBM wanting to take on Intel head to head.

---

I personally see AMD having too many disadvantages at this point in time to justify anyone buying them out/taking them over.

Well, IMHO and only IMHO, if IBM gobbled up AMD, it would be the end of sub $400 CPU’s. Or, if they held the line on price, you would only be able to run them eight hours a day.

If AMD were to be bought, it would probably be a large Asian consumer of chips. An in house chip fab and development infrastructure would be a good fit for them.

Forgot to add one thing to this comment:

Quote:

---

Definately. Unless there are some surprises, Intel and IBM will be duking it out for the server market.

---

Intel would have an advantage in getting vendors to use their chip. Why? Because IBM is also a vendor - Dell & Co. wouldn't want to support the competition by using the competition's chips in their own servers. This is quite similar to the problem that Pepsi had by owning Taco Bell, Pizza Hut, and KFC. Other fast food restaurants chose Coke so they wouldn't help out the competition.