Intel Ivy Bridge discussion thread.

[greenhawk]

Impressive that the top-bin TDP will be 77W. Thoughts on why that might be?

that is where the 20% performance increase of the 3D transisters and 22nm improvments went to. Just into TDP reduction. Currently expecting (going by some posts) that clock speeds are basically staying at the same level (bar the "100Mhz" change intel does over time).

 

[Edrick]

that is where the 20% performance increase of the 3D transisters and 22nm improvments went to. Just into TDP reduction. Currently expecting (going by some posts) that clock speeds are basically staying at the same level (bar the "100Mhz" change intel does over time).

I agree with that. We will be lucky to see 4Ghz chips ship at stock.

However, with the increased multipler for IB and lower power requirements, I expect to see some 6Ghz overclocks.

 

[boxleitnerb]

What is the maximum overclock? 63x100=6.3 GHz? I would expect Intel implementing some kind of limitation. If most chips could do 5.5-6.0 GHz, that would be so close to the limit. But awesome.

 

[lau808]

6ghz? i thought oc headroom was only 5% increase? so the guys getting 5ghz could reach 5.2/5.3?

 

[Puppies04]

^ that was everyone's expectation/assumption since time-zero, but the skuttle-butt that made the rounds was that there was something intrinsic about uefi that precluded the possibility of taking a eufi that was SB compatible and making it IB compatible.

I don't know jack about it other than the observation that when those rumors were circulating there were no tangible arguments made at the time for why the rumors would be BS.

Check the manufacturers websites. I saw a link to the asus one in a thread around here somewhere and everything above "budget" SB boards were listed as compatible with IB with a bios update, pretty sure all the GEN 3 PCIE ones were compatible straight out the box aswell.

 

[gevorg]

6ghz? i thought oc headroom was only 5% increase? so the guys getting 5ghz could reach 5.2/5.3?

The SB max multiplier is 57 while IB is 63, so according to multiplier the headroom increase is roughly ~10%.

6GHz+ will most likely be reserved only for extreme overclockers, but I won't be surprised if getting 5.0-5.2 on IB would be as easy as getting 4.6-4.8 on SB.


EDIT:
IB might be *the* chip that will let the "casual" overclockers with modest cooling and modest voltages beat the 5GHz barrier. Just speculation. 🙂

 

[Nemesis 1]

Are we mixing up Haswell with IB here?

No were not. IB IGP is said to come in 3 flavors. GP I II III. Won't belong and you will see IB at the show . Its IGP if I go by what guys say about llano is correct . Your in for a shock and the reason why AMD cancled GF 28. Things are going well for IB.

 

[Nemesis 1]

How did you go from that to graphics on Haswell? Sometimes your posts are really confusing to the readers.

Good example on the GT3 on ULV/mobile vs GT2 on desktop from another thread:



Integrated graphics are FAR more important on laptops due to the difficulty of being able to upgrade the components.

Well if you would point out were I mentioned Haswell in my post . I wouldn't be confused by your post . I suspect your using a Red herring here . As Intel makes horriable IGP . Just ask any of the arm chair engineers that post here.

Also I never posted this point here.

Also, are the A4 APUs true from-the-fab dual cores, or are they full quads with disabled cores? Strange that the desktop A4s are 2 cores/160 SPs, and the mobile ones are 2 core/240 SPs. Makes me go hmmmmm

Your post had 1 thing correct your confused.

 

[Nemesis 1]

Didn't was see a leaked document that said the level-3 graphics would be on the ULV CPU? I refuse to believe the fastest GPU will be on the ULV chips.

My guess is that it is a sort of situation where perf/watt for that part is the highest out of all of them, and so the model number (at least the internal model number) is the highest.

Thats what the leaked document said on a post at XS. I think we will see these GPIII on any number of notebooks .

 

[Idontcare]

I agree with that. We will be lucky to see 4Ghz chips ship at stock.

However, with the increased multipler for IB and lower power requirements, I expect to see some 6Ghz overclocks.

I was thinking 6GHz for top-end IB OC's as well...then I realized that in Dec 2006 I had my 65nm QX6700 quad at 4GHz, and now we are nearly in Dec 2011 and my 32nm 2600K quad tops out at 5GHz (I'm talking stable OC's), although we really should count that as Dec 2010 since the chips were nearly out back then an equally capable for OC'ing.

So that's 4yrs, 3 process nodes later, and 5 microarchitecture revisions...and a gain of 1GHz in the top-end clocks.

So yeah, 6GHz seems like "irrational exuberance" at this time for IB. If history is any indicator, we shouldn't be expecting 6GHz capability for 24/7 OC's until ~2015 on 14nm process tech (rockwell?).

 

[IntelUser2000]

Well if you would point out were I mentioned Haswell in my post . I wouldn't be confused by your post . I suspect your using a Red herring here . As Intel makes horriable IGP . Just ask any of the arm chair engineers that post here.

It's you who's confused. GT3 SKUs are supposedly only coming with Haswell. Even Anand himself made it pretty clear on his Ivy Bridge overview at IDF.

that is where the 20% performance increase of the 3D transisters and 22nm improvments went to. Just into TDP reduction. Currently expecting (going by some posts) that clock speeds are basically staying at the same level (bar the "100Mhz" change intel does over time).

20% reduction in TDP does not equal 20% improvement in performance. Due to voltage, you can increase performance/watt by keeping it the same clocks versus increasing performance. The usual new node claim is 30% power reduction or 20% gain in performance. Here they are claiming 50% power reduction or 20% higher performance. Even with TDP reductions clock speed can still increase.

How about ultrabooks with "real" dual cores ~3GHz instead of the ultralow voltage 1.6GHz chips we're seeing now.

You do know the top ULV part has a base clock of "only" 1.8GHz but can have sustained 2.4GHz Turbo speeds? With both cores active? I assume the gap will become even closer in the future.

 

[Nemesis 1]

It's you who's confused. GT3 SKUs are supposedly only coming with Haswell. Even Anand himself made it pretty clear on his Ivy Bridge overview at IDF.

Ya I am aware of this . That doesn't erase the fact there is a supposed leaked PDF showing GT3 on the single core IVB. So thats all I have read on it . It matters not to me . As I will use NV graphics alongside Intels IGP just as I am now . I just brought up what was shown in a thread @XS. As far as I am concerned I won't need a new gaming cpu for some years to come. Just NV graphics update when they release their 28nm chips.