More Sandy Bridge performance numbers (+Rumoured Pricing)

[Mad1723]

yeah i figured it would be only around a 20% increase over now. good for them, sb is more of a revision than a new uarch right?

Sandy Bridge is a new uarch in fact. The i7 in 45nm are the new arch, the Nehalem family, and the 32nm (Westmere if I'm not mistaking) is a die shrink. Sandy Bridge is a new arch, Ivy Bridge will be a shrink, Haswell will be a new arch and so on

[flippin_waffles]

So a new arch and only a 10% performance increase, or averaging only 20% if overclocked with turbo? Wow not too exciting. Ill have to dig up that other post that claimed a 50% performance increase was lackluster for a new arch. 10% plus turbo? Seriously that's it? Was expecting a LOT more by the way some people were talking!

[spursindonesia]

It's pretty good, certainly not world changer ala AMD K8 or Intel Core 2, but a nice step forward in technology.

[Nintendork]

Really, I can't see SB as a new arch (I feel more hype for Haswell). At least not as "core2 / corei". More like core2 E6000 --> E8000 (and their quad brothers). Seems too similar.

Anyone having an i7 Nehalem should just wait to Haswell.

[Iconyu]

Extreme performance without extreme pricing. Sounds good to me.

The price is probably low because Intel have to push AVX as the standard, last time they tried to push their 64bit standard Microsoft told them to piss off. So if they price too high they may end up having to adopt another AMD standard and that would be sour.

Also would I be fair in thinking that AVX optimisations will give a even bigger boost, so it would be like comparing the original P4 benchmarks before the benchmark makers added SSE2?

[Sn0wm@n]

we dont know anything about this new turbo mode, and I really doubt the frequency is much different from clarksfield w/ turbo when stressing all cores..

23/24% faster in int/FP at the same base clock is good either way..

new instruction set on top of that....

Really, I can't see SB as a new arch (I feel more hype for Haswell). At least not as "core2 / corei". More like core2 E6000 --> E8000 (and their quad brothers). Seems too similar.

Anyone having an i7 Nehalem should just wait to Haswell.

core 2 was the new arch ... and all the cpu's after it were all minor tweaks ... haswell is the next new arch from intel ....

[kl0012]

new instruction set on top of that....

No.It's not a new instructions. It's a core optimizations. There is no software yet using AVX (which can bring up to 2x performance improvement).

core 2 was the new arch ... and all the cpu's after it were all minor tweaks ... haswell is the next new arch from intel ....

In many ways its a new architecture. AVX (256-bit vectors) and CPU-GPU fusion (not just a two different cores on the same die but a tight connection through L3 cache using a ring bus) alone are a big step forward.

[kl0012]

The price is probably low because Intel have to push AVX as the standard, last time they tried to push their 64bit standard Microsoft told them to piss off. So if they price too high they may end up having to adopt another AMD standard and that would be sour.

Also would I be fair in thinking that AVX optimisations will give a even bigger boost, so it would be like comparing the original P4 benchmarks before the benchmark makers added SSE2?

AMD already decided to adopt AVX so there is no other AMD standart on the horizon therefore there is no reason for Intel to lowering prices because of AVX.

[terrace215]

anyway, it seems some people here were wrong with their previsions of nearly zero gains in IPC....

*cough* saaya *cough*

These parts will obviously rape llano, whenever AMD manages to get those to market, and the 6- & 8-cores will do a number on BD.

Even with the memory subsystem clearly not tuned right, the scores are impressive. And check out the HT improvements in multi-threading. Guess that extra load port helps a bit.

There are also some issues of Win 7 vs Win XP , different memory sizes & speeds...

[savantu]

So a new arch and only a 10% performance increase, or averaging only 20% if overclocked with turbo? Wow not too exciting. Ill have to dig up that other post that claimed a 50% performance increase was lackluster for a new arch. 10% plus turbo? Seriously that's it? Was expecting a LOT more by the way some people were talking!

So you're comparing a 50% increase in performance ( with 33% more cores ) at socket level to a 23-24% increase at thread level ?

If Intel ships SB EX with 16 cores what will we have compared to Westmere EX ? 60% more cores and 20% more performance per core....hmm 2x the performance ? Where does that leave Interlagos ?

IPC or Turbo boost? There sure is 15-20% combined jump,there is no doubt about that.And that is what the buyer gets which is pretty good.

Well, both of them have Turbo boost. The Nehalem being clocked so low used max step increase. We don't know what Turbo levels are active on the ES SB.

