X800 PRO launch on 4th of May + specs

[Alkali]

A quote from the Inquirer...... I dont know how reliable the specs are below, but its a shock to me, I thought the X800 Pro was clocked a bit higher...

ATI to launch X800 PRO on 4th of May - X800XT to follow in two weeks

By Fuad Abazovic: Friday 16 April 2004, 12:56

EVEN THOUGH ATI doesn't want us in Canada without signing one of its famous NDAs, we always have our ways of getting information about its future plans. Back at CeBIT we heard that new chip will be called X800 but simply could not confirm it at that time and we also heard that this chip will be introduced at System Builder Summit.

The date that is apparently being suggested is the 5th of May, and it could be that ATI will launch this chip during the SBS event.

Sources close to ATI claim that it can actually ship this card at the same day of launch and we know that there are quite some samples ready as we speak.

As for the specs, the R420PRO or should we call it X800PRO will be clocked at 475 MHz, 25 MHZ higher then we suggested since yields are better, while the memory will be clocked at 900MHz as we suggested before.

Everything is packed into 180 millions of transistors and the card has 12 pipelines as we suggested. The R420XT, Radeon X800XT will be higher clocked card with 16 pipelines which came as a huge surprise to us.

This card is meant to fight Geforce 6800 non Ultra - the one with 12 pipelines and should be in retail as soon as ATI launches it.

This means that you should be able to buy one on the 4th of May but let's wait and see if ATI can deliver on that day.

Faster Radeon X800XT will come just after E3, and soon after the 14th of May we are told. I cannot remember if I said this before, but the R420 is chip based on R300 marchitecture and you can clearly see this from its 96 bit precision support and lack of PS 3.0.

Performance wise it will give NV40 a good run for its money. µ

 

[NFS4]

:beer:

 

[Adul]

 

[InlineFive]

I wonder if the lower end NV40 cards will have the encoder also. It's starting to look to me that nVidia might win this round.

-Por

 

[gauravsharma311]

X800 isn't shader model 3, and therefore less complex circuitry, more speed per MHz. remember NV30 had more complex circuitry to work its extra features and speed per MHz was less than a similar clocked radeon because the radeon simply did base DX9. If X800 just does DX9 base (like r9800) and does it faster (as i'd expect, not least because of higher clocks), then i don't think anyone will be complaining.

 

[InlineFive]

Originally posted by: gauravsharma311
X800 isn't shader model 3, and therefore less complex circuitry, more speed per MHz. remember NV30 had more complex circuitry to work its extra features and speed per MHz was less than a similar clocked radeon because the radeon simply did base DX9. If X800 just does DX9 base (like r9800) and does it faster (as i'd expect, not least because of higher clocks), then i don't think anyone will be complaining.

Does this remind anyone of Intel and AMD?

-Por

 

[JBT]

Originally posted by: PorBleemo
I wonder if the lower end NV40 cards will have the encoder also. It's starting to look to me that nVidia might win this round.

-Por

I heard that ONLY the Ultras will have the decoder.

 

[wizdum]

As for the specs, the R420PRO or should we call it X800PRO will be clocked at 475 MHz, 25 MHZ higher then we suggested since yields are better, while the memory will be clocked at 900MHz as we suggested before.

900MHz = 450x2

-or-

900MHz = 1.8GHz memory. :Q

?

Groov

 

[gauravsharma311]

intel's pretty much (internally) admitted defeat there really - their new pentium 5 (2006) is basically a new design based around the centrino P-M chip (which is one hell of a chip, must add). nv/ati situation different, changes all the time, e.g. radeon 8500 had dx8.1 but ti500 still beat it sometimes due to efficiency, and ati's r500 will be more complex than nv40 all over again. thing with ATi is that they're sticking with MS, and have/had perfect parts ready for DX9, DX10 launches, whereas nvidia's being more experimental early on and seem to optimise for efficiency later in product life cycle.

 

[gauravsharma311]

900MHz effective.
remember that radeon9700s with 550mhz are still kicking around 1000mhz memory nv30s, and in future games will be less dependent on memory bandwidth and more on shader efficiency/parallelism (anand done nice piece on this on HL2 article). X800xt id imagine has 8 vertex lines and will win over nv40 regardless of mem speed in games like HL2. if they've improved the PS2 even further on radeons, then nvidia have far from won anything yet.

