I.nfraR.ed

Ok, looks bad on stock - 800/1130. Will see if it scales.

Does not need vmem mod? I see chips on mine are Qimonda.

Excellent Just got one Asus myself, will try it soon.

Access is denied for the XTU background on the stage page: http://hwbot.org/competition/gbt_targetoc/stage/1766_xtu:_159_marks/ Link leads to http://news.hwbot.org/Competitions/1008_GigabyteTargetOc/target_paticipants_1920x1080_1.jpg

Not sure if it's locked, more like on "auto". 990FX-UD3 rev3 has tWCL control, I think. Last time I tried, it was way more inefficient though. I could try it tomorrow. tWCL doesn't seem to have a big impact anyway.

Just what I expected. The power of Gigabyte 970A-UD3. It lacks tWCL in bios (board sets it at 8) and can't run 1T at 2600, but still faster. It is almost as fast as Crosshair III Formula with K10 and it seems not bad with K15 too. Also a very very tricky board. Shaved one second off.

I could do the same mem clocks on air. Cold is required for higher mem and cpu-nb frequency. CH5 is not very efficient though.

Bullant, ram was on air, yes. Not a problem for the PIs, they should be good for at least 2666 tight on Intel. I put the cpu on SS to clock the CPU-NB higher, because 2900 on air was not enough. The problem on AMD is that at high frequency it starts to drop a channel or if I manage to boot in dual channel it's not stable. To be honest, I haven't played much 32M on Vishera, so can't tell much details. Not sure if board or cpu related. I have no Samsungs to try though and BBSE was a no go. Also tried Hypers for 2400 8-8-8-24 tWCL6, but didn't manage a full run. From all the mems I have, Pi's were the best. It was painfully slow before Stilt's BDC and I really preferred 32M on K10. Now when it's a little faster and there's some activity ion the thread, it is time to play . Wish I see you both here, but I know it is too sloooow for you...waiting 15min every time BTW, there's one of my runs with ram on LN2 http://hwbot.org/submission/2426500 but not sure this was the optimum. I didn't like that board, it had some serious problems with bios..at least for me. Apart from that, we had some other runs with ram on LN2, but on a different board 1444 7-11-7 seems more like it... http://hwbot.org/submission/2463549_i.nfrar.ed_aquamark_radeon_hd_7970_363212_marks or 1330 6-10-6, but he already sold that cpu http://hwbot.org/submission/2465491_more_aquamark_radeon_hd_7970_314300_marks As far as I remember 1400 6-10-6 was also possible with the first cpu, but don't remember subtimings. Definitely going to try with that board again, although for that kind of frequencies CR=2T is required.

Nice! Not sure if I can handle 4 pots myself...except the case when most of them go full-pot

Yeah, it was very hard..took me 1 sec to figure it out.

No, the wallpaper is exactly for XTU. <- Ah, yes..got what you mean. You can find the XTU wallpaper on the XTU stage page. Wallpaper for next stage will be "revealed" after that 2 XTU hours submission window, or during that time. Wallpaper for 3rd stage will be revealed during submission time for 2nd stage and so on...you get the idea. At least that's what I understood. Not that I don't have all wallpapers already To staff: Better change them (guess I'm not the only one that figured it out) and restrict access to new files, otherwise the whole idea is lost.

Other reports I saw were using 7x0G chipset too, not sure if possible on 790FX and my Sempron doesn't even unlock to try.

Yes, as in other rankings, the first person to submit wins if there are 2 or more submissions with the same score. I find it a little bit annoying, but that's how the rules are and we play by the rules. Rules are defined by those who give the money

My goal was 2666, but not gonna happen. Board is loosing 2nd channel or very unstable at over 2600. Maybe it's the IMC, or it needs 2T, dunno. Also tried BBSE - no go. I may try again on a Gigabyte board without UEFI.

14m 14.297s | G.Skill Pi 2000C6 @ 1300 8-12-8-26 1T tWCL6 90ns http://hwbot.org/submission/2645764_i.nfrar.ed_superpi___32m_fx_9370_14min_14sec_297ms CPU frequency is ~4998MHz, like in this screenshot: http://i.imgur.com/qCwtv7D.jpg. Just a cpuz glitch at the moment of taking the screenshot. newlife, you did an excellent job! Took me some time to beat your score.

which means Sandy and newer Intels. Of course we can compete in other stages and I will participate anyway.

Why XTU? This leaves AMD users and Intel guys with older platforms out of the lucky draw. It makes more sense giving a chance to people that still don't have current-gen Intel platforms to win a new generation intel mobo.

Sorry, but this is bugged. Just look at Fur, Fluid and Raymarch.

Yes, I have used this method for 7970 Lightning, since I never managed to get the ABX (if you open some of my 7970 submissions you can notice ArtMoney ). So it really works and see no reason it doesn't for other cards if there's no other limitation and the voltage is programmable with software in real time. My advice is try to do it with clocks slider first to get familiar with the process, although it's not hard at all. When you find the correct address to modify, just name it e.g "VGPU" and save it. Next time just load the saved address and only change the value.

Not sure what is the max without mods, I had it modded from some of the previous competitions. I think it was 272 without mods and 284 with mods Maybe I should try older drivers, that might be the problem.

Well, might be it just needs some more caps or to replace the existing ones, but I had that thing on 3 cards. And I remember the same was happening on GT3 when benched 3DM03: http://hwbot.org/submission/2310105_i.nfrar.ed_3dmark03_geforce_3_2134_marks And 40MHz is a really big difference, maybe your card is just that great or I'm missing something. Did you try another card on cold? All I have tried on air usually get into the 270-280 range. Looking at the charts, only poparamiro managed 290 out of it and it's a non-reference design. It even has a power connector. Maybe that's the card to look for.

1.7-1.75, higher was making it worse, but with epower it could handle 1.8+ I think.

Yes, I put boost and normal clocks and voltages to the same level, but it was freaking out with higher voltage, although temps were not high. It was the same on chilled water. And I had to drop the mem clocks to 1500, when the gpus were clocked high.

Yes, out of 5 different models GF3's I have only one that does above 600. Maybe a specific chips to look for, but I don't have enough knowledge about these cards. Speaking about gpus though, mine could run 330-340 nature on LN2, but other tests were limited to 290-ish. Any higher and the screen goes into "color blocks on black screen" mode.

Respect to Greeks and congrats. I actually thought we would come a little closer. I'm disappointed from myself, because I could do better in some (AMD) stages.