Page 5 of 6 FirstFirst ... 3456 LastLast
Results 41 to 50 of 51

Thread: AMD Radeon Gallium3D Catches Up To Catalyst For Some Linux Games

  1. #41
    Join Date
    Jan 2014
    Posts
    4

    Default

    I've been thinking about trying Linux on my older PC, which has a HD 4870. Is it worth sticking to an older distribution just so I can use the catalyst driver?
    One game I'd like to test and compare to Windows' performance is Metro Last Light

  2. #42
    Join Date
    Dec 2010
    Location
    MA, USA
    Posts
    1,376

    Default

    Quote Originally Posted by Kivada View Post
    If anyone is anti community here it's you apologists that chased away many former readers of the site. I again offer you the brand check challenge of asking about Phoronix on any other site out there.
    To my knowledge, many of the former readers likely were visitors of phoronix when it was basically just a blog about Michael's linux experiences. The website has obviously changed a lot since, so those former readers you speak of have either got what they wanted, or, they have the same false expectations like you. I've only been to this site since I think 2009, but as far as I'm aware, the premise never changed. Michael may have put a little more time and effort into articles a few years ago, but as far as I'm aware, the content and purpose of the articles was never any different than it is today.

    His reviews are terrible, the only relevant test he's even running is Unigine. Everything else is as crap as a site that runs just crap synthetic benchmarks. It's as stupid as running every test on default settings then wondering where all of those performance enhancements he's reported on for the last few months have gone when he does a review.
    And yet again, you have managed to repeat yourself and ignore the reason behind these tests. Until you wake up and realize why the chosen tests are done and why Michael doesn't choose what you desire, you're never going to stop complaining, so you might as well save yourself the effort and stop making your requests. They're not going to happen no matter how much you personally think they should.

    Games with a built in benchmark aren't randomized, FRAPS is used for end users to grab video while they are gaming and see what their current framerate as they are playing. It is not used for benchmarking. Due to the overhead of FRAPS it isn't even useful for settings tweaking.
    ....what? Plenty of games with built-in benchmarks are randomized..... I'm sure if you took any of the PTS game tests and did a time-based performance graph, you'd get something pretty bumpy (as long as the test isn't CPU limited). While I personally think FRAPS makes a decent (but not great) performance monitor when you aren't recording, YOU are the one going into irrelevancy here since to my knowledge, something like that doesn't exist for linux. I'd like to be proven wrong. But even if it did exist, it'd still be utterly useless in most games. The average game doesn't have an easily repeatable automated scene that uses in-game graphics. Some games could take as much as 10 minutes just for 1 test.

    I would really like to see you spend the time to write ONE article involving all the tests you want to see on phoronix. Prove that you can accurately benchmark everything that you request within a timely manner, with a useful conclusion, and I think you'll have every right to bash the GPU articles on this website. Until you realize the time and effort it takes, you REALLY need to stop whining.

    Perspective buyers don't care about how a half assed modified decades old game that nobody plays does. They want to know what the software they are actually going to run does. We've al seen Larabel's office setup, would it really kill him to hit start once every 5 mins grab the other mouse and click "Start" on the HL2:LC torture test then take a screenshot of the framerate that it spits out at the end, rinse and repeat a few times and plug the numbers into PTS's spreadsheet so it can make him a graph?
    AGAIN, you don't understand the point of the tests. They're not about "perspective buyers" - how have you not understood that yet? How the hell is an article like the one we're posting on supposed to appeal to perspective buyers when it's not even about an individual product? The article is about a milestone in the progression of gallium for AMD GPUs - it has NOTHING to do with buying a new GPU, recommending a GPU, telling users how to tweak their system, and so on. If it was about those things then I would wholeheartedly agree that the article would be a piece of irrelevant crap, plus some additional things that it would need that you didn't mention.

    On a side note, unless it's an average, taking a screenshot of the framerate at the end is a painfully inaccurate and nonrepresentational "benchmark". That's like eating a pie and judging how it tastes based on the edge crust.

    We're Linux users and hardware geeks, most of us that are interested in building our own systems don't give 2 shits about the default settings as we know how to change them to get what we want out of our setups, this is the mindset of the vast majority of people that read hardware review sites. If we where going for the default we'd have bought a fucking Dell.
    Yes, we are linux users and hardware geeks. But I would summarize this particular article as a driver review. Linux graphics are not up to par and never have been. Even Nvidia has it's issues. Benchmarking something for the purpose of letting linux users know what to buy is a stupid campaign for obvious reasons. On the other hand, if that was what the articles were about, Michael would probably get a LOT more free hardware. Anyway, Michael has to use default setups because if he uses something too obscure, he has to cover everything that is obscure (which he doesn't have the time for), so, using a more generic setup is easier for people to compare to.

    You're an idiot if you don't think that Google devs don't check their results against the competition for every single product that they make. If they worked in a complete vacuum they'd have failed as badly as "beer" like Miller, Coors and Bud.
    I figured you were going to mention that - of course Google has to test their competitors, but that wasn't my point. I worked for IBM I've seen Dell, HP, and Sun servers in a room dedicated to IBM. You're a little too arrogant for me to get into detail. As for the "beers", I'm not denying those brands are complete crap, but they're still very successful, so I wouldn't say they failed.

