The Sandy Bridge Preview
by Anand Lal Shimpi on August 27, 2010 2:38 PM ESTA New Architecture
This is a first. Usually when we go into these performance previews we’re aware of the architecture we’re reviewing, all we’re missing are the intimate details of how well it performs. This was the case for Conroe, Nehalem and Lynnfield (we sat Westmere out until final hardware was ready). Sandy Bridge, is a different story entirely.
Here’s what we do know.
Sandy Bridge is a 32nm CPU with an on-die GPU. While Clarkdale/Arrandale have a 45nm GPU on package, Sandy Bridge moves the GPU transistors on die. Not only is the GPU on die but it shares the L3 cache of the CPU.
There are two different GPU configurations, referred to internally as 1 core or 2 cores. A single GPU core in this case refers to 6 EUs, Intel’s graphics processor equivalent (NVIDIA would call them CUDA cores). Sandy Bridge will be offered in configurations with 6 or 12 EUs.
While the numbers may not sound like much, the Sandy Bridge GPU is significantly redesigned compared to what’s out currently. Intel already announced a ~2x performance improvement compared to Clarkdale/Arrandale, and I can say that after testing Sandy Bridge Intel has been able to achieve at least that.
Both the CPU and GPU on SB will be able to turbo independently of one another. If you’re playing a game that uses more GPU than CPU, the CPU may run at stock speed (or lower) and the GPU can use the additional thermal headroom to clock up. The same applies in reverse if you’re running something computationally intensive.
On the CPU side little is known about the execution pipeline. Sandy Bridge enables support for AVX instructions, just like Bulldozer. The CPU will also have dedicated hardware video transcoding hardware to fend off advances by GPUs in the transcoding space.
Caches remain mostly unchanged. The L1 cache is still 64KB (32KB instruction + 32KB data) and the L2 is still a low latency 256KB. I measured both as still 4 and 10 cycles respectively. The L3 cache has changed however.
Only the Core i7 2600 has an 8MB L3 cache, the 2400, 2500 and 2600 have a 6MB L3 and the 2100 has a 3MB L3. The L3 size should matter more with Sandy Bridge due to the fact that it’s shared by the GPU in those cases where the integrated graphics is active. I am a bit puzzled why Intel strayed from the steadfast 2MB L3 per core Nehalem’s lead architect wanted to commit to. I guess I’ll find out more from him at IDF :)
The other change appears to either be L3 cache latency or prefetcher aggressiveness, or both. Although most third party tools don’t accurately measure L3 latency they can usually give you a rough idea of latency changes between similar architectures. In this case I turned to cachemem which reported Sandy Bridge’s L3 latency as 26 cycles, down from ~35 in Lynnfield (Lynnfield’s actual L3 latency is 42 clocks).
As I mentioned before, I’m not sure whether this is the result of a lower latency L3 cache or more aggressive prefetchers, or both. I had limited time with the system and was unfortunately unable to do much more.
And that’s about it. I can fit everything I know about Sandy Bridge onto a single page and even then it’s not telling us much. We’ll certainly find out more at IDF next month. What I will say is this: Sandy Bridge is not a minor update. As you’ll soon see, the performance improvements the CPU will offer across the board will make most anyone want to upgrade.
Facebook
Twitter
RSS
200 Comments
View All Comments
Anand Lal Shimpi - Friday, August 27, 2010 - link
I don't believe any of these apps have AVX support, they're all too old for that.Take care,
Anand
ESetter - Friday, August 27, 2010 - link
Thank you for the quick answer. It would be great to include some software with AVX support in the full review, when Sandy Bridge launches. Probably the Intel Math Kernel Library will be updated in time.darckhart - Friday, August 27, 2010 - link
1. i'd like to see some temp numbers. along with, does intel stock hsf actually do the job here? (which they have been getting better at really)2. i didnt see anything about accelerated hd video playback using the on die gpu?
