ASUS Maximus II Formula - new P45 king?

[CryptiK]

Nice work. Its crazy though having to push our memory so hard (600+Mhz) to break 10k on theses intel chipsets. I miss hitting 11k+ with nforce @ 400-450mhz. I never even tried anything any higher than link sync. I wonder what 600+mhz would have done. I think i'll order another SIIF and see. I miss SLI and playing everest.

Thanks. Yes, the P45 doesn't produce as high a numbers in everest as chipsets like the 680i, but the superior Everest read & latency performance of the nvidia chipsets does not translate to faster SPI32M times. Clock for clock in SPI32M, the P45 is faster, so is the P35/X38/X48.

Looks like the P45 more than held its own in the memory performance comparison with the X48. Interesting results, and not what I would have expected given what people have previously reported.

Nice!

Thanks mate, I had to really push the board hard to do that, the boards MCH was as tight as it could go, the only way I could get better performance is to increase vdimm and tighten the ram timings, oh and if Asus had made pull-ins that actually work instead of being useless, perhaps I could improve on it a little.

However the X48 was doing that with relative ease, and had more in it for sure, eva just didn't have the time to fully optimize.

now thats impressive! .. thanks for all your testing and information about that!
... but 1.45vNB isn't that good
Big differences when SRC/DRT is disabled?

Thanks. Yes, SRC/DRT makes a very noticeable improvement, I would really recommend enabling them. Also Clock Twister on 'stronger' drops the latency significantly and gives a small bandwidth boost. 1.45vNB is pushing hard, I had it up to 1.49v with other testing and the NB was getting really hot.

Have any of you guys played with the NB GTL Ref ??
I added +20mV to mine today and it has allowed me to lower my VFSB and VNB by a HUGE amount, with a Q9650.
If i put it back to Auto, with lower values for those two settings, the machine will BSOD.
Give it a try and post results here

I have had a brief play with it, and got the best results leaving it on auto, much to my displeasure. However now I using different ram/chipset settings I will have another go.

Well guys, I've just about had it with this board...

I cannot prime blend for more than 30 minutes without an error.

I have tried 2 different sets of G.Skill 2x2GB F2-8500CL5D-4GBPK PC8500
As well as a set of 2x1 GB A-Data PC6400 that I know performs at 1066.
All 3 sets passed memtest but all fail at the 30-35 minute mark on prime blend.

Even stock clocks, cpu and ddr2....same story. This exact rig was 24hr prime stable on a 780i board @ 474x9.5 & 1:1

Unless you have a dodgey board, perhaps your rams are having compatibility problems. Do you have any others, or can you even borrow a kit from a mate and try them? It should not be unstable at 453 FSB 1:1, I can get my board stable at 534 FSB 1:1 and higher, so the 'average board' can do that no problem.

Try different ram, if still a no go, perhaps swap the board.