I.nfraR.ed

Yeah, I think this is max on air/water for me and stock volts on the VGA. It's 22C here, but still "colder" than 30C before

He really pushed it to the max. 0.99V is pretty insane! [ATTACH]2412[/ATTACH] Also going through bios options up and down like a pro, with settings on auto .

Why last few squares on the right column and the whole left column are black/not rendered? I think it's bugged, sorry. There's no magic tweak to compensate 400MHz difference in CPU frequency. Never seen this, curious that it gave you a score at all.

Noticed the same while browsing through some 3000+ submissions. My sub is not displayed in the rankings. In fact none of my 3000+ submissions are there. http://hwbot.org/hardware/processor/athlon_64_3000_venice_s939/ [ATTACH]2395[/ATTACH] Also the search engine does not find anything when I search for SuperPi 1M/I.nfraR.ed/3000+ Venice s.939 for benchmark/user/cpu model respectively. Points in the category are not skewed/recalculated though. I still get the points, just the entry is missing in the table and slower results are shifted 1 place up. Athlon64 6000+ 1M is missing in the general K8 rankings too (and a lot of other entries from different users) when switching from 3000+ rankings to K8. It works when viewed from 6000+ rankings. [ATTACH]2398[/ATTACH] [ATTACH]2399[/ATTACH]

Wonderful! Was waiting for this result, but show us the maximum

Great one. As always . Keep up the good work.

I blame Massman for this Despite this is one of the cheapest new systems you can buy, it still requires a high-end nvidia card to be competitive in the competition. Wish it was the iGPU only... Anyway, good job!

Damn, I can't fight with you guys without 780(Ti)

Same happens for me if I use "capture.exe". Noticed that during MOA qualifiers.

Both seem fine, but are on the high side. For the same money you get this: http://www.ebay.co.uk/itm/310672415584?_trksid=p2059210.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT

Here it is: 5V of a molex connector. Not easy to test something else at the moment, but they are good enough.

I think it was this one: http://www.ebay.co.uk/itm/201085740430 Problem for me was that many sellers don't send to Bulgaria, so I had to buy the cheapest that had shipping to my country. Let me check the accuracy...

No, throw these in the trash. They have 1% accuracy, which means 1 digit after the decimal sign. You have a potentiometer on the back side to adjust the reading according to DMM, but from 1.16 to 1.25 it will show 1.2, from 1.26 to 1.35 it will show 1.3 and so on. I got fooled from a seller's picture and description, but fortunately got a refund afterwards, because the info was wrong. They have corrected it since then. Look for 0.1% accuracy or better. Otherwise you can connect red line to any 3 to 28V source.

Since I gave up on class B with my weak cpu, I had nothing to do but watch the show in EMEA A and class B. I believe the Iranian guys didn't do that on purpose, but in the end both results are not valid in terms of defined rules. Not sure about UH, but it doesn't matter if it as a bug or done on purpose - still not a valid score. Well, it matters to some extent, because if it was done on purpose the user deserves some kind of punishment, but in both cases the result needs to be removed. For 3DMark 03, even if it is completely valid, the missing validation link renders it invalid for the competition. So it doesn't matter if you provide all the proof in the world. Rules are rules. Now, if you didn't count on sandbagging and had uploaded some lower, but valid scores earlier, you could still get to the finals when your top scores get removed. And if these placeholder scores were not valid, people would report them earlier and you would have had time to correct these flaws. You got screwed by the sandbagging. Just my 5c. About cyclone: I didn't pay attention to the screenshot and now it's gone, but it's really disappointing to see such thing from a respected overclocker. Don't know what he was thinking... We all know what is he capable of. When you don't have a good hardware to compete, just give up and save your good reputation. That's a shame if true . I still would like to see that screenshot though.

Oh, man. I don't know how you got that performance, but I'm 10 min behind and tried everything I know. Wazza doesn't seem to help. Good job.

I didn't see a difference, just like with other chips. Haven't tried some really low memory clocks (e.g. 1333), but single or dual channel, 2200 or 2666 - doesn't matter.

RAM doesn't matter in R15.

No, there are only 2 benchmarks for 2 days left. Based on my experience - 20L of LN2 would be enough for 10-12 hours straight benching (maybe more). It should be enough to max out your system for the given benchmarks - 32M and Cinebench R15. You can leave other benchmarks for the next days.

Strong cpu! I bet there's some more in it. MOA ticket is waiting for you. Good luck in the rest of the comp.

Imagine if I had 6.1+ chip Spent a lot of time, but first core is weaker than 2nd and couldn't run higher regardless vcore/vccin/other voltages. Maybe next time, who knows. CB was good: -135.

I don't even have a card yet (still traveling from UK), but it is pointless with that cpu - bad IMC, cores are not strong enough, basically 200MHz behind from current top results and final bombs will be even higher. Good luck to everyone, looking forward to see the final scores.

I've already posted my best scores and I'm out

Wiz, is that for CB11.5 only or in general? You told me before, but I don't remember. Because it does way higher frequency for other benchmarks, while for CB11.5 higher vcore makes things worse. For example it crashes on the same 6480 with 1.8V. It also feels very stable - I can do whatever I want in OS - unpack zip files, run CB15 all day long, but it simply crashes in CB11.5. Spent a lot of time to get that 6480 run. HT Ref is not a problem, because I can boot even at 500 and most of my scores are done with HT Ref set in bios and direct boot to OS. For example that 6940 validation was done with direct boot at 496 HT Ref. Couldn't make 497x14 no matter what - tried direct 497 boot or with setfsb in os - black screeb and restart. I set multi with K10stat, ofcourse. For 7GHz I would need 500x14 and that was not possible. Maybe with 1.95+ V on the cpu...I don't know.

Decided to try one more time, smaller potato this time. Same batch L418C276 (no other available here since the beginning ) Finally 5GHz 32M 1.44V VID on auto according to MPOWER is 1.120. If it has the same -135 CB should be better than the previous one. PS: G3258

Heavy guns, others are crying, lol. Great one!