Main link: Skylake Overclocking: Regular CPU BCLK overclocking is being removed
Intel has, for many years now, disabled overclocking on all but a select few, highly-priced CPUs, by fixing the maximum clock multiplier. (A practice not limited to Intel, as AMD has also done so on some series). The base clock was technically modifiable, but since it drove not just the CPU, but also RAM and PCIe clocks, you were lucky to get even a few megahertz out of it.
With their newest generation of chips, codenamed "Skylake", the PCIe domain is on a separate clock generator. While Intel officially only supported overclocking on their designated CPUs, and only on their highest-end chipset, SuperMicro, ASRock, and several other motherboard vendors produced motherboards using low-end chipsets that allowed base clock overclocking on any processor. Since this could allow extremely cheap systems to be performance-competitive with much higher-cost systems, albeit with higher cooling requirements and greater risk of failure, Intel was obviously upset.
The story is still developing (no parties have yet been willing to talk on the record, least of all Intel), but the latest BIOS update for several ASRock motherboards includes a firmware update and disabling the BCLK overclocking (the two are believed to be connected - the latest firmware prevents BCLK modification). Additionally, all marketing surrounding this unofficial-official overclocking support has been pulled. SuperMicro and other vendors have not yet done so, but unnamed sources are indicating that they will.
While it may be tempting to put the blame solely on Intel, this was clearly not a feature they intended to support, and the motherboard vendors should have been more cautious about making a feature out of bypassing a limitation on the CPUs, regardless of whether it was an artificial limitation or not. That said, I for one hope AMD's next line of CPUs is both fully competitive with Intel, and fully overclockable across the entire range. Maybe that is what is needed to force Intel to compete on price/performance again.
(Score: 1, Funny) by Anonymous Coward on Thursday February 11 2016, @12:07AM
Using Supermicro is punishment enough. Let them overclock it.
(Score: 2) by dyingtolive on Thursday February 11 2016, @08:02AM
Yeah, I saw Supermicro and Assrock, and I thought to myself, "The Dane Cook of motherboards? Who's overclocking on those?"
Don't blame me, I voted for moose wang!
(Score: -1, Troll) by Anonymous Coward on Thursday February 11 2016, @12:16AM
>>AMD
>>competitive
Good joke. *golf clap*
(Score: 2, Interesting) by gman003 on Thursday February 11 2016, @12:52AM
I'm actually pretty optimistic about Zen. It's too early to know whether it will succeed, but it has all the right pieces to do so.
(Score: 0) by Anonymous Coward on Thursday February 11 2016, @01:18AM
It's too early to know whether it will succeed,
No, it's never to early to know that AMD will fail like they always do.
(Score: 2) by gman003 on Thursday February 11 2016, @01:33AM
Oh, crap, am I old enough to play the "you damn kids must be too young" card finally? I guess so.
I remember the Athlon (which went blow-for-blow with the Pentium III and first-gen Pentium 4s) and the Athlon 64 (which kicked P4's ass). Even the Phenom wasn't a clear loser - it was 50/50 for each market segment which was better, K10 or Core/Nehalem.
And while I'm too young to actually remember the K6 or preceding, I can at least read the Wikipedia articles and know that it happened. Historically speaking, AMD's current position of being a *losing* underdog is the unusual thing for them.
(Score: 3, Informative) by requerdanos on Thursday February 11 2016, @01:48AM
I remember a glossy magazine ad for the AMD 486DX4/100 (a clock-quadrupled 25MHz 486) that boldly proclaimed: "Pentium Performance." -- It basically matched the performance of the first Pentium chips, with which they were competing. Admittedly iirc the first Pentiums were a dirt-slow 60MHz (yes, MHz), but they were Intel's fastest thing going and AMD was keeping up with them with a 486.
(Score: 1, Interesting) by Anonymous Coward on Thursday February 11 2016, @02:04AM
Admittedly iirc the first Pentiums were a dirt-slow 60MHz (yes, MHz)
Having cut my teeth on a Commodore 64, I have to mention that the C64 was a lowly 1 MHz and only 8 bit with only 64 KB of RAM (and not all of it was usable without special tricks). It was insane how much work people were able to squeeze out of that machine, though. Nobody puts that kind of time into optimization anymore.
I do remember moving to an IBM PC clone and having all those MHz and RAM was just a godsend. But Windows 3.1 managed to suck that up pretty quick, and I've been wanting for RAM pretty much ever since. I don't think Apple is helping, either, charging ridiculous amounts for memory when it's actually quite cheap these days.
I never did get into overclocking myself - too expensive if you screw up and I'm not a gambler.
(Score: 2) by fnj on Thursday February 11 2016, @05:41AM
You had 64K in your C64? Well I had to make do with 1K in my Altair 8800.
(Score: 2) by Hyperturtle on Thursday February 11 2016, @06:16PM
you guys. I cant even remember what I used back then, and I think I liked it!
I always wanted a black speaker box that had a computer voice that followed me eveywhere, like in wargames. I never was clear on how that same voice box existed everywhere broderick went.
if everyone at the WOPR facilitiy had one then imagine all of those computer operators having these discussions with the WOPR all at the same time in that big room for mission control. it'd be chaos! Like tricking your xbox to turn off or one person saying HEY GOOGLE and everyone's PC suddenly searching for whatever. Then again, the one guy did ask if the other guy liked vodka. It'd help in a call center environment like that.
(Score: 0) by Anonymous Coward on Friday February 12 2016, @03:31AM
We had a few Acorn computers in my high school and those had some predefined words they could say if you typed the right command. So we entered commands to sleep for 5 minutes, then say something funny to whoever was in the viscinity. Of course we'd stick around to watch their reaction. Good times. :)
(Score: 0) by Anonymous Coward on Thursday February 11 2016, @04:42AM
Oh, crap, am I old enough to play the "you damn kids must be too young" card finally? I guess so.
