Ivy Bridge Discussion - Page 7

Phil · December 20th, 2011, 01:51

With Fab moving to 22nm, it's more cost effective for them to pay of the new tooling with mass manufacturing. Doing so also increases the probability of higher binning.

Once Ivy hit's it's stride, it will quickly become the main focus, SB proper will diminish.

__________________

Vision is the art of seeing what is invisible to others

runawayprisoner · December 20th, 2011, 02:35

And I'm looking forward to seeing my ******* run all PCSX2 games at full speed.

All of the technical aside, I'm pretty amazed with Sandy Bridge's performance on mobile devices. If anything, I kind of expect Ivy Bridge to close the gap between a highly overclocked desktop and a stock standard laptop.

Phil · December 20th, 2011, 03:33

PCSX2 emulation (with cottons hacks) should move into the 2ghz range. Good progress.

__________________

Vision is the art of seeing what is invisible to others

runawayprisoner · December 20th, 2011, 03:44

Actually, it's cotton's MTVU that gave that boost to quad-core chips. The rest of the hacks can be turned off completely and some of the heavy hitters like God of War and Shadow of the Collossus would still be playable on a laptop, which is quite something. In fact, I think that once MTVU is perfected and properly optimized, most more-than-two-core Ivy Bridge with Turbo Boost would walk through PCSX2 like nothing.

In fact, I think I'm starting to run more into GPU limitations with some of the higher-end games (GT4 comes to mind). Most everything runs full speed with just MTVU and no other speed hack.

That's also understandable since the lowest quad-core Sandy Bridge can Turbo Boost to 2.9GHz, and the highest ones can go up to 3.7GHz.

runawayprisoner · December 26th, 2011, 16:52

Oh my God...

And my guesses weren't off the mark. Mobile Ivy Bridge chart:

http://www.engadget.com/2011/12/07/i...e-cpus-abound/

If my experience right now with PCSX2 doesn't degrade, it's safe to say that the newer Ivy Bridge quad-core laptops and M*****k would definitely be able to handle PCSX2 at full speed in most situations and even up to some demanding games with MTVU.

Good to see we've finally had hardware fast enough to handle PCSX2 on mobile.

Core i7 3870QM at 2.7GHz -> 3.7GHz single -> 3.5GHz dual -> 3.4GHz quad is crazy!

Princess Garnet · December 27th, 2011, 00:04

Now that's interesting. If they were able to raise clock speeds on the mobile lineup that much, where thermal and power is a bigger issue, that means the headroom is hopefully there on the desktop side to go higher than it does even though they didn't raise them.

__________________
"The heart has it's reasons that reason knows nothing of."

runawayprisoner · December 27th, 2011, 15:06

If there is no headroom, then it's safe to say around 4GHz is the roof, and mobile is catching up with desktop pretty fast.

But it's fine. Intel can afford the 4GHz roof because AMD has yet to introduce any viable alternative.

Phil · December 27th, 2011, 18:11

They should watch out for VIA though. They make some mean processors.

__________________

Vision is the art of seeing what is invisible to others

runawayprisoner · December 27th, 2011, 18:54

Quote:

Originally Posted by Phil

They should watch out for VIA though. They make some mean processors.

Since a certain company never uses VIA, and would likely only use Intel procs until they switch to ARM completely, I wouldn't worry about it.

On a more serious note, I think VIA is going to ride on ARM instead, so that's one less worry for Intel. On the other hand, the mobile market is quickly shifting to ARM's side, even though Intel still has most of the pie.

I personally don't care either way... as long as Ivy Bridge can do Gran Turismo 4 on PC at full speed.

**@ruantec** · December 27th, 2011, 20:15

Intel should release an Ivy Bridge powered Tablet PC and i will create the UI + Emu-Frontend + On-Screen Gamepad Overlay

.... i know Phil will love it

__________________

Current development tools:

Visual C++.net, Visual C#.net
Visual VB.net, Visual Webdeveloper.net
Bloodshed Dev C++, Borland C++
Visual Basic 6

Princess Garnet · December 27th, 2011, 22:56

Quote:

Originally Posted by runawayprisoner

If there is no headroom, then it's safe to say around 4GHz is the roof, and mobile is catching up with desktop pretty fast.

But it's fine. Intel can afford the 4GHz roof because AMD has yet to introduce any viable alternative.

