Early Ivy Bridge Sample Tested

[Cold Fussion]

Hm not very impressive. I mean 3770K was 100 MHz faster and scored ~ 9% better. Hope it overclocks much better otherwise I will stick to my SB-CPU.

20 watt lower TDP and it's not much more than a die shrink, so theres no reason to expect vastly better performance. April can't come soon enough.

[C3]

Thats true. Obviously I was expecting too much.

[bamtan2]

Thats true. Obviously I was expecting too much.

as the chip manufacturing gets smaller, it gets harder, which we all know. but intel is way ahead of everyone else at it. so whatever happens, it looks like intel will continue to have the highest performance chips, AND the highest performance:power chips. whether intel beats its own overclocked 2600k in april, or later with IVB-E, it is almost selfish for us to ask. we're spoiled with performance. how many years might it be before AMD beats an overclocked 2600k? don't answer that.

[radier]

Intel raised multpier from 57x in Sandy to 63x in Ivy. I think not without reason.

Maybe each K series CPU will do 5GHz on BOX cooler

[the finisher]

Man, that's what I've been hoping. A reliable 5+Ghz hyperthreaded quad would satisfy cpu needs for sometime.

Keep in mind I tote this thing all over the country for some real gaming fun +going to crunch on hotel power.

[EniGmA1987]

as the chip manufacturing gets smaller, it gets harder, which we all know. but intel is way ahead of everyone else at it. so whatever happens, it looks like intel will continue to have the highest performance chips, AND the highest performance:power chips. whether intel beats its own overclocked 2600k in april, or later with IVB-E, it is almost selfish for us to ask. we're spoiled with performance. how many years might it be before AMD beats an overclocked 2600k? don't answer that.

1 year, 2 tops. In multithreaded benchmarks at least. Single threaded is a long way off. All AMD has to do is get the manufacturing process tweaked properly to fix the power draw problem and higher speeds. Optimizng the core will be here this year too so that will bring a tiny improvement too.



I hope Ivy does well though. Im not hoping for a lot but with a die shrink generally comes a little bit higher OC headroom so I am hoping for an average OC on the chips to be around 5GHz on air

[sdsdv10]

<snip>



I hope Ivy does well though. Im not hoping for a lot but with a die shrink generally comes a little bit higher OC headroom so I am hoping for an average OC on the chips to be around 5GHz on air

Remember this is more than just a die shrink. This is the first commercial 3-D transistor (Fin-Fet, Tri-gate whatever you want to call it) technology available to the masses. Pass results may not be directly comparable (in a good way I hope).

[Hornet331]

Hm not very impressive. I mean 3770K was 100 MHz faster and scored ~ 9% better. Hope it overclocks much better otherwise I will stick to my SB-CPU.

Its pretty much what intel always offers with a die shrink ~5-10% performance (clock for clock) increase over the last cpu.

[Cyras]

8 Core IVY ~ 3,3GHz 3,6GHz Turbo, 130W TDP
10 Core IVY ~ 2,8GHz X,XGHz Turbo, TDP

[R101]

Now we're talking .

[flyck]

Remember this is more than just a die shrink. This is the first commercial 3-D transistor (Fin-Fet, Tri-gate whatever you want to call it) technology available to the masses. Pass results may not be directly comparable (in a good way I hope).

Those '3-d' transistor won't have any effect on ocing because they only have an advantage at low voltages. Ofcourse the smaller node still has an advantage.

So they gained about 400Mhz on a first 22nm attempt. Thats pretty nice, thats about 16% higher clocks.
(current E7 10 cores 2.4GHz @ 130W TDP)

[Mech0z]

I want to see mobile chips battery life tested !

[informal]

Its pretty much what intel always offers with a die shrink ~5-10% performance (clock for clock) increase over the last cpu.

Yep,IB is ~5.7% faster than SB at the same clock(1.089x3.4/3.5) according to these tests from the unnamed guy. Not bad at all, but nothing special either. More important are the power savings and how they will affect the clocks at ULV segment.

[PatRaceTin]

My 2500K hit wall at 5.55 (x55Multi)

[gallag]

Intel are holding back, imagine what the reviews will look like, they will further their lead in both performance and power usage, bulldozzzer is about to look even worse so why would intel play their full hand?

