AMD

[jaydub]

[ QUOTE ]
[ QUOTE ]
[ QUOTE ]

1. IMO the multicore approach is basically an admittance of an architectural limitation.


[/ QUOTE ]

Your opinion is in the minority and without elaboration on your part, I strongly disagree.

Jason,

Please understand that the Yahoo message boards are almost always populated by people who are full of [censored]. If they told me the sky was blue, I'd go out and check for myself.

J

[/ QUOTE ]

The x86 architecture has been around 25+ some years. I don't think there have been significant advancements in the architectural space in x86 for a while. A majority of the improvements have been die shrinks, pipeline/prefetching tweaks, and moving more components onto the die all in the fight against heat and latency. The physical limitations are already starting to show which is why the transition to multicore was so quickly adopted. I guess I should have rephrased my statement to say that the limitation is both architectural and physical.

Just curious, but what's the majority opinion on the reasons for the transition to multicore processors?

QFT on the last part.

[/ QUOTE ]

JJ,

Max largely covered it but I wanted to respond as well.

Regarding arch improvements, 286 -> 386 was pretty damn significant as is x64.

J

[jaydub]

[ QUOTE ]
Theres only one reason why they went to multicore. Its not a matter of opinion open to debate.

Multicores are there because having two cores provides better performance. Making a single core run fast enough to provide similar performance is too difficult given the heat and power problems.

Multicores were less viable before because the processor size would have been too large and negatively affected yields.

So yes, it is somewhat of an admittance of process (I think this is what he meant instead of x86) limitations.

This doesn't have that much to do with the original topic other than to lend credibility to the fact that the product is becoming increasingly complicated while prices are falling, while growth rate of the marketplace is slowing.

Max

[/ QUOTE ]

You may be focused on the home user market. On the server and workstation side the multi core systems allow double (or quad) the number of active processes to run nearly as fast. This is huge.

Putting it back on topic, AMD's approach to multi core is technically better than Intel's but Intel was smarter in that they hacked together an inferior quad core to beat AMD to market.

J

[disjunction]

I have traded AMD and Intel since 1999, very successfully. In order to trade AMD successfully you have to have to have to understand the technology, both the "whose chip is better" as well as the manufacturing technology.

If you don't stay on top of both of these things, which basically turns into a full-time job in and of itself, then you should not trade AMD.

There are always funny things going on with the stock and options price. It's hard to know what to make of them. In the end, the technology trend wins out.

[cbloom]

[ QUOTE ]

You may be focused on the home user market. On the server and workstation side the multi core systems allow double (or quad) the number of active processes to run nearly as fast. This is huge.

Putting it back on topic, AMD's approach to multi core is technically better than Intel's but Intel was smarter in that they hacked together an inferior quad core to beat AMD to market.


[/ QUOTE ]

Sounds like you're thinking of multi-processor not the new multi-core stuff like Cell. But maybe not. For the record, multi-processor is sort of the old way, where you have several large full processors with reorder buffers and their own caches and you hook them together in various ways. AMD had a very cool system for that which was good for servers. Multi-core is the new thing in which you take one processor with only one L2 cache and one memory controller and you stick a ton of simple in-order cores on it (each with its own L1).

I had a dream in which I dreamed this : "Intel has a future multi-core chip for the home PC which is similar to Cell and supposedly very awesome".

However, I just don't see high-end CPU's like that really commanding the huge premiums that they did in the past. I agree with a previous poster who said CPU's are becoming commodities like memory chips which means the profit margins will fall and it will become a tougher and tougher business. I don't like Intel or AMD as a long term buy & hold stock. Also it seems like IBM may become a bigger player in the processor market.

In the short term - the market is really really dumb about technology companies, whenever somebody makes a big announcement the stocks go nuts. So as others have said you can definitely make money playing these things in the short term.

[cbloom]

Anyone who's traded AMD must remember the K6 debacle. And pretty much everything else AMD has done which has always been a giant debacle. They keep seeming to have the chance to catch up to Intel and then they blow it somehow.