If 4GHz is the roof, what's with probably literally all of Sandy Bridge "K" processors being able to going over, sometimes well over, that. I'd quickly guess that around 4.4GHz to 4.8GHz seems to be where most fall. If the same limitations apply, it's certainly not 4GHz.

Actually, that's not just Sandy Bridge either. To quote someone else, "4GHz is so passe". 4GHz has been old news since Wolfdale, and honestly, that's even easy for an earlier Pentium 4 (LGA 775) and perhaps Pentium D too once we later got the right motherboards and cooling. Outside of Conroe, it's almost like if you have a relatively recent (last five or six years or so) Intel CPU itself that can't do 4GHz, it sucks; that's how common 4GHz can be.

The "roof" is purely self imposed by Intel, as you said, due to lack of competition. Power draw and thermals are important in the mobile sector, and if they can release those sorts of speeds on the mobile front, they must be running much cooler (good news for me with a Sandy Bridge that thinks it's a Core i7 920 or something). That's why I asked, the only question left now is if they overclock higher. Even if they aren't stable, a large many of Sandy Bridge will at least boot at or close to 5GHz (I know mine boots into Windows at at least 5.2GHz; haven't tried higher), so I'm holding out hope.

__________________
"The heart has it's reasons that reason knows nothing of."

Last edited by Princess Garnet; December 27th, 2011 at 23:01..

Princess Garnet · December 28th, 2011, 09:16

I agree with most of that, the part of a desktop having no need for them aside. People who are power users will always go with a dedicated solution from another brand (nVidia or AMD) as they are leagues better, but Intel still knows it needs to up the ante of it's dedicated stuff on the desktop if it's to make a real push with more of an APU approach, and Llano shows why. That's not something that's exclusive to the mobile sector. This is why much of Ivy Bridge's focus was on improving the IGP. Llano does, what, two (roughly?) times better than the Sandy Bridge "K" SKU models (HD 3000 series), and most of the stuff out there has less than that (HD 2000 series, or half the performance), so Intel knows it needs to push there. AMD isn't competing in raw performance, so they've no need to push too hard there, but Intel isn't stupid and knows they still have to improve, as they are still competing with.... themselves! They need to convince users to upgrade. Making Ivy Bridge just better enough, and being a drop in upgrade, makes it good for Sandy Bridge owners and owners of older platforms, and the push on the IGP front catches it up there too.

__________________
"The heart has it's reasons that reason knows nothing of."

runawayprisoner · December 29th, 2011, 17:07

Quote:

Originally Posted by Princess Garnet

If 4GHz is the roof, what's with probably literally all of Sandy Bridge "K" processors being able to going over, sometimes well over, that. I'd quickly guess that around 4.4GHz to 4.8GHz seems to be where most fall. If the same limitations apply, it's certainly not 4GHz.

Actually, that's not just Sandy Bridge either. To quote someone else, "4GHz is so passe". 4GHz has been old news since Wolfdale, and honestly, that's even easy for an earlier Pentium 4 (LGA 775) and perhaps Pentium D too once we later got the right motherboards and cooling. Outside of Conroe, it's almost like if you have a relatively recent (last five or six years or so) Intel CPU itself that can't do 4GHz, it sucks; that's how common 4GHz can be.

The "roof" is purely self imposed by Intel, as you said, due to lack of competition. Power draw and thermals are important in the mobile sector, and if they can release those sorts of speeds on the mobile front, they must be running much cooler (good news for me with a Sandy Bridge that thinks it's a Core i7 920 or something). That's why I asked, the only question left now is if they overclock higher. Even if they aren't stable, a large many of Sandy Bridge will at least boot at or close to 5GHz (I know mine boots into Windows at at least 5.2GHz; haven't tried higher), so I'm holding out hope.

Being able to go over 4GHz doesn't mean it's a long-term stabilization thing. Sure, I actually was able to run an E8600 at 4.5GHz way back in the days for... a whole year without any issue at all. None. Then after that, it started showing signs of degradation.

Same thing happened to a i7 920. Ran it at 4.2GHz for a whole year. No problem. Then it started getting unstable, and finally settled in at 4.0GHz.

For non-overclockers, it wouldn't matter either way because Intel obviously "imposed" the 4GHz roof on itself. There is no Intel CPU that can Turbo Boost to 4GHz. The highest I have seen is 3.8GHz, but, at the same time, they are also making mobile chips that approach that limit, so the gap with desktop is lessening by the day.

Did you know a Sandy Bridge laptop can now cleanly beat a medium high performance workstation from 2 years ago? In every aspect? That's how close the gap is.

