SI
SI
discoversearch

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: Charles R who wrote (107322)4/22/2000 3:04:00 AM
From: Joe NYC  Read Replies (2) | Respond to of 1572942
 
Chuck,

Now, an interesting question (for me at least): Who will debut a mass market part with memory interface embedded into the CPU? What time frame? What will that memory interface be?

The Cyrix MediaGx suffered from the lack of L2, and tiny L1. Their more mainstream chip Mx (or Mxi?) with 6x86 core and 64K L1 never made it to the market because the whole chip did not scale. THe chip that was supposed to scale well M3 or Jalapeno (not what via is selling now as Cyrix 3) was a great concept, but never made it out of the design center. The next CPU with on-chip memory controller will be Timna, but the memory interface is Rambus.

I think Timna is based on P6 core, but I don't know if it has any L2. The performance will be crippled by the Memory translation to SDRAM.

The approach I would like AMD to take, if they ever go in this direction is to integrate only the memory controller, nothing else. For a mainstream market CPU that will never be used in multiprocessing environment, this would provide optimum performance.

Joe



To: Charles R who wrote (107322)4/22/2000 3:42:00 AM
From: pgerassi  Respond to of 1572942
 
Dear Charles:

I Believe the Sledgehammer Dual Core may include all the pieces of a SMP Node consisting of two Athlon Cores, one LDT Connected Hub, and a DDRDRAM Controller or two. This would be equivalent to a section of the Athlon SMP Presentation given at the 1999 Microprocessor Conference. This would reduce the complexity of Large SMP Systems. Such a setup could easily beat Williamette SMP easily.

Pete