On-package GPU and Graphics Turbo

Arrandale and Clarkdale are two-die packages. There's the 32nm CPU die and next to it is a 45nm DirectX 10 GPU die (no DX11 support until possibly Larrabee).

This isn't Larrabee (yet), it's a direct descendent of the graphics in G45. While G45 was built on a 65nm process, the 'dale graphics is built on a 45nm process.

The smaller transistors enable much higher performance. While G45 had 10 shader cores, the 'dale GPU increases that to 12. A number of performance limiting issues have now been resolved, so we should see much more competitive performance from Intel's graphics.

The memory controller has been moved off of the CPU die and is on the GPU die instead. It's still on-package so you get decently low latencies, but it shouldn't technically be as low as on Lynnfield. This is a temporary problem that fixes itself once the CPU/GPU are on the same die with Sandy Bridge.


Sandy Bridge brings on-die graphics

I've already explained turbo mode quite a bit so I won't rehash it here. The technology basically allows you to run your CPU at the fastest possible frequency regardless of how many cores are active. Westmere has this.

Arrandale will support graphics turbo modes, while Clarkdale won't. Clarkdale graphics is already running as fast as possible regardless of TDP.

If the GPU demand is higher than the CPU demand, the CPU will allocate more of its TDP to the GPU and vice versa.

AES-NI: Encryption/Decryption Acceleration Quad Core Performance From Two Cores?
POST A COMMENT

96 Comments

View All Comments

  • Arbie - Friday, September 25, 2009 - link

    Is this true? Right when apps will begin to really use the quads, Intel is going to stop making new ones?

    That's amazing, but good news because it will give AMD a great opportunity to recover. They can sell a quad against any Intel dual as long as the base clocks have the same first digit. Nobody (statistically) is going figure that even a double-throwdown super-trick over-and-under hyperthreaded handshaking pair will beat four of a kind, across the board. It might be true, but it won't sell at Best Buy.

    Arbie
    Reply
  • Arbie - Friday, September 25, 2009 - link

    Oh - now I see. No new quads until Sandy Bridge. Well, that will at least give AMD some time to make hay. They need it, and we need them. Reply
  • grimpr - Friday, September 25, 2009 - link

    Unfortunately, Clarkdale is NOT the Conroe succesor, despite all the stiring of enthusiasm and feelings about Conroe that our dear Anand likes to provoke. As TAH152 points out, its a half baked product like Lynnfield, everyone thats taking a serious look at the platforms and products of 2011 will clearly see the complete and mature solutions that will arise in that year, cpu and platform wise, Sandy Bridge, the TRUE Conroe successor from Intel and a completely new architecture from AMD.

    Dont bite for just a piece of blue sky.

    Reply
  • gwolfman - Friday, September 25, 2009 - link

    Ooooo, I want! I'll replace my Intel G45FC without even having to think about it! :) Reply
  • gwolfman - Friday, September 25, 2009 - link

    @Anand

    Any news on those 4 SATA + eSATA ports supporting RAID5? I hope so!
    Reply
  • vshin - Friday, September 25, 2009 - link

    The final piece of the puzzle has been revealed and now we can map out the right path to upgrade for the budget enthusiast PC gamer market (<$250 CPU budget, no SLI).

    Clarkdale is not the answer. This is a CPU for the casual non-gaming PC user and the HTPC market. On board graphics and 2 cores do not make this the ideal upgrade choice for the majority of us.

    Bloomfield was the answer. It's still a great CPU on a great chipset but it will cost more than a Lynnfield without a proportional payoff in gaming performance. Yes the X58 platform might be "future-proof" since Gulftown will be a drop-in upgrade. But Gulftown is rumored to start at $1000. This makes X58 a dead-end money-pit. Even if Gulftowns eventually drop in price, by that time Sandy Bridge will be out, along with USB 3.0, PCIe 3.0, and SATA III so you'll want to upgrade your motherboard anyways.

    Lynnfield may be a "crippled" Bloomfield but this is the least expensive option for us until Sandy Bridge. There might be a 32nm version but probably not until mid 2010. In any case, all we need is to build a system that will perform well until around 2011-12 since we'll want to upgrade our motherboards by then anyways. So go with a low budget system for now, perhaps a Lynnfield 750 (or 860 if you can find one for under $250), a non-SLI 1156 mobo (under $120), with 4GB DDR3 1600 ram (<$90). Throw in a decent heatsink and overclock this sucker as high as you can.

    I just want to make clear that I don't hate the Bloomfield. It's still better than Lynnfield but less bang-for-your-buck to the typical budget gamer. I suppose if you can spare the cash (>$250 for CPU) or plan on building an SLI setup, you wouldn't be wrong to go with a Core i7 920, especially if you can pick up a D0 stepping at the local Microcenter for $200. ASRock has a socket-1366 motherboard selling for $170 with SLI/Crossfire, which is probably the best deal you're going to get for this platform. I think decent 6GB DDR3 1600 triple-channel kits can be found for around $120.

    TLDR version:

    If you want to spend <$250 on a CPU, then go with Lynnfield.

    If you can spend >$250 on CPU and want to SLI, fold proteins, or encode video, then go with Bloomfield.
    Reply
  • bobbrb - Friday, September 25, 2009 - link

    Is there any info available on sockets/chipsets spec for future 22nm processors? I've been waiting to build a monster rig for few months now and keep postponing with news of better chips coming out... but I set the deadline for Jan'10 and it looks like Clarkdale will be the winner for me. But... it looks like it will come out with only 2 cores (what's up with that???). So I'll be looking for good mobo with plenty room to grow and 2-core Clarkdale hoping to upgrade the chip once 22nm processors come out - that's if sockets/chipset support will allow. Reply
  • chizow - Friday, September 25, 2009 - link

    Just seems like its going to be really short-lived. It'll capture the low-end, maybe $150 and lower because above that you have Lynnfield. On the high-end, 6-core and $1000 for Gulftown puts it out of reach for even most X58 users, and no 4-core Westmere part means no upgrade path for current Bloomfield owners.

    Then you have Sandy Bridge in late 2010....meaning Westmere will have a shelf-life of about a year? Or will 32nm Westmere low-end parts not see a refresh until Ivy Bridge? Almost seems like Intel is going away from its straight refresh approach and alternating high and low-end parts on their Tick and Tock and coordinating with the process shrinks.

    Better for the consumer I suppose, as it offers 2 year life cycles for their preferred performance bracket, but seemingly worst for Intel's bottom-line. Unless they think their bottom-line is best served pacing their progress due to lack of any real competition from AMD.
    Reply
  • strikeback03 - Friday, September 25, 2009 - link

    Enthusiast irrelevance maybe, but it will probably be huge in laptops (which will probably make up over 50% of processor sales for most if not all of the life of these chips) and will also fill in the cheaper desktops from HP, Dell, etc. So yes, the Westmeres seem to be skipping the midrange market, but I imagine they will still sell plenty of them.

    Too bad, I'd love a 32nm 4 core/8 thread part for under $300, but they took so long releasing Lynnfield after Bloomfield that obviously they are not anxious to turn right around and replace the 45nm i7s with 32nm.
    Reply
  • chizow - Friday, September 25, 2009 - link

    Yep but again, that all hinges on whether or not there's a 32nm Sandy Bridge refresh of this low-end, or not. I'm guessing not. 11-month life span would be short even by Intel's standard. I guess their new plan will be to space out generations further, especially as they venture into greater core plurality. Last I heard 8-core was going to be the baseline core # for Sandy Bridge. Reply

Log in

Don't have an account? Sign up now