Also from a mobile standpoint, Intel is more competing with ARM than it is with AMD. The success of Apple's Macbook Air has already created a wind tunnel for Ultrabooks to draft along, so AMD's APU netbooks are already getting irrelevant, and fast. Give manufacturers a year or two and they'll have ultra thin notebooks in the $500 - $800 price range. On the low end, Intel would have Atom in the $200 - $500 price range. That leaves no room for AMD APU. On the other hand, tablets are fast growing, and if Intel wants a piece of that pie, they'd want to maintain the same level of performance but cut power consumption to around 1/5 of what it is right now on their low end chips. That doesn't really sound good for them.

Princess Garnet · December 29th, 2011, 18:48

You possibly had two CPUs degrade. It doesn't mean anything so far as the ceiling being 4GHz. All that really proves anyway that overclocking can cut into a chip's lifespan to varying degrees, which is known. You still ran them there stably, so those speeds were valid.

__________________
"The heart has it's reasons that reason knows nothing of."

runawayprisoner · December 29th, 2011, 22:15

Of course they are valid, but that doesn't mean 4GHz still isn't the "roof" of stability where anything beyond will simply degrade the chip over time.

If we are talking about the maximum frequency where something can sustain long enough to be benchmarked or run something else, you can just use liquid nitrogen to cool the CPU and push it to 6GHz or so...

In fact, Pentium 4 was able to achieve stability at 4GHz way before we even had Wolfdale or Yorkfield to play around with.

I'm not saying that going higher isn't valid. I'm merely saying that Intel is intentionally holding back at 4GHz, and for better or worse, even if you try to overclock their chips, you don't stray that far away from that frequency. And by not too far away, I mean 1GHz over 4GHz is... what? 25%. Even if you were to sustain 5GHz stable clock 24/7, it's still only a 25% improvement over stock.

Princess Garnet · December 30th, 2011, 01:11

I'm pretty sure someone who is into values and numbers would consider 25% to be a rather large margin or error for claims.

All you're really doing now is saying what I already did; that Intel is holding back, but what does that have to do with ceiling (not) being 4GHz?

Yes, the Pentium 4 also did 4GHz, and I also mentioned that too, but it wasn't until later boards/cooling that this was well known.

There's also a difference between suicide runs and something that can be caled even half stable and practical.

__________________
"The heart has it's reasons that reason knows nothing of."

runawayprisoner · December 30th, 2011, 04:41

All I'm saying is that it is possible Intel is intentionally holding their chips back at below 4GHz "and" at the same time preventing others from going much higher than that. In other words, 4GHz is the magic number, one you can get any chip to, and not too many chips stray far away from it.

Also, I think taking a Q6600 from 2.4GHz to 3.6GHz (50% overclock) a more drastic improvement. Compare that to taking a 2600K from 3.8GHz (Turbo Boost) to 4.6GHz (21%), and you'd see that we are just scaling back, and not boosting up.

I was active a few years ago when we could easily achieve 50%, 75%, or even 100% and 125% overclock... nowadays it's much less exciting when you look at those statistics even if the chips themselves go faster.

Princess Garnet · December 30th, 2011, 07:22

Okay, but what you're saying now is besides the point of what we were discussing, which is that 4GHz is or isn't the ceiling.

As for percentages, I don't really care much. The final results are what matters. If anything, that also goes a bit against the part of them intentionally holding back as much anyway.

__________________
"The heart has it's reasons that reason knows nothing of."

runawayprisoner · December 30th, 2011, 17:07

4GHz is not the "ceiling" in that it's the absolute limit that no CPU can pass. It's the "ceiling" in that frequencies can go up to it, and stretch a bit more, but then will scale back to it eventually.

The absolute "roof" that I think you are talking about is the limit at which something becomes unstable enough to not last long, but that's a different kind of limit altogether.

Think of a rubber band. There is a certain amount you can stretch it so that it won't pull back. That's what I meant. If you try to stretch it more, it'll obviously work just fine, but if you release it, it'll pull back eventually. Now, not being able to boot, or stay stable, that's where you snapped the rubber band.

See the distinction?

Also the result is likely only around 10-20% improvement over stock. I guess if you want the absolute best experience you can get out of your money, then sure, it works. But personally, I like it at a number where I can comfortably say that any error or warning arising from the computer is due to my own or the software's error, and not something caused by the CPU.

