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: Björn who wrote (35697)	4/13/2001 9:32:56 PM
From: combjelly	Read Replies (2) \| Respond to of 275872

"please give me a hint how the 200/266 diff in FSB frequency affects the system performance." Well, this is a question that belongs up there with Pascal's Wager. Because of the large L1 caches, the Athlon is not as sensitive to memory bandwidth as it might otherwise be. But for those things that exceed the caches, the more bandwith the better. Personally, I like to overclock the FSB as much as possible to take care of those situations where the memory bandwidth dominates. Getting up to an FSB of 280MHz or 300MHz is no big trick. With PC150 memory, 320MHz or greater is real easy. YMMV. KT133A motherboards seem to be the best for this. The ECS motherboards have multipliers up to 400MHz, which is the highest I know of. Cheap, too.

To: Björn who wrote (35697)	4/14/2001 12:16:20 AM
From: ptanner	Read Replies (2) \| Respond to of 275872

Bjorn, Re: 200/266 FSB -> system performance Well, from recollection DDR systems were praised for a 10-12% performance gain depending on application. Until the KT133A the benefit from higher FSB versus DDR memory was unclear. Now it appears that most of the gain was from the higher FSB - so figure 8-10% depending on application for 266 vs 200 with a tad more for DDR. Of course, some apps are more interested in raw MHz and others more in memory bandwidth. So if you have a specific application that you are wanting to optimize for then I would try and hunt down some of the comparative reviews at Tomshardware, Anandtech, or Aces. I don't think the difference from KT133 to KT133A is compelling for an upgrade but KT133A is the best choice for a new board. There have been some compatibility problems with trying to use a 200 FSB CPU @ 266 but there seem to be plenty of workarounds. I can dig up links but let me know specific areas of question. FWIW, I just ordered an Asus A7S-VML (Sis 730 integrated uATX) today after considering Sis 730 based offerings from lesser known vendors. I am also planning to order an Iwill KK266 (KT133A) in a day or two (or possibly an Abit KT7A but onboard sound is a selling point for me but Abit is fully jumperless). Both systems will host OC Durons... unless my supplier gets some more 266FSB T-Birds... apparently they are selling better than the supply (which was reported as "we get a few and they are gone."). -PT EDIT: Both Anand and Tom have reviewed a variety of KT133A boards. Tom thought they were all pretty good since they basically evolved from mature KT133 designs. Anand also gave similar marks to all but a slightly higher to the Iwill KK266 which is physically very similar to the Abit KT7A. As for dual Athlon mobos... I have decided to move on and just be surprised. Seems like May/June but I wonder then there will be multiple products available for comparison and price competition. It would also be nice if some of the boards were targetted more towards the enthusiast rather than the workstation or server markets.

To: Björn who wrote (35697)	4/14/2001 10:12:30 AM
From: fyodor_	Read Replies (2) \| Respond to of 275872

Bjorn: Anyone, please give me a hint how the 200/266 diff in FSB frequency affects the system performance. How's the memory bandwidth affected? Ok, a couple of additions to what the other guys have said: One thing that's important to remember is that memory (bus) frequency and FSB frequency should be matched. That is, if you use 2x100 (which is what the 200MHz FSB is), the ideal motherboard would use either PC100 memory or PC1600 memory (the latter of which runs at 2x100MHz, the former at 1x100MHz). Because PC133 memory is one of the "check list items" that OEMs just need to have, the ideal motherboard doesn't exist for PC100 memory. AMD's original Athlon chipset (AMD750) was ideal, in the sense used here, but was overall not quite as fast as the competition - and more expensive. Luckily, VIA just introduced the ideal motherboard for PC133 memory: the KT133A (where the "A" is very significant!). In the DDRSDRAM realm, there are a couple of "ideal" solutions. Right now, the main ones seem to be AMD's 760 and VIA's KT266. The latter has gotten very mixed reviews, but this is apparently because of some problems with the initial boards and shouldn't be an issue (for long?). In any event, I would make sure to check some reviews before committing to a KT266-based motherboard. Both chipsets should be matched with PC2100 (266MHz) DDRSDRAM when using the 266MHz bus. One last note: Never sacrifice frequency for latency! For example, CAS2 PC100 performs virtually identical to CAS3 PC133. You really need to get CAS2 PC133 to get the performance increase. The same goes for PC1600 and PC2100. (CAS is, simply put, an expression of the memory's latency - lower is better). -fyo