Radeon RX 560/570/580 vs. GeForce GTX 1060/1650/1660 Linux Gaming Performance

[nukedathlonman]

Very nice the see the RX 570 join in as it gives me a better performance referance (thats what I have in my desktop).

[higuita]

This kind of tests should also use some other CPUs ... having cheap cards with a high end, expensive cpu is a little lie. Please do the same test with 2 more CPUs, one middle range and one low range CPU (or even 4, 2 CPU from amd, 2 from intel) so we can also check the card performance based on the cpu performance too.
Some games are cpu limited, other are GPU, but combining both may have interesting results and can give more real world results for most people

[skeevy420]

This kind of tests should also use some other CPUs ... having cheap cards with a high end, expensive cpu is a little lie. Please do the same test with 2 more CPUs, one middle range and one low range CPU (or even 4, 2 CPU from amd, 2 from intel) so we can also check the card performance based on the cpu performance too.
Some games are cpu limited, other are GPU, but combining both may have interesting results and can give more real world results for most people

Yeah. Like how my Westmeres don't have AVX support. I'm sure that effects my results.

[castlefox]

This kind of tests should also use some other CPUs ... having cheap cards with a high end, expensive cpu is a little lie. Please do the same test with 2 more CPUs, one middle range and one low range CPU (or even 4, 2 CPU from amd, 2 from intel) so we can also check the card performance based on the cpu performance too.
Some games are cpu limited, other are GPU, but combining both may have interesting results and can give more real world results for most people

I know that you are saying here but I would disagree with this...

The results would have less value since there is more than one variable being tested at one time. Not to mention much more work to be done by ML.

[nomadewolf]

This kind of tests should also use some other CPUs ... having cheap cards with a high end, expensive cpu is a little lie. Please do the same test with 2 more CPUs, one middle range and one low range CPU (or even 4, 2 CPU from amd, 2 from intel) so we can also check the card performance based on the cpu performance too.
Some games are cpu limited, other are GPU, but combining both may have interesting results and can give more real world results for most people

These kind of tests show the true power of each GPU by removing the CPU bottleneck.
Sure, a midrange CPU would be a more real-world representation, but which midrange CPU?... There are so many...
Also, these tests also are done in clean systems without anything else running. Is that a real world representation also?
What programs should be running on the background, then?
The tests are done with the least 'noise' possible, so they are also easy to verify.
But it's up to the buyer to carefully analyse what they actually mean, and the circumstances they were done in...

[skeevy420]

These kind of tests show the true power of each GPU by removing the CPU bottleneck.
Sure, a midrange CPU would be a more real-world representation, but which midrange CPU?... There are so many...
Also, these tests also are done in clean systems without anything else running. Is that a real world representation also?
What programs should be running on the background, then?
The tests are done with the least 'noise' possible, so they are also easy to verify.
But it's up to the buyer to carefully analyse what they actually mean, and the circumstances they were done in...

I'd go with the CPU that's fastest w/o AVX, fastest with AVX1, & fastest with AVX2. That should be good enough to show architecture and/or instruction set bottlenecks.

[Termy]

These kind of tests show the true power of each GPU by removing the CPU bottleneck.
Sure, a midrange CPU would be a more real-world representation, but which midrange CPU?... There are so many...
Also, these tests also are done in clean systems without anything else running. Is that a real world representation also?
What programs should be running on the background, then?
The tests are done with the least 'noise' possible, so they are also easy to verify.
But it's up to the buyer to carefully analyse what they actually mean, and the circumstances they were done in...

generally i agree with you and i don't think that every test should include multiple CPUs. But now and again it would be nice to see how much overhead the drivers produce ^^

[gukin]

Hi. What is your experience with RX 570 in terms of system stability?

I'll answer this one even though I wasn't directly questioned. I have a RX 570 and it is very stable. I was having some unrecoverable GPU hangs while playing some of the Batman games on the GPU on my Ryzen 2400g but when played on my RX 570 I had no problems.

I played Batman Arkham Knight all the way through and had no crashes or hangs and Arkham Knight has been a notoriously buggy game. I've got a PowerColor card that was on sale a few months ago for $140 from Newegg. I'd never really paid any attention to Powercolor but it's a dandy car.

[skeevy420]

Typo:

the RADV Radeon performance was coming in a bit short wit hthe RX 580 losing to the GTX 1060

Nah. That's correct. I have a 580 and it is a high temperature heat exchanger.

"wit" is a proper English word. An example of use is "gettin' jiggy wit it".

[alex79]

Using twice as much power as nVidia is nothing to be happy about.