GF100’s Gaming Architecture

Although it feels like ages ago, NVIDIA announced the Fermi architecture back in September of 2009, focusing on the compute abilities of the GPU that would be GF100. Today’s announcement is about filling in the blanks – where does the graphics hardware fit in to the design that NVIDIA revealed earlier.

As a quick refresher, the GF100 is composed of 512 SPs, which NVIDIA now calls CUDA cores. Each CUDA core is organized in to a block of 32 cores which we call a Streaming Multiprocessor (SM), which in turn are laid out so that 4 of them are in a Graphics Processing Cluster (GPC). Finally, there are 4 GPCs in a single GF100 chip. 32x4x4 = 512 CUDA cores.

  GF100 GT200x2 (GTX 295) GT200 (GTX 285) G92 (9800+ GTX)
Stream Processors 512 2 x 240 240 128
Texture Address / Filtering 64/256 2 x 80 / 80 80 / 80 64 / 64
ROPs 48 2x 28 32 16
Core Clock ? 576MHz 648MHz 738MHz
Shader Clock ? 1242MHz 1476MHz 1836MHz
Memory Clock ? GDDR5 999MHz (1998MHz data rate) GDDR3 1242MHz (2484MHz data rate) GDDR3 1100MHz (2200MHz data rate) GDDR3
Memory Bus Width 384-bit 2 x 448-bit 512-bit 256-bit
Frame Buffer ? 2 x 896MB 1GB 512MB
Transistor Count 3B 2 x 1.4B 1.4B 754M
Manufacturing Process TSMC 40nm TSMC 55nm TSMC 55nm TSMC 55nm
Price Point $? $500 $400 $150 - 200

When NVIDIA first unveiled GF100, we did not know anything about the ROPs, texture units, or any of the fixed-function graphics units that are customary in a GPU. Today we now have that information, and can discuss where it goes.

While GF100 resembles GT200 in a number of ways as a compute GPU, as a gaming GPU it’s very close to being a complete departure from GT200. The big change here is that the single block of fixed-function hardware as we know it is gone. Virtually every piece of hardware has been split up and moved down a level, and can now be found as part of a GPC or as a SM. NVIDIA’s GF100 block diagram does a good job of showing this.

What’s left of the fixed function hardware is the GigaThread Engine, which is what NVIDIA is calling their scheduler. To understand where our fixed function hardware went, we must meet the newest execution units: The PolyMorph Engine and the Raster Engine.

We’ll start with the Raster Engine. The Raster Engine is the combination of all the raster related hardware that we would have found previously in the fixed function pipeline. It does edge/triangle setup, rasterization, and z-culling in a pipelined manner. Each GPC has its own Raster Engine. NVIDIA has very little to say about the Raster Engine, but very little has changed from GT200 besides the fact that there are now 4 of them (one for each GPC) as opposed to 1 larger block. Each rasterizer can do 8 pixels per clock, for a total of 32 pixels per clock over the entirety of GF100.

The much more important of the two new execution units is the PolyMorph Engine, which is what NVIDIA is calling the execution unit that handles geometry for GF100. The PolyMorph Engine is responsible for handling Vertex Fetch, Tessellation, Viewport Transform, Attribute Setup, and Stream Output. It’s here where the Tessellator is located, one of the biggest changes that DX11 is bringing to GPU design. Unlike the Raster Engine, each SM has a PolyMorph Engine, giving the GF100 16 PolyMorph Engines altogether.

While the PolyMoprh Engine may sound simple in its description, don’t let it fool you. NVIDIA didn’t just move their geometry hardware to a different place, clone it 15 times, and call it a day. This was previously fixed-function hardware where a single unit sat in a pipeline and did its share of the work. By splitting up the fixed-function pipeline like this, NVIDIA in actuality created a lot of work for themselves. Why? Out of order execution.

OoO is something we usually reserve for CPUs, where high-end CPUs are built to execute instructions out of order in order to extract more performance out of them through instruction level parallelism. OoO is very hard to accomplish, because you can only execute certain instructions ahead of other ones while maintaining the correct result for your data. Execute an add instruction that relies on a previous operation before that’s done, and you have problems. GF100 isn’t a full OoO design, so we’re not going to cover OoO in-depth here, but if you’d like to know more please see this article.

