Ivy Bridge: Intel goes 22 nm and DirectX 11

Today we can see a preview of Intel’s upcoming Ivy Bridge at Anandtech: http://www.anandtech.com/show/5626/ivy-bridge-preview-core-i7-3770k

Once again, Intel has pulled off the seemingly impossible: they increased the IPC of the Core i7 series. They have also made the turbo more aggressive for even better single-threaded performance (more on that in an upcoming post). The short story is this: you get more performance at lower power consumption, for the same price.

But I am more interested in the graphics side of things. And Intel has something nice in store on the graphics front as well: DirectX 11. Granted, they are rather late in the game, especially since AMD has already released the first DirectX 11.1 GPUs, but for Intel it’s a nice update nonetheless.

The performance is also a nice step up from the IGP in Sandy Bridge (which in itself was a major upgrade). In games it is 20%-50% faster than Sandy Bridge, making it a viable option for low-end gaming. It is still not enough to catch up to AMD’s Llano in graphics performance, but the gap has closed considerably. Since Intel has not dedicated as much silicon to the IGP as AMD has in Llano, it is not at all bad. If Haswell can make another step like Ivy Bridge and Sandy Bridge before it, the gap with AMD may be closed altogether.

But what is most amazing is the DirectCompute performance:

Intel actually manages to outperform AMD’s Llano here. Which is quite a feat for Intel’s first-ever DX11 architecture. It seems that GPGPU is a thing that Intel pulls off quite well. I’m still waiting for the day that Intel decides to build discrete GPUs. They’ve come a long way in recent years, and combined with their manufacturing process, Intel has the potential to offer quite competitive GPUs.

About these ads
This entry was posted in Direct3D, Hardware news and tagged , , , , , , , , , , , , , . Bookmark the permalink.

13 Responses to Ivy Bridge: Intel goes 22 nm and DirectX 11

  1. Sharikou says:

    But the soon to be released Trinity APU will double LLano’s GPU performance, leaving Intel at 50% of AMD again.

    • Scali says:

      Perhaps, but since the CPU performance is a joke, it will have limited appeal.

    • k1net1cs says:

      Yes, let’s just hope Intel has a more efficient CPU architecture and manages to catch-up on GPGPU even when they don’t have a dedicated GPU department from swallowing another company.

      Oh wait.

      • Scali says:

        Yea… trouble with AMD and faster GPUs will be that they will become increasingly CPU-limited. AMD’s GPUs take up a far greater piece of the die than Intel’s, so AMD is currently limited to quadcores (which perform in the same ballpark as Intel’s dualcores).
        Even if Trinity’s GPU will be twice as fast on paper, I doubt it’d work that well in practice. After all, it will be paired with a 4-core bulldozer architecture, which may even be slower than the current Llano CPU. It seems to have CPU-limits written all over it. And integrated GPUs are never as efficient as they seem on paper anyway. Based on the specs, people thought Llano would perform like a Radeon 5570. It doesn’t work that way, since you share resources with the CPU, and system memory is generally not as fast as dedicated GDDR anyway (https://scalibq.wordpress.com/2011/06/15/amd-releases-fusion-a8-series-codename-llano/).

        Intel does not compromise on CPU performance, and GPU performance is getting to the ‘acceptable’ level. Even Crysis is playable.
        Not bad for a CPU where the integrated GPU is mainly designed to keep the cost and power consumption down (more for laptops than for desktops really).

    • Scali says:

      As I predicted, Trinity isn’t that great, because the CPU is holding its GPU back. Gaming performance is not that much better than Llano. It even loses out to Intel’s HD4000 in various games: https://scalibq.wordpress.com/2012/05/15/amd-introduces-trinity-when-is-an-improvement-not-an-improvement/
      50% you say? Lies I say!

  2. nickysn says:

    If Trinity is going to be slower than Llano, then why is AMD claiming it’ll be 25% faster? Best case scenario?

    • Scali says:

      Are you really going to trust AMD’s performance predictions? After what they’ve done with Barcelona (Randy “40% faster” Allen) and Bulldozer (John “IPC will be higher” Fruehe)?

      Let’s look at this logically:
      Llano has a CPU that is basically a slightly improved Phenom II X4, right? Just a few % faster.
      Trinity will have a CPU based on a 2-module/quadcore version of Bulldozer, right?

      Now we know that the FX4100 will get outperformed by the faster Phenom II X4s, and even Llano A8 CPUs (which are clocked considerably lower):
      http://www.legitreviews.com/article/1766/

      So how exactly is Trinity suddenly going to be that much faster?
      They’d need to make the current Bulldozer architecture 25% faster just to become competitive with Llano/Phenom II X4. And then another 25% on top of that? Yea, sure.
      AMD will do that…

    • zorg says:

      Very simple.
      The top Llano is clocked at 3 GHz (no turbo). The top Trinity will be clocked at 3.8 GHz with 4.2 GHz all integer core and 4.5 GHz half core turbo. Even in the mobile market, the Trinity will support up to 1.5 GHz turbo. Especially that mobile Trinity ES I have uses an 1.4 GHz turbo, with 2 GHz stock clock.
      The IGP is awesome. I have tested Shogun 2 in DX11, and a 35 watt mobile Trinity is 4 times faster than a desktop Sandy Bridge i7-2700K. Ofcourse the Intel IGP only able to run the game at DX10.1 mode.

      • Scali says:

        Oh dear, the MHz-myth strikes again!
        Let me point out this review again: http://www.legitreviews.com/article/1766/
        The FX4100 there is clocked at 3.6 GHz and 3.8 GHz turbo, yet it loses out to the Llano at 2.9 GHz, and not just a little.
        Clocking it at 3.8 GHz and a bit more turbo is not going to make a drastic improvement. It may edge out Llano, but it’d still be trailing behind the faster Phenom II X4 chips.

  3. Pingback: AMD introduces Trinity: When is an improvement not an improvement? | Scali's blog

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s