Gigabyte Board Sets Intel X79 Overclocking Record 113
MojoKid writes "Renowned overclocker 'Hicookie' achieved a new high clock speed on the Intel Core i7 3930K processor by cranking the chip past 5.6GHz using a Gigabyte GA-X79-UD3 motherboard, the first mobo in the world to achieve a mulitplier of 57x. There was a bit of a scandal with Gigabyte recently when a YouTube video showed one of its X79 boards going up in smoke. Gigabyte released a BIOS update for several of its X79 boards to prevent such incidents from happening, and there were outcries that the new F7 BIOS would ... [reduce] overclocking performance; Hicookie's achievement should erase those concerns."
Finally (Score:5, Funny)
A way to compile Gentoo in a reasonable amount of time.
Re: (Score:2)
Re:only 2 cores enabled. (Score:5, Insightful)
For some applications 2 cores at 5.6GHz are better than any number of cores at 3GHz.
Re: (Score:2)
Re: (Score:3)
So, what's the equivalent in software? (Score:1)
Re: (Score:2)
The demoscene is sort of like that, with stuff like contests for producing impressive 3d graphics in under 4k executables.
Re: (Score:2)
yea its a lot less impressive when its using a metric ton of open GL and directX
Re: (Score:1)
Re: (Score:2)
The title implies that this is some sort of record and it's significantly less meaningful if AMD and Intel have both bested it by a couple gigahertz prior to the attempt.
Re:thats all you got? (Score:4)
8585.05 is is a bigger number than 8320.4. Do you see how that works?
Swing and a miss (Score:1)
Re: (Score:3)
Re: (Score:2)
Under my kind of load (compiling under Linux) it keeps up with the 2600K. Yes it's not a gamer's CPU... until game SW catches up.
Re: (Score:2)
The 2600k's power draw would be less though. :*(
Come on Piledriver!!
How is this even news? (Score:1)
*facepalm*
Someone's going to go well beyond 5.6ghz using the 125MHz BCLK strap.
And they won't even need 57x multiplier.
Underclocking (Score:4, Interesting)
Re: (Score:2)
Given that you can cool an i3 more or less passively, why would you want to underclock? Power saving? Get a refurbished laptop board...
Re: (Score:2)
Re: (Score:2)
The problem is, some machines already run for 10 years straight (maybe with a mainboard battery swap) - so you might just be giving away performance that wouldn't have reduced your hardware's useful lifespan anyway.
Re: (Score:2)
I should probably add: For particularly failure-prone hardware, that makes sense, of course. My graphics cards are always underclocked and undervolted as far as possible during idle... because I've had 3 die on me already.
CPUs on the other hand? Meh :p
Re: (Score:3)
I used to do this back when processors didn't have good power management.
Now you don't have to. This is what frequency scaling (SpeedStep / Cool'n'Quiet) is for. It automatically underclocks and undervolts the processor when it's idle, then spins it up to full when you're ready to use it.
Interestingly, it actually draws less energy to go to full speed for a moment when it has work to do. It uses more power for the burst, but it gets back to sleep sooner for a net savings. For this reason there isn't muc
Re: (Score:2)
Sandy Bridge and Bulldozer are down to 3 or 4W idle due to extensive clock and power gating. Experience has shown that slowing down a CPU with low idle power is counterproductive. You can undervolt modern CPUs at stock clock quite a bit though.
Re: (Score:2)
I am too lazy to find the benchmark, but IRIC at that clockcycle the PIII still is their fastest architecture...
No way that can be true.
The P3 could achieve a peak throughput of 3 micro-ops per clock cycle with carefully selected integer instructions, when also following the 4-1-1 rule, while even shitty Athlon64's can pull up to 6 micro-ops per cycle, and without the silly asymmetric limitations that the P3 had.