No, but I expected you to pull out the "But the Athlon64!!" as if 13-year-old processors have any relevance to today and you didn't disappoint.
(Score: 2) by damnbunni on Thursday February 11 2016, @05:27AM
If an AC says AMD historically fails, 'pulling out' historic non-fails is entirely relevant.
(Score: 0) by Anonymous Coward on Thursday February 11 2016, @03:35PM
Whatever makes you feel better, fangirl.
(Score: 3, Informative) by Gravis on Thursday February 11 2016, @12:55AM
overclocking is unsupported by definition. the fact that they give you the capability to overclock the CPU does not imply it's a supported feature, just that it's possible and at your own risk. there isn't a warranty in the world that will cover your losses when it goes up in smoke.
(Score: 3, Informative) by gman003 on Thursday February 11 2016, @01:07AM
Intel sells an additional warranty that covers replacement of a CPU that fails when operating outside Intel specifications. [intel.com]
$35 for no-questions-asked replacement on a $1000 CPU is actually a pretty good deal - I would definitely get one if overclocking an LGA2011-3 CPU, and I might even grab one for the 6700K I'm thinking of using in my next build.
(Score: 3, Insightful) by Gravis on Thursday February 11 2016, @02:22AM
if you have to pay for it, it's an insurance policy.
(Score: 1) by anubi on Thursday February 11 2016, @02:20AM
Had I been designing the chip, I sure would not have had this control available to the end user.
He misuses it, blows my thing up, then blames me.
I learned my lesson on those old CRT monitors, where one ( as an expensive practical joke ) could tamper with the horizontal timing registers, screw up the horizontal output circuit, and smoke the monitor.
There should be no HCF ( Halt and Catch Fire ) instruction in any machine released to the public.
I do my damndest to design my stuff so that software bugs ( or even deliberate attempts ) won't fry the hardware. For me, there was only one exception to this, which was to permanently disable write lines into an area of memory to insure that the code I put into it stayed there - unaltered. If I sold the thing under my guarantee, I needed the confidence to release it knowing that my customer wasn't gonna change the code and hold ME responsible for the end result.
"Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]
(Score: 1, Insightful) by Anonymous Coward on Thursday February 11 2016, @10:42AM
The problem isn't when it goes up in smoke but when it doesn't.
(Score: 5, Insightful) by requerdanos on Thursday February 11 2016, @02:10AM
I could not disagree more. If I purchase an item, whether an electric motor or a CPU or any widget-gizmo, I OWN IT. I therefore can do in the privacy of my mad scientist laborarory WHATEVER I WANT TO with it, without begging permission, hat in hands, of its manufacturer. That whole idea is simply nuts. Wrong wrong wrong.
In fact, "bypassing a limitation" is often the point of tinkering, hacking, mad-scientisting. There should be more tools for doing this, not fewer.
(Score: 0) by Anonymous Coward on Thursday February 11 2016, @02:25AM
Sure you can. But nothing makes it their responsibility to make that easy for you now is it? Make your own BIOS for their board that supports OC. They don't have to.
(Score: 0) by Anonymous Coward on Thursday February 11 2016, @10:04AM
This is why you should not use proprietary software. Only use software which respects your freedoms, and then you wouldn't have masters.
(Score: 3, Insightful) by gman003 on Thursday February 11 2016, @03:29AM
On your own device? Sure. Anything you want to do, go ahead and do. I don't care what the lawyers say their license agreement says - the nature of the transaction was a sale, and so as long as they have your money and you have their product, it ought to be yours to do with as you see fit. (The law may not be on this side, but as far as I am concerned, ethics is on mine here). If you want to buy up one of those motherboards now, and flash the old BIOS images onto it, that is (or at least should be) your right as a consumer. (And actually not a bad idea, come to think of it)
However, this feature was added by motherboard manufacturers. For them, the nature of the transaction actually is a license - they pay $X per year or per item to have permission to make these motherboards, using Intel's information and trademarks. They did not reverse-engineer Intel's socket and chipset and firmware - they signed a contract and paid for it. And the terms of that contract, presumably and apparently, included something to the terms of "if you make life hard for Intel, we can take this license away".
Intel's marketing gimmick of making only certain processors overclockable, well, it's not illegal, and it actually has at least a bit of basis in reality (they only mark the highest-binned CPUs as K-series, so even if you could overclock any chip, the ones currently sold as overclockable would clock higher than the others). I'm not going to call it outright wrong. So if Intel wants that to be their policy, well, it is their prerogative to do so. They can't (or at least shouldn't) be allowed to stop you if you find a way to go around it, but if a motherboard "partner" tries to bypass it as a way to drum up their own sales, that's a different story.
Because a license is an ongoing thing that can be broken, Intel could just threaten SuperMicro et al. with revoking their license if they didn't comply. Being unable to compete further in the motherboard market is not worth marginally increased sales for the current models, so of course they acquiesced.
My point in the summary was that that whole sequence of events was predictable for them. SuperMicro couldn't possibly have thought that Intel would just let it slide. Maybe they really were willing to go through all that trouble just to get extra sales for a few months. Maybe they somehow thought it was a deliberate feature. Maybe they just have really bad lawyers.
Honestly, with that line I was mainly just trying to prevent the comments from devolving into AMD fanboys bashing Intel. But since the only dumb comments were instead Intel fanboys bashing AMD, that may not have been necessary.
(Score: 2) by requerdanos on Thursday February 11 2016, @11:53AM
Though I may not like it, I do see and acknowledge your point. Thanks for pointing this out.
(Score: 2) by goody on Thursday February 11 2016, @04:06AM
Well, now you can tinker and bypass the limitation of the limitation.