Princess Garnet · December 30th, 2011, 20:26

No, I am not talking about the absolute limits, and you're really getting more into personal opinion on the matter, which, by the way, I am also more conservative about these days, but it's a bit besides the point.

Granted, I have no sources here, but I'd take a big guess and say that if we take averages into account, it's probably not 4GHz, but rather beyond that, where the average for Sandy Bridge would fall. It's not 4GHz, as I'd say that's more like 3.6GHz or 3.8GHz for a Core 2 Duo E8x00. I'd say around 4.5GHz or so is about what 4GHz was to Wolfdale (and even then, it depends; my C0 stepping Core 2 Duo E8400 was a totally different thing than my E0 stepping Core 2 Duo E8600 [and that was just a stepping change, not even a die shrink]). Afterall, 4GHz is like 200MHz or 300MHz higher than stock Turbo, right? Even if they're not get 100% overclocks, you have to think they have a bit more than that in them. Most will do it, or if not fall just shy of it. If mine wasn't like a Bloomfield in how warm it runs, even mine would otherwise do it. I could probably get 4.4GHz, but I feel the drop to 4GHz isn't losing me much for the cooler temperatures I get (and maybe need). I think you're confusing your preference for the "sweet spot" with the "new averages" though.

As for your last statement, that's the risk with ANY overclock. My Core 2 Duo E8600 was at "only" 4GHz, and it was the culprit of a BSOD after about a year of use with no issues. It turned out it was likely never stable to begin with (I knew of less programs and admittedly didn't put it through much). There's a lot of variables involved. So, just saying "there's diminishing returns" doesn't mean average overclocks haven't been going up. They clearly have been. Bulldozer, although with it's lesser performance, follows the trends too (from what little I've seen, anyway).

__________________
"The heart has it's reasons that reason knows nothing of."

Thread Tools
Show Printable Version Email this Page
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

December 20th, 2011, 01:51	#121
Phil 339.9 Join Date: Oct 2007 Location: USA Posts: 14,777	With Fab moving to 22nm, it's more cost effective for them to pay of the new tooling with mass manufacturing. Doing so also increases the probability of higher binning. Once Ivy hit's it's stride, it will quickly become the main focus, SB proper will diminish. __________________ Vision is the art of seeing what is invisible to others

December 20th, 2011, 02:35	#122
runawayprisoner Level 9998 Join Date: Nov 2006 Location: Java Posts: 9,377	And I'm looking forward to seeing my ******* run all PCSX2 games at full speed. All of the technical aside, I'm pretty amazed with Sandy Bridge's performance on mobile devices. If anything, I kind of expect Ivy Bridge to close the gap between a highly overclocked desktop and a stock standard laptop.

December 20th, 2011, 03:33	#123
Phil 339.9 Join Date: Oct 2007 Location: USA Posts: 14,777	PCSX2 emulation (with cottons hacks) should move into the 2ghz range. Good progress. __________________ Vision is the art of seeing what is invisible to others

December 20th, 2011, 03:44	#124
runawayprisoner Level 9998 Join Date: Nov 2006 Location: Java Posts: 9,377	Actually, it's cotton's MTVU that gave that boost to quad-core chips. The rest of the hacks can be turned off completely and some of the heavy hitters like God of War and Shadow of the Collossus would still be playable on a laptop, which is quite something. In fact, I think that once MTVU is perfected and properly optimized, most more-than-two-core Ivy Bridge with Turbo Boost would walk through PCSX2 like nothing. In fact, I think I'm starting to run more into GPU limitations with some of the higher-end games (GT4 comes to mind). Most everything runs full speed with just MTVU and no other speed hack. That's also understandable since the lowest quad-core Sandy Bridge can Turbo Boost to 2.9GHz, and the highest ones can go up to 3.7GHz.

December 26th, 2011, 16:52	#125
runawayprisoner Level 9998 Join Date: Nov 2006 Location: Java Posts: 9,377	Oh my God... And my guesses weren't off the mark. Mobile Ivy Bridge chart: http://www.engadget.com/2011/12/07/i...e-cpus-abound/ If my experience right now with PCSX2 doesn't degrade, it's safe to say that the newer Ivy Bridge quad-core laptops and M*****k would definitely be able to handle PCSX2 at full speed in most situations and even up to some demanding games with MTVU. Good to see we've finally had hardware fast enough to handle PCSX2 on mobile. Core i7 3870QM at 2.7GHz -> 3.7GHz single -> 3.5GHz dual -> 3.4GHz quad is crazy!