The fact that Phenom II's are still limited to that same peak performance of 3 instructions (2 micro-ops each) per cycle that the Athlon64's had makes AMD look fairly mediocre in the modern arena.. but i
Garbage board? (Score:1)
LN2 cooling (Score:2)
It's briefly interesting that they can hit such numbers with LN2, but I fail to see much value in the exercise. To the best of my knowledge, increasing the core multiplier doesn't have any impact on the motherboard, it's all internal to the CPU. As long as the board's power circuitry can deliver enough voltage, you just need to dissipate the heat, hence the LN2. You could replicate this on a $100 board, as long as it's not a Biostar.
Re: (Score:2)
You get far more powerful VRMs on higher end boards. A garden variety won't be able to supply to 200+ amps.
Understand why overclocking risks instability (Score:3)
The reason you can overclock without raising the voltage is that there is a voltage "guard band," which is like a safety margin. Some of my research has been about finding ways of reducing that guard band, because it's wasted energy. But that guard band is there for a good reason. Typically the critical paths in the chip (those with the longest propagation delay, which limit the safe clock speed) are a bit faster than the clock period. But that's only true whe the voltage is stable. If the voltage droops, then the propagation delay of those paths will increase, possibly too much, and you get incorrect computation. Voltage droops occur when circuits suddenly start switching a lot, demanding more current, or in other words, the effective impedence of the circuit drops, and by V=IR, for the current being supplied by the voltage regulator at that instant, the voltage inside the chip will drop. The regulator cannot respond instantly, so a guard band is provided so that the maximum droop never brings the instantaneous voltage below a certain margin. If you overclock without raising voltage, then your CPU will work fine most of the time, but certain workloads will cause wide swings in current demand, and if you execute one of those, you may crash your system.
This is why memory tests are worthless for stability testing, because due to cache miss latency, the current demand is relatively low and stable. Prime number generators are also not so great, because their current demand is relatively high and stable. I know that some of the SPEC and PARSEC benchmarks have some wild behavior, like FFT, for instance, or anything that has a lot of barrier synchronization. For the regular user, what's likely going to happen is that you'll get random such events where variation in cache hits and vector computation phases will cause significant spikes in current, and your game will crash.
Re: (Score:2)
Prime95/mprime's torture test cycles between in-cache "small FFT" and larger FFTs that need memory accesses, that's why it's such a great test.
I keep wondering what supercapacitors could do for CPUs that are not far from using kilo amps.
Re: (Score:2)
For the supercapacitors to do much, they'd have to be on-chip, which is impossible. Expensive voltage regulators already have big capacitors, which really keep the EXTERNAL voltage very smooth. But there's a fair amount of inductive decoupling between the pins and the silicon that makes a lot of internal voltage fluctuations invisible on the outside.
Sorry but not impressed. (Score:2)
Okay, recreational air/water cooling overclocks? I can dig that.
They're sustainable overclocks, something you can run the system at every day.
Throwing a liquid nitrogen pot on top of the CPU and getting some stupid-high speed while destroying the chip, the board, and most of the components?
Yeah. Not sustainable, and not impressive beyond a half second or so of "gee whiz".
Re: (Score:2)
It's exactly the same as drag racing.
Re: (Score:2)
My point exactly. It appeals to the dumbest common denominator.
I don't care how good it is... (Score:2)
...if it still puts HPAs on hard drives without warning and offers no way to permanently disable this behavior, I don't care if it runs twice as fast as everything else on the market and mixes me a drink at the same time while cooking me a gourmet dinner, I don't want it even if it's free.
bad FET (Score:2)
looks like it took out a Fire [4qd.co.uk] Emitting [fieldlines.com] Transistor [4qdtec.com] in the power supply section of the board...
Good times, seeing what happens when you let out the Magic Smoke.
