DarkMatter
New Member
- Joined
- Oct 5, 2007
- Messages
- 1,714 (0.27/day)
Processor | Intel C2Q Q6600 @ Stock (for now) |
---|---|
Motherboard | Asus P5Q-E |
Cooling | Proc: Scythe Mine, Graphics: Zalman VF900 Cu |
Memory | 4 GB (2x2GB) DDR2 Corsair Dominator 1066Mhz 5-5-5-15 |
Video Card(s) | GigaByte 8800GT Stock Clocks: 700Mhz Core, 1700 Shader, 1940 Memory |
Storage | 74 GB WD Raptor 10000rpm, 2x250 GB Seagate Raid 0 |
Display(s) | HP p1130, 21" Trinitron |
Case | Antec p180 |
Audio Device(s) | Creative X-Fi PLatinum |
Power Supply | 700W FSP Group 85% Efficiency |
Software | Windows XP |
I think the tmu was a bottleneck in rv670, the memory bandwith is good for high res with high aa. More shader powa is necessary as always, especially if gpu physics come into the picture.
I'm not certain if the rops are a bottleneck...
My point is that if TMUs were the bottleneck, they would have done 2 x Shader Power AND dunno 3x Texturing power? and not 2x-2x. If textures were the bottleneck, this flaw has been carried over to the new series. Since I don't think that possible, as I don't think they are so stupid, my only conclusion is that it wasn't a bottleneck. I am based in other concerns to reach tht conclusion too, like the efficiency at which they are capable of using the VLIW + SIMD SP arrays, for example. I have defended since day one that R600 was limited by it's shader power, it's just lately and after hearing most people complain about its 16 TMUs that I had to "admit" or adopt the idea that the bottleneck is on TMUs.