Massman

2x CPU indeed.

If the Vreg chip is different, you'll need to look up the datasheet of that IC.

Overclocking doesn't kill hardware, people do. Leeghoofd has blackmailed me into doing another round of PCMark on the Crosshair Extreme. Soo little time ...

Nothing so far . If they use the same Vreg and the stock resistance is the same - yes.

Hyperthreading is a feature, so of course it's allowed.

No. Yes.

They all do from 1.0B31 forward.

Bios 1.4b6 isn't the easiest one it seems. Took me quite some time to get it running ... Result is quite okay without copy-waza and for 3x1GB

Very weird. Post SPD file here, maybe MSI R&D can have a look at it.

do you have the bad write performance at any clock frequency? even below DDR3-2000? Can someone dump the STT spd here?

No retail 980X in the house here. I don't really see how that could be the issue, though.

Hypers worked perfectly here. One thing I always do with the MSI is allow the board to set the subtimings as it wishes to do. So, whenever I start testing from scratch, I do this: - clear cmos - boot into bios - F10 without setting anything - reboot into bios - disable all features I don't need (not yet adjusting in cell menu) - F10 (save and quit) - reboot into bios and start adjusting the settings No idea if that would help?

I'm surprised 99 and 00 don't scale much more with high-end cpu. Especially 2k! 2001, our most beloved benchmark, is the only one that scales so dramatically with C2D

Running now. Using the same memory as on MOA EU: [hwbot=1022831]submission[/hwbot] Tested several boards? Tested several CPUs? What kind of error do you have? For 4000/2000/4000 I use 1.35V VTT and 1.7 Vdimm. For 4300/1975/4300 I need 1.45V VTT. Did you clear the bios or load bios profiles?

# submissions with 980X: 12329 # submissions with 32nm i5: 1343+818+262+453+560 = 3436 Overall figures, yes, but I'm sure you'll see something similar BM-specific. //edit: also, these numbers don't represent the amount of chips that have been tested before the good ones were found.

Problem with serial numbers is that it doesn't solve hardware sharing, it just forces people to buy (or find) more than one sample. Eg: a team has 10 samples of GTX480 and sends around just one 1.3G sample for LN2 results. You will have 10 serials, but it's still hardware sharing. As long as you know someone who works in a HW store you can also use serials from sold hardware pieces.

The 980X was an example, we will not take away points from scores posted months ago. i5 ES = i5 retail silicon. Also, I've seen retails that can match ES. Consistency is relative to the amount of tested pieces ... if someone tests 100 ES and find one or two golden ones it's normal you think retail is not catching up if only 1 out of 75 is golden

Hey guys and gals, Just a heads-up for the upcoming Sandy Bridge release. In Rev4, we'll no longer accept non-retail revisions to participate in the end-user leagues (in the rev4 document named 'XOL'). So, there's no need to hunt early killer batches (if there actually are any) if you're not going to participate in the MFC league. In terms of 980X, this would mean: - A0 only in MFC league - B1 ES allowed in all leagues.

What kind of crashing do you get? Freeze up or error messages?

This was the last run before I said it goodnight

Yeap, GTX2 Hypers. I could try some PSC (3x1GB) if needed. Don't have that much memory lying around here. Ehr, how? Just ... set settings and run. 12K sounds like B2B was enabled.

That's a thoroughly evil post there, K404!

It's not the most appealing solution in terms of marketing exposure. Europe consists of a lot of small countries, each having their own language and press. To maximize exposure in Europe, you need to have each country represented in the European final. I think for the European continent, a country-based pre-selection is the best choice. The more nationalities in a regional final, the better .

It's possible that the team for the UK was pre-selected, I'm not sure about that. This kind of global events is quite a hassle to set up. Not only in terms of cost, but also in terms of organisation. So, mostly the HQ lets local offices know there will be a live final somewhere and then the local offices can decide whether they want to join. Sometimes the local office doesn't want to join at all, resulting in no competitors from a certain country. Sometimes they want to join, but don't have ANY resources whatsoever to set up a competition, in which case you might see a pre-selected team join the competition. Sometimes the local office has the time/resources to set up a small competition and there's pre-selections. Obviously we're pushing as much as possible in the direction of pre-selections, but in the end there's always the financial problem. The best way to resolve that is work together with the local office to set up something. I know many local offices where the marketing people don't really understand how extreme overclocking works, so sometimes they need guidance in how to set up a cost-effective competition

Last year, there were pre-selections for the EU final. In fact, I had to go through an online qualifier and a live final before going to the EU final