A smoking board is a "scandal?" (Score:2)
Re: (Score:2)
Re: (Score:3)
More CPUs More MHz (Score:2)
Re: (Score:2)
Re: (Score:2)
Given the choice of those two chips I would go for the x3 (assuming there are no plans to overclock). Three cores at 3.3GHz are going to beat two cores of the same architecture at 2.2GHz in pretty much any workload. Add that to the fact that afaict the athlon in question is a newer and better architecture than the phenom in question and I really see no reason to go for the x4. Generally you should only start looking to more cores after you hit the point of diminishing returns in speed of individual cores.
Th
Re: (Score:2)
Re: (Score:2)
As for me, my laptop's CPU and HDD still aren't fast enough to do my work stuff as fast as I'd like. Judging from the task manager, the CPU is the bottleneck for the compiling and the HDD is the bottleneck for the packaging part, and CPU+HDD for various tortoisesvn[1] stuff. And it doesn't have enough memory (firefox, chrome, email, IM, multiple visual studio instances, remote desktop, putty, etc it all adds up, go
Re: (Score:2)
Re: (Score:2)
But why are you trying to do all of that on a portable anyway?
Coz that's what the company provides. Seems like almost everyone is on laptops here.
Anyway, I recently created a development environment in a virtual machine on a server, and compiling and building there is actually faster than on my laptop. So I might start to do development on that more often.
I'm tempted to buy a 256GB SSD for my work laptop, out of my own pocket. But that's more for fun than need :).
Re: (Score:2)
Re: (Score:2)
Most of us (at least on Slashdot) don't picture the ideal computing experience as Apple slapping our ass and calling us a good girl as it does us from behind.
Yes we like control. Only reason I don't overclock is because it's destructive and my CPU is good enough (I think an i7 940 is "good enough"), but if I want a performance boost when my computer is close to needing an overhaul I might go for it.
Re: (Score:1)
I picture you giggling to yourself as you typed this, idly scratching your neckbeard and staring doe-eyed at the Linus Torvalds poster on your wall. You don't have control in your life, so you seek it in computers. It's textbook. You need to justify the time you invest, so you feel compelled to tweak things and to invent evil enemies to fight against l
Re: (Score:2)
Wrong on all points except the giggling and the fact that I watch anime.
Re: (Score:2)
Re: (Score:1)
I'm not an Apple shill. I'm just anti-neckbeard. Slashdot has become the most out-of-touch tech community on the internet. Note how you had no problem with his post, but when I respond with the same kind of snark, suddenly there's an issue.
Re: (Score:2)
Slashdot has become the most out-of-touch tech community on the internet
Well that's debatable... But some of us consider that part of the fun :D
Re: (Score:2)
Re: (Score:2, Interesting)
Obviously you're misinformed, since a gentle overclock, say a few hundred Mhz, is well within the average chip capacity, with no appreciable impact on chip lifespan compared to the gain.
Really, by the time you hit the edge of the lifetime, your chip's performance would be worth replacing anyway.
Re:Stoopid. (Score:4, Insightful)
Such a small overclock will also have no appreciable impact on performance, either, outside of benchmarks.
Re: (Score:1)
Why do you care? It's not your money nor your hardware. Suprisingly enough there are modern games CPU bound and not video card bound yet. Is this really any different from someone trying to get more power out of their car, despite 90HP being more than enough to travel at modern highway speeds?
Re: (Score:1)
Overclocking my CPU would be inviting destruction operation for fart pipe-like performance gains. Which was my point.
Re: (Score:2)
kills its value, and I am glad I am not the only one calling it a fart pipe
Re:Stoopid. (Score:5, Interesting)
Overclocking rarely causes damage. If you gradually increase the speed, you'll hit the point where your computer isn't stable. However, this is well before the point where you cause permanent damage. As long as you stay reasonable (don't change voltages) you're getting a good performance gain for free. Why not get a 30% performance boost?
Re: (Score:2)
Who the hell overclocks without Prime95 (or Furmark) and a temp monitor? :)
Re: (Score:2)
are there linux equivalents for furmark and prime95? I'm mainly interested in furmark. I have all sorts of ways to test memory and cpu stability in linux.
