Ryzen 7 2700 / Ryzen 7 2700X / Core i7 8700K Linux Gaming Performance With RX Vega 64, GTX 1080 Ti

[Azpegath]

Wow, what is going on at 4K for the Vegas? marek, any idea? Michael, could you test that with the official AMD Vulkan driver as well?

[schmidtbag]

The 8700T/regular/K is a beast of a CPU. Kicks even the previous Extreme Edition, the 6850K. Incredible performance for a great price.

Not really that incredible relatively speaking. When compared to the 8700K, the 6850K is nearly 2 years older, and the max turbo frequency is 900MHz lower. 900MHz for 12 threads is a BIG difference. Clock-per-clock, I wouldn't be surprised if the 6850K is actually faster, due to having an extra 3MB of L3 cache (that, and because Intel hasn't really upgraded their architecture since Haswell). The main benefits of the 8700s are the lower TDP and lower price. And on that note, the pricing just proves how Intel has been gouging us due to lack of competition. I'm not saying the 8700s aren't good or worth getting, but they're a lot less impressive once you clear away the smoke and mirrors.

[humbug]

oh wow, the Ryzen is getting a spanking.

Why is the 8700k beating out the 2700X by so much?

something is wrong with how people are porting games to Linux. On windows it seems that modern games are more multithreaded which allows Ryzen to keep up.

Something happens when porting and it becomes more reliant on single threaded performance. When you use 4K which is a more GPU than CPU bottlenecked scenario the difference goes away.

I dunno maybe it's the same reason every game takes a 20% performance hit when porting, even on Intel CPUs.

[fernie]

For any real world scenario, there is not that much difference (nobody plays 1080p low on high end system). But the price, currently 2700x lists 340€, for 8700k 360€ but you would need a cooler, so thats a 70€ for noctua nh-u14s = 430€. 90€ difference. Considering 2700x beats 8700k in almost everything else, thats not a difficult choice

[VikingGe]

Frecuency is most determinated, most games dont use core quantity cpu as ryzen

I know, but the same applies to Windows. Hardly any game benefits from 8 over 6 cores at all. Yet Intel still doesn't win by as much as they do here.

[leipero]

something is wrong with how people are porting games to Linux. On windows it seems that modern games are more multithreaded which allows Ryzen to keep up.

Something happens when porting and it becomes more reliant on single threaded performance. When you use 4K which is a more GPU than CPU bottlenecked scenario the difference goes away.

I dunno maybe it's the same reason every game takes a 20% performance hit when porting, even on Intel CPUs.

Not really (tho ports do suck), I did not see tests where they tested with low settings, and here, that 60% difference is on low, not on high settings. One could argue that type of test indicates "future gen GPU's" but that really isn't the case. With low settings, the amount of textures and data whole system handles is (should be) much lower, potentially stressing different parts of the system (outside of CPUand GPU), maybe even comunication type of Intel system have benefits, in non-scientific terms "sending brick by brick at 100mph vs. sending brick by brick at 50mph, where one system can send up to 2 bricks at 100mph while other can send 4 at 50mph, making them literally sending at the same speed, however, when you do not have 4 bricks, and waste time on sending one by one, the sending system at 100mph will be twice as fast", this is ofc. analogy that may or may not be true for this situation. At the end, low settings really do not show "real picture", and you can see that on Portal example where both CPU's hit at around 1000+ FPS with slight difference in favor of Intel system (mainly because of clock advantage) at what is probably "maximum setting" (since my ancient PC can push well over 200 FPS average at those settings).

We can test that theory simply by testing Portal on low settings as well, if theory is correct, Intel system should push ahead at least 20% over current difference (we should count in game "engine" difference, where "source engine" was really good in managing CPU resources on AMD hardware, even old k7 systems and latter FX series).

[dungeon]

oh wow, the Ryzen is getting a spanking.

Why is the 8700k beating out the 2700X by so much?

By how much? Sometimes diff is zero and sometimes up to 28%

Similar like Vega64 vs GTX 1080Ti diff. Only that sometimes as i see Vega is faster by 12% than 1080Ti, while sometimes Ti is faster by even up to 40%

Joke aside, drivers needs to improve these worse case and common scenarios, only that it needs to figuring out what worse cases happens to be common

[dungeon]

Most interesting here to me is that Vega64 beats 1080Ti, but only on Xonotic and Tesseract.

So what is a difference there, both uses older GL and are opensource games (else are not, with an exception of Openarena), that even leads me to believe how steam libs might collide again with mesa (to not say llvm)

Else could be that nvidia driver profile these or just porters (Feral, etc...) use something special for nvidia

[Jahimself]

Was waiting for this one. I was expecting r5 2600 to see if there is a difference with windows (2600 is faster than 2700 on windows for gaming)
We have a lot of triple A title now on linux. It would be interesting to bench heavy title such as Hitman and Dawn of War 3 and also to include Dirt Rally.

[Dr. Banana]

Those Dota 2 benchmarks look flawed. May I ask you exactly what your testing methodology (for Dota 2) is?