Sandy Bridge Nerfs Overclocking

Intel plans to deliberately limit Sandy Bridge overclocking | bit-tech.net

Looks like overclocking records won't be beat with the SB architecture...

 

Very unfortunate.

 

Don't make any conclusions yet.

 

Yeah, just like how EE CPU's can be overclocked, not your run of the mill processor. In the video the guy mentioned making it a little cheaper for people who don't want to pay $1000 for an EE CPU, but the question is how much cheaper, what will be overclockable, and how far will they be allowed to overclock. From the first link we see that there is a possibility of overclocking via traditional methods to be dead. Without that, we are more at the mercy of Intel with regards to overclocking. For all we know, it is just dumb as a stump turbo-boost renamed to overclock-boost or something stupid like that. I think it is safe to assume that I can no longer buy low end processors and overclock the bejeezus out of them like I have been doing for the past 8 years or so.

 

My next desktop computer is soooo red team

 

... Huh.

NOW that "pay to unlock more features" thing is an issue. I hope this at least means we can pay to unlock the multiplier of any CPU...

 

I'd say these are a good weather vane to follow:
Newegg.com - Intel Core i5-655K Clarkdale 3.2GHz 4MB L3 Cache LGA 1156 73W Dual-Core Unlocked Desktop Processor BX80616I5655K
Newegg.com - Intel Core i7-875K Lynnfield 2.93GHz 8MB L3 Cache LGA 1156 95W Quad-Core Unlocked Desktop Processor BX80605I7875K

It looks like the days of picking up cheap CPUs like the E5x00 and throttling them way up are gone, but I'm really don't think there's much cause to worry about overclocking being dead between that demonstration in the vid, and current pricing of non-EE CPUs with unlocked multipliers.

 

I lol' a little. It's the truth though. Intel isn't leading by enough in performance that an OC AMD chip wont be a better value than a stock Intel chip. So this move makes very little sense....

 

Old news, and hi AMD

 

I'm guessing that eventually somebody will get an asynchronous setup working.

And yes, I'm with Syberia, hi AMD

 

Wow, haven't seen those. But still, I bought my i7 860 for $200. If I couldn't overclock it, and had to pay $330 for one that could be overclocked, I would have definitely gone AMD. Intel loses a lot of value for the money if you need to pay a large premium to overclock their chips, and AMD becomes attractive again just like in the good old days.

 

here's the other thread on this topic (good discussion inside)

http://forum.notebookreview.com/hardware-components-aftermarket-upgrades/519047-pay-50-unlock-more-cache-hyp-threading.html

 

Yes. I agree 100%. Phenom II N930 anyone?!

 

old news is old

And what remains to be seen is by how much you can OC partially unlocked SB CPUs (probably not much but it's not clear yet), and how much you'll pay in premium for a K series ; intel has asserted they'll be priced lower than current ones.

 

this is an old article.. got a debate about this... but i doubt intel is stupid to limit OCing.. it will force many of its gamer customers to go to AMD.. and i certainly wouldn't hesitate if this happened.

 

meh, Intel will learn soon enough. Just leave overclocking alone, and add more cores! And speed! And Reduce power! Don't drive the market away with stupid gimmicks.

 

Limiting overclocking on the first release of the SB processors is completely reasonable. Look at the complexities of the new internal layouts (including the ring data buss) and you'll understand.

Perhaps on the refresh OC-ing will be unlocked. But until Intel gets the new architecture completely debugged, patience might be a virtue.

 

So what are you trying to say, that Intel is doing us a favor? I don't see what any new type of layout or complexity has to do with anything, there are always going to be more clock cycles to eke out of any cpu. Overclocking is not something generally warranted by the manufacturer. It is all at your own risk, so what does Intel care?

But that is just wishful thinking. If they don't have it when it is first released, why would they have it later on? Everything we currently know about Intel's plan does not agree with your assessment. Intel does have the architecture debugged, or else they would not be releasing it, and it is going to be better than Nehalem from the get-go. New revisions almost without fail supersede previous ones both stock and overclocked. That is just the way things are, but that in no way means previous revisions are crap or that they should not be overclocked. If you want to play that game you are going to sit around until the end of time. This is not in anyone's favor except for Intel's, so they can wring out ten million of extra $ from people who want to overclock.

