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: jim kelley who wrote (39558)	4/10/2000 8:33:00 PM
From: jim kelley	Respond to of 93625

Re: Processor& Memory speeds The first level of high speed memory is the tightly coupled cache memory which is usually on the Processor chip. A second level of memory could be the cache memory -more loosly coupled-that is near the processor chip. DRAM sits at the third level of the memory hierarchy and usually is on a DIMM or a RIMM and communicates across a bus some distance from the processor. The L1 cache is the fastest, followed by the L2 cache and the L3 memory is the slowest. The cache memories are usually static memories and ideally no processor wait state need to be inserted to synchronize the processor read/write/fetch operations with the memory. Apparently, the frontside bus interfaces with the L3 DRAM memory. The current Intel speeds on this fsb are 100 and 133MHZ. Putting memory that can access faster than 133 MHZ is pointless. However, no DRAM currently accesses at either of those speeds since the address must be latched before the access. Then the data must be stable before being read from or written into memory. I believe Intel is planning to go to a 400 Mhz FSB with their new core contained in the Williamette processor. Also, they start the speed of the processor at 1.4 GHZ. The speed will increase with successive genrations and processes as they have in the past. This should match very well with RDRAM capabilities as the speed can scale with RAMBUS design. Eventually, the Williamette core will find its way into the low end processors the same way as the PII core found its way into the Celeron product. The clock speeds on the low end products will rise and they will be able to make effective use of the RDRAM type memory. So this needs to be viewed as a technology transition consisting both of architectural and process changes. One can cool down the L1 and L2 and processor and then overclock the system. So if you want to do this go buy a refrigerator and some mineral oil. :)

To: jim kelley who wrote (39558)	4/10/2000 11:18:00 PM
From: Tenchusatsu	Read Replies (2) \| Respond to of 93625

Jim, Tom Pabst is using his "uberclocked" 440BX test as a predictor of potential performance using an Intel chipset and PC133 SDRAM. What people don't realize is that you can't draw a connection between an overclocked 440BX and real-world PC133 performance. Unfortunately, some news agencies were reporting that RMBS' drop from 400's to 200's were due to Tom's articles. Even though that was just bogus speculation, I'm sure Tom's ego has inflated tremendously as a result. So despite the invalidity of his tests, he's still going to milk the anti-Rambus controversy for all its worth. Tenchusatsu