Ivy Bridge models and clocks leaked

[Nemesis 1]

Anandtech previously said in an article that Ivy Bridge is going to have IPC increase of 4 to 6%, and GPU is going to be up to 60% faster.

Seems that they are focusing more on GPU performane and power consumption rather CPU performance. However it will still be better than AMD Crappydozer

AT also said that SB is only 15% faster than last gen . its more like 20-25% faster . Sure there are apps its not . But if you averageraged every app out . and NOT every game it would be 20-25% faster . Games are to GPU bound , were they are cpu bounf the % is good.

If this is all correct . and I want a low power chip . My next build will be ARM /Nv and SB-e on the high end.

 

[lol123]

If we should believe these rumors, then stock clock of IB is pretty much the same as SB. So the 18% or so higher clock speed Intel touted are same as the Bulldozer IPC claims by AMD. Or (just speculation):

(1) These rumors are baloney and real deal is 18% better clock speed.
(2) Intel has/had issues with finfet, they can't raise the clock speed but have to release a product. Too much leaks? They absolutely want to maintain 77 watts and they hit a thermal wall? Cosmic rays (only half joking)?
(3) Intel crippling their own chips so the SB-E line stays top dog until IB-E.
(4) No competition from AMD, why bother?

First of all, increased transistor performance does not directly translate into higher clock frequency of the entire CPU. Second of all, higher clocks enabled by the process does not require Intel to implement them in the SKU's they sell. With AMD, as you say, offering no competition at all in the performance segment, it makes sense to limit clock frequencies to maximize yields (and also to lower TDP). Overclockers can then take their unlocked chips as far as they can take them.

It's entirely possible (even likely) that Intel are having initial problems with their tri-gate process, but it's also very unlikely that they would have gone through with it if there were fundamental problems with the design that would prevent it from running at high clocks.

 

[tweakboy]

Wow great thread. OMG 65w TDP ,, oh my,,, Im @ 168 TDP with a 65nm.

What are these quad cores with HT. I mean come on , release a 8 core processor for desktop already. Soo many people have to go with server boxes so they get Xeon 8 core for their Audio Workstation or Video station etc....

 

[toyota]

its 77 watts TDP for the normal Ivy i5/i7 quads and your Q6600 has a TDP of 105 watts.

 

[Nemesis 1]

First of all, increased transistor performance does not directly translate into higher clock frequency of the entire CPU. Second of all, higher clocks enabled by the process does not require Intel to implement them in the SKU's they sell. With AMD, as you say, offering no competition at all in the performance segment, it makes sense to limit clock frequencies to maximize yields (and also to lower TDP). Overclockers can then take their unlocked chips as far as they can take them.

Well just because I went against the flow with the AMD hype . Doesn't mean AMD cpus aren't fast enough . With Intel fixing trim for Raid0 An ARM nv chip should be a great browser and I will have zero problem with going that direction . If intel can't push harder screw them . At = clocks IDF said 50 decrease in power usage . Screw intel with that line of LIES. 50% of 95watts is 47.5% thats what intel promised at the same clocks . LIES LIES LIES.

 

[Nemesis 1]

http://www.xtremesystems.org/forums/showthread.php?277242-Early-Ivy-Bridge-Sample-Tested

 

[Olikan]

For people who care about power/performance it would. And also people who care about the +60% IGP performance increase.

yeah...new chip for 20 watt less... in 2 years of full use it might pay for it self.

buy a 28 nm gpu, WAY better deal

edit:

mmm... almost 10% increase in cinebench, woot
now i wanna see overclock

 

[ViviTheMage]

I am very curious to know how well that HD 4000 does, compared to the 3000 in laptops.

 

[Edrick]

buy a 28 nm gpu, WAY better deal

Yea, cant wait until the end of 2012 for one of those.

 

[Zink]

We can't really complain that performance won't increase over SB. Intel can't release chips that clock higher than SB-E. These 1155 IB cpus are what your parents will buy in $700 Dell systems, they're not for CPU enthusiasts.

