I.nfraR.ed

Wish I had better efficiency on that board, though... otherwise the card is great. Have even 446MHz gpu with the bugged driver.

Congrats for the DDR record, I'll leave you on #1 for couple of days

Challenge accepted.

That's good news. It's pretty hard to make a screen with all 4 cpu-z instances and wait for both cpuz-cpu and cpuz-mem to update after fsb change. You saw how unstable it is...

Is one cpu-z instance with mem tab enough? If so, it will make my life easier.

Not sure, impossible to control and read correct temp with tinfoil, but I belive something between 0 and -10, no more. If too cold, there's a blue screen - restart sequence. Done literaly for 10 min, have to try harder next time. Your cpu seems fine, it's just the board that is tuned for BH-5/UTT and higher capacity modules (1GB). My Expert doesn't clock well even BH-5.

Ok, back in my possession. Will try for higher freq soon.

I agree with slamms. Me and other small fish in hwbot ocean that work hard for their money and then buy everything from the store or ebay at regular prices, when see such resuts and behavior from top and great overclockers, we simply get demotivated. In the end, we don't feel hwbot as a fair playing field. At least me... can't speak for the others. And I don't care about PCMark, so that's not a valid forgiving reason for me .

Imo, it's either allow it or disallow it for everyone, including the old scores. Yeah, I know some of the top scores will fall, but according to the rules back then, for me it was also not legit, because it's causing the benchmark to not render some parts of GT2 and GT3, which gives a big boost (~300p). If it gets forbidden now and old scores remain, how we are supposed to break them? It's like tolerating a "cheat" driver for some and forbidding it for others. If I'm not mistaken the bug is with GF4 only, but I was not able to install it on GF3 to verify (black screen and monitor going "out of range" before desktop) and my FX5900XT is now with LN2 clocks, because I forgot to change the bios >.<. GF2 can only run GT1, which is not being rendered incorrectly, so no point in testing that. I also think the answer is rather late. I have a single run with that driver, but won't submit in the competition if noone else does. I asked about it more than 1 month ago and got no official answer, except those from Knut.

I have higher "validation only" This stick is crazy and actually scaled a little with cold, while my Adata (415MHz max) didn't.

Yeah, very hard to find these boards here as well.

I.nfraR.ed_absolutely_reliable, that's what I am

Once I did 3812.9 in dual channel 7-7-7, so that's mediocre at best...

Files that are already validated will be "invalidated" or this is only for future submissions? I'm sure the valids are ok, but I forgot to add "_reliable" after my nickname.

It seems that it's still working... This one is an older Windows installation and didn't notice that I'm using cpuz 1.60 on it. Didn't pay attention when it validated first time. Christian could say if it's ok to use as long as it validates. Otherwise I can rerun, but it took me 3-4 hours to get there.

Yeah, this stick is crazy. The other one is no way near this and it's actually worse than my best Adata Vitesta.

Nice start Good luck in the future LN2 sessions.

Benching alone 3DMark 03 for the TeamCup

You guys are killing me. This is nothing compared to my absolutely awful performance. I'm at much higher clocks (365/262) and got only 442.

It's obvious, no one wanna play in easy games ;p

Don't know what is that SPECTRUM UTAMA board, but seems to be bugged. Believe it or not, this cpu has a locked multiplier which can't be increased more than 16.5 (which is the stock multi) and that frequency is impossible on water!

Those Gainward cards look good.

That board can run 1200+ memory . However lacks AGP freq control. My performance is bad and I had very little LN2 left. May rerun it later.

This CPU has a locked multiplier and can be lowered only (from stock). No way you can set 31x multi. If it's real then you should be able to show us 16.5x300 = 4950MHz on water That must be like a walk in the park. 16.5 is the default multi of this CPU.

Nah, for me overclocking is not pure frequency and being the first to submit, it's optimizing every aspect of the system that matters in the given benchmark. In other words: squeezing the last ounce of performance from the system. Just my understanding. I'm not complaining (much) for those "lost" 1.6p, I just feel the proper ranking is in reverse order by cpu frequency (it's a cpu benchmark after all, although other factors play role as well). I can easily beat it with better settings. CPU-Z validations are okay to rank by submission time. Only frequency matters there. Seems to be this way in the bot, but don't see how that matters for pi, wprime, ucbench, pifast. What is more valuable - being fast and in a hurry, picking a better cpu and running at higher frequency or submitting later with worse cpu, but same score (spent more time on tweaking)? I know the answer for me, but others might think different. That;'s just a hypothetical question, not saying my score is uber efficient. Actually I feel it's quite the opposite . Not applied here, scores are ranked by submission time. Bottom line is I'm behind Massman, when we all know he sucks