Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : Advanced Micro Devices - Moderated (AMD) -- Ignore unavailable to you. Want to Upgrade?

To: ptanner who wrote (87417)	8/21/2002 10:28:56 PM
From: Joe NYC	Read Replies (2) \| Respond to of 275872

Patrick, Very odd. I missed that the first scan through. Why so much finagling with HT links resulting a wide range of product components? Will there be that much difference in the price of a CH vs SH that it would be worthwhile to create an asymmetric dual board? Or can you put a CH in socket 940? I have the same questions. With the differences in the number of HT links, cache sizes, there could be way too many processors: 1. CH, 256k L2, 1 HT 2. CH, 512k L2, 1 HT 3. SH, 512k L2, 2 HT 4. SH, 512k L2, 3 HT 5. SH, 1MB L2, 2 HT 6. SH, 1MB L2, 3 HT I would like to see configurations 2, 3 and 6 in that order, with possibility of adding 1 later on as a replacement for K7 (unfortunately, I think AMD will start with 1, and 2 may not see the light of day until .09u) Joe

To: ptanner who wrote (87417)	8/21/2002 10:38:53 PM
From: Dan3	Read Replies (3) \| Respond to of 275872

Re: Why so much finagling with HT links resulting a wide range of product components? Well, if you're in a position to do it, you best maximize revenue by segmenting the market. AMD either has, or thinks they have, a good chance of gaining a tier one partner to market hammer. It would have to be SUN, IBM, or HPQ. I'd guess the best chance is HPQ/Compaq. Intel really looks like they've screwed the pooch on their server strategy. For the coming years, Itanium's positioning results in it being far too low in volume to allow any significant economies of scale - in either software or hardware. And workstations and servers need to be 64-bits, now. Because hardware has to reserve a number of blocks of the physical address space, and the operating system and software interrupts reserve more, the effective maximum memory for a 32-bit system is 3 gigabytes. That's less than $1,000 at current memory prices, and less then $500 next year. Any database or web server will benefit from more memory than that, and any workstation working with large image, video, or CAD files will benefit from more memory than that. Any workstation or server expected to sell for more than $2,000 next year will have to be 64-bit. So, Intel has limited itself to entry level and enterprise servers for next year, with nothing to offer the low mid-range, mid-range, or high end buyer. In workstations it's even worse - there will be no such thing as a 32-bit workstation next year, and Itanium hardware/software is too peculiar and obscure to be appealing to anyone but a handful of eccentrics. So AMD and its partner(s) must think they'll be able to sell $1,500 3-link chips, $750 2-link chips, and $400 split link / single link (dual 8-bit HT) 64-bit chips. I hope they aren't getting ahead of themselves, but AMD will probably sell more of their 64-bit Hammer chips in the first 5 minutes of taking orders from OEMs than Intel will have sold in 2 years of marketing, trying to sell, giveaways, and even paying people to use Itanics.