TaPaKaH

I reckon you would have done a lot faster using GTX2 at 8-8-8 ish timings (+tighter subs, +gain from 2Gb modules)

am I right, highest GPU clock ever achieved ?

well, that's why you don't see the low end conroes being benched at 650+ fsb very often ... otherwise it would have been too simple

Have you had any luck with 6-6-5 or 6-5-5 ? GTR should have no issues running that.

what's the batch/PCB of this chip ?

new Gulftowns suck for SuperPi (even a 6.2GHz 06CPUtest chip can barely do 1M at that speed) so I guess people prefer to waste their money buying SB that max out 52/53x lol

welcome to the grinders' club

Most likely it's the FSB wall, Conroe chips have that between 450 and 600 on air depending on the chip. In most cases the FSB wall moves when going subzero - ~30MHz gain from air to dryice/singlestage and 50-60MHz gain from air to LN2

still no batch info ?

what's the batch/pcb code of this chip?

I have four 2500K here. Batch L051A869, SN are 2L047225A1607,1608,1609,1610. For 24/7 use they all clock exactly the same (4800 1.34v, 4900 1.38v linx) but max bootable multiplier is different on all four chips (using exact same settings): x52/x52/x54/x51 my personal view: batch for maximum overclock = witchhunt

why are there two P67A-UD7 ?

anything over 1/2 cards (depending on the benchmark) is massively cpu-bottlenecked, not much point in going subzero on graphics

what batch is that ?

unreleased 990X scores had to wait for months for points but unreleased Gigabyte board can have points straight away?

I think that the only thing we can separate between the two is the die shrink (55 vs 65 nm) which IMHO isn't enough to make two separate categories since overclocking is the same. Also, if you flash a BIOS from a 65nm card into a 55nm card, how will GPU-z detect it? P.S.Massman, if you're so excited about the 9800 cards, why don't you fix the P4-2.4 stuff?

I just came back from my Moscow trip .. here's a picture of the actual chip that was used for the run:

facepalm .. I don't think that anybody got lost in the current rev3 page layout. I'd rather spend time on optimising and fixing current things, not creating more things to go wrong and complaining about increased running costs.

is that 9-9-9 being misread or you have the new ultimate benching memory?

may be it has something to do with very benching-friendly memory you're using? (16gb = too much stress for imc)

...on Hypers

is that real vcore?

mine just stops at 940 stable, at 952 it passes memtest but fails on reboot and at 960 i get no post at all same story with multiple kits I tried

I think people tested this with PCIe riser ages ago, the reason why this never caught on is the extra latency caused by slot extention.

is that your old chip you've sent me not so long ago ?