Though VIA is moving forward with their processor plans, there does not seem to be much enthusiasm for it in Europe or North America. There doesn’t seem to be much news out of Asia with regards to popularity, but it can’t be too significant or we would likely be seeing a lot of press releases. Transmeta also seems to be doing poorly with their Crusoe chip, though they have announced a few design wins recently. It would be my guess that the fierce competition between AMD and Intel, along with the aggressive speed ramps and price reductions, will leave both VIA and Transmeta to suffer the same fate as Cyrix and IDT had several years ago, since neither seems capable of offering a significantly better price/performance part in the near future.
On the desktop, Athlon/Duron seems to be the popular choice, though P4 has reportedly begun to pick up some steam. All indications point to the PIII as the processor that is getting affected by P4 sales, rather than Athlon. The current P4 implementation (Willamette/Socket 423) is scheduled to top out at 2GHz later this year, however there seems to be a chance that Intel may leave 2GHz for Northwood (Socket 478). Until Northwood appears, Athlon/Duron will probably still have the momentum in this market. However, there is no doubt that Intel is pushing P4 with all their might.
Northwood will debut around November at 2GHz, and probably will reach 2.2GHz by the end of the year. Palomino (next generation Athlon) will probably be released around September at 1.5GHz, and likely won’t get any faster than 1.7GHz by the end of the year (some have said 1.6GHz is the top end this year). The real limitation for both of these processors is the power and heat dissipation, so getting to a .13u process is essential for continued speed ramps. This is probably the reason we are not seeing the rapid release of speed grades that we saw late last year – both companies realize that .13u high-end processors are not going to be produced in volume until sometime in 2002.
Intel’s current .13u offering (Tualatin) is being de-emphasized in the desktop market, and is instead targeting the mobile and server markets. The 1.26GHz part has reportedly been removed from the desktop roadmap, as Intel does not want anything to prevent the P4 from dominating in this space. Apparently, 1.2GHz is the fastest Tualatin that the desktop market will see, at least until it moves down into the low-end segment and becomes the Celeron, and even then it likely won’t go beyond 1.2GHz until 2003. Further evidence of this is that the desktop version will have only 256K of L2 cache, while both the mobile and server parts will have 512K cache and will achieve speeds greater than 1.2GHz.
In the mobile market, Intel still has the major share, however the recently introduced Athlon 4 may provide some sales for AMD. HP has announced a laptop based upon the Athlon 4, and others should follow in the near future. Unfortunately, the high power requirements for Athlon/Duron will be no match for Tualatin w/512K cache (now called Pentium III Processor M) at .13u, and AMD doesn’t seem to have any real solution for the problem this year. One minor positive for AMD is that Tualatin will not be introduced this month, as Intel had expected earlier this year. The launch has been pushed back to the end of July or early August as Intel focuses instead on the Server market.
Intel seems to have drawn a line in the silicon, and it is in the workstation and server market. Based upon recent product positioning, and the acquisition of the Alpha resources and IP from Compaq, there can be no doubt that Intel views this as their most important market for the future. At the high end is IA-64, with Itanium now and McKinley in early 2002. AMD has nothing to counter this until mid-2002 when they release their x86-64 Hammer family of processors. Even then, it is questionable whether the server market will seriously consider anything but IA-64, particularly now that Alpha is no longer a viable option.
In the ‘General Purpose’ server segment, as Intel defines it, Tualatin will apparently hold the fort at speeds up to 1.4GHz, until the P4 ‘Prestonia’ becomes available early next year. For rack servers, which Intel defines as high-density and ultra-dense, Tualatin again becomes an important strategic part. Due to very low power dissipation, it is a perfect candidate for this segment. Eventually, P4 (Prestonia) will move into this segment, but likely not in any significant way until mid-2002. Again, AMD has no answer to the Intel solutions until midway through 2002 at least. Much seems to be hanging on the Hammer processor line for AMD.
The workstation market should have some interesting choices, including Northwood, Prestonia, Itanium and AthlonMP. This is one high-end segment that AMD appears to have a decent shot at gaining a toehold. The current P4 is not dual processor capable, and it won’t be until early next year that Intel will have anything other than a PIII Xeon that is – except, of course, the very high-end and expensive Itanium chip. If AMD can prove the reliability of their AthlonMP processor and 760MP chipset, we may see some customers jumping to this platform just for the exceptional price/performance. This is very important for AMD, as it would also give them some automatic customers for their Hammer processors.
What this all comes down to is that Intel has decided that the time for the Pentium 4 is now, and that Tualatin is useful only as a placeholder until the Northwood becomes available. AMD is now in a position where their Athlon is getting a bit long in the tooth, and they will not get to .13u until next year – perhaps not until Q2. Looking at things objectively, AMD really has very little to counter upcoming Intel products until early next year when they get to .13u, giving Intel a chance to take back some of the market share they lost and solidify their mobile and server markets. It seems absolutely imperative that AMD get their Hammer processor line out on time.
Be the first to discuss this article!