Re: (Score:2)
Oooof, that's a very good question. It's not like there's many games with hardcore graphics to stress test with either...
Install Windows? :p
Re: (Score:2)
hmm good idea... although the little 2.5" 250GB drive in here is 2 partitions already, a little /boot and a rest of the drive lvm one. Also that would then cost me a windows license....
There is Unigine Tropics/Heaven/etc demos, but my understanding is that they(games) don't stress the GPU like furmark. Maybe someone could write a openCL app that does.
(before you ask there is a server that serves up all the media on the network as well, and the 250GB drive was a cost thing on this build as the build came wit
Re: (Score:2)
Dunno about furmark (never heard about it) but there's an equivalent to Prime95 called mprime, also on the mersenne download page [mersenne.org]. Same torture effect on the CPU AFAICT.
Re: (Score:2)
Right, like I said all sorts of ways to stress disks, memory, network cards, and CPUs in linux, but no good way to do GPUs. Maybe there is an openCL app that does.
Re:Stoopid. (Score:5, Insightful)
Because some of us would rather pay more to get a 30% performance boost without fiddling about trying to gauge system stability, and others of us are happy enough with the out-of-the-box stable systems that we have by default.
I've done my share of overclocking (having first overclocked a 386SX from 33MHz to 40MHz, a P100 to 120, and then some K6-2s from 300 to 350), but lately I'd rather just have a system that is both reliable and that doesn't need fucking-with.
YMMV.
Re: (Score:2)
Meh. None of that: AMD K6-2 300's running at 350 for years and years with perfect stability, using whatever random cooler, with months or years between planned reboots, with nothing more than a change in clock speeds: From 100x3 to 100x3.5.
Maybe I was just lucky,
Re: (Score:3)
There's not much fiddling about these days. If you're buying Intel stuff new, you really don't have much choice other than to get an unlocked multiplier - and setting a multiplier isn't exactly rocket science. If you stick to the basics (either undervolt at stock frequency or crank up the frequency at stock voltage), it should take about 10 minutes including reboots and stress testing.
AMD on the other hand might be a bit more complicated - I know I never got a stable overclock back during my Socket939 days
Re: (Score:2)
If your benchmark for stability is "I overclocked a [random thing] a few days ago and it still seems to be happy," then I don't think we really have anything discuss right now.
Let me know in a year or three how it works out.
(I've overclocked my Q6600 SLACR and had it appear to be stable for a few days, but then the crashes came. Much tweaking and posturing later, I've found that I'm far happier with a slower system that does not crash than I am with a fast system that sometimes crashes: I want my systems
Re: (Score:2)
My indicator for stability is that the machine has been running 24/7 at full tilt (encoding video via Handbrake) for about a week now... I've never seen an unstable system do that :)
Is the crashing system prime-stable? If not, it's just overclocked too far, or not getting enough voltage... or the motherboard is having problems coping... or the power supply isn't up to snuff... there's lots of possible causes.
I'll reply to this in a year or so with the amount of crashes I've had... ;)
Re: (Score:2)
Your Handbrake benchmark is impressive. It makes me ponder whether I should investigate overclocking modern-ish CPUs more than I have been.
But meanwhile (again), I've really enjoyed stability: Plug it together, configure it to spec, and it just works. It's very nice in ways that weren't always common, when a random ISA card might be unhappy with the bus speed of a (stock) 10MHz XT, or when dealing with the various hairy tribulations of VLB.
As to your future report: Please do, in a year or so -- by all m
Re: (Score:2)
To be honest: If this were a mission-critical machine, I wouldn't be overclocking it either. Then again, if it was mission-critical, it'd be a Thinkpad or Thinkstation with next-business-day on-site support, and those generally aren't overclockable in the first place ;)
For my day to day work machines, I agree 100%: Stability, longevity of components, low power usage and a general sense of a lack of risk are just too good to pass up.
Then again, I've had many non-overclocked systems exhibit instability too...
