2D performance?

[Goi]

Many of you might have read FiringSquad's Building the Ultimate High-End Gaming Workstation: Stage 1 article. I know I did. What piqued my interest however was Page 6 of their article which claimed that the Radeon 9800 series offered significantly lower 2D performance than the GeForce FX5900 series.

Now its been a while since any review site actually benchmarked 2D performance of graphics card, and like many of you I'm sure, I've just about written off 2D performance as being identical for the various graphics cards. That article seems to indicate otherwise, and its true that the last time I saw a 2D performance benchmark chart, the Radeons were in fact trailing the GeForce/TNTs. At lower resolutions this might not matter, but with an increasing amount of people getting larger displays and using higher resolutions this might start to become a problem again, especially with multi-monitor setups. I was wondering if anyone had any insight as to whether this is a valid concern?

Edit: Fixed the broken link

 

[Jeff7181]

If all you want is 2D... buy a Matrox card and forget about it =)

 

[Goi]

Well, not all Matrox cards are built the same. The G450 with its 64-bit memory bus for example is really slow at 2D.

 

[Pete]

Erm, you copied-and-pasted your post twice, I think.

All I can say is my Radeon 9100 seems fast enough with 2D apps at 12x9. I can't imagine needing more speed in 2D rendering, and I really can't imagine nV's cards being THAT much faster that you can notice the difference in webpage rendering! FS may be right, but I really doubt the difference is (that) significant, but I don't have experience with a discrete nVidia card, just my nF1's IGP.

I wonder if there's a difference in 9600 Pro and 5600U "2D speed," as measured by OfficeBench (not sure what that bench is, BTW). Both have the same core speed, but different memory bandwidths. So is the difference due to drivers, memory bandwidth, or just 2D hardware? Core speed is out, as the FS bench shows the 5900U performs the same at "2D speed" (300MHz, IIRC) as at "3D speed" (full speed). The 5900U definitely has faster memory than a 9800P, but I can't imagine desktops needing anywhere _near_ 20-30GB/s. So are nV's drivers simply better-optimized, or is nV's 2D hardware simply better?

 

[Pete]

Originally posted by: Jeff7181
If all you want is 2D... buy a Matrox card and forget about it =)

Actually, if you read the original 2D article FS linked in their current one, the Matrox G450 was getting whupped by a Radeon.

 

[Goi]

Yes Pete I noticed that which is why I edited it, but thanks for pointing it out. It took 2 edits to sort it out =p

You raise a couple of interesting points, and I follow your train of thought, but don't have answers either, which is why I started this thread. IIRC, both hardware and drivers support can cause differences in certain feature support in the Windows GUI(or other OSes). Of course, pure core speed's gonna cause a difference too, but I suspect not as much.

 

[zephyrprime]

Nobody's ever said that all cards have the same 2d performance. Rather, it's believed that the 2d performance is more than good enough so that it's no bottleneck at all for any 2d program.

 

[Goi]

Well apparently to FiringSquad editors its not "more than good enough" if they can qualitatively and quantitatively tell the differences among the nvidia/ATI cards.