Radeon 9700 Benchmarks...

[Indiana]

Originally posted by Pace
Is Anand implying that boosting the DAC speed improves the sharpness?

Yes, the way he puts that really sounds as if he had absolutely no idea whatsoever..
But if you read further in the text it's not really that bad anymore. It sounds as if the R9700Pro would be on par again with the old Radeon (the 8500 was below) and that quality is really acceptable.

The best thing comes (again) from Lars Weinand:

Look at the RAM clockings, the bus width and the bandwidth he claims for the parts:
GF4 Ti: 325MHZ, 128Bit DDR -> 10.4GB/s
R9700: 325MHz, 256Bit DDR -> 19.8GB/s

Huh? Can't that guy even calculate? Those are the axact same parameters, just the bus width of the R9700 is double that of the GF4, so the final bandwidth is what? Yes, double, of course.

I'd really like his explanation how he gets his 19,8 GB/s for the Radeon anyway. Let's see: 325(MHz) * 32(256Bit) * 2(DDR) = 20800 MB/s.

These are the results when you simply copy & paste data from manufacturers papers without understanding them. Most likely NVidia calculates with 1.000 and ATI with the more correct (but rarely used) 1.024 for a Kilobyte. But when you put those values in a table beneath each other you have to correct this of course!

[SanGreal]

Originally posted by Indiana


Yes, the way he puts that really sounds as if he had absolutely no idea whatsoever..
But if you read further in the text it's not really that bad anymore. It sounds as if the R9700Pro would be on par again with the old Radeon (the 8500 was below) and that quality is really acceptable.

The best thing comes (again) from Lars Weinand:

http://www.indiana.claranet.de/Weinand.png

Look at the RAM clockings, the bus width and the bandwidth he claims for the parts:
GF4 Ti: 325MHZ, 128Bit DDR -> 10.4GB/s
R9700: 325MHz, 256Bit DDR -> 19.8GB/s

Huh? Can't that guy even calculate? Those are the axact same parameters, just the bus width of the R9700 is double that of the GF4, so the final bandwidth is what? Yes, double, of course.

I'd really like his explanation how he gets his 19,8 GB/s for the Radeon anyway. Let's see: 325(MHz) * 32(256Bit) * 2(DDR) = 20800 MB/s.

These are the results when you simply copy & paste data from manufacturers papers without understanding them. Most likely NVidia calculates with 1.000 and ATI with the more correct (but rarely used) 1.024 for a Kilobyte. But when you put those values in a table beneath each other you have to correct this of course!

The band with figure is correct. Its the core/mem speeds he messed up on. The Radeon 9700 has a core speed of 325MHz and a Ram Speed of 310MHz, not the other way around as he lists. 310*32*2=19840

[PeTe]

Originally posted by Wombat
Because they wanted an AGP-spec compliant card, for one.

Btw, it was said in the [H] review [ if i remember right.. have read atleast 5 reviews ] that Radeon 9700 IS AGP-spec compliant, but its cutting so close that they included the extra power connector just to make sure.

Pe-Te

[thop]

you can run the card without the power connector in 2D but not stable in 3D.

[PeTe]

Seems that the core is easily o'ced to 345-350Mhz.. giving even more POWER.

Mem for some reason doesnt seem to go over ~350DDR, but with some cooling it should do the same that most GF4Ti4600's do.. 360-370DDR... giving R300 the bandwith of whopping 22-23G.

Pe-Te

[Indiana]

Originally posted by SanGreal

The band with figure is correct. Its the core/mem speeds he messed up on. The Radeon 9700 has a core speed of 325MHz and a Ram Speed of 310MHz, not the other way around as he lists. 310*32*2=19840

Oops, you're correct, he messed that up.
But while this surely is an error that can happen, he also states the same thing in his written text:

Dank nun 256 Bit DDR Busbreite ist diese beim R300 nun bei gleichem Takt fast doppelt so hoch.

Rough translation: Thanks to the 256Bit wide bus at the same clockings the R300s bandwidth is nearly double that of the GF4.
The fact that he doesn't even seem to notice the stupidity of this statement renders him a total idiot.
That's as if he would say: 4*6 is nearly double 2*6....