Futuremark's New 3DMark 11: Performance Explored

Last week, Futuremark released the latest version of 3DMark.  We've taken the new benchmark for a spin using CPUs from Intel and AMD as well as GPUs from AMD and NVIDIA. The new version, dubbed 3DMark 11 (in reference to its level of DirectX support, not the upcoming year) includes a fresh set of tests, a game demo sequence, and measures CPU performance rather differently than its predecessor, 3DMark Vantage.

We ran the benchmark through multiple tests, comparing it across AMD and Intel CPUs as well as AMD vs NVIDIA GPUs.

Via:  HotHardware
Comments
realneil 4 years ago

I was wondering how this benchmark was going to evolve and it looks to be worth buying it again for the newer version.

Thanks for the good review and I think that the Jeep is '80+Yellow'.

countcristo 4 years ago

This makes me sad that I cannot run it on my older GPU ~ Radeon 4870.

Awesome review though Dave! I love it and your graphs are organized in descending order (easier to read). MMMM, tasty graphs; it's like a nerd's new toy. Loved how it showed differences between AMD 6 cores and Intel's 4 cores. I'm wondering if this is due to instruction set as well other than like it said "clock for clock" perf.

*needs new pants* (sorry is this a PG forum?) Go ahead and modify/delete if mod sees this.

Joel H 4 years ago

*coughs lightly* ;) 

Dave, of course, approved the content but it was my nimble fingers slaving to the bone. ;). 

Regarding AMD vs Intel performance:  I wanted to evaluate how much CPU performance impacted 3DMark 11 but remember that in a real game, CPU choice doesn't usually have that much impact. For whatever reason, the performance gap is wider than I would've expected and not necessarily of what you'd see in a shipping title. 

countcristo 4 years ago

This is often true, but as new games come out with similar tech/algorithms using DirectCompute/Tessellation/etc. (DX11), then 3DMark11 will earn credibility?  I don't even know if that made sense.

Joel H 4 years ago

One of the things I like about 3DMK11 is that it corrects several issues that prevented Vantage from being as cross-comparative a test as Futuremark might have liked.  It's also one of the only DirectX 11 programs to test a wide range of DX11 capabilities. Most of the games that support DX11 only use it for certain specific tasks and can fall back to DX10/10.1 to handle those same effects if necessary. 

One of the things I dislike about the current Unigine benchmark is that the higher detail levels aren't really DirectX 11 performance tests--they're DX11 tessellation tests. Tessellation is one of the main selling points, it's very important—but benchmarks that start off as balanced and turn into tests of essentially one DX11 feature need to be labeled as such.

The TLDR version: I like DX11 because I think it improves the current state of DX11 benchmarking. I'd still prefer to see premium titles take full advantage of what the API can do.

MSgtGunny 4 years ago

Try testing the AMD CPU on a modern AM3 motherboard.  You could be running into issues from the DDR2 memory being used compared to the ddr3 of the i7 920

inspector 4 years ago

LOL Count... have you read this forum? its no where near PG xD

Nice review Dave. LOL that antec jeep :P

SammyHayabuza 4 years ago

[quote user="countcristo"]

Awesome review though Dave! I love it and your graphs are organized in descending order (easier to read). MMMM, tasty graphs; it's like a nerd's new toy.

[/quote]

[quote user="Inspector"]

Nice review Dave. LOL that antec jeep :P

[/quote]

Stick out tongue

[quote user="Joel H"]

*coughs lightly* ;) 

Dave, of course, approved the content but it was my nimble fingers slaving to the bone. ;). 

 [/quote]

LMFAO!!!

Joel H 4 years ago

I. AM. NOT. RODNEY. DANGERFIELD.

/Wrath

Der Meister 4 years ago

I ran it on my i7 with my 5870 and got a p4500 score... if i remember correctly 

Tbagger01 4 years ago

I just wanted to make a comment about the review. I think people are assuming that everyone knows that an ATI 5970 is dual GPU, but it should never be assumed. It's very important that people know that you are comparing a dual GPU board against a single GPU board, that your results in no way really reflect a true one on one comparison. If I were new to this, I would be thinking that the ATIs are way faster than the current crop of Nvidias, when that simply is not true.

 

I have seen too many hardware sites make a comparison like this without noting that it is a crossfire vs non SLI comparison you are really doing. And before anyone calls me a fanboy, I would make the same complaint if it was a dual GPU Nvidia, vs a single GPU ATI.

 

countcristo 4 years ago

I think the only people that read these sites are the ones that do the research or are generally interested in such sites, articles, reviews, etc.  The people that read these sites usually know what a bus, pci-express, (computer hardware vocab), as well as the "new, hip" thing to have where new technologies come out.  It's a nerdtastic place to be and I think that new people who do venture this site and don't know what new tech is out there would be so lost that they wouldn't even know where to begin.

That being said, if people did have an interest and found such sites, then hopefully they know how to search and have started building computers or learning where to start on building a computer.

Joel H 4 years ago

Tbagger has a good point and I amended the article to include that information. Unfortunately the only cards on hand were 5970s and NV doesn't yet support SLI for 3DMark 11. That's why the comparison looked the way it did.

Edit:  RealNeil - I did some checking on this--the dual GPUs on the 5970 are connected on one board the same way they'd be connected via Crossfire across two physical slots. Scott over at TR refers to this as "Crossfire on a stick" and that's a pretty good explanation. Any evaluation of a 5970 *is* a tacit evaluation of Crossfire, in any context.

realneil 4 years ago

well shut my mouth

Post a Comment
or Register to comment