[disjunction]

[ QUOTE ]
Disclaimer: I work for intel.

I think your conclusion is correct that you shouldn't buy AMD. However I think your reasoning isn't really based on anything substantial.

Your reasoning should follow this line of thought...

Max

[/ QUOTE ]

Don't be so pessimistic. I was this pessimistic 8 years ago.

There are plenty of tricks left. And plenty demand for those tricks. Image, 3D, and AI/Learning applications need as much CPU as you can give them. I think the only danger for AMD/Intel is that down the road the technology that makes for the fastest chips will more fall into some other company's expertise.

[maxtower]

I am sure you know how much intel's stock has fallen since 8 years ago.

How could you be anything but pessimistic about a company whose only viable product is becoming less and less profitable to make each year?

[jaydub]

[ QUOTE ]
[ QUOTE ]

You may be focused on the home user market. On the server and workstation side the multi core systems allow double (or quad) the number of active processes to run nearly as fast. This is huge.

Putting it back on topic, AMD's approach to multi core is technically better than Intel's but Intel was smarter in that they hacked together an inferior quad core to beat AMD to market.


[/ QUOTE ]

Sounds like you're thinking of multi-processor not the new multi-core stuff like Cell. But maybe not.


[/ QUOTE ]

No, I am quite specifically referring to processors with multiple cores. To be precise; Clovertown, Woodcrest, Tulsa et al.

[ QUOTE ]

For the record, multi-processor is sort of the old way, where you have several large full processors with reorder buffers and their own caches and you hook them together in various ways.


[/ QUOTE ]

Old way?

Multi core does not remove the need for multiple processors. It just allows larger numbers of cores without the isssues involved in machines with larger numbers of processors (density, power, heat, cost). An 8 way system now is far cheaper to purchase and maintain than a couple years ago.

[ QUOTE ]

Multi-core is the new thing in which you take one processor with only one L2 cache and one memory controller and you stick a ton of simple in-order cores on it (each with its own L1).


[/ QUOTE ]

This is a simplification and not entirely correct but that is not really relevant and well off topic.

However, to bring it back to point, the details which have been simplified here get towards why AMD is technologically better.

J

[NajdorfDefense]

[ QUOTE ]
Do I just assume anything that sticks out in a balance sheet is already worked into a stock price?

[/ QUOTE ]

Absolutely not. Both APT models and EMT assume there are traders taking positions to correct incorrect valuations. That doesn't happen instantaneously, even if you believe all that foolish EMT-hand waving.

Secondly, people routinely overlook/underlook at balance sheets for years on end. Enron would be an easy example. Any company guilty of channel-stuffing, it shows up on the balance sheet and other statements before they announce the earnings miss.

I've owned/followed AMD off and on for 15 years. It will break your heart.

[DesertCat]

[ QUOTE ]


AMD bought ATI Tech in 2006 and at first it was receieved well by wall street and the price soared. However in 2007, AMD has been murdered by INTEL. A big reason for this is a rather controversial issue which involved INTEL handing Dell money (alledgedly) under the table in order to get intel into the computers. The bottom line is that intel has been very scrappy and been doing whatever it takes to get into manufacturers, and AMD is struggling on the marketing side even though its chips are said to outperform Intel's.


[/ QUOTE ]

I think you got a lot of good responses on the state of current processor design, an area I have long had no interest in. But one point, Dell was getting money above the table from Intel, it's the Intel Inside marketing program, it's well known and they've been doing it for a long time. There is some controversy over how Dell accounted for it, and it looks like they did so in a misleading way, some heads might be rolling at Dell when the investigation is complete.

Because Dell took this money from Intel, they refused (or weren't allowed) to use Intel processors. This is one of the reasons Dell has struggled recently, esp. last year when AMD processors were clearly cheaper/better.

And to repeat what has already been said, ignore internet rumour mongers. I really doubt any hedge fund is committing huge sums of money to manipulate Intel options. That is typically the whine of a loser in a bad trade looking for someone else to blame instead of themselves. Hedge funds might be trading the options heavily but it's because they think some news is coming that's going to reward their positions.