December 27th, 2011, 00:04	#126
Princess Garnet From Love and Limerence Join Date: Nov 2004 Location: Ohio Posts: 6,555	Now that's interesting. If they were able to raise clock speeds on the mobile lineup that much, where thermal and power is a bigger issue, that means the headroom is hopefully there on the desktop side to go higher than it does even though they didn't raise them. __________________ "The heart has it's reasons that reason knows nothing of."

December 27th, 2011, 15:06	#127
runawayprisoner Level 9998 Join Date: Nov 2006 Location: Java Posts: 9,377	If there is no headroom, then it's safe to say around 4GHz is the roof, and mobile is catching up with desktop pretty fast. But it's fine. Intel can afford the 4GHz roof because AMD has yet to introduce any viable alternative.

December 27th, 2011, 18:11	#128
Phil 339.9 Join Date: Oct 2007 Location: USA Posts: 14,777	They should watch out for VIA though. They make some mean processors. __________________ Vision is the art of seeing what is invisible to others

December 27th, 2011, 20:15	#130
@ruantec Crazy GFX coder Join Date: Nov 2002 Location: Dominican Republic/Austria Posts: 8,111	Intel should release an Ivy Bridge powered Tablet PC and i will create the UI + Emu-Frontend + On-Screen Gamepad Overlay .... i know Phil will love it __________________ Current development tools: Visual C++.net, Visual C#.net Visual VB.net, Visual Webdeveloper.net Bloodshed Dev C++, Borland C++ Visual Basic 6

December 28th, 2011, 09:16	#132
Princess Garnet From Love and Limerence Join Date: Nov 2004 Location: Ohio Posts: 6,555	I agree with most of that, the part of a desktop having no need for them aside. People who are power users will always go with a dedicated solution from another brand (nVidia or AMD) as they are leagues better, but Intel still knows it needs to up the ante of it's dedicated stuff on the desktop if it's to make a real push with more of an APU approach, and Llano shows why. That's not something that's exclusive to the mobile sector. This is why much of Ivy Bridge's focus was on improving the IGP. Llano does, what, two (roughly?) times better than the Sandy Bridge "K" SKU models (HD 3000 series), and most of the stuff out there has less than that (HD 2000 series, or half the performance), so Intel knows it needs to push there. AMD isn't competing in raw performance, so they've no need to push too hard there, but Intel isn't stupid and knows they still have to improve, as they are still competing with.... themselves! They need to convince users to upgrade. Making Ivy Bridge just better enough, and being a drop in upgrade, makes it good for Sandy Bridge owners and owners of older platforms, and the push on the IGP front catches it up there too. __________________ "The heart has it's reasons that reason knows nothing of."

December 29th, 2011, 18:48	#134
Princess Garnet From Love and Limerence Join Date: Nov 2004 Location: Ohio Posts: 6,555	You possibly had two CPUs degrade. It doesn't mean anything so far as the ceiling being 4GHz. All that really proves anyway that overclocking can cut into a chip's lifespan to varying degrees, which is known. You still ran them there stably, so those speeds were valid. __________________ "The heart has it's reasons that reason knows nothing of."

December 29th, 2011, 22:15	#135
runawayprisoner Level 9998 Join Date: Nov 2006 Location: Java Posts: 9,377	Of course they are valid, but that doesn't mean 4GHz still isn't the "roof" of stability where anything beyond will simply degrade the chip over time. If we are talking about the maximum frequency where something can sustain long enough to be benchmarked or run something else, you can just use liquid nitrogen to cool the CPU and push it to 6GHz or so... In fact, Pentium 4 was able to achieve stability at 4GHz way before we even had Wolfdale or Yorkfield to play around with. I'm not saying that going higher isn't valid. I'm merely saying that Intel is intentionally holding back at 4GHz, and for better or worse, even if you try to overclock their chips, you don't stray that far away from that frequency. And by not too far away, I mean 1GHz over 4GHz is... what? 25%. Even if you were to sustain 5GHz stable clock 24/7, it's still only a 25% improvement over stock.

December 30th, 2011, 01:11	#136
Princess Garnet From Love and Limerence Join Date: Nov 2004 Location: Ohio Posts: 6,555	I'm pretty sure someone who is into values and numbers would consider 25% to be a rather large margin or error for claims. All you're really doing now is saying what I already did; that Intel is holding back, but what does that have to do with ceiling (not) being 4GHz? Yes, the Pentium 4 also did 4GHz, and I also mentioned that too, but it wasn't until later boards/cooling that this was well known. There's also a difference between suicide runs and something that can be caled even half stable and practical. __________________ "The heart has it's reasons that reason knows nothing of."

