Announcement

Collapse
No announcement yet.

Sapphire Radeon HD 4650 512MB OC

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

  • #11
    Originally posted by bridgman View Post
    The glxgears benchmark tests only the most basic rendering functions; coloured triangles with no textures and no shaders. As a result the numbers end up being misleading in a couple of ways :

    1. Older cards often score higher than newer cards because modern cards spend much more of their transistor budget on shaders and texture engines rather than ROPs. It's not unusual for an older card to outrun a newer card in the same range; usually the older card has more ROP throughput than the new card.

    2. In many cases the benchmark runs CPU limited rather than GPU limited; in that case you end up measuring is the driver efficiency on a trivial path which is mostly used by glxgears and screen savers.

    In the case of 7300gt vs 4670 both cards have 8 ROPs, memory width/type is the same and the clock rates aren't that different, so their theoretical performance on glxgears should be relatively similar. Shader power is a diferent story; the 4670 has somewhere between 5x and 8x the shader power of the 7300gt depending on how you do the math. Something like the Unigine demos (in the Phoronix test kit IIRC) would probably show off that difference.

    By the way you mentioned "fglrx 8.12 and mesa 7.2" which I didn't understand; fglrx doesn't use mesa for 3d, it has its own 3d driver stack.
    Yes, I expected something like this. I installed the latest stable mesa because I got these scores in the tests and I noticed phoronix tested their card with mesa 7.2 so I gave it a try but it just lowered the scores.

    It is clear that these cards are nearly in the same level but I would expect more performance when I use a new card in higher price range than a 2 years old cheap 7300gt. OK it has almost the same performance then let it be.

    But in the phoronix test with nearly the same drivers they got much more fps (double that much)in a higher resolution. I have a dualcore athlon 64 so I don't think the cpu hold back my vga. And I'm running Ubuntu Ibex like they did. I don't see the reason for my card low performance.

    Comment


    • #12
      Originally posted by Ginko View Post
      Yes, I expected something like this. I installed the latest stable mesa because I got these scores in the tests and I noticed phoronix tested their card with mesa 7.2 so I gave it a try but it just lowered the scores.

      It is clear that these cards are nearly in the same level but I would expect more performance when I use a new card in higher price range than a 2 years old cheap 7300gt. OK it has almost the same performance then let it be.

      But in the phoronix test with nearly the same drivers they got much more fps (double that much)in a higher resolution. I have a dualcore athlon 64 so I don't think the cpu hold back my vga. And I'm running Ubuntu Ibex like they did. I don't see the reason for my card low performance.
      I will answer my own question. I reinstalled the whole system (it was dist-upgraded 3 times) and voila nearly doubled my scores.

      Comment


      • #13
        Oh good. I was starting to run out of ideas
        Last edited by bridgman; 14 December 2008, 02:18 PM.
        Test signature

        Comment

        Working...
        X