X1800 XT looks like a beast

[Helevitia]

625MHz Core
1.5GHz memory

Check out the marketing sell sheets here.

The X1300 and X1600 info are there as well.

[rylan]

Certainly is... check out the heatsink and fan. They must've really overclocked the $hit out of it.
At least that it doesn't require an external power brick.

[Helevitia]

A few things to note:

1. Dual DVI Links
2. X1800 XT blows away the 7800 GTX in Fear by 80%
3. 3 HDR modes(I wonder what will become of this?)
4. New Adaptive AA mdoe that looks awesome
5. 48GB/s memory bandwidth!

[Helevitia]

Certainly is... check out the heatsink and fan. They must've really overclocked the $hit out of it.
At least that it doesn't require an external power brick.

The rumour is, you still have room to OC it, even at 625MHz! If that's true, this will be one badass card.

[Elie]

And I may get a demo to test out

[DGhost]

*yawn*

oh well, at least they are getting close to the sort of architecture that NV introduced with the GeForceFX series. let's see how well they optimize and balance it.

[Tjalfe]

from the looks of it, they are all working towards a general purpose CPU with all the programmability they are introducing

edit: spelling

[DGhost]

DX10/WGF2 is based around the Unified Shader Model. the difference between ATI and NV at this point is that NV has 3 years of experience shipping products with architectures that have decoupled shader, where as this is the first product ATI will have shipped with anything close to it.

[Helevitia]

Why so much Nvidia praising and ATI bashing DGhost?

[Helevitia]

And then there's the whole series 6 vs. R4x0. Series 6 had SM3 support for so long.

In 2 days ATI will switch from "SM3 is useless" mode to "SM3 is a must have improvement over SM2 so get our newest radeon now" mode lol.

Uh, no. ATI said that SM3 was useless for the R4XX series and they were absolutely right. And it still is useless since there isn't anything out there that actually uses it except FarCry and that is more like a technology demo in game.

[Helevitia]

Well, its all marketing man. marketing.

Look at the sheets "SM3.0 done right"

They are talking about Full FP32 bit processing and AFAIK Nvidia hasn't accomplished that yet?

[DGhost]

They are talking about Full FP32 bit processing and AFAIK Nvidia hasn't accomplished that yet?

thoughts in no particular order...


I would be extremely suprised if there were not limitations on that. note, they also don't say that they enabled FP32 across the whole pipeline - just just it internally always handels FP32 processing. shit, the GeForce6 series did internal FP32 processing - the major drop out was in texture filtering.

the fact that they list the 64bit Floating Point/integer as their maximum rage for HDR should give you a fairly good idea that it simply does not do FP32 processing across the whole pipeline.

what, exactly, is an "Ultra Threaded Dispatch Processor"?

the whole thing is almost entirely marketing speak. of course it looks good - thats what it is for. the simple facts of the matter is that 1) they list their own pixel fillrate as on par with the GeForce 7800 series, 2) they run their clock speed a good 200Mhz or so higher than the 7800 parts, and 3) they measure their pixel fillrate using the pixel shaders instead of the actual ROP's. the fastest way to measure that is to assume 1 op per pixel and run it as a pixel shader function. NVidia has put a large amount of effort into the G70's pixel shading units being able to handle common scenarios that cause a decoupled pixel shading pipeline to hang up - will this handle things as well?

i would be willing to bet the 7800 does actually have a higher IPC than the X1800XT...

ATI has yet to demonstrate their ability to get good yields out of the 90nm process.

why does it have the same hoover attachment that the original 5800's shipped with? what, exactly, are they doing that it requires that? Note, the 7800GTX runs fine on a single slot cooler. could the 90nm process not be as good as they claim?

ATI still has yet to fix their supply problems. will we even see these things hit market? and how stripped down are the X1300 and X1600 parts going to be?

edit: one more thought. wasn't this supposed to be a derivative of the Xenos GPU?

[Marshmallowman]

Another little thing I would like to see clarification on, are those benchmark figures "temporal AA" or proper AA ?.

[GT98]

edit: one more thought. wasn't this supposed to be a derivative of the Xenos GPU?

I dont think so...the Xenos GPU has embedded memory on it

[DGhost]

I dont think so...the Xenos GPU has embedded memory on it

there were other, more significant architectural changes that i would have hoped could have been ported over.

yeah, the X1800XT card might be an awesome card, but all that has been """""officially""""" confirmed is all marketing fluff. if a car manufacturer was claiming 70MPG in city driving, would you believe it without seeing it? also.... just like the 5800, then the 5900 Ultra, then the 6800 Ultra, and the 7800GTX - high end cards are such a small percentage of their sales it's not even funny. companies will push their high end cards to the point that it is almost impossible to mass produce them just so that their initial reviews are better. NVidia has done it. ATI has done it too. what makes you think this is different?

on a different note, i just saw this at HardOCP:

ATI CrossFire MIA:
There has been a lot of hoopla about ATI and their CrossFire master cards lately, and I figured I would drop my 2 cents about all of this. Let me start by saying that we checked today with one of the Nets largest online computer parts retailers and they explained to me that they did not anticipate having stock on ATI CrossFire Master cards till November. Now keep in mind this is a mere 5 to 6 (depending how you count) MONTHS after their initial introduction of the product back in May...of this year.

Instead of waxing on for pages, I am going to cut to the chase and say what everyone is thinking and what many of us in the technology press are feeling.

1. ATI, HardOCP is not your personal PR bitch, so don't treat us like one again. You're on thin ice and you should be ashamed of your actions this year. We at HardOCP don't want to be a mouthpiece for vaporware in order to keep your stock price up. So from now on, if it is not the real deal, just keep it to yourself.

2. Continue with the "If NVIDIA does it, we can do it too!" methodology that has been working for you guys for a while now. But from now on, keep a tighter timeline.

3. Repeat after me. Paper launches, NO. Product launches, YES.

4. NVIDIA is setting the new standard by having their card builders selling the products on the day that they are launched. Learn it, know it, live it. And NVIDIA, don't change up on us now.

The bottom line is that the enthusiast hardware community wants products, not Canuck promises. ATI is quickly burning through a solid and loyal fan base that they built back in the days of the 9700/9800. Now is the time for ATI to step up with honest damage control, not another half filled basket of half baked promises.

And oh yeah, Quake 4 just went Gold. WooHoo! I think I just found a reason to go ahead and go SLI. (Always been a big Quake fan...) Too bad CrossFire is not a real option, although I am unsure whether or not I would consider it at this time, just due to having a fear of lacking support and service by ATI.

Kyles always been a bit judgemental, but still...