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: Paul Engel who wrote (34966)	7/22/1998 1:37:00 AM
From: Petz	Read Replies (3) \| Respond to of 1573004

Paul, Ali predicted SDRAM would have negligible impact and it WAS negligible, 1-2%. I suggest everyone read the article at tomshardware.com SDRAM does practically nothing to improve memory latency, but only increases memory bandwidth. Thus the effect on performance is MUCH less than the theoretical bandwidth ratio of better than 2:1. Petz

To: Paul Engel who wrote (34966)	7/22/1998 11:56:00 AM
From: Ali Chen	Read Replies (2) \| Respond to of 1573004

Paul, you are confusing few things. <today the Pentium II is a better chip than the K6.> You are confusing here a CPU and a system. We were talking about CPU core performance, and my point was that the P-II core is exhausted while the K6 core will show up with apropriate support from the system. It was Intel intention to limit Socket7 features to separate it from PPro/PII/Slot1 segment. They made Socket7 systems artificially inferior than they could be and took a significant chunk of system on PPro chip implementing non-blocking L2 cache. Clear example - lousy 64 MB cacheability limit on the top Intel 430TX chipset. AMD had to play this game initially to penetrate the market. <You stated there WAS NO IMPROVEMENT using SDRAMs - and when you were proven wrong you had to twist things> What you are talking about? Who proved me wrong? You? Any website will show you that there was virtually no improvement in system-level benchmarks when switching from old 440FX to 440LX SDRAM boards. There is only one synthetic benchmark (STREAM) you stuck with where huge but primitive vector operations are exercised. So what? This artificial type of data pattern is not used in business application in any significant quantities, and that's why there is no visible gain in overall system performance between EDO and SDRAM. The same goes for 66-to-100MHz move on P-II platform with 440Lx to BX chips. Speaking about industry switching to "more expensive" SDRAM, you are wrong here again. That was a matter of scale of production. If you check today's prices, 64MB of SDRAM is cheaper than EDO. The reason for industry to switch was easier handling of just one piece of DIMM memory rather than dealing with two 72-pin SIMMs, plus connectors, plus extra tracing... <YOU ARE STILL WRONG !> Let me suggest here something. I see you are trying to make you right by shouting your point in bigger letters. If you could use html switches to even bigger fonts, maybe it will make you more right? Look at any html source to figure out which control characters to use :-)