Architecting Fermi: More Than 2x GT200

NVIDIA keeps referring to Fermi as a brand new architecture, while calling GT200 (and RV870) bigger versions of their predecessors with a few added features. Marginalizing the efforts required to build any multi-billion transistor chip is just silly, to an extent all of these GPUs have been significantly redesigned.

At a high level, Fermi doesn't look much different than a bigger GT200. NVIDIA is committed to its scalar architecture for the foreseeable future. In fact, its one op per clock per core philosophy comes from a basic desire to execute single threaded programs as quickly as possible. Remember, these are compute and graphics chips. NVIDIA sees no benefit in building a 16-wide or 5-wide core as the basis of its architectures, although we may see a bit more flexibility at the core level in the future.

Despite the similarities, large parts of the architecture have evolved. The redesign happened at low as the core level. NVIDIA used to call these SPs (Streaming Processors), now they call them CUDA Cores, I’m going to call them cores.

All of the processing done at the core level is now to IEEE spec. That’s IEEE-754 2008 for floating point math (same as RV870/5870) and full 32-bit for integers. In the past 32-bit integer multiplies had to be emulated, the hardware could only do 24-bit integer muls. That silliness is now gone. Fused Multiply Add is also included. The goal was to avoid doing any cheesy tricks to implement math. Everything should be industry standards compliant and give you the results that you’d expect.

Double precision floating point (FP64) performance is improved tremendously. Peak 64-bit FP execution rate is now 1/2 of 32-bit FP, it used to be 1/8 (AMD's is 1/5). Wow.

NVIDIA isn’t disclosing clock speeds yet, so we don’t know exactly what that rate is yet.

In G80 and GT200 NVIDIA grouped eight cores into what it called an SM. With Fermi, you get 32 cores per SM.

The high end single-GPU Fermi configuration will have 16 SMs. That’s fewer SMs than GT200, but more cores. 512 to be exact. Fermi has more than twice the core count of the GeForce GTX 285.

  Fermi GT200 G80
Cores 512 240 128
Memory Interface 384-bit GDDR5 512-bit GDDR3 384-bit GDDR3

 

In addition to the cores, each SM has a Special Function Unit (SFU) used for transcendental math and interpolation. In GT200 this SFU had two pipelines, in Fermi it has four. While NVIDIA increased general math horsepower by 4x per SM, SFU resources only doubled.

The infamous missing MUL has been pulled out of the SFU, we shouldn’t have to quote peak single and dual-issue arithmetic rates any longer for NVIDIA GPUs.

NVIDIA organizes these SMs into TPCs, but the exact hierarchy isn’t being disclosed today. With the launch's Tesla focus we also don't know specific on ROPs, texture filtering or anything else related to 3D graphics. Boo.

A Real Cache Hierarchy

Each SM in GT200 had 16KB of shared memory that could be used by all of the cores. This wasn’t a cache, but rather software managed memory. The application would have to knowingly move data in and out of it. The benefit here is predictability, you always know if something is in shared memory because you put it there. The downside is it doesn’t work so well if the application isn’t very predictable.

Branch heavy applications and many of the general purpose compute applications that NVIDIA is going after need a real cache. So with Fermi at 40nm, NVIDIA gave them a real cache.

Attached to each SM is 64KB of configurable memory. It can be partitioned as 16KB/48KB or 48KB/16KB; one partition is shared memory, the other partition is an L1 cache. The 16KB minimum partition means that applications written for GT200 that require 16KB of shared memory will still work just fine on Fermi. If your app prefers shared memory, it gets 3x the space in Fermi. If your application could really benefit from a cache, Fermi now delivers that as well. GT200 did have an L1 texture cache (one per TPC), but the cache was mostly useless when the GPU ran in compute mode.

The entire chip shares a 768KB L2 cache. The result is a reduced penalty for doing an atomic memory op, Fermi is 5 - 20x faster here than GT200.

A Different Sort of Launch A More Efficient Architecture
Comments Locked

415 Comments

View All Comments

  • SiliconDoc - Thursday, October 1, 2009 - link

    The R600 was great, you idiot.
    Of course, when hating nvidia is your real gig, I don't expect you to do anything but be parrot off someone else's text and get the idea wrong, get the repeating incorrect.
    -
    The R600 was and is great, and has held up a long time, like the G80. Of course if you actually had a clue, you'd know that, and be aware that you refuted your own attempt at a counterpoint, since the R600 was "great on paper" and also "in gaming machines".

    It's a lot of fun when so many fools self-proof it trying to do anything other than scream lunatic.

    Great job, you put down a really good ATI card, and slapped yourself and your point, doing it. It's pathetic, but I can;t claim it's not SOP, so you have plenty of company.

  • papapapapapapapababy - Wednesday, September 30, 2009 - link

    because both ms and sony are copying nintendo...

    that means, next consoles > minuscule speed bump, low price and (lame) motion control attached. All this tech is useless with no real killer ap EXCLUSIVE FOR THE PC! But hey who cares, lets play PONG at 900 fps !
  • Lonyo - Wednesday, September 30, 2009 - link

    Did you even read the article?
    The point of this tech is to move away from games, so the killer app for it won't be games, but HPC programs.
  • SiliconDoc - Thursday, October 1, 2009 - link

    I think the point is - the last GT200 was ALSO TESLA -- and so of course...
    It's the SECOND TIME the red roosters can cluck and cluck and cluck "it won't be any good" , and "it's not for gaming".
    LOL
    Wrong before, wrong again, but never able to learn from their mistakes, the barnyard animals.
  • Zingam - Thursday, October 1, 2009 - link

    Last time I bought the most expensive GPU available was Riva TNT!
    Sorry but even if they offer this for gamers I won't be able to buy it. It is high above my budget.

    I'd buy based on quality/price/features! And not based on who has the better card on paper in year 20xx.
  • SiliconDoc - Thursday, October 1, 2009 - link

    Well, for that, I am sorry in a sense, but on the other hand find it hard to believe, depending upon your location in the world.
    Better luck if you're stuck in a bad place, and good luck on keeping your internet connection in that case.
  • ClownPuncher - Thursday, October 1, 2009 - link

    Or maybe he has other priorities besides being an asshole.
  • SiliconDoc - Thursday, October 1, 2009 - link

    Being unable, and choosing not to, are two different things.

    And generally speaking ati users are unable, and therefore cannot choose to, because they sit on that thing you talk about being.

    Now that's how you knockout a clown.
  • Lord 666 - Wednesday, September 30, 2009 - link

    That actually just made my day; seeing a VP of Marketing speak their mind.
  • Cybersciver - Friday, October 2, 2009 - link

    Yeah, that was cool.
    Don't know about you guys, but my interest in GPU's is gaming @ 1920X1200. From that pov it looks like Nvidia's about to crack a coconut with a ten-ton press.
    My 280 runs just about everything flat-out (except Crysis naturally)and the 5850 beats it. So why spend more? Most everything's a consul port these days and they aren't slated for an upgrade till 2012, least last I heard.
    Boo hoo.
    Guess that's why multiple-screen gaming strating to be pushed.
    No way Jose.

Log in

Don't have an account? Sign up now