Friday, December 30th 2011
Gigabyte X79-UD3 and Core i7-3930K Set Records with New F7 BIOS
It's been a tough week or so for Gigabyte marketing. It had to admit that quite a few people had burned their motherboards, to which it offered free replacements, and isolated the problem to faulty firmware. There has been talk that this firmware cripples overclocking by throttling CPU clock speed under extreme stress. Gigabyte set out to do some myth-busting. Renowned overclocker and Gigabyte PR guy HiCookie set up a test-bench using Core i7-3930K, an "infamous" X79-UD3 motherboard running the latest version F7 BIOS, and a typical extreme-cooling bench.
HiCookie achieved 5643.2 MHz clock speed, using a base clock of 99 MHz, 57.0x multiplier, and CPU voltage of 1.584V. The rest of the test-bench consisted of Kingston HyperX Genesis memory, and Corsair AX1200W PSU. To demonstrate that this isn't a hit-and-run feat, HiCookie put the overclocked bench through Super Pi 1M and 32M and PiFast benchmarks, with record-setting scores. The CPU-Z validation can be found here. Details of the HWBot record scores can be accessed here.
A video of the benchmarks follows.
Source:
Tom's Hardware
HiCookie achieved 5643.2 MHz clock speed, using a base clock of 99 MHz, 57.0x multiplier, and CPU voltage of 1.584V. The rest of the test-bench consisted of Kingston HyperX Genesis memory, and Corsair AX1200W PSU. To demonstrate that this isn't a hit-and-run feat, HiCookie put the overclocked bench through Super Pi 1M and 32M and PiFast benchmarks, with record-setting scores. The CPU-Z validation can be found here. Details of the HWBot record scores can be accessed here.
16 Comments on Gigabyte X79-UD3 and Core i7-3930K Set Records with New F7 BIOS
Nice try Gigabyte, Im not buying it
It would be great if GB released a hybrid WC/air heatsink like they had a little way back, might just be what is needed.
sure yay its a great OC, but it still pails in comparison of the 'plug and play' style maturity and stability that Asrock and Asus UEFI patsburg boards have... I could go on for hours but cbf. so ill stop here
Kingpin works for EVGA
Hicookie works for GIGABYTE
You would believe in any of the rest, but you wouldn't believe in the GB one? Seriously?
BTW there is no use in having a fan point over the VRM B/C LN2 cools the VRm from the inside out, and the circumstances aren't irregular, go over to HWBot and take a look at some of the resources these non-paid but heavily sponsored OC teams and users have.
But mainly, this doesn't address the throttling issue, it just kind of sidesteps it. Yes, he got records for clock speed and SuperPi, but was the CPU still throttling? What happens with all the cores enabled? Does the CPU throttle itself to the point that the overclock settings you set in the BIOS are pointless? That is what people are concerned about. It isn't about being able to set the settings in the BIOS and have them kind of work in Windows with 4 cores disabled long enough for a SuperPi run to finish, it is about achieving 24/7 clocks that aren't just numbers on the screen that aren't accurate because the board is throttling the CPU in the background under load.And using LN2 is easy for people with cases?Pretty much all the records are set by disabling as many cores as possible, standard practice for a world record run.
FYI fan blowing in that video is required for Ln2 OCing, not to cool anything but rather keep air moving so it doesn't condense as fast and turn into water droplets.
I mean I do understand where they are coming from. I mean if you goto OCN it is the same 90% are practical and 10% are extreme, for them it isn't an OC unless it is on all cores and all threads and prime95 or IBT stable, but that stability has to run for days, and they want to check your temps too just be sure. lol. Sometimes it is so different how different overclockers are, and what they deem to actually be an overclock. Of course then you can goto the benching sections and cooling sections of the same forum and 90% will be privy benchers and 10% will be those who want to learn. Personally I didn't really care for the extreme stuff, I mean yea i had a phase change until years back, but when i started doing LN2 that is when the real fun starts, you get a sense of satisfaction that someone who hasn't done it doesn't understand. You also begin to realize how much more this sport has to offer. Then you gotta learn OS optimization and software tweaks, and then you gotta learn what benchmarks like what OS and what hardware levels. Highest clock isn't always best, and that is hard to understand for many, someone with 100mhz less could kick your ass easily if you don't know what you are doing. Don't forget it isn't easy to control a CPU with a Cold Bug and a Cold boot bug, you have to pour so easily, you have to use different pots for different CPUs. You also gotta learn the secrets, like what thermal paste really works best, and it isn't artic ceramique.
After you have lost a few CPUs, and you don't know why, you thought it might have been voltage or it might have been the board, then you learn oh you aren't supposed to screw down the pot that hard. Then you are like damn i just messed up a lot of money, but you come out a bit smarter. Of course you had some fun too.
Even for those not extreme, seems pretty good. Getting baseline benches, seems faster than F4.:eek:
Gonna be OC testing soon.
Very few "extreme" clockers here. Personally, it's a lot of work, and I got a decent amount of respect for the guys that do it. But in the end, I'll leave that to them. I don't have hardware to burn at all. IF I did...:roll:
I think you are off the mark by implying our complaints about this being a poor demonstration of a 'fix' are misinformed. A quick read over the source shows that it was Gigabyte who provided this info to Toms Hardware in response to the VRM troubles, so why would we not be critical that the 'proof' is not sufficient? What does a LN run prove to us?
So, this BIOS made the board consume MORE power...kinda contrary to the idea there are issues with the VRM, isn't it?
Waiting on a couple more things, a) Cadaveca's review and b) the new stepping chips in February. Then I'll decide.