I've been out of the loop for some time, but I'm looking to upgrade sometime in the next 3 months when Penryn hopefully makes quad core processors a bit cheaper.
However, when trying to decide if I should spend the extra money on 1333 or even 1600 MHz DDR3 memory, i came to think about the fact that (if I remember correctly) the FSB is quite limiting in terms of bandwidth. When using a 1333 MHz FSB cpu/MB along with 1333 MHz memory you will ideally have twice the bandwidth from the memory to the NB compared to the bandwidth from the NB to the cpu due to the FSB being 64 bits and the momory interface being 2x64 bits. I am aware of the fact that the theoretical bandwidt can never be achieved due to latency issues. However, when comparing fast DDR3 or DDR2 memory at 1066 MHz to DDR 3 memory at 1333 MHz, the latency seems to be about the same, so why would 1333 MHz or even 1600 MHz be necessary except for their potential lower latency (only potential, as the memory makers could just as well manufacture very low latency 1066 MHz RAM)?
Does the trend with matching the FSB and memory frequencies have anything to do with better performance when running the two synchronous like on old nforce 2 chipsets, or is it simply a way to spend money without getting anything in return?
Thanks in advance
EDIT: And what about populating more than two memory slots? Will this still force you to run at slower settings - e.g. 2T instead of 1T?
However, when trying to decide if I should spend the extra money on 1333 or even 1600 MHz DDR3 memory, i came to think about the fact that (if I remember correctly) the FSB is quite limiting in terms of bandwidth. When using a 1333 MHz FSB cpu/MB along with 1333 MHz memory you will ideally have twice the bandwidth from the memory to the NB compared to the bandwidth from the NB to the cpu due to the FSB being 64 bits and the momory interface being 2x64 bits. I am aware of the fact that the theoretical bandwidt can never be achieved due to latency issues. However, when comparing fast DDR3 or DDR2 memory at 1066 MHz to DDR 3 memory at 1333 MHz, the latency seems to be about the same, so why would 1333 MHz or even 1600 MHz be necessary except for their potential lower latency (only potential, as the memory makers could just as well manufacture very low latency 1066 MHz RAM)?
Does the trend with matching the FSB and memory frequencies have anything to do with better performance when running the two synchronous like on old nforce 2 chipsets, or is it simply a way to spend money without getting anything in return?
Thanks in advance
EDIT: And what about populating more than two memory slots? Will this still force you to run at slower settings - e.g. 2T instead of 1T?
Facebook
Twitter