Announcement

Collapse
No announcement yet.

FireGL V5100 or FireGL V5200?

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

  • FireGL V5100 or FireGL V5200?

    Wiki data table http://en.wikipedia.org/wiki/Compari...#FireGL_series

    I lack knowledge to decide, which one of those two cards will be better for my needs. I intend to use it for gaming with open drivers. I made my mind to game on this and nothing will change it.
    Basically when I can't hit high FPS, I set lower resolution before I turn details down. So I will prefer a GPU that lets me play full-detail low-resolution rather than low-detail high-resolution. Of those two, which one fits my needs better? Which parts of hardware are important for me?

  • #2
    Originally posted by Hirager View Post
    Wiki data table http://en.wikipedia.org/wiki/Compari...#FireGL_series

    I lack knowledge to decide, which one of those two cards will be better for my needs. I intend to use it for gaming with open drivers. I made my mind to game on this and nothing will change it. Basically when I can't hit high FPS, I set lower resolution before I turn details down. So I will prefer a GPU that lets me play full-detail low-resolution rather than low-detail high-resolution. Of those two, which one fits my needs better? Which parts of hardware are important for me?
    The two boards you mentioned are both from 2005, and pre-date the move to unified shaders where the same shader hardware was used for vertex and pixel processing. In terms of "which parts of hardware are more important", your preferences suggest that vertex processing is more important than pixel processing.

    The 5100 is based on a high end r4xx GPU, while the 5200 is based on a mid-range r5xx GPU with fewer pixel pipes but a higher clock. Given that the available memory bandwidth seems to be the same, I would lean towards the 5200 since it should have higher vertex processing throughput. That said, the 5100 has maybe 2x the pixel processing power so you would need to be sure that your idea of "low resolution" was low enough for the hardware to keep up and not end up bottlenecked on pixel processing anyways.

    That said, one could argue that moving to a slightly newer GPU with unified shaders would be your best bet. Any reason for choosing FireGL boards over the consumer Radeon boards ?
    Test signature

    Comment


    • #3
      Originally posted by bridgman View Post
      The two boards you mentioned are both from 2005, and pre-date the move to unified shaders where the same shader hardware was used for vertex and pixel processing. In terms of "which parts of hardware are more important", your preferences suggest that vertex processing is more important than pixel processing.

      The 5100 is based on a high end r4xx GPU, while the 5200 is based on a mid-range r5xx GPU with fewer pixel pipes but a higher clock. Given that the available memory bandwidth seems to be the same, I would lean towards the 5200 since it should have higher vertex processing throughput. That said, the 5100 has maybe 2x the pixel processing power so you would need to be sure that your idea of "low resolution" was low enough for the hardware to keep up and not end up bottlenecked on pixel processing anyways.

      That said, one could argue that moving to a slightly newer GPU with unified shaders would be your best bet. Any reason for choosing FireGL boards over the consumer Radeon boards ?
      My previous Radeon one day ceased to boot up. Right now I am stuck with an integrated NVidia that doesn't even support my native desktop resolution. I have little money to spare and on my hunt in an auction house I decided to go with FireGLs because of their age. Additionally, the workstation GPUs tend to be cheaper in this price sector. But it's not the only reason I decided to stick with the Fires. I have free space of one slot in my chassis and all Radeons worthy of my interest have huge double-slot fans. The consumer cards also tend to use more Watts and be of lower build quality. And I love quality. If I had the money I would happily spend it on something like FirePro V7900. Thats why I prefer the free driver - it is of higher quality than the blob, including the support for newest kernels.

      I forgot about the transition to the unified shaders. Is this the reason for change of name to FirePro? I have found a cheap FirePro V3700, but one thing holds me back - bus width of 64 bits. Does it scale more with resolution or quality?

      Comment


      • #4
        Makes sense.

        Originally posted by Hirager View Post
        I forgot about the transition to the unified shaders. Is this the reason for change of name to FirePro? I have found a cheap FirePro V3700, but one thing holds me back - bus width of 64 bits. Does it scale more with resolution or quality?
        AFAIK we changed the brand name from FireGL to FirePro because an increasing number of workstation applications were using DirectX rather than OpenGL, and the FireGL brand name suggested "OpenGL only".

        Clock rates have been going up so the narrower bus isn't as much of an obstacle as it might seem (say 2/3 the bandwidth instead of 1/2) but it is still going to make some difference. Between rv530 and rv620 I'm not sure which is going to give you better gaming performance, unfortunately.
        Test signature

        Comment


        • #5
          Originally posted by bridgman View Post
          Makes sense.



          AFAIK we changed the brand name from FireGL to FirePro because an increasing number of workstation applications were using DirectX rather than OpenGL, and the FireGL brand name suggested "OpenGL only".

          Clock rates have been going up so the narrower bus isn't as much of an obstacle as it might seem (say 2/3 the bandwidth instead of 1/2) but it is still going to make some difference. Between rv530 and rv620 I'm not sure which is going to give you better gaming performance, unfortunately.
          My current understanding is that the bus width is the amount of data that is moved between GPU and CPU. So, considering the fact that the radeon driver is not as optimized as fglrx, the narrower bus can easily become a bottleneck, right?

          Comment

          Working...
          X