NVIDIA's GeForce GTX 580: The SLI Update
by Ryan Smith on November 10, 2010 10:00 AM ESTNormalized 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.
82 Comments
View All Comments
tigersty1e - Wednesday, November 10, 2010 - link
I've seen you guys review high-end pre-built gaming systems before. It's not like you had to give the pre built system high praise. 1 page in the review wouldn't have taken away from anything.piroroadkill - Thursday, November 11, 2010 - link
Come on, people moan at AnandTech for all kinds of reasons. It's best to keep it clean and get a retail sample from elsewhere.erple2 - Friday, November 12, 2010 - link
That's a slippery slope to start going down. Ultimately, it's up to Anandtech to decide what they can/can't will/won't review, not up to the whims of some chipmaker (despite that there are only 2 chipmakers these days).Reviewing a pre-built gaming system takes a lot of time and effort to do properly, and anything less degrades the reputation of the review site, even if it's just to jump through hoops with some snooty vendor. "If you're willing to jump through THIS hoop to do what we want here, can we withhold a new card for you on the next launch unless you give us complete editorial powers on that review?"
Ultimately, the separation between the review site and the vendor is extremely important to help ensure unbiased, thorough reviews (something that I've appreciated from Anandtech over the years) complete with "here's a bunch of numbers, but what does this all mean in context" (aka "analysis"). Part of that is not bowing to a vendor's wishes (outside of a standard and acceptable product launch NDA).
LSWilson - Wednesday, November 10, 2010 - link
You guys get to play with the best toys, I could not be more jealous. Keep up the good work.Oxford Guy - Wednesday, November 10, 2010 - link
In Unigine Heaven, the 480 beats the 580, except at 2560x1600. The difference is very dramatic at 1680x1050 and considerable at 1920x1080.http://techgage.com/reviews/nvidia/geforce_gtx_580...
http://techgage.com/reviews/nvidia/geforce_gtx_580...
http://techgage.com/reviews/nvidia/geforce_gtx_580...
Why is that?
I would like to see this matter addressed, as minimum frame rates can make a big difference in perceived performance.
Frankly, I'd like to see an article called "Minimum frame rate shootout" that would compare all the mid-range through high-end cards.
erple2 - Friday, November 12, 2010 - link
Your initial statement is deeply flawed - the 480 beats the 580 IN MINIMUM FRAME RATES, except at 2560x1600.What I'd like more to know is whether the "blip" in minimum frame rates is a temporary spike/drop, or whether it's endemic. But that's something that the average tells you, too.
Oxford Guy - Friday, November 12, 2010 - link
Take a look at the subject line, smart guy.Oxford Guy - Wednesday, November 10, 2010 - link
I believe the poster was referring to the Asus card. Take into account that Asus did a 10 Mhz "overclock" which itself is rather cheesy.keitaro - Wednesday, November 10, 2010 - link
I'd like to request that y'all start testing Eyefinity and Vision Surround setups and resolutions for the games that supports them. Granted, there is a vastly difference in how it is supported in hardware, it'd be nice if any hardware capable of this can be tested.Hauk - Wednesday, November 10, 2010 - link
With GTX 580, 5970, and the 68xx series as exceptions, I've ran every other config on that GPU load temp chart, and have lowered temps drastically using a combination of a mobo that allows card spacing, a proper case, decent airflow, TIM replacement, and an Afterburner fan profile. I'm not talking jet engine; there's a tolerable middle ground with any card/system.As an example, I have GTX 470s right now loading to 70-73c (both cards) in Crysis & Metro 2033, and GTX 480s loading to 78-81c. I don't game on Furmark so it's irrevelant.
It's a shame to see such high numbers knowing that reviewers have done ZERO to find a tolerable medium. I have respect for what they do; but geez, people around the world see this and take it for gospel. You could at least show that temps can be improved..