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: Ali Chen who wrote (4142)	8/9/2000 6:28:15 PM
From: Petz	Read Replies (1) \| Respond to of 275872

Ali, I guess the point about the code fragments I cited being nearly twice as fast on DDR is: This is the same type of code Intel tries to load benchmarks with for demonstrating the superiority of RDRAM. For example MPEG encryption/decryption, filtering huge graphics files and speech processing, all of which have nearly infinite continuous datasets. So, whatever Intel does to benchmarks to make RDRAM look good will also make DDR look good. I thought I would take a look at Intel's benchmark results to see which benchmarks are highly dependent on main memory maximum data transfer rate. But looking at the Intel website I notice that only 3 of 13 benchmarks run faster on the 1600 MB/sec i820 than they do on the 1033 MB/sec i815. Those three benchmarks are Video 2000 -- 1.5% faster on 1820 Winbench 99 FPU Winmark -- 0.2% faster on i820 JMark 2 processor test (Java) -- 0.7% faster on i820 On the other 10 benchmarks, the i820 with 1600MB/sec RDRAM is, on average, 2.7% slower than the i815 with 1033MB/sec PC133. So there's practically nothing that can take advantage of the an increase of bandwidth beyond 1033 MB/sec obtained by ordinary PC133. Clearly, something besides better peak data transfer rate is messing up the i820 results. Yeah, I know, latency. Maybe the best way to estimate the maximum performance advantage of PC2100 to PC133 is to compare PC133 to PC100 (both with CAS 2) and then triple the difference. Thats being generous, because latency for PC2100 is no better than PC133. Anyone want to take the challenge? Of course make sure you compare PC100 to PC133 using the same manufacturers chipset. (Don't compare Irongate PC100 to VIA PC133; DO compare Intel i815 to Intel BX PC100 or KX-133 with PC100 to KX-133 with PC133 and same CAS settings.) Petz