Which of these two memory set ups is faster ?
- Thread starter silvercut
- Start date
Two main things involved with memory are
Latency is harder to explain but the lower it is the better, in an ideal world you want higher bandwidth and lower latency, for the average user lower latency would be more noticable than higher bandwidth with an average load, low latency = low latency system and makes the computer very *snappy* and responsive!
In your two examples the DDR2-1066 is superior in everyway way to DDR2-800 with the exception maybe of cost! ££
DDR-800 CAS 4 10.0 ns Dual (128-bit) 6,400 MB/s
DDR-1066 CAS 5 9.4 ns Dual (128-bit) 8,528 MB/s
- Bandwidth
- Latency
Latency is harder to explain but the lower it is the better, in an ideal world you want higher bandwidth and lower latency, for the average user lower latency would be more noticable than higher bandwidth with an average load, low latency = low latency system and makes the computer very *snappy* and responsive!
In your two examples the DDR2-1066 is superior in everyway way to DDR2-800 with the exception maybe of cost! ££
DDR-800 CAS 4 10.0 ns Dual (128-bit) 6,400 MB/s
DDR-1066 CAS 5 9.4 ns Dual (128-bit) 8,528 MB/s
), and the Low latency is very snappy, so +1 to that Thanks for the responses, big wayne - I think it's pretty easy to decide that option two is better is better if memory is running 1:1 in both scenarios, i'm just wondering did your calculation take into account that the second option is running at 3:4 ?
Last edited:
The only thing that matters is Memory Frequency and Timings . . .
The [1:1] memory multiplier is IMO the worst LGA775 config possible and the [2:1] memory multiplier the best . . . sadly for reasons I don't understand it doesn't seem possible to get DDR2 memory frequency to keep up with the System Bus once you start cranking the FSB frequency although this doesn't stop people lusting over high FSB's, I guess if ones Data-Set can be contained within the CPU's internal superfast memory (cache) it doesn't matter but in the Real-World most of the benefits of high FSB are lost in a mismatched set-up once the data is transmitted to system memory which just slows things down a lot the slower the Memory is running . . .
If you just try and keep your effective memory frequency running as fast as your effective FSB (system bus) your doing well!
266MHz-FSB
1066MHz-System Bus
[2:1]
DDR2-1066
333MHz-FSB
1333MHz-System Bus
[2:1]
DDR2-1333
400MHz-FSB
1600MHz-System Bus
[2:1]
DDR2-1600
500MHz-FSB
2000MHz-System Bus
[2:1]
DDR2-2000
The [1:1] memory multiplier is IMO the worst LGA775 config possible and the [2:1] memory multiplier the best . . . sadly for reasons I don't understand it doesn't seem possible to get DDR2 memory frequency to keep up with the System Bus once you start cranking the FSB frequency although this doesn't stop people lusting over high FSB's, I guess if ones Data-Set can be contained within the CPU's internal superfast memory (cache) it doesn't matter but in the Real-World most of the benefits of high FSB are lost in a mismatched set-up once the data is transmitted to system memory which just slows things down a lot the slower the Memory is running . . .
If you just try and keep your effective memory frequency running as fast as your effective FSB (system bus) your doing well!
266MHz-FSB
1066MHz-System Bus
[2:1]
DDR2-1066
333MHz-FSB
1333MHz-System Bus
[2:1]
DDR2-1333
400MHz-FSB
1600MHz-System Bus
[2:1]
DDR2-1600
500MHz-FSB
2000MHz-System Bus
[2:1]
DDR2-2000