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


To: Ian Anderson who wrote (38773)3/23/2000 12:54:00 PM
From: Bilow  Read Replies (1) | Respond to of 93625
 
Re DDR being relatively slower on second accesses. Not so. Given the same peak bandwidth, a DDR system will suffer the same added delay as an RDRAM system. If the DDR has higher bandwidth, as in the Dolphin, then the DDR will have lower delay for the second (and consecutive) accesses. This, of course, assumes data read from the same page, as would be the case for the bytes of a cache line.

SDRAM / DDR allow burst accesses that are designed to support cache fills with excellent efficiency. The figure to look at is peak bandwidth. That will give you how much delay the 2nd, 3rd, etc., accesses take, after taking into consideration the (actual) granularity.

Example. A memory system with a peak access rate of 3.2GB/sec that outputs 16 bytes at a time. (As per certain DDR memory configurations.) Each consecutive 16 bytes read takes 16B/(3.2GB/sec) = 5ns.

-- Carl



To: Ian Anderson who wrote (38773)3/23/2000 11:15:00 PM
From: Scumbria  Read Replies (1) | Respond to of 93625
 
Ian,

Sorry I meant CAS cycle not RAS cycle.
The reason RDRAM is faster on the second transfer is that it operates concurrently, transferring 128 bits of data and reading the second 128 bits from the DRAM core at the same time. Thus it is ready to start the second transfer almost as soon as the first one finishes. The CAS-to-dataavailable latency is hidden for second and subsequent bytes.


SDRAM does the same thing. You can pipeline accesses to consecutive addresses with no missed clocks. The core technology of DRDRAM and DDR is very similar, and both require the same CAS cycles. As a result, DDR and DRDRAM both provide the same bandwidth.

Scumbria