ITotals: October`2002
Processors
The month turned out to be quite controversial. On the one hand, no products were actually released. On the other hand, MicroProcessor Forum held in October had certain consequences. But let me start with the current events instead of the future plans. AMD finally released its Athlon XP 2700+ and 2800+, but it was only an official announcement - the first processors with a 333MHz system bus rolled out at the end of the month in Japan only as preproduction samples. Again in Japan, two months after the release of the Athlon XP 2400+ its first real solutions hit the streets in the second half of October. Announcement of a processor is not followed straight away by deliveries of real solutions today. It concerns both Intel and AMD, but AMD has it even worse. And nothing is going to change here in the near future: the 2800+ will arrive only in the first quarter of next year!
Athlon XP 2800+
However, the poor Duron has lost its favor entirely: in this quarter production of this family will be ceased. Well, it's quite fair since it's wasteful to maintain a separate cheap line taking into account the current prices for the junior Athlon XP. Moreover, the Opteron will undoubtedly forces down the prices for the Athlon XP line even
more. But it might have problems with its release as both the Opteron and its analog for desktop PCs are going to get the mass support only in summer 2003 instead of the initially scheduled Q4 2002. I can't help mentioning once more the mine AMD put under the Hammer which is a built-in memory controller Intel is so delighted with. The upcoming DDR-II makes them make changes in the processor's design. If a new specification arrives tomorrow, the design will have to be changed again. But from the architectural standpoint (as RDRAM) the idea looks good which is proved by preliminary tests of the 2GHz sample.
Intel, however, didn't have a lot to boast of the last month, though it has a great heap for November, including the well known processor of the senior x86 line - 3 GHz Pentium 4 to be released on November 14, and Celeron 2.1 and 2.2 GHz expected on November 10. Now it's clear why AMD had to give up the Duron line, though such radical overfulfillment of the plans (the processors were scheduled for the first and second quarters next year) looks strange. But no secret that the new Celerons are not very successful because of a too small L2 cache; and the idea to lift up their performance at the expense of the clock speed seems to be nice. The most pleasant thing for customers here is that in the middle of November both lines will have their prices lowered marginally. End-users will notice the effect later, and the effect will really
be strong enough.
Also, Intel had tremendous plans for the server processors - it got first working samples of the Madison (0.13 micron Itanium 2) in October which is to be announced next summer, almost simultaneously with the desktop 0.09 micron Prescott. The "Celeron from Itanium" - Deerfield, which traditionally has its L2 cache cut down is expected by that time as well. But the processor will remain a server solution with 3 MB of L2 cache. By the way, once we touched upon the senior server line of Intel, I must say that the last orders for the Itanium (Merced) will be accepted in April 2003 - after that the market will have only the McKinley to offer.
But the processor makers didn't have much time indeed as they were busy summing up the results of the third quarter. AMD had the sales volume of $508M with the loss of $254M which is worse in both cases compared to the second quarter. The expected capital expenditures this year are $750M. In 2003 the company is going to make its business profitable again at the expense of reducing the costs and increasing the sales volume.
And here the Hammer can be of much help in the second half year, at least, there is certain interest toward it today - Dell dancing around it might have a real interest or just try to force its old partner Intel bring down its prices; besides, AMD is noticed by the Chinese Ministry of Education (the volumes must be impressive then); moreover,
Cray promises to build up by 2004 one more super computer for the US government within the ASCI program on the Opteron processors which is predicted the first or second position in the world table of ranks of the world super computers. Well, the level is going to be high enough provided the AMD is not too late.
The company's share on the market of x86 processors amounted to 11.6 %. It's quite a failure compared to the second quarter (15.6 %) and the worst outcome since the Q2 1998.
Those lost 4% were immediately picked up by Intel whose share thus increased from 82.8% to 86.8%, and the outcome of the last quarter amounted to $6.5 billion of the sales volume and $686 million of income. Both sums look much better compared to the second quarter. The capital expenditures in 2002 are around %4.7 billion (that's much lower than the planned %5.5 billion). One of their achievements is Fab 11X set going in October with 0.13 micron 300 mm production estimated at $2 billion; next year it is switching to the 0.09 micron production.
One of the failures in October is the victory of Integraph in the court, and Intel has to pay off the forfait of $150 million and then $100M for purchase of the disputed technology. The company has appealed to the court but the Intergraph's victory is almost undoubted.
One more player on the market, Sun Microsystems, has its scores between AMD and Intel, still nearing AMD. Its sales volume amounted to $2.7 billion with the losses being $111 million. The sales volume is a little worse compared to the previous quarter but the losses are smaller. That's the result of the proper restructuring. The company keeps on following this policy which may soon bring in 20% reduction of the staff.
And finally, we have Transmeta whose figures look much more decent - the sales volume is just $6.4M and losses are $21.8M. The tendency is like in case of Sun - the sales are falling down but the expenses are being reduced. The company is going to get profit by the end of 2003 but it seems that by that time the Crusoe will be buried by Banias, and the company will reach the zero balance.
VIA is now mostly working for its future - at present it is agitating the manufacturers for the Eden whose clock speed will reach 1 GHz next year, get the SSE support (Nehemiah core), and I have a feeling that VIA will focus entirely on this family having phased out production
of the Socket-370 processors without even starting the Socket-478.
VIA uncovered its roadmap at MPF which was quite a risky undertaking this year. Although Intel unleashed the new Xscale at IDF held at the same time, it's not very important as the details are much more significant. Intel showed two first processors of the new concept - system-in-a-package which combine the XScale PXA250 core and
Intel StrataFlash memory on the same chip. The Manitoba will add the DSP core, which is going to be an ideal combination for new cell phones. But it will be only in 2004. Apart from the RISC in the mobile field Intel also announced the 0.13 micron 400 MHz ULV Celeron, 3.4W.
AMD responded instantly, though only orally, with its new trade mark AMD Alchemy Solutions, - such processors will belong exactly to the same niche the new XScale is in now. But they are the processors of future, like those announced at MPF and meant for the same market niche - 800 MHz NEC Sapphire and 1.2 GHz Samsung Halla - and this is the good old ARM.
By the way, at MPF ARM Holdings displayed its microprocessor cores of the ARM11 family. They support DDR, consume just 200 mW at 533 MHz, has an integrated FPU and a
lot of new buses. The ARM11 will be used for the most new processor cores including the XScale. But the MIPS is not given up yet - at the end of October Toshiba released a new core based on this architecture, this is a new offering for digital TV sets, printers etc. However, the RISC is used not only for PDA and peripherals - just remember the PowerPC and Apple. At MPF they showed us a new PowerPC: PowerPC 970, the first 64-bit sample of this line, which is actually an analog of the x86 Opteron. The old 32-bit core polished with the 64-bit superstructure. This is a good marketing trick of Apple in the future competition against the Pentium 4.
 |
Top Stories: |
 |
 |
|
MoBo:
| |
|
|
VGA Card:
|
|
|
|
CPU & Memory:
|
|
