3D, AMD, Business, Companies, Graphics, Intel

ATI and Nvidia cards for 2009 will be monsters

As the 2008 is drawing to a close, our thoughts are turning towards 2009 and what incredible hardware will come at our doorsteps. Upcoming year will bring a breeze of competitiveness, with AMD and Intel fighting for enthusiasts hearts and minds in the world of CPUs. GPUs will see a tough three-way battle between AMD GPG (ex-ATI), Nvidia and newcomer Intel with its Larrabee cGPU.

hynix_gddr5But one of main building block was launched yesterday, in 2008. Hynix introduced a chip with a friendly and “easily understandable” name: H5GQ1H24AFR. Even though the name looks like something that ENIGMA would encrypt, we’re talking about 128MB (1Gbit) memory chip that operates at the clock of 1.75 GHz in QDR mode, resulting in 7 GigaTransfers per second (7 GT/s or 7 “GHz”). Currently, ATI Radeon 4870 and 4870X2 come with 900 MHz chips that offer 3.6 GT/s, so we’re talking about doubling the memory bandwidth per chip.

This means that a GPU with a 256-bit memory controller would have roughly 219 GB/s of bandwidth, while 512-bit memory controller and these Hynix chips would result almost  A GPU with 256-bit memory controller and 438 GB/s. These numbers are astonishing and quite frankly, will open the doors for higher performance jump than previously imagined.

Best thing of them all: due to new manufacturing process, Hynix 2nd Gen GDDR5 chips at 1.75 GHz works at 1.35V rail, and consumes less power than initial 900 MHz chips (3.6 GT/s ones). Yep, the power consumption will go down, and performance per chip is now doubling. Who says you can’t have “wolves stuffed, and all sheep numbered” as the old Croatian saying go (english version: have your cake and eat it too)?

Now you know. Nvidia’s GT212, or the 40nm shrink of GT200 chips consumes around 25% of power eaten by the original 65nm chip, can have double the bandwidth and GDDR5 memory that eats less power than GDDR3 memory present on GTX280 cards. As far as ATI is concerned, the upcoming RV870 will be in the same boat as Nvidia.

Can you say, 8-pin power connector is going the way of do-do birds? Well, I would say yes, but don’t forget that GPU makers will use these power savings to clock their cards to absolute physical limits.
H1 2009 will see $299 parts that enable 1920×1200 in 16x AA/AF at 120 fps with no sweat.
If you thought that GTX280 and 4870X2 are incredible… well, we haven’t seen anything yet. Now, will the game designers finally follow the path set by Race Driver GRID, Unreal Tournament III, Far Cry 2, Fallout 3 and offer absolutely fantastic gaming experience without constant crying that “hardware isn’t powerful enough”. Or at least, prove that it really isn’t.

P.S. Before you ask.. this is still single-ended GDDR5. Still waiting for that Differential GDDR5 to show up…of course, we need Differential GDDR5-capable memory controllers too.

  • Mass production of those chips won’t start until H1/09 which could be january or may. 😉

    Qimonda already announced 6.4 Gbps GDDR5 back in july. With voltage dropping from 1,5V to 1,2 V and later to 1,0 V.

  • Well, a Hynix bird told me they’re tying the launch of their memory to a launch of a “new graphics part”, since they won the contract for it.

    Is it ATI with RV870 (just like was the case for the 4870 and 4870×2 – at least 1st batch of BBA boards for partners) or nV with GT206 or GT212, only time will tell. I would say ATI, but that would be rumor-mongering.

  • MB

    Either way, I suspect one or both companies will ref clock them at 6250mhz (1562-3mhz QDR). Either the 6.4gbps chips or the 7gbps chips, I imagine both Nv and AGG will have parts that shoot for that nice round 100gbps/200gbps/400gbps (128-bit,256-bit,512-bit).

    That really is quite amazing. A 256-bit bus using these chips having 25% more bandwidth, although that is evolution that’s needed I suppose, as these smaller chips will probably use a smaller bus. Nvidia will probably go to 256-bit with gt212, and I can’t help but wonder if each ATi core will be 128-bit…

  • MB

    *25% more bandwidth than the gtx280 GDDR3 on a 512-bit bus.

  • Going back to 256-bit and 128-bit? Sorry, ain’t going to happen. This industry is not known for going backwards, sorry 😉

    All that can happen is that ATI overcomes packaging issues and adapts 512-bit memory controller that the company introduced and shunned with 2900XT.
    Intel LRB is actually based around the same idea of bi-directional bus, 1024-bit internal and 512-bit external, thus LRB will be 512-bit part. Nvidia will not go back with GT212, that is not much more than a die-shrink of GT200 architecture, perhaps some small adjustments like the “coming-soon” 55nm part (GT206). My $0.02 is that GT212 will be paired with these GDDR5 modules and targets 400GB/s of memory bandwidth.

    400GB/s just might be the ticket for RV870, even though 5870X2 with 512-bit memory controller on a single PCB is close to impossible – there is no room to fit the chips and connections with 1024-bit memory and all the power converters… even though, ATI might pull a nice sueprise there.

    What is good is that the world is splitting in a positive way, and we actually got a class of graphics cards that performs like high-end, but costs just a bit north of $200 mark. This was not the case in the past, so speaking car-wise, we went from Mercedes C-Class, E-Class and S-Class to:

    S-Class: High end: 512-bit or 2×256-bit
    E-Class: Odd one out (448-bit) or 256-bit
    C-Class: 256-bit or 128-bit
    A-Class: 64-bit or 128-bit.

    What is good here is that some C-Class parts compete with E-Class, and some E-Class parts compete with S-Class. We haven’t seen this since GeForce 4 Titanium 4200 and Radeon 9500Pro.

  • Mr puter

    yeah something we all need to look into larger bit memory controller and faster GHZ GPU speed 1.5GHZ and above that need to act like CPU speed? remember thoes days where 1.0GHZ 1.2 GHZ doesn’t cut it for games and HD movies because of the lacking bottleneck or speed transfer feed data that doesn’t fly right into our CPU to GPU to chipset to system memory controller mabe we need to relook at chipset and GPU that need to act like CPU speed.we are still slow.lol