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.

Politics : Formerly About Advanced Micro Devices -- Ignore unavailable to you. Want to Upgrade?

To: Scumbria who wrote (49309)	2/11/1999 7:12:00 PM
From: Yougang Xiao	Read Replies (3) \| Respond to of 1572687

K7 to be released in EARLY 2nd half: realworldtech.com AMD Still Fighting In a discussion with AMD Channel Sales representative Alex Willingham, the message was delivered loud and clear that AMD intends to continue the fight against Intel on all fronts. Alex was very open about the roadmap AMD has laid out, and while it is really no different than what has been given for the past several months, he was able to answer some questions that until now have been subject to much conjecture. The discussion covered the K6-2, K6-III and the K7. K6-2 - The K6-2 will continue to be manufactured in volume through 1999, and possibly into mid-2000. Intel has forced AMD to position their K6-2 against the Celeron, and AMD has accepted that role. Though plans had called for the K6-2 to top out at 400MHz, recent events have convinced AMD to produce a 450MHz version, which should be available sometime in Q2. When asked whether AMD was committed to the Socket 7 platform through 1999 and beyond, Willingham responded that this was the current plan. While no firm commitment was given to produce a 500MHz K6-2, he did not rule it out. Market demands and Intel's aggressive stance will likely force AMD to make whatever process improvements are necessary to achieve this. K6-III - While Intel is trying to force AMD to play exclusively at the low end, AMD has other plans in mind. Willingham assured us that K6-III will be officially released on Feb 22nd - on schedule. The retail channel, however, will not likely see any product until mid-Q2, since Gateway and other OEMs will take most, if not all, produced. This is actually fairly good news for AMD as more OEMs look to capitalize on the momentum AMD has built up during the last year. The K6-III will be positioned directly against the Pentium III (thus explaining the name), and will debut at 400MHz, followed soon after with the 450MHz. Both the K6-2 and K6-III will be made at the same production facility, since they can be manufactured on the same production line. This means that the only reason for supply problems will be incorrect forecasting or extremely high demand, since it takes approximately 12 weeks to produce a wafer. There are currently no solid plans to transition the K6 line to .18 micron technology, though it may happen late in 1999 or early 2000 if faster speeds are necessary. K7 - The K7 will be positioned against the Williamette, which is still over a year away according to Intel's current roadmap. This is AMDs best chance to raise their ASP (Average Selling Price), giving them at least 6 months before Intel will be able to release a comparable processor. This could be used to subsidize the low-end processors, much as Intel subsidizes their Celeron line with the Xeons and high-priced Pentium IIIs. K7 will be released in early 2nd half of '99, with most going to OEMs initially. The retail channel will not likely see any until sometime in Q3. The first chips will be using .25 micron technology, transitioning to .18 micron and copper technology late Q4 '99 or early Q1 2000. Though much as been said in the press regarding AMDs troubles competing against Intel, they have not wavered from their stated goal. Remember that last year at this time AMD appeared to be in grave trouble, but the K6-2 turned things around in quite a dramatic fashion. The K7 may provide the same impetus if AMD can get there in one piece. The one difference between this year and last year is that Intel has now recognized the AMD threat, and is reacting with force and decisiveness. It remains to be seen if Intel actually has any other tricks up their sleeves to fight against the K6-III. The one thing that may cause AMD substantial trouble is if Intel decides to push their Pentium III roadmap up as quickly as they have the Celeron roadmap. Though that would likely cause Intel to lose substantial revenue in 1999, it could be the final nail in AMDs coffin. At this point, it is obvious that the focus is upon the K7, as this will give them the breathing room necessary to get back into the black for this year. If AMD can struggle through the next 4 months until the K7 is released they have a chance to solidify their position in the market and become a true competitor to Intel over the long haul realworldtech.com An AMD spokesman indicated in a conversation that the Dresden fab has already produced first silicon. The plant will primarily be producing K7 chips. Initially, these will be .25 micron but will be transitioned to the .18 micron and copper technology during Q4 '99 or Q1 '00. The K6-III will be initially released at 400MHz, with a 450MHz scheduled to follow in a short time. the K6-2 will also be pushed to 450MHz, and AMD indicated that all production problems have been solved. Since it takes about 3 months to make a wafer, expect volumes to increase for both K6-2 and K6-III about midway through the 2nd quarter

To: Scumbria who wrote (49309)	2/11/1999 7:52:00 PM
From: Elmer	Read Replies (1) \| Respond to of 1572687

Re: "High frequencies are hard to test, because most testers can not run at high frequencies." I made this same argument about the K7 with a highspeed backside L2 cache bus. It's untestable with todays production testers. Others here insisted the test capability exists (it doesn't). Assuming you are correct about Intel lowering the Rambus speeds, and I haven't seen Intel make that statement, do you think it is a problem with Intel testing their chipsets or the RDRAM manufacturers testing their RDRAM modules? EP