But something else is more interesting : BD was supposed to ship at the same time if not earlier than SB.
-Yet SB was demoed last fall, BD taped out last month.
-We have clear dates and pricing alread for SB, BD is " on track for somewhere in 2011"
-We have accurate and varied benchmark data for SB compared to vague claims of 50% more throughoutput at socket level
-The 50% claims were before they had real silicon it seems, based on projections. The question : is the silicon up to the expectation? The ambiguos nature of AMD's marketing screen offers no answer to this.

[Movieman]

So you're comparing a 50% increase in performance ( with 33% more cores ) at socket level to a 23-24% increase at thread level ?

If Intel ships SB EX with 16 cores what will we have compared to Westmere EX ? 60% more cores and 20% more performance per core....hmm 2x the performance ? Where does that leave Interlagos ?




Well, both of them have Turbo boost. The Nehalem being clocked so low used max step increase. We don't know what Turbo levels are active on the ES SB.

But something else is more interesting : BD was supposed to ship at the same time if not earlier than SB.
-Yet SB was demoed last fall, BD taped out last month.
-We have clear dates and pricing alread for SB, BD is " on track for somewhere in 2011"
-We have accurate and varied benchmark data for SB compared to vague claims of 50% more throughoutput at socket level
-The 50% claims were before they had real silicon it seems, based on projections. The question : is the silicon up to the expectation? The ambiguos nature of AMD's marketing screen offers no answer to this.

C'mon, deep breath huh?
A suggestion from me: Go out, hit the local 7-11 and grab a 6 pack of Mike's Hard Lemonade.. Suck down the first 2-3..Nurse the 4th and then sit back all nice and mellow and just laugh when you see things you disagree with..
Trust me it works, and your blood pressure stays nice and low..

[Kuroimaho]

Another comparison could be done by using a recently published Geekbench result of a 1.6 GHz Sandy Bridge CPU. So I compared it to a Core i7 also running at 1.6 GHz and made following table with overall results and a diagram showing the differences in detail.



So the average performance increase with those CPUs at the same base clock, but with different Turbo Boost implementations, is about 20%.

Was that 1.6G bench done with disabled turbo or with it ?

Considering how little time tick tock left with them especially compared to the belated bulldozer I would be surprised for such a performance jump, would expect that from haswell, or bulldozer with the new arch and shrink together.

[Mumak]

I don't think that's a good performance comparison.
Turbo Mode on SNB will be pretty different (let's say the next level/gen. ) over NHM, will allow to go even higher, etc.. Sorry, can't give out more details..

[savantu]

C'mon, deep breath huh?
A suggestion from me: Go out, hit the local 7-11 and grab a 6 pack of Mike's Hard Lemonade.. Suck down the first 2-3..Nurse the 4th and then sit back all nice and mellow and just laugh when you see things you disagree with..
Trust me it works, and your blood pressure stays nice and low..

Well, I'm still young and have good blood pressure, but I do follow your advice. There's no animosity in it, no sleep lost. Maybe 10 years ago, but now I'm cool. It's pointless to argue over the internet and let if affect you when there are so many things you can do outside.
Last night was the bday of one of my friend's wife and we went out to the local brewery, some heavy food ( steaks, beer sausages,etc ) plus lots of unfiltered beer = double plus good. The guy used to be a die hard greener when younger. Now he moved to the dark side, but still feels deep down for the underdog. Human nature I guess.
The same happened when we were going out with local forums members. When sipping liquor or some beers, things were always cool on the CPU side. THe GPU guys however were far more rabid. Nvidia and ATI fans would eat each other , while we, the Intel/AMD bunch which were GPU agnostic, were laughing from the side.
Unfortunately, time has passed and life has other priorities now. Still trying to follow what's going on in the industry though, at least on CPUs.
*Enough off-topic; back to the thread now.

[Movieman]

Well, I'm still young and have good blood pressure, but I do follow your advice. There's no animosity in it.
Last night was the bday of one of my friend's wife and we went out to the local brewery, some heavy food ( steaks, beer sausages,etc ) plus lots of unfiltered beet = double plus good. The guy used a die hard greener when younger. Now he moved to the dark side, but still feels deep down for the underdog. Human nature I guess.
The same happened when we were going out with local forums members. When sipping liquor or some beers, things were always cool on the CPU side. THe GPU guys however were far more rabid. Nvidia and ATI fans would eat each other , while we, the Intel/AMD bunch which were GPU agnostic, were laughing from the side.
Unfortunately, time has passed and life has other priorities now. Still trying to follow what's going on in the industry though, at least on CPUs.
*Enough off-topic; back to the thread now.

