Announcement

Collapse
No announcement yet.

Why I Love AMD and Why You Need To Stop Whining

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #51
    AMD ---a fucking company with fucking products with fucking bugs !

    Originally posted by MWisBest View Post
    I have had a lot of experience with AMD products. I go back to the ATi Radeon X300 SE, which if my memory serves me right was one of the first PCI Express x16 GPUs. I have had a lot of experience with Linux as well, starting a year or two after the X300 SE. More recently, I have had experience with OpenGL programming. The current systems I have with AMD products are:

    Dell Dimension 8400 (with the mentioned ATi Radeon X300 SE)
    HP Pavilion dv6-6135dx (AMD A8-3500M APU which contains AMD Radeon HD 6620G GPU and also has a dedicated AMD Radeon HD 6750M GPU for Switchable and/or Dual Graphics)
    HP Envy m6-1205dx (AMD A10-4600M APU which contains AMD Radeon HD 7660G GPU)

    and I have had a few other AMD laptops. I have helped pick out for others or purchase for myself a total of 6 all-AMD laptops since October 2009 and I haven't had an issue with them nor have the ones I helped pick out.

    The main reason I go with AMD is the price to performance ratio, or "bang for your buck." For example, I picked up that HP Pavilion dv6-6135dx for $699, and that was bought the day it was released. To get an Intel system with comparable performance at that time, I would've had to spend at least double that amount. The system is also completely overclockable (CPU and both GPUs if you know what you're doing), even on Linux with a bit of programming and Linux kernel knowledge (it can't be overclocked via BIOS settings, it is overclocked via software). I'm able to run new release games with 90% of the bells and whistles turned on and have a playable framerate (40+ FPS).

    Now, I'll get into OpenGL and driver stuff. This is the stuff I see a lot of people complaining about around here.

    General Driver Complaints:
    - Doesn't support new X or Linux Kernel releases:
    If your Linux distro is using a new X or Linux Kernel release it is their responsibility to patch the driver to function on them properly. AMD doesn't rush officially supporting these newer releases because they take the time to ensure that there is no drop in driver performance or compatibility or other issues brought up by these new releases. They want to make sure the product they are putting their name on doesn't have those issues. I myself use Kubuntu 13.04 for Linux, and I use the xorg-edgers PPA repository which recently has had the fglrx packages updated quickly after new releases. They have them patched to support the newer Linux kernels and X releases. It works without a hitch on both of my aformentioned AMD laptops.

    - Poor 2D performance:
    This has been 100% resolved since the 12.6 release of the drivers.

    - I HAVE _issue_ AND IT ISN'T FIXED:
    Have you filed a bug report, including proper stack traces? Without doing this, there is no way for AMD to fix the issue.

    - _application_ WORKS WITH NVIDIA BUT NOT AMD:
    Nvidia's implementation of OpenGL is not to proper specifications. Application developers using Nvidia can end up having issues with AMD users because AMD properly implements OpenGL to the actual specifications, so the developer is using this non-standard OpenGL and expecting it to work everywhere. Complain to your developer and/or Nvidia. In my experience with OpenGL programming, I've ran into issues with Nvidia users because their OpenGL implementation doesn't behave properly! I'm forced to then check if a user is using an Nvidia GPU, and if so, having to use hackish workarounds or even disabling certain features for them. I have to do this for Intel as well, but not as often as I have had to in the past. Intel has definitely come a long way with GPUs recently and I applaud their efforts. Competition is good for everybody.

    Specific Driver Complaints:
    - Cannot resume from sleep:
    I have issues resuming from sleep when using the open-source radeon driver. When I install fglrx these issues go away. This has been reproduced on at least 3 AMD systems I've worked with so I have a hard time believing the people mention this are really being truthful or actually using fglrx.


    If you have other issues you'd like to bring up, I'll offer a rebuttal. This is only what I can think of at the time, I just can't stand to read another thread page full of trolling AMD on this forum. Even the articles themselves are biased here!

    I think AMD should work hard to remove the bugs of the driver, I was very disappointed with AMD Video cards
    AMD dirver for linux with a lot of bugs should be fixed!

    Comment


    • #52
      DirectX is dead tho...

      Here's my take on this whole driver issue. AMD is quoted as saying to the effect that "DirectX is dead" and "There will be no DirectX 12". That being said with Mac gaining ground among 'at least' in casual gaming, OpenGL is going to come around. With full force more than likely. All this obsesion with workability and benchmarks when we are, in fact, in a state of regression as far as software utilization goes. AMD has YET to announce their road-map for 2014. Very hush, hush. Which may mean a paradigm in the coming steppings. What now?... among the assembler programers there is gossip that Intel's hyper-threading is going to prove a nightmare in scale, because of the way is mishandles bits, there is ERRATTA as we descend in lithography. AMD HANDLED IT RIGHT in implementing ONE (1) thread per core. This will boad well as we ascend (descend?) further towards 8nm? Even further? What now futurist? What happens when the electron is too big?? I'll tell you what... we're going to be carving out resonance channels!!! That means we get the MUCK out of this digital nightmare and back to where we belong... ANALOG computing! Were we can truly describe the psychical universe mathematically. C++ WILLl NOT SAVE YOU, lol! We've been stepping (digitally) to the temple, only to see the mountain (analog) behind it. It's gonna change ALOT of what we perceive as possible or optimal .... in the coming decades. (30+yrs)? Mimzy?