Re: (Score:2)
Re: (Score:2)
I'm sure they have better tools for checking stability than Prime95 and Furmark though :D
Are you talking about traders as in... stocks?
Re: (Score:2)
My FX-8150 ran torture tests under Linux (mprime) at 4.2GHz on all cores just with improved cooling (Noctua NH-D14), without raising the voltage. With slightly higher voltage is does 4.5GHz flat out (all 8 cores, no turbo, no throttling.) I have ECC RAM and VT-d and for this kind of performance I'd have to pay at least $2k for motherboard plus Xeon/Opteron CPU as opposed to $500. Yes It took a couple hours of testing to figure out the limits of the system but it was fun.
Re: (Score:2)
That's a good point many people tend to forget... myself included. Overclocking is fun, especially if you start seeing the benefits quickly. Overclocking a graphics card that's just on the verge of making your game playable... and then having the game run butter-smooth afterwards. Or seeing 3 hours shaved off of your 12 hour encoding time... :)
Re: (Score:2)
Re: (Score:2)
There have been reports about degraded or dead CPUs after overclocking but it only happens after severe overclocking/overvolting. Overclocked CPUs run stable for years as long as you keep it reasonable and keep the temperature down.
Re:Stoopid. (Score:4, Informative)
If you do a gentle OC without touching the voltage, you can't damage the chip. Two things damage a CPU: Heat and Voltage. Even if it locks up, it's not going to hurt anything. I squeezed a +10% OC and an 8% under-voltage on my GPU. 10% faster at full load and runs cooler than stock settings.
Re: (Score:2)
That is why i added copper heatsinks to the MOSFETs on my motherboard... Granted I'm running a 125W 1055T X6 in a 95W compatable motherboard. I under volt-ed the X6 down to the 95W 1055T levels and all has been good for the past 9 months many many chrome compiles, and libx264 encodes later. In fact those are about the only things that get the temps and the CPU fan speeds up.
Re: (Score:2)
Re: (Score:2)
It's not your money nor your hardware.
Exactly my thoughts to the OP.
However, from the summary:
There was a bit of a scandal with Gigabyte recently when a YouTube video showed one of its X79 boards going up in smoke.
I thought that this was totally part and parcel with overclocking. Although I was never into it too much, I was pretty sure that you started at normal speeds and slowly worked up higher and higher while keeping an eye on CPU and motherboard temperatures and bits?
Re:Stoopid. (Score:5, Funny)
Re: (Score:1)
Re:Stoopid. (Score:5, Interesting)
Actually, back in the day, it was the opposite of stupid. I bumped my old K6-III by 25%. I used it for video transcoding and a 25% increase in performance was huge. It could take a full day to do a high quality multi-pass of an hour of video. Shaving 5 or 6 hours off that was kind of a big deal. Sorry you couldn't figure out how to do it right.
Re: (Score:2)
K6's were some durn flexible chips, my favorite thing was to under-clock them, some poor sap would walk in with a pentium and pocket lint to spend, heck yea I will take this used K6/2 400, clip a pin (if the target was socket 5) clock that fucker at 366, charge them a song and a dance and get called a hero for (sometimes) tripling their speed, and making a customer. (also being honest and telling them the bullshit I was about to pull helped as well)
Re: (Score:2)
H264 high profile without GPU acceleration (Handbrake) is similar. 30% overclock => multiple hours won, at least on last-gen (~IntelCore2) hardware.
Re: (Score:2)
my undervolted AMD 125W 1055T X6 already does 480P video at ~60FPS in H264 level 4.1 high profile using libx264 and the ffmpeg lossless_slower profile with dual (5.1 and downmixed 2.0) AAC audio tracks.
I don't have any 720P or 1080P source to try out. I guess i could grab a random youtube video to try out.
Re: (Score:2)
Go ahead and try two-pass high profile 1080p... encoding for mobile phones is fast on my setup too, and I only have two CPU cores :p
Re: (Score:1)