A 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.

A New Name A New Socket and New Chipsets
Comments Locked

200 Comments

View All Comments

  • Anand Lal Shimpi - Friday, August 27, 2010 - link

    You're correct, I didn't feel a dual vs. quad-core comparison was fair which is why I focused on the 760. I'll clear up the text though :)

    Take care,
    Anand
  • Anand Lal Shimpi - Friday, August 27, 2010 - link

    fixed :)
  • mastrdrver - Friday, August 27, 2010 - link

    If we go with what Anand has said and use the roadmap to guess pricing I just have one question then:

    Why in the world would anyone spend ~$300 for the 2500 and ~$500 on the 2600 then use the on chip gpu with no plans on some kind of discrete?

    If the difference between a $600 HP is Llano and Sandy Bridge, Llano has a possibly huge advantage since I think its safe to assume that the gpu side will start at 5450 performance.

    Its like Intel would be trying to tell you that SD Xbox 360 is better than HD Xbox 360 (Llano). Are you serious? If Llano can hit a pc at that price point and have a full shader count, Sandy Bridge is dead in the consumer market.

    I know that's a lot of ifs and time between here and then but Intel doing what it has always done with graphics (suck) is going to haunt it. I think Intel let the door wide open and its head between it and the frame. All AMD has to do is shut it.
  • DanNeely - Friday, August 27, 2010 - link

    There are people whose workloads are heavily CPU bound but who don't need a heavy duty GPU. Higher end servers and a lot of workstations fall into this category.

    Beyond that unless Intel made a GPUless die or deliberately disabled the onboard GPU there's no reason not to include it. While we'll have to wait until Intel shows off labeled die shots I doubt that the GPU is a large enough chunk to justify the engineering effort just to save a little on the manufacturing side.
  • mastrdrver - Saturday, August 28, 2010 - link

    You are correct but my point was meant to be on "Best Buy" systems and not server or workstations. Sorry if I didn't get that clear.

    On the server front this will have to go up against Bulldozer which is an entirely different topic.

    While it would be foolish for Intel to make a gpuless die since integration with the cpu side is inevitable, Larabee or what ever better be good. Then there is the driver thing. That Dragon Age Origin picture sure doesn't look right. For drivers that still have work to do, that picture looks exactly like the one from when Clarkdale was released. I'd be a little surprised if much driver work is left if those two pictures are actually different.
  • arh2o - Friday, August 27, 2010 - link

    How much will these new 1155 motherboard prices be? Will they be in the same price range as the current 1156 motherboards?
  • Rajinder Gill - Saturday, August 28, 2010 - link

    I'd imagine the P67 boards will be priced between $150~$200.
  • odin607 - Friday, August 27, 2010 - link

    What about temps =(
  • cmdrdredd - Friday, August 27, 2010 - link

    I'm never buying an Intel CPU or Motherboard ever again. This is one area that made them what they are today. The ability to take a mid range part and clock it up is what made the Core 2 series such a success with gamers and other performance enthusiasts. Not all of the success is attributed to overclocking, but a good bit of the popularity came from a $200 CPU being able to clock up to levels that the $700+ cpus hit. Now, if the unlocked parts can hit big overclocks and aren't overpriced then maybe it'll work out. However, it's all to easy for Intel to give us the finger and price a $200 CPU at $600 because it's unlocked and say "tough crap, if you want to overclock then pay up!". I am hopeful it doesn't come to this.

    Anyway, quads are old news IMO...I'm looking at 6core for my next one.
  • JumpingJack - Saturday, August 28, 2010 - link

    "but a good bit of the popularity came from a $200 CPU being able to clock up" ...

    Reading the preview, it looks like the 2500K may fit this description.

    From the articles:
    "...and the 2500K will replace the i5 760/655K ($205 - $216). ..."

    Even the 875K, when it launched, wasn't as you claim... it actually came in 200 bucks cheaper than the 870.

    http://techreport.com/articles.x/18988

    It would seem to me that Intel has been planning this change for sometime and went out to address this....

Log in

Don't have an account? Sign up now