      So with my prior argument that hyper-threading being bunk, let's look at real potential before we criticize.

      Disable it & build your OS from source. Spend a year on it. GNU is dough, pi crust, and mold it to your micro-architecture (-march=native)... and that's money. (Hurd that!) So let your bench-marked fortran and linear algebra dictate your limits and inconsistencies in the family, step, & fab. Then cook it in. Compile glibc, gcc, binutils ad-nausium. Then repeat. Know the kernel, it will teach your system as you compile. Python is amazing. I've found a whole new outlook, and I'm not a coder. I'm building my OS today to port it over to my machines tomorrow.

      So that being said, with debug & tracing, who will produce a superior OS? Intel or AMD? And 'NO' I don't believe speed is superior... in states of regression you need to scale and transcend. Hell, Intel even released IA64 as a freaking RISC processor (*epic*face palm*) and we're trusting them to lead the way? Not that I hate intel by any-means... but they milked the P4 and we are still paying for it to this day. Soooo.... as we keep looking to the here and now, especially with GPU, we're are still dealing with ATI legacy in radeon and catalyst.

      So I say wait, exciting things are in the -pipe !

      Comment


      • #53
        Originally posted by Vim_User View Post
        So you are telling me that they have not released a driver that supports anything newer than a 9 months old kernel (I don't count 3.5 here, since it is already EOL) because they don't rush things? You must be kidding.
        You are forgetting that a business developing software and going through the proper SDLC processes takes time. Open Source developers can drop a new release every other day, and deal with bugs as they arise. That's not how the world of commercial software works. After all, Red Hat Enterprise Linux 6.4, the very latest and greatest release, used by corporations around the world for back-end business processing, is still on the 2.6 kernel. Do a little homework into how commercial software development works and you may learn something.

        Comment


        • #54
          Originally posted by torsionbar28 View Post
          You are forgetting that a business developing software and going through the proper SDLC processes takes time. Open Source developers can drop a new release every other day, and deal with bugs as they arise. That's not how the world of commercial software works. After all, Red Hat Enterprise Linux 6.4, the very latest and greatest release, used by corporations around the world for back-end business processing, is still on the 2.6 kernel. Do a little homework into how commercial software development works and you may learn something.
          Nvidia supports their workstation products and are able to deliver support for the latest kernel/Xorg versions. So reading your description Nvidia does not deliver commercial software, I would think.
          Commercial products 101: If you are not able to keep up with your competitors you will go down the drain, unless you have some kind of cult status. AMD is lacking the cult status and they don't keep up with Nvidia in Linux driver development, guess what follows: Dozens of formerly satisfied customers state in forums that they will not buy AMD videocards again, unless those issues are solved.
          I am one of them and I repeat it again: AMD, do something about your shitty driver policies or you will not get me back as customer.

          Comment


          • #55
            Originally posted by Vim_User View Post
            Nvidia supports their workstation products and are able to deliver support for the latest kernel/Xorg versions. So reading your description Nvidia does not deliver commercial software, I would think.
            Commercial products 101: If you are not able to keep up with your competitors you will go down the drain, unless you have some kind of cult status. AMD is lacking the cult status and they don't keep up with Nvidia in Linux driver development, guess what follows: Dozens of formerly satisfied customers state in forums that they will not buy AMD videocards again, unless those issues are solved.
            I am one of them and I repeat it again: AMD, do something about your shitty driver policies or you will not get me back as customer.
            Replying just to reply, I like to cite my prior post and state that it is the commercial sector who is in a pickle over the hyper-threading bug. From what I've heard it's the HUGE roll-out of hyper-threading in our back-end databases is where the mishandling of bits by these hyper-threading enabled processors is manifesting. Those HUGE 20+ petabyte installations are going to show inconsistencies that make small ext4 errata seem like a crack in the sidewalk compared to San Andreas. GOOD LUCK WITH THOSE MIGRATIONS!

            Tape silos anyone? Two months DR (disaster recovery) time & plans. It's expected in several fortune 500s... and even then, records inproperly truncated will be inadequate. How much is the TaCo on months (or years) worth of R&D in some polymers lab worth? Now please reproduce those in the lab...SAP didn't handle it quite right...ON INTEL! LOL Muhahahahha... <jk>

            AMD *rah rah rah* !!! Goooooo Opteron!

            Comment

            Working...
            X