Chuckerp.
You might as well try a mild overclock, these boards, and those chips, seem quite happy with it so long as you're sensible.
Try 266Mhz FSB, that'll keep everything in sync and be sure to force your memory to DDR2/533 so it runs 1:1 with the FSB. I've a feeling that keeping a 1:1 ratio is quite important to stability on these...
That same advice goes to KeeleySam, going by that screenshot you have the memory set for 667Mhz, or maybe as "auto" in BIOS. I seem to get better performance having the memory at 1:1 as above, so for these kind of FSBs that means ensuring the board sees the memory at 533Mhz. You also should see lower latencies, which is always a good thing...
The reason I got 667Mhz memory is that I can then overclock the FSB and still keep a 1:1 ratio (by setting the RAM as 533Mhz in BIOS, as before) , which works a charm here. CPU-Z reports my memory at 290Mhz, same as my FSB and everything is happy. Memory timings here, 4-4-4-12-2T
I've not tried any benchmarks with memory performance since updating the BIOS to 1.80, but I seem to recall, with the 1.40 bios, the 533Mhz memory benchmarks (especially with forcing "enable) on a GTQRL or whatever its called pipelining in advanced memory settings) was better than the 667Mhz benchmark anyway.
Anyway, according to Asrock regarding my previous query about PCI/PCIE bus locking, AsRock are adamant that the bus is locked, and it seems that SiSoft Sandra, which I was using to guage my PCI/PCIE bus speed is incorrect in reporting this.
Groovy! Still happy here, nice, stable and quick, and not a single problem to report now except for that weird AGP 7600GS 32-bit 2D performance thing.
One thing I'm investigating, anyone here come across the Microsoft Dual Core hotfix?
http://support.microsoft.com/kb/896256
Not only does it appear to help with power states, it also appears to improve load balancing across both CPU cores. For example, before this hotfix WinRAR would appear to do most of its work on the one core, after the hotfix, it appears to use both more evenly. Another test, SuperPI single-thread benchmark would use 50% of total CPU time, with all that being 100% on CPU 0. After the hotfix, it still uses a total of 50% CPU, but with that being 25% over both cores.
Interesting.. I've yet to have the time to try some proper benchmarks/tests to see if this patch improves things wholesale, but has anyone tried this and has got any benches/experiences to report themselves?
Another thread (from notebook review, but this patch seems to improve things across all DC systems, not just notebooks...) hailing the benefits of installing this patch.
http://forum.notebookreview.com/showthread.php?t=60416
Interesting....
