Articles: Mainboards
 

Bookmark and Share

(41) 
Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 ]

To prove my point I would like to offer you some performance results obtained at different FSB frequencies with the CPU multiplier set to 8x.

The benchmark results show clearly that the performance of the memory subsystem drops when we shift from 420MHz FSB to 425MHz FSB. The memory bandwidth gets about 9% lower and the latency increases by about 19%. This trick allows reaching high overclocking results in terms of the clock speed, but reduces the performance of the overclocked systems. In particular if we look at the SuperPi and Quake 4 results, we will see that by surpassing the 420MHz FSB barrier we lose from 120MHz to 240MHz of the processor speed (for Core 2 Duo).

This is exactly why we wouldn’t recommend using high FSB frequencies on ASUS Striker Extreme. It makes sense to use them for overclocking of slower processor models from the Core 2 Duo E6000 family with low default frequency multiplier, which cannot be overclocked in any other way. Moreover, this overclocking will only make sense if the FSB frequency is raised beyond 440-450MHz.

To prove this point let me compare the performance of Core 2 Duo processors overclocked on ASUS Striker Extreme up to 3.6GHz with different clock multipliers:

The table shows that as the bus frequency increases, so does the processor performance, however once we hit 450MHz FSB speed, the situation falls out of the picture. 8 x 450MHz is not good for overclocking at all; the results are awful in this case.

We didn’t want to end the overclocking discussion on a sad note like that, so in conclusion we would like to draw your attention to the fact that pseudo-asynchronous memory clocking of the nForce 680i SLI chipset allows overclocking DDR2 SDRAM absolutely independently of the FSB overclocking.

For the sake of experiment we decided to check how far we could push the frequency of our TWIN2X2048-8500C5D memory modules at 2.4V voltage setting. ASUS Striker Extreme showed a really great result here.

The memory was working stably at 587MHz. In other words, this mainboard proved a good platform not only for CPU overclocking, but also for DDR2 SDRAM overclocking. Even though the asynchronous memory clocking leads to relatively large increment for memory frequency adjustment.

 
Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 ]

Discussion

Comments currently: 41
Discussion started: 02/07/07 06:52:26 AM
Latest comment: 08/27/08 07:16:01 PM

View comments

Add your Comment