Massman

Yeah, that's a mistake. Just pulled the result out of the competition .

Not sure. It's actually a feature of the highend board (UP7), so ... who knows. I don't think it will ever be on the entire line-up as there's an additional cost to adding a switch on a board.

Haha - well ... it's not as bad as Hicookie experiences

Hicookie did the 7.8G validation with it - previously 7.5G. Afaik, they also did some tuning on the digital PWM, but that requires a firmware flash and the tools for that are under NDA . Still looking for a way to get that stuff without violating the NDA, though. Hicookie mentioned he got about 50-100MHz with the LN2 mode. The rest comes down to slim XP instead of big W7. [hwbot=2329722]submission[/hwbot]

Germany going for the title this year?

That's the latest one? The UD7 is EOL, so I'm not sure if there will be a new BIOS that fixes the issue. Anyway, good to hear you have no longer issues

As everyone who has tested Trinity and Piledriver before knows, both are very hote when overclocking. In fact, the pot would be boiling even when idling at 7.5GHz and high voltage. Because both architectures are very much temperature dependant when it comes to overclocking, you need to reach as low as possible temperature to get the highest frequency. For the Z77X-UP7, Gigabyte already has an "LN2 mode". This feature is not really "LN2 mode", that's just the marketing name, but is just a simple P-state modification. When enabled, the CPU will be forced to the lowest P-state, which has the lowest CPU ratio. For Trinity and Piledriver, this is more useful than Ivy Bridge as the low P-state will allow for colder temperatures in idle. The way to use the LN2 mode is: - Step 1: go in OS and open CPU-Z - Step 2: set LN2 mode enabled - Step 3: set bclk / ratio you want by software - Step 4: wait for a couple of seconds to reach the lowest possible temperature - Step 5: disable LN2 mode => CPU will go to highest P-state - Step 6: important! wait for a couple of seconds in the highest p-state mode (*) - Step 7: hit F7 to make validation - Step 8: go back to step2 and repeat. (*) necessary because CPU-Z has 3-second rule for validation requirement. More info: http://hwbot.org/forum/showthread.php?t=30365 At the AMD event, Hicookie used LN2 mode to achieve higher clocks than before without the LN2 mode. He was so friendly to share how to do the LN2 mode on these boards ... yeey! It's actually quite simple: link the P-state soldering point to a switch and hook up the switch to gnd. Connect P-state point to gnd to enable LN2 mode, switch to the other side to disable. It's so easy! Example: LN2 Mode modification for A85X-UP4 LN2 Mode modification for 990FXA-UD3 (ps: maybe another pro-tip. To achieve the lowest possible temperature in idle, don't forget to also reduce the NBclk and lower the HT link and NB voltages ...) // end of sharing // //edit: by the way, they didn't want to share what pin to mod so it's not possible to copy the mod to other boards as well. But, just like with the Clarkdale CB mod, it's just a matter of reverse engineering. Just follow the P-state point to the socket pin and check what the pin is in the AMD reference datasheets. I don't have much time left this week, but I might have a look at it next week. If someone finds it ... please share so we can copy this mod to all boards .

Cool! What BIOS are you on, by the way?

Fix it to something not "extreme".

Sigh ... Well, congratulations anyway.

Fixed the bug .

Did you set loadline to 'extreme'? Try 'very high'.

Haha ... why not do it in public?

Oh, okay. Well ... that error makes it "hard" to bench in XP then ...

Hey man - HWBOT would definitely host that benchmark

Yes, allowed. By the way, did you update to the latest Systeminfo?

Hey man - that's awesome! Really like it!

Yes - posting in the thread at SF3D.fi is mandatory.

Sure! Be creative ! Yes, allowed

It will be interesting to see how the community feels about the new 3DMark. @Jarnis: since you are releasing new DX9, DX10 and DX11 benchmarks, does that mean you will discontinue the existing DX9/10/11 benchmarks?

SF3D is in charge of this competition - feel free to head over to SF3D.fi forums too to check up on the wallpaper

Because: - processes mature - you having a bad chip from AMD doesn't disprove them pre-scanning

No - AMD's been seeding binned chips. I think the media samples are, in general, pre-scanned too.

I might have some setup later this week to test. Don't expect too much from it though: bioses are no longer checked for compatibility with Hyper. Besides, on FM2 you're better off with high frequency (DDR3-2600+) than tight timings .

Just a quick heads up for everyone who's playing with the UD3 and Piledriver. GBT recently released the UEFI BIOS for AMD on the 990FX series and I had the opportunity to quickly get a hands-on testing. As you can imagine, going from regular to UEFI BIOS comes with a whole lot of issues and problems and the first release didn't do so well for overclocking. I've quickly tested this combo for overclocking capability and the latest FBd BIOS fixes most of the issues I had before. The board's playing quite nicely HTT-wise - better than the UD7 in fact. There's still some stuff missing (DDR3-2400 divider for example) and some bugs (HTT overclocking when XMP enabled might not always work), but in general it's now working fine for overclocking . If you're on FA or FBa BIOS ... definitely flash . Didn't have the time to do much testing, so just ran a quick 3DMark11 Physics test on air cooling. Bios download: http://downloads.hwbot.org/downloads/BIOS/990FX/9FXAUD33.FBd Have fun!