3. sure these cpu look great from price point performance gain....until you realize you need a full platform upgrade to go along with it...which if we assume mainstream mobo around the 100$ mark and ram to match since they're taking away the bclk deal... and every 2 yrs is a bit too soon for full platform upgrade imo.
4. hardware virtualization parts? i know the current i3 vs i5/7 chips had some stuff disabled. will these SB chips follow the same profile?
5. mobile versions? we know the mobile ones are usually cut back to fit low tdp profile. will the same cuts apply like the current mobile i3/i5 parts (eg, no real quad core parts)? otoh, what about the quad core mobiles? the current i7 mobile quads are laughable at their performance and heat output (i'm looking at you first gen hp envy). do you think these SB quad mobiles will actually be decent?
DanNeely - Friday, August 27, 2010 - link
Wikipedia lists both 2 and 4 core mobile parts. Not definative but they generally do a good job of keeping up with the latest leaks for things like this.http://en.wikipedia.org/wiki/Sandy_Bridge_(microar...
hamitaltintop22 - Friday, August 27, 2010 - link
I hope there is a price drop for the i5 750 to around $150 when this comes out or i7 920 to $200 (no microcenter here).DesktopMan - Friday, August 27, 2010 - link
I'm not sure about this, but I seem to recall having read that aes-ni instructions use the GPU, at least partially. Makes sense as the gpu is excellent at parallel tasks. If this is the case, would the 6 EU part perform differently than the 12 EU part at AES?Any news on when the inevitable Q67 would launch? I guess it's likely that Q67 will use AMT 6.0 as it was a pretty recent upgrade.
With sata III support at launch you'd imagine they'd also support sata III on their gen 3 SSDs. Time will tell I guess.
overclocking101 - Friday, August 27, 2010 - link
wow bummer. welcome to the end of intel Bus speed overclocking. I will not be adapting the new sockets unless something happens and intel changes their minds. overclocking is not as easy as switching multiplyers even EE cpu's of nowadays show that. 90% of the high overclocks with EE cpus show that a mixture of multi and bus speed is needed. i sense though that with the higher end socket intel will allow it. if not i think its a very bad move on their part.starfalcon - Friday, August 27, 2010 - link
I don't think any of the sandy bridge graphics will be able to get to GT 240 levels.This one trades blows with the 5450 as we can see, and just looking at 3DMark06 scores the 5450 scores about 3500 or so, while the GT 240 does maybe 9000 or 10000.
If the more powerful sandy bridge graphics can get up to 4000 or 5000 or so that would be great, that would be beating the 9400 GT and closing in on the 9500 GT, not getting to GT 240 levels though. Wonder what the next integrated graphics after this will be like.
TETRONG - Friday, August 27, 2010 - link
I take it this means it will soon be the optimum time to purchase current-gen technology at significantly reduced prices?Just wanting to build a no nonsense system at slightly below the current price/performance sweet-spot.
Seems Intel are only interested in toying with consumers.
They've wasted die space that could've been used for a more capable CPU. How many years have we been chained under 4Ghz frequency? 5 years or so?
Nine women can't make a baby in one month! Not every problem is parallelizable - we need greater frequencies/efficiencies.
Now they are locking processors and playing games with the sockets. No USB 3.0!!?
Garbage, No Thanks!!!
Seems you are giving them a free pass Anand. Very convenient timing to steal AMD's thunder, eh!
I love you man - big fan since the beginning, but you should read Scott Wasson over at Tech Report. Those value scatterplots are very helpful to me - these regurgitated press releases, not so much.
Sorry;(
To be so harsh, but we deserve better than these kiddie chips!
Only you can hold them accountable for these failures of imagination.
wyvernknight - Friday, August 27, 2010 - link
I am a bit disappointed. Seems like since intel is wiping the floor with AMD, decided it was OK to screw us all over with this socket thing. I will still buy an intel processor if AMD has no cards to play, but i wont be pleased.