At any rate, internally each PolyMorph Engine is still a simple in-order design. NVIDIA hasn’t gone so far as to make a PolyMorph Engine an OoO design – but because there are 16 of them when there used to be just 1, OoO hazards can occur just as they would in an OoO executing CPU. NVIDIA now has to keep track of what each PolyMorph Engine is doing in respect to the other 15, and put the brakes on any of them that get too far ahead in order to maintain the integrity of results.

To resolve the hazards of OoO, GF100 has a private communication channel just for the PolyMorph Engines that allows them to stay together on a task in spite of being spread apart. The fact of the matter is that all of the work that goes in to making a design like this work correctly is an immense amount of effort, and NVIDIA’s engineers are quite proud of this effort. They have taken the monolithic design of prior GPUs’ geometry units, and made it parallel. We can’t overstate how much of an engineering challenge this is.

However all of this work came at a cost, and not just the significant engineering resources NVIDIA threw at GF100. The other cost was time – we believe that the PolyMorph Engine is the single biggest reason that GF100 didn’t make it out last year. It’s the single biggest redesign of any component in GF100, and is something that NVIDIA had to start virtually from scratch on. When NVIDIA told us that designing a big GPU is hard, this is what they had in mind.

Now why did NVIDIA put themselves through all of this? Because in their eyes, they had to. The use of a fixed-function pipeline in their eyes was a poor choice given the geometric complexity that a tessellator would create, and hence the entire pipeline needed to be rebalanced. By moving to the parallel design of the PolyMorph Engine, NVIDIA’s geometry hardware is no longer bound by any limits of the pipelined fixed-function design (such as bottlenecks in one stage of the pipeline), and for better or for worse, they can scale their geometry and raster abilities with the size of the chip. A smaller GF100 derivative will not have as many PolyMorph or Raster units as GF100, and as a result won’t have the same level of performance; G92 derivatives and AMD’s designs both maintain the same fixed function pipeline through all chips, always offering the same level of performance.

Speaking of performance, we’ll get to that in a bit, but for now we’ll finish our discussion of GF100’s graphics units. Each SM has 4 texture units, which have changes somewhat from the GT200. In GT200, the texture units were at the TPC level; here they are per SM. The texture units are also affected by the cache reorganization of GF100; for L1 cache they still have the same 12KB of texture cache per set of 4 texture units, while they all share GF100’s 768KB L2 cache, which is now used by all execution units and not just textures. Each unit can compute 1 texture address and fetch 4 texture samples per clock, a significantly higher texture fetch rate than on the GT200. Furthermore in conjunction with DX11, NVIDIA’s texture units now support DX11’s new compressed texture formats. Finally, texture units are now tied to the shader clock and not the core clock. They run at 1/2 the shader clock speed of GF100.

Last but not least, this brings us to the ROPs. The ROPs have been reorganized, there are now 48 of them in 6 parttions of 8, and a 64bit memory channel serving each partition. The ROPs now share the L2 cache with the rest of GF100, while under GT200 they had their own L2 cache. Each ROP can do 1 regular 32bit pixel per clock, 1 FP16 pixel over 2 clocks, or 1 FP32 pixel over 4 clocks, giving the GF100 the ability to retire 48 regular pixels per clock. The ROPs are clocked together with the L2 cache.

That leaves us on a final note: clocks. The core clock has been virtually done away with on GF100, as almost every unit now operates at or on a fraction of the shader clock. Only the ROPs and L2 cache operate on a different clock, which is best described as what’s left of the core clock. The shader clock now drives the majority of the chip, including the shaders, the texture units, and the new PolyMorph and Raster Engines. Specifically, the texture units, PolyMorph Engine, and Raster Engine all run at 1/2 shader clock (which NVIDIA is tentatively calling the "GPC Clock"), while the L1 cache and the shaders themselves run at the full shader clock. Don’t be surprised if GF100 overclocking is different from GT200 overclocking as a result.

