|
|
Question : Problem: What is the significance of supply voltage in RAM?
|
|
I'm upgrading my AMD 3200+ system with RAM. The ram it can only take is PC3200 or less / DDR 400mhz
I've noticed that there are several RAM types meeting the above standard, however, they have different voltage amounts.
http://www.play.com/PC/PCs/4-/3432614/Kingston-1GB-184-Pin-DDR-400MHz-PC3200-Memory-Module-For-Dell-Optiplex-GX270-SX270-Series/Product.html
This Kingston one for example offer 2.6v and CAS latency of 3
http://www.play.com/PC/PCs/4-/931233/Computer-Memory-1GB-PC3200-DDR-400Mhz-184pin-DIMM/Product.html This Hynix one costing much less has a higher voltage of 3.3v and a CAS of 2.5 (im already using one of these 1gb sticks).
Aside from needing to match two pairs the same spec to get dual channel advantage, does a higher voltage RAM stick equate to higher speeds, and the same for lower CAS timing?
Also why is it that a 1gb stick of DDR2 memory costs nearly one third if not half the cost of my above memory. I thought all latest RAM modules would be more expensive (leaving aside the economics of it i.e. more output of the newer types).
Thanks.
|
Answer : Problem: What is the significance of supply voltage in RAM?
|
|
"... CAS LATENCY -- that is how fast the ram needs to be refreshed to stay alive ..." ==> NO!! While both CAS latency and refresh rate are important parameters, the CAS latency is NOT the refresh rate.
CAS latency is the time between when a column address signal ("strobe") is placed on the memory bus and the memory subsystem is ready with the data from that strobe. Clearly a smaller number is better (i.e. a bit faster) => but it's not the only delay associated with memory access. The Row addressing, the RAS-CAS delay requirement, and the precharge delays all factor into access time as well. The CAS latency is probably the most important of these delays, because memory is often accessed sequentially ... so many cells from the same row are accessed. In these cases, the CAS latency determines how much time between those sequential accesses.
But, for example, if a module is specified as 2.5-3-3-5 timing, then the initial access to a new memory address will take 2.5+3+3+5 clock cycles ... i.e. 13.5 clock cycles plus the command rate delay (which is usually not specified). So with a 400MHz clock that's 33.75ns. If the same memory had a CAS latency of 3, it would then be 35ns => not much difference. But if there were 4 bytes being read from that row (not uncommon), the next 3 would be read in either 6.25ns each (CAS latency 2.5) or 7.5ns each (CAS latency 3) => a more notable difference in % terms.
But also remember that any "cache hit" in the CPU doesn't even reference your memory ... it happens at CPU speeds (basically one clock cycle). So yes, CAS latency impacts performance ... but except for memory benchmarks the real-world difference is not all that large.
But I've digressed ... Bottom Line: The voltage doesn't matter; the timing does. Just buy the modules with the best timing parameters and you'll be fine :-)
|
|
|
|