The Best, Most Efficient Graphics Cards For 1080p Linux Gamers

[giacomo160790]

Hello Michael, congratulations for the article, I am very interested (if you will) to review 4k

A question, the games dying light, serious sam and X rebirth not allow the benchmark mode?

Why they are the only ones I've played on linux... All in 4k with a R9 290 seemed friedly linux

Looking at the counter fps steam into play I was about this performance

Dying light @ 3840x2160 and low detail 27-40fps
X Rebirth @ 3840x2160 and high details 28-30fps
serious sam 3 BFE @ 3840x2160 and everything to maximum 70-80fps

I have a i5-4440 - R9 290 - 8Gb ram - AOC U2868PQU 4k @ 60hz, Linux arch kernel 4.1.6, fglrx 15.7

Thanks again to Michael, good evening to all

[Michael]

I would love to see a dual bar for proprietary and floss driver for each card in one chart. While not the purpose of the of the article for card buying reasons it might matter for some. I am excited to see these tests. All you need to do is put some amazon links for the cards and should be a good ethical revenue stream.

Yeah I would love to see that also !!! I think I would finally subscribe w/ that feature b/c I see to always be looking for that also.

Great work on this article !! I really hope AMD picks up their game buy the time I finally build myself a new PC.

Here you go! http://www.phoronix.com/scan.php?pag...-cat-wow&num=2

You can subscribe @ http://www.phoronix.com/scan.php?page=phoronix_premium

Or if you're talking about ALWAYS seeing a open vs. closed bar in each graph/article, that's not possible due to frequently running new hardware/software combinations in the time required to run the tests, etc. Plus that like in this article, some of those tests don't yet run properly on RadeonSI, there is the R9 Fury lack of re-clocking, etc.

[Michael]

A question, the games dying light, serious sam and X rebirth not allow the benchmark mode?

Complain to the developers that they should allow for properly automated benchmarking and if they have questions to contact me.

[Deavir]

Here you go! http://www.phoronix.com/scan.php?pag...-cat-wow&num=2

You can subscribe @ http://www.phoronix.com/scan.php?page=phoronix_premium

Or if you're talking about ALWAYS seeing a open vs. closed bar in each graph/article, that's not possible due to frequently running new hardware/software combinations in the time required to run the tests, etc. Plus that like in this article, some of those tests don't yet run properly on RadeonSI, there is the R9 Fury lack of re-clocking, etc.

Had to check to see if I was still a member (I am). Yeah like I said not really the purpose of this article. I was thinking more like a linux buyers guide type graph.

[Michael]

Had to check to see if I was still a member (I am). Yeah like I said not really the purpose of this article. I was thinking more like a linux buyers guide type graph.

How so? The article earlier in the week had the open source results. But those tests were somewhat different based on driver requirements, the gtx 900 series doesn't have hardware acceleration, the r9 285 / r9 290 are crippled until PM/reclocking, etc. so really trying to understand what's requested versus what has already been done. The least common denominator of the few cards that work well on the few tests between the two drivers?

[jmsq]

This article really should have included open source radeon driver results as well, at least in the benchmarks where it works. I've been testing catalyst against radeonsi a bunch lately (290x) and the number of games that radeonsi performs very well on far outstrips the number of games with significant speed gains from catalyst. I know it's not counted here but radeonsi crushes catalyst (and even nividia on a 970) on wine games such as Starcraft 2 at 4K.

[Deavir]

How so? The article earlier in the week had the open source results. But those tests were somewhat different based on driver requirements, the gtx 900 series doesn't have hardware acceleration, the r9 285 / r9 290 are crippled until PM/reclocking, etc. so really trying to understand what's requested versus what has already been done. The least common denominator of the few cards that work well on the few tests between the two drivers?

Maybe I am thinking of an unrealistic usercase. Lets say I am looking to buy a new videocard. I want to use Floss driver if I can get away with it, if not I will use closed. I play CS GO and with one chart it would be nice to see if I can get away with the floss driver or go closed for AMD and see if it compares to nVidia and by how much. Currently I would have to look for the CS GO benchmark for AMD open vs closed, then look for nVidia vs AMD closed (smallest violin in the world for me right). I don't think I've even seen another site that benchmarks games over many cards as well as you do now, so it is honestly not criticism just something I always thought would be interesting.

[Michael]

Maybe I am thinking of an unrealistic usercase. Lets say I am looking to buy a new videocard. I want to use Floss driver if I can get away with it, if not I will use closed. I play CS GO and with one chart it would be nice to see if I can get away with the floss driver or go closed for AMD and see if it compares to nVidia and by how much. Currently I would have to look for the CS GO benchmark for AMD open vs closed, then look for nVidia vs AMD closed (smallest violin in the world for me right). I don't think I've even seen another site that benchmarks games over many cards as well as you do now, so it is honestly not criticism just something I always thought would be interesting.

So you're really talking about then is like a Radeon HD 6000 ~ Rx 200 series vs. GeForce GTX 600~700 graphics comparison on both drivers but on the Nouveau side just being partially re-clocked. Could extend the Radeon comparison to the R9 Fury / R9 285 but then too extremely clock-limited. For doing a price comparison is hard since it's hard to find even these older cards for sale from like NewEgg/Amazon. But if I can get proof of some new subscribers or donors I'll run such an article, even though it's been all indirectly handled many times over in past comparisons.

[Luke]

To be "the best" it would have to be a card whose firmware is not signed-or which will run unsigned firmware while supporting power management and at least OpenGL2.0 (enough for Cinnamon or GNOME) or preferably 3.3. I will never buy one of those Maxwell cards unless someone cracks the firmware signing requirement, and then only if Nouveau gets automatic reclocking working. By them I'm guessing it would be a $15 card out of a basket at a hamfest-and no longer supported by Nvidia's blob. Hell I pulled over 40fps out of Scorched3d while testing a GTS450 (fermi-no reclocking yet) swapped in place of my HD6750 (80fps same game but at double the clocks) , main purpose of that test though was to see if a kdenlive/movit render to 0 byte file bug was specific to the AMD driver, it was not.

[Jonathan Silverblood]

I always chuckle a bit about how bad these FPS/$ comparisons really are, given that they *always* gets it completely wrong.

Micheal, can you re-do the math on this, but cap every FPS number to the VSync of the screen you're running on (or just simply capping them at 60 FPS or some other arbitrary number) s? those looking at a particular game and wondering what to buy could actually get something that matches their needs, rather than than the artificial "this card is cheaper because it can run the game at 12000 FPS, so has lower $/FPS score"?

I've for a very long time recommended the low-to-midrange cards to clients after asking them what their intended use has been, and they've always been happy with the cheaper solution since it still delivers the FPS required, but do so without costing a buttload of unnecessary money.