 

[Idontcare]

Well just because I went against the flow with the AMD hype . Doesn't mean AMD cpus aren't fast enough . With Intel fixing trim for Raid0 An ARM nv chip should be a great browser and I will have zero problem with going that direction . If intel can't push harder screw them . At = clocks IDF said 50 decrease in power usage . Screw intel with that line of LIES. 50% of 95watts is 47.5% thats what intel promised at the same clocks . LIES LIES LIES.

What's with all the rhetoric?

50% the power consumption at same clocks if the chip itself contains no microarchitecural changes.

IB is not SB shrunk to 22nm. The xtor counts will not be identical, etc.

Obviously IB is going to have more xtors than SB, not surprising then that the end product will not have 50% the power consumption.

 

[Nemesis 1]

We all know the xtors in the IGP are there . Are you now saying that the cpu also has more xtors . Rhetoric is rhetoric . depending on who says what . so its wait for IB-e to compare to SB-e to see if intel missed their target

 

[Fjodor2001]

In the original post there is one column named "FMB" with values such as 2011B, 2011C and so on. What does FMB and the column values mean?

 

[AtenRa]

Overclockers can then take their unlocked chips as far as they can take them.

Only 2 CPUs out of 18 can be OCed(K Models), pathetic

 

[exar333]

Only 2 CPUs out of 18 can be OCed(K Models), pathetic

IB CPUs will likely be able to be OCd much more easily than SB SKUs. Not sure if a new MB will be needed, but its possible. 'Limited OC' is what I have heard. Not sure if someone else can confirm...

 

[OS]

I am very curious to know how well that HD 4000 does, compared to the 3000 in laptops.

yeah the better process/power use and better IGP will make this a really interesting mobile product.

 

[Edrick]

Only 2 CPUs out of 18 can be OCed(K Models), pathetic

You do realize that all SB chips can OC, just not that high. I expect the same with IB.

 

[cbn]

50% the power consumption at same clocks if the chip itself contains no microarchitecural changes.

IDC, 50% reduction in power consumption for a die shrink (without change in clocks) seems very impressive.

What would be the typical reduction in power for a simple die shrink (without change in clocks) for a planar xtor die shrink? Has this trend remained constant over the last decade?

 

[nyker96]

amazing 77w tdp for the unlocked chips!

 

[Idontcare]

IDC, 50% reduction in power consumption for a die shrink (without change in clocks) seems very impressive.

What would be the typical reduction in power for a simple die shrink (without change in clocks) for a planar xtor die shrink? Has this trend remained constant over the last decade?

It is impressive, which is why Intel wanted to make sure everyone understood this new 3D xtor tech was more than just marketing PR, it actually does good stuff at the engineering level.

(don't get me wrong, Intel letting us know is itself marketing PR, but the technology that is the basis of their marketing efforts is not just marketing jargon, it really is the best thing since sliced bread)

Typical reduction would be about half that, 20-30%.

The question we have yet to answer as laypeople is how well do these things scale. Will 14nm 3D xtors have 50% lower power compared to 22nm 3D xtors? Or is this merely a one-time bump in performance/watt after which we are back to the same old scaling trend?

 

[Subyman]

I, for one, am very impressed with the power reduction. Losing 20W over SB for the K series is something to be excited about. Hopefully, these guys overclock well. 5.0Ghz on daily air is something to look forward to.

 

[SithSolo1]

If the price is right the i5-3330s could be nice budget 1155 quads.

 

[soccerballtux]

Wow great thread. OMG 65w TDP ,, oh my,,, Im @ 168 TDP with a 65nm.

What are these quad cores with HT. I mean come on , release a 8 core processor for desktop already. Soo many people have to go with server boxes so they get Xeon 8 core for their Audio Workstation or Video station etc....

this is the good thinga bout HT its the equivalent of unlocking cores on an AMD you get 2x the processing power thats why ht makes it have 8 cores in task manager

 

[ed29a]

this is the good thinga bout HT its the equivalent of unlocking cores on an AMD you get 2x the processing power thats why ht makes it have 8 cores in task manager

Only if your motherboard is of dark colored PCB.

gg and gl!

 

[cbn]

Typical reduction would be about half that, 20-30%.

Thanks for confirming.

I made a guess of 25% power reduction based on Intel's claim that Finfet's 50% power reduction was equal to two regular node jumps.