Replies to post #4696 on Advanced Micro Devices Inc (AMD)

[Haddock]

The ant-swarm scenario would only work if AMDx86-64 is implemented in Prescot, so that a quick decision by Intel to activate it (and make chipsets available for it) would send--in a matter of weeks--all Intel's might into the space that AMD was so nice to clear for them. If it's a matter of fabbing new chips--that takes too long. If they don't already have a chipset for it, it may also be too late.

Many on this board know the business alot better than I do. Is this feasible?

They don't need new chipsets, since Xeon is already capable of addressing more than 4Gbytes of RAM. PAE is an ugly solution, but the uglyness is all inside the CPU as far as I know, so externally it's just a simple 36-bit system (40 in Prescott?).

I think it's feasible to have a secret x86-64 mode in Prescott that can be just switched on, but I don't personally find it very likely.

[sgolds]

blauboad, Yamhill by Christmas? No, it won't happen because even Intel has to distribute Beta versions, get iterative feedback, make corrections, rework masks, etc. in order to get a reliable product. It would take at least a year in the field before it becomes a product. Once you start distributing something of that magnitude then the rumors will be everywhere.

If Prescott has Yamhill in it then it will be turned off in the first commercial release of Prescott, and that release will serve as the Beta for the Yamhill code (switched on by a motherboard mod, for instance). The earliest it would be available for Xeon sales would be late 2004.

Let me address another part of your statement:

To go with the "Ant Hill" codeword, Intel's strategy might be to let AMD venture out first, watch if it finds a market, then send out a swarm of chips to take that market away. Like the one ant you see in your kitchen Monday becomes an invasion on Tuesday. AMD takes the risk, Intel stands ready to benefit. But can it work?

We have the benefit of the 3DNow! and SSE extensions as guidance. Intel's strategy was to watch AMD with 3DNow! and then come up with something that duplicated (most) of the functionality but was totally incompatible with 3DNow!. Effect was to blow AMD out of the waters because everyone coded for the huge Intel market.

No, if Intel licensed x86-64 from AMD then they would validate AMD's design and be correctly perceived as followers. (They would need aHT, too.) Instantly everyone would be porting their code to x86-64. There would be a landslide of support in the market, and by the time Intel got their version ready then they would be facing a formibly entrenched enemy in the server room.

I don't think Intel would like that.