 

AMD Have mobile CPUs?

Intel are really begin stupid.

 

You know damned well that Intel has a track record of their product refreshes being more capable than the initial release as well as solving bugs that the public may or may not have been aware of.

You also know (or should know) that Intel has said that the SB internal architectures, in particular the mix of async and sync caches and the internal ring network, are a significant departure from everything the Core and iCore have delivered. Read the IDF documentation on SB architectures if you care to.

You also know (or should know), that Intels track record with Core and iCore is exactly that. Initial releases were locked down, people screamed, wrung their hands, and threatened to jump to AMD. When the refreshes came out with higher thermal margin and unlocked multipliers (enabling easy and successful overclocking), all was forgiven.

 

Seems a lot of bugs are not fixed or have workarounds instead. Also take bugs like fulltime IDA (both cores running 100% load at IDA frequency), that was never fixed or made public by Intel. But yes, new stepings with changes do happen.

The overclocking seems to be made harder for each new architecture. My thought is to wait until SB comes out and see what can be done. ie I'll cross that bridge when I come to it as apposed to giving up before even starting. If however all you want is unlocked multi's then what about the 2920XM? Probably going to be too expensive for my liking though.

Still, it should be interesting times. I'm wondering how long it will take the software houses to make use of the new AVX instructions and what impact that will have.

 

Yes, that is true. However, that in no way means that the unrefreshed steppings are in any way bad. They are only not as good when later comparing them to what superseded them, just as how they are more capable than what came before them. People were having a grand old time overclocking the socket 1366 i7 920 when it first came out. Later steppings invariably were able to overclock higher and use lower voltages, but that doesn't mean the early steppings were bad at all.

How does this in and of itself pose any barriers to overclocking? It doesn't.

This is false, there was nothing like that whatsoever. The first 65nm and 45nm Core 2's and the first Nehalems were better than anything that came before them and nobody "wrung their hands" about them. They weren't "locked down" in any way either compared to later steppings. Also EE CPU's have been made out of each stepping, and they are priced out of reach of all but fanatics and those with money burning a hole in their pockets, so I don't know what you are talking about there.

1. Everything points to overclocking the processor being stopped, except through changing the multiplier on select high-dollar processors.
2. Intel has absolutely no track record of restricting anything on the first steppings of different processor cores, only to allow overclocking with later steppings.
3. Initial steppings are only poor in hindsight relative to later steppings. At the time they were the bee's knees and excellent overclockers.

 

I actually feel like overclocking has never been easier, even compared to when we had unlocked multipliers. Today even your basic motherboards have a slew of overclocking options. It is rare to find one that doesn't. We also have far less variables to worry about. I remember when I had to match the ram to FSB and had to worry a great deal about timings and whatnot. Voltages on everything was a big deal too. I actually used to volt-mod my motherboards. These days things can pretty much go whiz-bang out of the box. There is very little to worry about and things operate far more smoothly. Another huge factor is that even bottom end chips are easily overclockable almost if not as much as top end EE processors. Also when you make the settings a little too optimistic, these days boards often boot with default settings while still saving your changes. Back in the day I had to clear the bios countless times when I overclocked. I could go on like this for hours.

I remember the days when I used to tinker for hours on end and fawn over every little detail. Nowadays doing that is just a waste of time.

 

Yes, a BIOS with good overclocking features and perhaps a mainboard with in built in modifications makes things easy however what has to be done to achieve that is IMO harder.

For instance from core2 to Nehalem introduces TDP and TDC limits. Well the board makers can get around this by faking the reported CPU current. Then there is the disappearance of VID control. Again the board makers may control Vcore by interfacing to the VRM instead of CPU direct control. Now in SB we have an integrated master clock generator.

Sure, the board makers may have made it easy for us but IMO that doesn't negate that the CPU hasn't been made harder to overclock just that the difficulty of doing it is hidden from us.

 

It's very valuable, Thanks for your sharing!

 

The cheap i5 2500k unlocked Sandy Bridge is still on sale at Fry's for 209+ this week. I picked one up yesterday. I still don't have a new board for it yet though :|