No, more off topic..
Here I was telling you to go get some Mike's hard Lemonade and then remembered that I'd bought a six pack a couple weeks back.
First liquor I've had in months..
oops, first bottle is done, time for number 2..

[nsegative]

So are we getting new core i5s? I thought the socket was one way to death...?

[savantu]

Was that 1.6G bench done with disabled turbo or with it ?

Considering how little time tick tock left with them especially compared to the belated bulldozer I would be surprised for such a performance jump, would expect that from haswell, or bulldozer with the new arch and shrink together.

??

Intel has multiple teams, some focused on uarchs while others on shrinks. This allows them to interleave products and offer a new uarch every 2 years while in between we have a shrink.
Main locations are Hillsboro Oregon ( Netburst, Nehalem, Haswell ), Haifa Israel ( Pentium M, Core, Sandy Bridge ) and Houston, Texas ( Atom ).

Haifa uarch team was working on SB ever since Core ( Core 2 products ) was finished in 2005. So they had plenty of time to refine it. This year they started working on the tock after Haswell's ( done by Oregon ) succesor scheduled for launch in 2015 and 15nm.

[duploxxx]

look at the blue fanboy strain..... rape this, huge increase, amasing you guys, you really should start asking Intel for money to keep up the fanboyism in forums.
so pls now all you stay in this thread and leave your magic .... in here and ready to shoot

hexa-core and octo-core models (high-end client & server 1p/2p) are coming later, more like Q3/2010...

no server chips are not q3/2010 that is few desktop and mobile , servers is q1/2011.

If I look into dresdenboy picture I only see increase into multithreaded performance, single thread is actually no gain accept for some code specific increases which suspect special instruction enhancements.....
And testing done is from very generic programs anyone who would be expecting an avarage 20% gain for the same clock might be very disappointed.

Sandy Bridge is a new uarch in fact. The i7 in 45nm are the new arch, the Nehalem family, and the 32nm (Westmere if I'm not mistaking) is a die shrink. Sandy Bridge is a new arch, Ivy Bridge will be a shrink, Haswell will be a new arch and so on

You can caal it an uarch and intel for sure will call it one since they offered it in there tick-tock strategy but in the end this SB is using the same baseline architecture as Nehalem.

[FlanK3r]

People, be realistic! Never seen in last years +50% better performance. CPUs are not GPUs. Its hard dramaticaly changing performance of CPUs (firstyl we should increase clock, then more cores CPUs with Atlon X2 dualcores, last years simillary architecture on smaller process with adding nstrutions and IMC). I think, some problems is in software at x86 and second, maybe cooperating between CPu and GPU with new programs will the future (maybe Fusion is this idea)

PS:gold K8 year and first x2 Athlons from AMD and next coming Core architecture was shocking, because few people foresaw so better scaling than greatly X2 Dualcores. But remember, this was not 50% more performance! In launch was the best E6700 and a bit later X6800.

[Kuroimaho]

??

Intel has multiple teams, some focused on uarchs while others on shrinks. This allows them to interleave products and offer a new uarch every 2 years while in between we have a shrink.

Sb is an evolution of Neha it's not like bulldozer, so if in 2 years they could increase the performance with 20% I would be really surprised, with the new turbo maybe.

[savantu]

Sb is an evolution of Neha it's not like bulldozer, so if in 2 years they could increase the performance with 20% I would be really surprised, with the new turbo maybe.

Well, both Nehalem and SB are tocks, in Intel's definition enough changes to warrant it being called a new uarch.

It's shares a distant resemblance with Nehalem, but inside just about everything is changes : data paths are wider, buffers are larger, new execution units, different latencies. I wouldn't be surprised they modified the pipeline, making it a bit longer so it can clock higher.

[FlanK3r]

big evolution for consumers was maybe K8 and Core (K10 and K10.5 is evolution K8 as Clark and Nehalem is evolution Core architecture

[YukonTrooper]

~20% performance increase plus a die shrink on the world's already fastest CPU architecture works for me.

[turtletrax]

~20% performance increase plus a die shrink on the world's already fastest CPU architecture works for me.

Boom. Nail on head

[Sam_oslo]

~20% performance increase plus a die shrink on the world's already fastest CPU architecture works for me.

Yep, ~20% increase on CPU is a lot. It doesn't happen every day, that's a 2-years events. I hope we can get the same ~20% on high-end SB when it gets out, then it will be really interesting, but maybe that's too much to ask.

The best part with these CB is the shrink. 32nm is much cooler and power efficient.