AnandTech Ivy Bridge Performance Preview

[soccerballtux]

IMO Ivy Bridge is just Intel buying time for Haswell.

Its very similar to sandy bridge which is why it will fit Sandy mobos with bios flashed.

If you have a Sandy now forget about IB and its 10 percent performance increase which you will never see in real world app....

If you dont have a Sandy, wait for Mandy,,, or I mean Haswell and do a nice 2013 upgrade. gl

Im holding up until this albino and GPU inside a CPU craze dies and they start focusing on adding more cores. I need cores for my Sonar X1 , and I need 128GB ram for all the samples I load in template, thx

64GB ought to be enough ram for anybody

 

[Diogenes2]

Ought, the lesser of the three To brothers .. The others being; Need and Had...

 

[Quantos]

64GB ought to be enough ram for anybody

We'll be quoting that in 2022

 

[Grooveriding]

from Kyle...


http://hardforum.com/showpost.php?p=1038465252&postcount=85

I don't have much faith in Kyle and his estimations on how a CPU will overclock. In the review he did on a 3960X, he just cranked voltage to 1.52V without doing other adjustments for a 4.7 overclock... Then went ahead to declare that as his OC result No time taken to make any other proper adjustments.

I've read that as the process shrinks the chips can't handle as much voltage, but also have the impression that higher clocks can be reached with the same voltages used on the previous process ? Sort of balancing it out and then some in favour of the new process ?

I am expecting IVB to give us probably 4.8 to 5 as the new easy OC from the 4.4 to 4.6 we see on SB. All layman guesswork.

 

[frozentundra123456]

I think the hd4000 will fine in most laptops since those ultrabook are all mostly 1366x768? res anyways and at that resolution that gpu will run almost any game at decent framerates.

Yea, too bad the laptop chips are more delayed. The most interesting comparison would be to me the HD4000 vs Llano A6 in a power constrained laptop scenario. And then of course HD4000 vs laptop Trinity.

 

[Idontcare]

64GB ought to be enough ram for anybody

I'm not interested in 64GB of ram. I want 64GB of L1$ (and without all the clockspeed vs latency tradeoffs of course )

 

[Makaveli]

I'm not interested in 64GB of ram. I want 64GB of L1$ (and without all the clockspeed vs latency tradeoffs of course )

lol be prepared to pay 10k for the chip

 

[IntelUser2000]

lol be prepared to pay 10k for the chip

LOL, more like your mansion. Let's just imagine the die size of a 64GB SRAM on 32nm.

Assuming a square die, that means the SRAM itself would end up 472mm on each side(220,000 mm2). You can't even print it out on a 450mm wafer!

 

[TidusZ]

If this overclocks worse than sandybridge I'm going to really regret waiting

 

[GammaLaser]

LOL, more like your mansion. Let's just imagine the die size of a 64GB SRAM on 32nm.

Assuming a square die, that means the SRAM itself would end up 472mm on each side(220,000 mm2). You can't even print it out on a 450mm wafer!

Making it an MCM would help, not that it would make it any cheaper though

 

[Arachnotronic]

If this overclocks worse than sandybridge I'm going to really regret waiting

Be prepared to regret, then...

 

[IntelUser2000]

Making it an MCM would help, not that it would make it any cheaper though

And each side is almost as big as side cover for full size ATX cases.

 

[jimpatrick]

Be prepared to regret, then...

why would that be?D:

 

[Arachnotronic]

why would that be?D:

http://hardforum.com/showpost.php?p=1038465252&postcount=85

 

[TidusZ]

what does that guy really know though? He said himself, he doesn't have an ivy bridge. Sounds like speculation to me.

 

[Nemesis 1]

I don't recall how long its been . But AMD has been screaming fusion and in DX11 compute test . Even though only 1 test was run Intel IVB is 10% faster than LLANO. . Bur stilll pretty weak. Haswell should be special . Intel with haswell will likely pass AMD on graphics never to look back again.

 

[Don Karnage]

http://hardforum.com/showpost.php?p=1038465252&postcount=85

Source

http://www.xbitlabs.com/news/other/...tel_s_Next_Generation_Ivy_Bridge_to_7GHz.html

 

[Nemesis 1]

Netburst -> Conroe seemed so awesome because people forget that the Pentium M existed in between them

People didn't forget. Only a handfull of people ran Dothan desktops. and of them I only knew of 2or three that got dothan stable @ 3ghz. It was also expensive .

 

[Olikan]

http://www.xbitlabs.com/news/other/...tel_s_Next_Generation_Ivy_Bridge_to_7GHz.html

at the same link...

It is unclear what cooling system was used

 

[Nemesis 1]

Well, what I said about three months ago and what was obvious: 10% faster, where 5% comes from higher IPC and the other 5% from slightly higher clock speed.

Nothing groundbreaking, but support for PCIe 3.0 is very nice.

Well rarther than argue with you over your miswording and failed attempt to downplay the IPC gains I will go along with ya . If 100mhz gets you 5% performance gains . IVB at 5ghz is going to be something vary special. Allmost doubling stock IVB performance once the 5% iPC is scaled also. IVB Igp drivers will bring good % increase in performance on some games not so much on others.
Its funny how AMD/NV drivers bring performance improvements but Intels don't . Its not funny its laughable.

 

[AtenRa]

I don't recall how long its been . But AMD has been screaming fusion and in DX11 compute test . Even though only 1 test was run Intel IVB is 10% faster than LLANO. . Bur stilll pretty weak. Haswell should be special . Intel with haswell will likely pass AMD on graphics never to look back again.

Triniti is around the corner and Gen3 AMDs APU will have GCN integrated. It took 15+ months for Intel to be 10% faster in one test (Still losing in gaming to Llano).

Have a look at the following slide from AT HD7870 review, observe how the HD7770(GCN) obliterates the HD5870(VLIW5 Llano) and HD6970(VLIW4 Triniti).

What makes you believe that Haswell will pass AMDs GCN APU in compute??

 

[Diogenes2]

...

Source

http://www.xbitlabs.com/news/other/...tel_s_Next_Generation_Ivy_Bridge_to_7GHz.html

Wow, and only 1.89 vcore !

Lots of headroom left ...

 

[TidusZ]

Wow, and only 1.89 vcore !

Lots of headroom left ...

http://xtreview.com/addcomment-id-15487-view-new-sandy-bridge-overclock-record.html

1.68v, 5863 mhz (sandy bridge). The record for SB is 6044 mhz at 1.728 v.

There's another picture floating around of 4.4 ghz IVB at 1.1v, so if 1.89 v didn't kill the processor instantly I think 5ghz+ sounds pretty reasonable.

 

[Vesku]

Hmm, this preview and the dribbling out of extreme OC numbers reinforces that some of Intel's toughest competition is what it has already sold.

A tad ironically, at least in the mobile space, it is the fact only recently Intel committed to offering more than the minimum as an integrated GPU that will help drive the upgrade bug.

 

[Edrick]

Assuming a square die, that means the SRAM itself would end up 472mm on each side(220,000 mm2). You can't even print it out on a 450mm wafer!

Naturally we will need to wait for 22nm!