Index Why NVIDIA Is Focused On Geometry
POST A COMMENT

115 Comments

View All Comments

  • marc1000 - Tuesday, January 19, 2010 - link

    hey, Banshee was fine! I had one because by that time the 3dfx api was better than DirectX. But suddenly everything became DX compatible and that was one thing 3dfx GPUs could not do... then I replaced that Banshee with a Radeon 9200, later a Radeon X300 (or something), then Radeon 3850, and now Radeon 5770. I'm always in for the mainstream, not the top of the line, and Nvidia is not paying enough atention to mainstream since Geforce FX series... Reply
  • Zool - Monday, January 18, 2010 - link

    The question is when they will come with mid range variants. The GF100 seems to be 448SP variant and the 512SP card will be only after A4 revision or who knows.
    http://www.semiconductor.net/article/438968-Nvidia...">http://www.semiconductor.net/article/43...en_Calls...
    The interesting part on the article is the graph which shows the exponecial increase in leakage power after 40nm and less. (which of course hurts more if u have a big chip and diferent clocks to maintain)
    They will have even more problems now that dx11 cards will be only gt300 architecture so no rebrand choices for mid range and lower.
    For consumer gf100 will be great if they can buy it somewhere in the future, but nvidia will bleed more on it than the GT200.
    Reply
  • QChronoD - Monday, January 18, 2010 - link

    Maybe I'm missing something, but it seems like PC gaming has lost most of its value in the last few years. I know that you can run games at higher resolutions and probably faster framerates than you can on consoles, but it will end up costing more than all 3 consoles combined to do so. It just seems to have gotten too expensive for the marginal performance advantage.

    That being said, I bet that one of these would really crank through Collatz or GPUGRID.
    Reply
  • GourdFreeMan - Monday, January 18, 2010 - link

    I certainly share that sentiment. The last major graphical showcase we had was Crysis in 2007. There have been nice looking PC exclusive titles (Crysis Warhead, Arma 2, the Stalker franchise) since then, but no significant new IP with new rendering engines to take advantage of new technology.

    If software publishers want our money, they are going to have to do better. Without significant GPGPU applications for the mainstream consumer, GPU manufacturers will eventually suffer as well.
    Reply
  • dukeariochofchaos - Monday, January 18, 2010 - link

    no, i think you're totally correct, from a certain point of view.

    i had the thought that the DX9 support is probably more than enough for console games, and why would developers pump money into DX11 support for a product that generates most of it's profits on consoles?

    obviously, there is some money to be made in the pc game sphere, but is it really enough to drive game developers to sink money into extra quality just for us?

    At least NV has made a product that can be marketed now, and into the future, for design/enterprise solutions. That should help them extract more of the value out of their r&d if there are very few DX11 games for the lifespan of fermi.



    Reply
  • Calin - Monday, January 18, 2010 - link

    If Fermi is working good, NVidia is in a great place for the development of their next GPU - they'll only need to update some things here and there, based mostly on where the card's performance lack (improve this, improve that, reduce this, reduce that). Also, they are in a very good place for making lower-end cards based on Fermi (cut everything in two or four, no need to redesign the previously fixed function blocks).

    As for AMD... their current design is in the works and probably too advanced for big changes, so their real Fermi-killer won't come faster than a year or so (that is, if Fermi proves to be so great a success as NVidia wants it to be).
    Reply
  • toyota - Monday, January 18, 2010 - link

    what I have saved on games this year has more than paid for the difference between the price of a console and my pc. Reply
  • Stas - Tuesday, January 19, 2010 - link

    that ^^^^^^^
    besides, with Steam/D2D/Impulse there is new breath in PC gaming. constant sales on great games, automatic updates, active support, forums full of people, all integrated with virtual community (profiles, chats, etc.). a place to release demos, trailers, etc. I was worried about PC gaming 2-3 years ago, but I'm absolutely confident that it's coming back better than ever.
    Reply
  • deeceefar2 - Monday, January 18, 2010 - link

    Are the screen shots from left 4 dead 2 missing at the end of page 5?

    [quote]
    As a consequence of this change, TMAA’s tendency to have fake geometry on billboards pop in and out of existence is also solved. Here we have a set of screenshots from Left 4 Dead 2 showcasing this in action. The GF100 with TMAA generates softer edges on the vertical bars in this picture, which is what stops the popping from the GT200.
    [/quote]
    Reply
  • Ryan Smith - Monday, January 18, 2010 - link

    Whoops. Fixed. Reply

Log in

Don't have an account? Sign up now