  3. #43
    Join Date
    Apr 2008
    Posts
    65

    Default

    Quote Originally Posted by schmidtbag View Post
    To my knowledge, many of the former readers likely were visitors of phoronix when it was basically just a blog about Michael's linux experiences. The website has obviously changed a lot since, so those former readers you speak of have either got what they wanted, or, they have the same false expectations like you. I've only been to this site since I think 2009, but as far as I'm aware, the premise never changed. Michael may have put a little more time and effort into articles a few years ago, but as far as I'm aware, the content and purpose of the articles was never any different than it is today.
    I'd be happy to sign under this. There's a lot of things I think Michael could be doing better, and as any kind of tech site, you have to read his opinions and statements with a grain of salt. That happens at Anand's and Tom's too, though. You have to read critically, not just consume the article.

    Quote Originally Posted by schmidtbag View Post
    I would really like to see you spend the time to write ONE article involving all the tests you want to see on phoronix. Prove that you can accurately benchmark everything that you request within a timely manner, with a useful conclusion, and I think you'll have every right to bash the GPU articles on this website. Until you realize the time and effort it takes, you REALLY need to stop whining.
    You know, I tried to benchmark laptops in the laptop shop I used to work at in 2008 - Michael might remember when I tried to put together a PTS livecd. It got boring very quickly, it's not a simple task, it's not something you can usually set and forget. I balk at the idea of benchmarking my own hardware, as it's just too time consuming.

    Quote Originally Posted by schmidtbag View Post
    They're not about "perspective buyers" - how have you not understood that yet? How the hell is an article like the one we're posting on supposed to appeal to perspective buyers when it's not even about an individual product?
    Needless to say, you can't buy half the products anymore...


    @ Michael, bridgman: do you have any idea how the R7 260 / R7 260X support is doing? Is it relatively in line with other radeonSI, or is it still a lot worse?

  4. #44

    Default

    Quote Originally Posted by PeterKraus View Post
    @ Michael, bridgman: do you have any idea how the R7 260 / R7 260X support is doing? Is it relatively in line with other radeonSI, or is it still a lot worse?
    Yes, I got a Radeon R7 260X a few days ago to review... It will probably come out next week in an article. RadeonSI on Mesa 10.1-devel + Linux 3.13 kernel performance was unbearably slow even though it reported it was using hardware acceleration and power management was initialized. Will try 3.14 soon. On Catalyst it seemed to work fine though it doesn't look like Dual Graphics work with Kaveri. More details in few days.

  5. #45
    Join Date
    Oct 2007
    Location
    Toronto-ish
    Posts
    7,514

    Default

    I think you need 3.14 to get dpm really working on CI parts... guessing that with 3.13 the 260 is probably showing the same "~20% of Catalyst" performance that SI parts were showing until recently.

  6. #46

    Default

    Quote Originally Posted by schmidtbag View Post
    Said some stuff.
    Send me the gear and I'll write the article. I'm not going to test my old stuff. I'll do Linux vs windows on the same hardware with overclock and underclock/volt results, power draw, noise levels, heat output etc. since those are the kinds of things that allot of people find relevant and will become more relevant with people looking to switch from Windows to SteamOS. If you want I'll even do dry ice and LN2 testing for shits and giggles.

    And yes, the tests of new hardware are always about perspective buyers. People want to know if what they are looking at is worth their money or if they will be better served with another piece of hardware.



    Look it up, the crap beers are becoming allot less popular ever since the recession, people have been moving to craft beers in a big way in the states. Now these days the macrobrew companies are trying to "fake the punk" by making up false microbrews or buying them out and doing everything they can to hide that they are wholly owned subsidiaries and because of this are forced to cut costs and quality.

  7. #47
    Join Date
    Jan 2010
    Location
    Somewhere in Kansas.
    Posts
    279

    Default

    Just a note:
    Feeding trolls doesn't do any good.

    Click on
    Settings -> Ignore list
    Insert the user name and hit "okay"

    And then you won't see the troll anymore.

  8. #48
    Join Date
    Feb 2011
    Location
    Ukraine
    Posts
    142

    Default

    Quote Originally Posted by Michael View Post
    Yes, I got a Radeon R7 260X a few days ago to review... It will probably come out next week in an article. RadeonSI on Mesa 10.1-devel + Linux 3.13 kernel performance was unbearably slow even though it reported it was using hardware acceleration and power management was initialized. Will try 3.14 soon. On Catalyst it seemed to work fine though it doesn't look like Dual Graphics work with Kaveri. More details in few days.
    R7 260X = hd 7790 + overclocking. I made many tests of my 7790, 3.13+mesa git shows impressive performance.
    http://openbenchmarking.org/result/1...SO-1401100SO00

    Unigine valley mesa+ogl 3.3 patch


    Unigine valley windows 8.1 catalyst 13.12

  9. #49
    Join Date
    Jan 2010
    Location
    Somewhere in Kansas.
    Posts
    279

    Default

    That's really interesting. They both had the same min framerate.

  10. #50
    Join Date
    Dec 2007
    Posts
    2,395

    Default

    You can also enable tiling for CIK parts if you are using a recent kernel, mesa, ddx, and libdrm. Just add:
    Option "ColorTiling" "true"
    Option "ColorTiling2D" "true"
    to the device section of your xorg config.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •