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 : Rambus (RMBS) - Eagle or Penguin -- Ignore unavailable to you. Want to Upgrade?

To: Scumbria who wrote (52055)	8/31/2000 12:39:09 AM
From: vc21	Read Replies (1) \| Respond to of 93625

Please keep posting all the negative links. You'll help bring on more naysayers that will help make RMBS longs even richer.

To: Scumbria who wrote (52055)	8/31/2000 12:50:40 AM
From: richard surckla	Read Replies (1) \| Respond to of 93625

Scumbria... You saw what the article said, "PHENOMENAL!" Take it up with The Register... they're the ones that presented the story!

To: Scumbria who wrote (52055)	8/31/2000 2:12:34 AM
From: NightOwl	Respond to of 93625

Hi Scumbria, I agree INTC's CPU doesn't show well on those benchmarks. Moreover, if they were in fact accurate representations of its capability, then we should have been seeing DRDRAM as the Center Ring #1 Attraction at this past IDF because the memory scores are the only thing that stand out on those 2cpu.com tests that I can see. But at the risk of incurring the *Wrath of Kharl* by sticking my Mom & Pop beak into the inscrutable calculations of you EE's <g>, I simply have to make some observations about these numbers. First of all the numbers may indeed be shocking and PHENOMENAL to a guy operating a WEB site called "2cpu"; but - for anybody around here who has been paying !ATTENTION! - they are, if anything, lower than I would have expected. The CLiBench Memory Throughput number given was 303,840kB/s. Assuming this is in kilo"bytes", rather than kilo"bits", by my <ugh> math that translates to 2,360MBytes/s. The SciSoft Memory Bench shown gives values in its "Stream" sweet of tests as ranging between 1255 to 1590MB/s and averaging 1407MB/s. Obviously I don't expect the guys at 2cpu read: electronicproducts.com [Which I was, unfortunately, forced at gun point to post on this list TWICE before!] Now I know that most of the vocal, er I mean "local" residents, ignored that URL since they "thought" it didn't show anything they "wanted" to see. But the more adventuresome reader, no doubt saw that the chart contained there clearly indicated that Dual Channel DRDRAM is fully capable of a maximum Peak Bandwidth of 3200MBytes/s assuming its used to full capacity and is kept far, far away from those nasty little random access "page-misses". Based on the design capacity of DRDRAM my guess is that the Clibench test is showing the systems maxmium possible throughput or Peak Bandwidth at 2360MBytes/s while the SciSoft Memory Bench is showing some sort of average rate on a test requiring minimal CPU calls for random 32bit packs of data, a/k/a a Stream test. The good news is that DRDRAM hit a BIG number at 1407MB/s. But the bad news is that these were "Stream" numbers, where those nasty little page-misses are about as low as they go in the PC environment. And finally, the very worst news is that 1407MB/s is less than 50 per cent of what the design should be capable of. Even the presumed peak bandwidth number is about 800MBytes/s under the DRDRAM's highest capability. All in all I suspect we are going to hear much FUD from RMBS and INTC as to who's technology is to blame for the relatively poor comparative benchmarks to come. If they don't get the Peak Bandwidth number up considerably DDR-SDRAM will eat their lunch, since Maximum Peak Bandwidth will be the illusory basis for all DRDRAM sales. Then again we have only the word of RMTR, the University of Maryland and the University of Michigan that DRDRAM is indeed capable of running at 3200MB/s Peak Bandwidth. Finding a real world application which uses and thus could prove this "capability" seems to be pretty tough to do. ...At least for the cut and pasters at "2cpu.com". Now I'm going to take cover before Kharl launches those Photon Torpedoes. <vbg> 0\|0