Basemark GPU v1.2 benchmarks with 21 GPUs

Nvidia has always dominated in OpenGL.

 

Nice results, could you please add GTX 1060 6G version on the tests?

I own one and it still one of the most popular cards around.

Best Regards,
Will

 

My Score...

https://powerboard.basemark.com/benchmark-result/714623

 

Score on Ubuntu Studio 19.10 with Ryzen 2700X MSI X570 gaming Plus , 32 GB Corsair Vengeance DDR4 3000 RTX 2060 Super Vulkan 7684, on OpenGL 7886 will look at Win 10 when I go into Windows for some games later

 

On Windows, yeah. AMD's Windows OpenGL drivers have always been medicore. On Linux with properly-ported titles , Nvidia doesn't have such a distinct win. For example, the 5700XT on average is slightly slower than the 1080Ti on Linux (sometimes faster, usually a little slower), but on these charts posted here, the 5700XT is practically an entire performance tier lower.

 

Linux is a whole different bag. I've only known one person IRL that was a die hard Linux user. His nickname was the "Root Goblin" And that was around 17 years ago.

 

Comment from HH conclusion
We might end up using Basemark GPU in graphics card test suite as it gives a really nice and quick comparison between APIs that you can test at home yourself as well.

What the actual frack??
It is clearly a ****** up benchmark and not reflecting real world performance between cards. Looking that AMD cards that are in reality on par with RTX2070 Super, are here losing to RTX 2060!

HH conclusion... Lets use that in our testing!

Mind blown.

 

Where exactly is it not reflecting real world performance? AMD's OpenGL drivers on Windows have been notoriously bad and thus you get scenarios where their cards lose to a 2060. That's real world. Simply google "AMD openGL performance" and literally every thread or even posts by developers like this one:

https://community.amd.com/thread/218932

Is that their performance is awful.

As far as Vulkan and DX12, quickly comparing to techpowerup averages between a few cards and Hilbert's other reviews - it looks be on par with everything else.

So I don't get why this would be a poor benchmark, looks fine.

 

1080 Ti ROCKS

 

Here's my score. Clearly, they have some work to do on their categorization because this is most definitely not a laptop and I could find no place to correct it.



The official test score :

 

^^That's a beast!

 

That may be, but the 5700xt is on par with Vega 64. A bit shocking they have not improved any from last gens cards.

 

Not if the bottleneck is feeding the card itself - which is typically what an API does. As far as I know AMD hasn't made any improvements to it's OpenGL implementation in years - instead focusing on Vulkan/DX12.

That being said I'm not sure it even matters.. I can't think of a single game I play or heard of that's OpenGL only, with the exception of titles on Linux and the AMD driver actually performs really well there. It just seems like AMD stopped supporting OpenGL entirely on Windows. This test among other reports from users by googling basically confirms it.

 

Such a strange benchmark, it runs super fast 3 times for me and then is all jittery on the actual benching portion of the Vulkan test.
Either way, gave it a few runs to try and get some heat into the GPU for realistic clocks.....................how is a ~20 second test supposed to get real numbers?

 

Makes sense. Cant recall any OpenGL game tested in recent memory either. I think this should be made clear whenever this bench is used, because its certainly got a lot of people riled up.

 

It only runs for 20 seconds? That not good.

That means there will be no time for the GPU to heat up... and all cards would run at full speed... which is not realistic for normal applications like games...

And so, any old crappy stock blower fan card would end up equivalent to the very best air/water cooled cards at same clock-speed, which do not down-clock due to overheating after a few minutes.

Whereas the crappy blower card probably would have to down-clock after a few minutes in real games.. anyone who has ever ran an extreme test like Furmark would know this. It may be unrealisticly extreme, but normal game operation does this too just slower.

 

Which makes me ask why use the benchmark when ~90% of of Guru3d forums are gamers?

I agree AMD needs to do a lot of work in a lot of areas. Hopefully with increased $$$ flow, they can address some of there sort comings.

 

Can we get a remake of 3DMark 2001 SE....I miss that benchmark

 

Likely because Guru3D audience is expanding beyond just being a gamers haven. If professional/creative types are seeking informative reviews on consumer cards for professional use why not check out some of the reviews here. Professional software review sites show inexpensive consumer cards as part of their evaluations since from a performance/cost perspective it makes sense. Seeing relative API performance across graphic cards is an interesting statistic from a gamer and developer perspective.