117 Gaming Benchmarks With NVIDIA GTX 1660 Ti / RTX 2060 vs. AMD RX 590 / RX Vega 56

[GruenSein]

60 fps is 16 milliseconds, not microseconds.

You are correct. That is what I meant.

[stingray454]

And even if it only lasts a few microseconds, it can be significant to the frame rate (60fps = 16us per frame in total).

60 fps is 16 milliseconds, not microseconds.

[humbug]

Just a few years ago, it was inconceivable that an open source GPU driver would perform on par or better than a proprietary one. This is a great time to be an AMD Linux gamer, for sure!

Yep, at the time I used to argue with lots of people on here that we should not settle for sub par performance on open source software just because it is open source.
Generally the driver that has the most amount of quality manhours put into it will perform better, regardless of whether it is open source or closed. But people had a different perception, they thought that the Nvidia performance is out of reach.

When I purchased my R9 290 like 4 years ago it was really bad on Linux. Open source was too slow for gaming and proprietary had bad 2d performance issues on the desktop. Now it really rocks on MESA although I still have to mess with the kernel to force it to use the newer kernel driver.

torsionbar28 - Well Intel GPU driver was quite good. The main problem was that the hardware wasn't high performance, they just made iGPUs. With the upcoming Intel Xe cards, we can see how good the Intel driver stack really is, against other drivers.

[torsionbar28]

Nice work by the AMD team! The performance is pretty competitive when compared to the NVIDIA GPUs.

Just a few years ago, it was inconceivable that an open source GPU driver would perform on par or better than a proprietary one. This is a great time to be an AMD Linux gamer, for sure!

[dimko]

Thank you Michael.

I wsih I was employed, so I could buy one of those cards and ttip you too!

Nice work by the AMD team! The performance is pretty competitive when compared to the NVIDIA GPUs.

[GruenSein]

btw...not sure if this has been done on linux recently - is there any investigation on how cpu heavy the Drivers are compared against each other. AMDVLK, RADV ; Nvidia, nouveau etc?

Sometimes, there are CPU usage charts. Unfortunately, I doubt their usefulness in determining CPU bottlenecking since their sampling rate appears to be relatively small. This means, you get some kind of average CPU usage over the last sample. However, a very short peak in CPU usage to 100% of a single core for a few microseconds is the only time when a better performing CPU would help. Otherwise you'd have enough computational reserves. And even if it only lasts a few microseconds, it can be significant to the frame rate (60fps = 16us per frame in total). Do detect such events, the CPU usage graphs appear to coarse to me.

[CochainComplex]

btw...not sure if this has been done on linux recently - is there any investigation on how cpu heavy the Drivers are compared against each other. AMDVLK, RADV ; Nvidia, nouveau etc?

[valici]

Wow 117 benchmarks. Nice.
Th 590 is looking good, but Vega 56 should be a little bit better. I wonder what s missing.