[nossy23]

I'm a bit disappointed. Doesn't look like I will upgrade my 2600K to one of these. Wasn't there another leak recently showing that Ivy Bridge won't really overclock more than Sandy Bridge? If that's true, less than a 6 % speedbump is not so good. Intel is spending more and more transistors on the GPU part, which is great for Ultrabooks, but does nothing for high end desktops. I would prefer they would make a separate line without the GPU stuff and give us more speed.

[Brice MJ]

Yep,IB is ~5.7% faster than SB at the same clock(1.089x3.4/3.5) according to these tests from the unnamed guy. Not bad at all, but nothing special either. More important are the power savings and how they will affect the clocks at ULV segment.

How will Piledriver get its projected 10 to 15 percent speed up? AMD says one-third will come from IPC improvements like structure size increases (so, three to five percent) and two-thirds will come from power optimizations that reduce consumption, enabling higher frequencies at a constant TDP (another six to 10 percent).

Just enough to keep Intel's IPC advantage over AMD offerings, it seems.
A nice speed bump over the SB processors they replace with ~20% TDP reduction. Not bad.

[informal]

Just enough to keep Intel's IPC advantage over AMD offerings, it seems.
A nice speed bump over the SB processors they replace with ~20% TDP reduction. Not bad.

I haven't mentioned AMD in my post at all. Piledriver will bring considerable boost over Llano in ULV and low(er) TDP brackets so AMD will catch up somewhat when CPU side is in question and maintain the lead when GPU is in question. On high end desktop front,things probably won't change much. This 3.5Ghz IB model will be <10% faster than 2600K which for most intel users won't be enough for a switch. For new intel users they will most likely pick this model instead of 2600K if price is not considerably higher. If AMD manages to get 15% perf. boost over 8150 or better yet,8170, then they may end up pretty close to 2600K or 2700K.ST performance will definetely be intel's stronghold for a longer while though,but we are at the point that most users already have plenty enough of ST power even in their Core2 or Phenom II chips so that more would mean nothing to them(I'm in this group too).

[dartaz]

Yep,IB is ~5.7% faster than SB at the same clock(1.089x3.4/3.5) according to these tests from the unnamed guy. Not bad at all, but nothing special either. More important are the power savings and how they will affect the clocks at ULV segment.

The problem with your calculation is that you are assuming that performance is going to scale linearly with clock speed in all benchmarks tested

i7 2700 has 3% higher clock speed than i7 2600, but in a lot of benchmarks the is advantage is going to be less 3%

Just an example, you can see from this review that i7 2700 is only 2.3% faster than i7 2600 in cinebench
http://www.legitreviews.com/article/1751/8/

[Brice MJ]

If AMD manages to get 15% perf. boost over 8150 or better yet,8170, then they may end up pretty close to 2600K or 2700K.ST performance will definetely be intel's stronghold for a longer while though,but we are at the point that most users already have plenty enough of ST power even in their Core2 or Phenom II chips so that more would mean nothing to them(I'm in this group too).

Seems like that whooping 10-15% (3-5% IPC) boost over Bulldozer will have to be enough to face Haswell derivatives too.
More ST performance (over a CPU with equal nº of cores) means more performance in most applications, you may like it not, but it is still very relevant.

http://images.anandtech.com/doci/549...20PM_575px.png

[informal]

Ok,so the difference between the perfectly scaled result and obtained result,in case of 2700K, is 1.03/1.023=1.006% which is much less than margin of error. If you run any workload more than 2 times on your own machine(without changing any of the HW settings) ,you will see the score will fluctuate a bit.

[sdsdv10]

Those '3-d' transistor won't have any effect on ocing because they only have an advantage at low voltages. Ofcourse the smaller node still has an advantage.

I'll wait to see how shipping silicon works out before accepting that conclusion...

[clayton]

Windows 8 + Ivy Bridge + Optimus = longer battery life!

[aussie-revhead]

Do we have an official thread on this yet or is this it?

[Mumak]

FAKEs.
Both fakes made of SNB-EP.

8 Core IVY ~ 3,3GHz 3,6GHz Turbo, 130W TDP
10 Core IVY ~ 2,8GHz X,XGHz Turbo, TDP