Normalized Clocks: Separating Architecture & SMs from Clockspeed Increases

While we were doing our SLI benchmarking we got several requests for GTX 580 results with normalized clockspeeds in order to better separate what performance improvements were due to NVIDIA’s architectural changes and enabling the 16th SM, and what changes are due to the 10% higher clocks. So we’ve quickly run a GTX 580 at 2560 with GTX 480 clockspeeds (700Mhz core, 924Mhz memory) in order to capture this data. Games that benefit most from the clockspeed bump are going to be memory bandwidth or ROP limited, while games showing the biggest improvements in spite of the normalized clockspeeds are games that are shader/texture limited or benefit from the texture and/or Z-cull improvements.

We’ll put 2 charts here, one with the actual framerates and a second with all performance numbers normalized to the GTX 480’s performance.

Games showing the lowest improvement in performance with normalized clockspeeds are BattleForge, STALKER, and Civilization V (which is CPU limited anyhow). At the other end are HAWX, DIRT 2, and Metro 2033.

STALKER and BattleForge hold consistent with our theory that games that benefit the least when normalized are ROP or memory bandwidth limited, as both games only see a pickup in performance once we ramp up the clocks. And on the other end HAWX, DIRT 2, and Metro 2033 still benefit from the clockspeed boost on top of their already hefty boost thanks to architectural improvements and the extra SMs. Interestingly Crysis looks to be the paragon game for the average situation, as it benefits some from the arch/SM improvements, but not a ton.

A subset of our compute benchmarks is much more straightforward here; Folding@Home and SmallLuxGPU improve 6% and 7% respectively from the increase in SMs (theoretical improvement, 6.6%), and then after the clockspeed boost reach 15% faster. From this it’s a safe bet that when GF110 reaches Tesla cards that the performance improvement for Telsa won’t be as great as it was for GeForce since the architectural improvements were purely for gaming purposes. On the flip side with so many SMs currently disabled, if NVIDIA can get a 16 SM Tesla out, the performance increase should be massive.

GTX 580 SLI: Setting New Dual-GPU Records
Comments Locked

82 Comments

View All Comments

  • Ryan Smith - Wednesday, November 10, 2010 - link

    Actually I'd like to. However I only have one 5970.
  • chillmelt - Wednesday, November 10, 2010 - link

    Yeah, it always bothers me that 5970 crossfired is never included in benchmarks. It's not like it's impossible to crossfire it to even a 5870 just for the sake of result.
  • the_elvino - Wednesday, November 10, 2010 - link

    Why do you only have one 5970?

    I only read the reviews here for entertainment purposes. I enjoy seeing how the reviewers struggle in their reviews to manage the balance between favoring Nvidia over AMD without showing their bias blatantly for everyone to see. However, they seem to be less and less successful at it...

    You have managed to find two GTX 590s pretty much at launch date but couldn't get hold of a second 5970 which was released a year ago? Really? Well, on the other hand it's understandable since Anand doesn't take any money from any company that it might just be over their budget, remember, this is a high-end part!
  • Minion4Hire - Wednesday, November 10, 2010 - link

    And do you remember that huge article from a while back showing the very poor scalability of mutli-GPU solutions when you get into three- and four-way GPU setups? I'm sure dual 5970s are not included because the performance would be severely unimpressive (not to mention all the screen tearing and the like) for the money you're throwing at it.
  • the_elvino - Wednesday, November 10, 2010 - link

    In an effort to avoid hurting AMD's feelings because the 5970 Crossfire (so you claim) doesn't scale well, it's not included?

    No, Ryan states he wanted to include it but doesn't have one, can't you read?

    Anyway, 5970 Crossfire seems to be doing fine:

    Far Cry 2, 2650x1600:

    5970: 54 fps
    5970 Crossfire: 98 fps

    That's an increase of 81%, so doesn't seem to be as severely unimpressive as you think.

    http://www.guru3d.com/article/radeon-hd-5970-revie...
  • tphillips63 - Wednesday, November 10, 2010 - link

    That has NOTHING to do witht the fact that they did not run HD 5970 CrossfireX, they "only" have one.
    I think AT needs to get another ASAP and test it too and stop ignoring AMD's top of the line soultion.
  • strikeback03 - Wednesday, November 10, 2010 - link

    I'm going to assume this post is in jest and you are not that much of a tool.
  • IceDread - Thursday, November 11, 2010 - link

    Yeah it's really annoying that hd 5970 x2 is not in the tests. After all, it's still after a whole year out there and it's also still the greatest gaming card.
  • Hemi345 - Wednesday, November 10, 2010 - link

    Com'on Ryan... you couldn't have submitted a request to any of AMD's AIB partners for a second 5970 loaner and held off publishing the SLI results for a few days if needed??

    Fair and balanced, my ass.
  • Servando Silva - Wednesday, November 10, 2010 - link

    Would you include GTX 460 SLI results? I think many people is still considering that option. Since it beats GTX 480, It should perform similar to a GTX 580?

    Thanks for the detailed review Ryan.

Log in

Don't have an account? Sign up now