December 30th, 2011, 04:41	#137
runawayprisoner Level 9998 Join Date: Nov 2006 Location: Java Posts: 9,377	All I'm saying is that it is possible Intel is intentionally holding their chips back at below 4GHz "and" at the same time preventing others from going much higher than that. In other words, 4GHz is the magic number, one you can get any chip to, and not too many chips stray far away from it. Also, I think taking a Q6600 from 2.4GHz to 3.6GHz (50% overclock) a more drastic improvement. Compare that to taking a 2600K from 3.8GHz (Turbo Boost) to 4.6GHz (21%), and you'd see that we are just scaling back, and not boosting up. I was active a few years ago when we could easily achieve 50%, 75%, or even 100% and 125% overclock... nowadays it's much less exciting when you look at those statistics even if the chips themselves go faster.

December 30th, 2011, 07:22	#138
Princess Garnet From Love and Limerence Join Date: Nov 2004 Location: Ohio Posts: 6,555	Okay, but what you're saying now is besides the point of what we were discussing, which is that 4GHz is or isn't the ceiling. As for percentages, I don't really care much. The final results are what matters. If anything, that also goes a bit against the part of them intentionally holding back as much anyway. __________________ "The heart has it's reasons that reason knows nothing of."

December 30th, 2011, 17:07	#139
runawayprisoner Level 9998 Join Date: Nov 2006 Location: Java Posts: 9,377	4GHz is not the "ceiling" in that it's the absolute limit that no CPU can pass. It's the "ceiling" in that frequencies can go up to it, and stretch a bit more, but then will scale back to it eventually. The absolute "roof" that I think you are talking about is the limit at which something becomes unstable enough to not last long, but that's a different kind of limit altogether. Think of a rubber band. There is a certain amount you can stretch it so that it won't pull back. That's what I meant. If you try to stretch it more, it'll obviously work just fine, but if you release it, it'll pull back eventually. Now, not being able to boot, or stay stable, that's where you snapped the rubber band. See the distinction? Also the result is likely only around 10-20% improvement over stock. I guess if you want the absolute best experience you can get out of your money, then sure, it works. But personally, I like it at a number where I can comfortably say that any error or warning arising from the computer is due to my own or the software's error, and not something caused by the CPU.

December 30th, 2011, 20:26	#140
Princess Garnet From Love and Limerence Join Date: Nov 2004 Location: Ohio Posts: 6,555	No, I am not talking about the absolute limits, and you're really getting more into personal opinion on the matter, which, by the way, I am also more conservative about these days, but it's a bit besides the point. Granted, I have no sources here, but I'd take a big guess and say that if we take averages into account, it's probably not 4GHz, but rather beyond that, where the average for Sandy Bridge would fall. It's not 4GHz, as I'd say that's more like 3.6GHz or 3.8GHz for a Core 2 Duo E8x00. I'd say around 4.5GHz or so is about what 4GHz was to Wolfdale (and even then, it depends; my C0 stepping Core 2 Duo E8400 was a totally different thing than my E0 stepping Core 2 Duo E8600 [and that was just a stepping change, not even a die shrink]). Afterall, 4GHz is like 200MHz or 300MHz higher than stock Turbo, right? Even if they're not get 100% overclocks, you have to think they have a bit more than that in them. Most will do it, or if not fall just shy of it. If mine wasn't like a Bloomfield in how warm it runs, even mine would otherwise do it. I could probably get 4.4GHz, but I feel the drop to 4GHz isn't losing me much for the cooler temperatures I get (and maybe need). I think you're confusing your preference for the "sweet spot" with the "new averages" though. As for your last statement, that's the risk with ANY overclock. My Core 2 Duo E8600 was at "only" 4GHz, and it was the culprit of a BSOD after about a year of use with no issues. It turned out it was likely never stable to begin with (I knew of less programs and admittedly didn't put it through much). There's a lot of variables involved. So, just saying "there's diminishing returns" doesn't mean average overclocks haven't been going up. They clearly have been. Bulldozer, although with it's lesser performance, follows the trends too (from what little I've seen, anyway). __________________ "The heart has it's reasons that reason knows nothing of."