Radeon 9700 Pro: ATI Ahead

Keefe writes "The epic battle between ATI and Nvidia wages on. While Nvidia awaits arrival of their near-fabled NV30 for redemption, ATI conquers all by introducing the fastest and most advanced graphics card to date. The next-generation ATI Radeon 9700 Pro marks the second time Nvidia cedes the performance crown to ATI (the first time being the brief glory when the ATI Rage Fury beat the Nvidia TNT). See how the ATI Radeon 9700 Pro stacks up at Techware Labs."

Comments

Legacy Gates

[Vegan Pagan]

I wonder how much extra circuitry this chip has in order to be backwards compatible. I rember reading that the Geforce 3 had to have some legacy circuitry that wasn't used in Direct Draw 8 games in order to run Direct Draw 7 games. Now that we're into DD8.1 and DD9.0, how much more legacy circuitry is in there?

Re:Legacy Gates

[Anonymous Coward]

There are no legacy circuits and there will never be any. Most of the 3d acceleration is done in hardware and every new version of DirectX has new facilities for games the use. The card manufacturers can choose to support these features in hardware of in their software through HAL. If a manufacturer thinks, that a certain feature can better be done in software, than that's the way it shall be.

Don't expect the stuff we had with all the Intel procs.

Actually...

[Tuxinatorium]

The Radeon 8500 outperformed the Geforce3 Ti500 and came out months before the Geforce4 Ti series. They're continually trading the performance lead. The only instance of nVidia ever keeping the performance crown continuously from one product generation to another was Geforce2-->Geforce3.

From the time the Geforce2 came out until the Radeon 8500 came out 17 months ago, ATI had the unquestionable performance crown, but since then it has been juggling back and forth, which is to be expected since each new product release is about 6 months after the competitor's last release and the technology improves as time goes on. nVidia has a habit of shooting for a holiday release but not actually shipping until the new year, and ATI has made their last two releases in August. So when either one of them makes a new release they have a 6-month lead over the other company's product, so you should expect them to always trade performance crowns unless one of them is more than 6 months begind the other in R&D, which would be saying quite a lot.

Re:Do the Linux drivers work yet?

[caino59]

do a search...yes this problem was fixed by ati.

grab the latest drivers...it was just a glitch in the first release.

or you could always flash the bios of your OEM card with the firmware from a retail card..

Buy a 9500 Pro instead

[SensitiveMale]

The only difference is 1/2 the memory bandwidth. While that may seem like a lot, the 9500 pro actually gives @ 70% of the performance of the 9700 Pro.

and....

The 9700 Pro, 9700, and the 9500 Pro use the exact same GPU. So download a new bios at www.3dchipset.com/temp/warp11.zip and you can overclock the GPU to get almost 90% of the performance of a 9700 Pro.

Read all about it here Firingsquad.com

Also make sure to get DirectX 9

and New Catalyst 3.0 Drivers

And the 9500 Pro is a cheap at $180 delivered.

www.pricewatch.com

Re:Eh?

[Gruturo]

386 Released with a math CoProcessor
reviews explaining the performance difference between the 386SX and 386DX here

Ehm.....
Actually the link between the math coprocessor and the SX/DX name is an 80486 thing.

80386SX (Singleword Xchange) had a 16bit bus and could be plugged into a 286 board.

80386DX (Doubleword Xchange) had a 32bit bus and needed a new motherboard design (but was way faster because of the wider data bus, and could directly address more memory)

When the 80486 was introduced, the SX/DX distinction remained, this time to indicate the presence of the built-in FPU.

Urban myth wants that 80486SX's were full-blown CPUs in which the FPU silicon had failed tests, or, later, was just disabled, even though perfectly working.

Even worse was the fact that the 80487 was actually a FULL CPU+FPU, and not just an FPU. Upon startup it would disable the main processor and do everything. What a waste of power....
I never knew if the Weitek 4167 did this too.

Re:Eh?

[TheCrazyFinn]

The 386sx could not be plugged into a 286 board. While they shared a bus design, the packaging was different, and therefore needed changes to the Motehrboard design. It also shared the 24 bit address bus of the 80286.

The 386DX had both a 32bit Data bus and a 32bit address bus.

The 486SX was a 486DX with the FPU disabled (This is not urban myth, comparisons of the die confirmed this), nobody knows why the FPU was disabled, apart from the fact it was.