 

[gauravsharma311]

jbt,
majority of videocards these days have hardware vid decoding?
if you meant encoding, it'll be on all the nv parts, but is tied to speed of those parts.

 

[XBoxLPU]

Originally posted by: gauravsharma311
900MHz effective.
remember that radeon9700s with 550mhz are still kicking around 1000mhz memory nv30s, and in future games will be less dependent on memory bandwidth and more on shader efficiency/parallelism (anand done nice piece on this on HL2 article). X800xt id imagine has 8 vertex lines and will win over nv40 regardless of mem speed in games like HL2. if they've improved the PS2 even further on radeons, then nvidia have far from won anything yet.

9700Pro memoy = 310 x 2 or 610

 

[Alkali]

As a side note:

I was reading a very intelligent post made by a guy on rage3d discussing Pixel shader 2.0 and 3.0.

Basically, PS3.0 does not actually make anything look any better. PS3.0 is designed to allow branch prediction and a whole load of other stuff which allows the GPU to process all the shaders faster than PS2.0.

However, if you have a card that can do a much higher PS2.0 it can theortically easily out-perform the 'optimised' PS3.0 code.

Sounds plausable to me.... gets more and more interesting as we move forward

 

[gauravsharma311]

i did say radeon 9700
(not pro).

 

[InlineFive]

Originally posted by: JBT

Originally posted by: PorBleemo
I wonder if the lower end NV40 cards will have the encoder also. It's starting to look to me that nVidia might win this round.

-Por

I heard that ONLY the Ultras will have the decoder.

Bummer...

 

[gauravsharma311]

ps3 is basically more complex shader programs with more flexiible syntax/branching. it can result in simpler coding for the same effect without performance drop. some effecs requiring too much effort using ps2 can be done feasably with ps3. there is space for IQ improvement simply because stuff's more doable with PS3. farcry examples are good for anyone wanting to see what the fuss is about - there's a reason why they didnt put those effects in for PS2 capable cards.

 

[AIWGuru]

Originally posted by: Alkali
As a side note:

I was reading a very intelligent post made by a guy on rage3d discussing Pixel shader 2.0 and 3.0.

Basically, PS3.0 does not actually make anything look any better.

Wow, so much crap in this thread, I don't know where to begin.
I'll just address this one for now:
This guy at rage3d is a retard.


PS 1.1

PS 3.0

Any questions?


If those specs are even remotely accurate, Nvidia has ATI completely, inarguably schooled this round...

 

[jdogg707]

I'll be looking forward to seeing what ATi can do, I would hate to go back to Nvidia now...even though I may have to if the 6800 Non-Ultra performs better than the X800 Pro. I will probably just be keeping my current card until PCI Express comes out since it would be a waste buy another AGP 8X card and then have to buy PCI Express onmy next upgrade.

 

[Alkali]

Originally posted by: AIWGuru

Originally posted by: Alkali
As a side note:

I was reading a very intelligent post made by a guy on rage3d discussing Pixel shader 2.0 and 3.0.

Basically, PS3.0 does not actually make anything look any better.

Wow, so much crap in this thread, I don't know where to begin.
I'll just address this one for now:
This guy at rage3d is a retard.


PS 2.0

PS 3.0

Any questions?

Due to the fact you obviously hav'nt read the posts about these screenshots, I'll forgive you for that.

The first screenshot is Pixel shader v1.1, and the second is pixel shader version 3.0 as confirmed by nVidia themselves after the presentation.

 

[InlineFive]

Originally posted by: AIWGuru

Originally posted by: Alkali
As a side note:

I was reading a very intelligent post made by a guy on rage3d discussing Pixel shader 2.0 and 3.0.

Basically, PS3.0 does not actually make anything look any better.

Wow, so much crap in this thread, I don't know where to begin.
I'll just address this one for now:
This guy at rage3d is a retard.


PS 2.0

PS 3.0

Any questions?


If those specs are even remotely accurate, Nvidia has ATI completely, inarguably schooled this round...

That looks more like detail setting changes then Pixel Shaders to me...

-Por

 

[AIWGuru]

You're right. But PS 2.0 still doesn't look anything like that.
"The following before-and-after images are from the CryEngine. Developer Crytek uses Shader Model 3.0 techniques (vs. 1.x shaders) to add more depth and realism to the scenes. Notice the more realistic look and feel when SM 3.0 is applied to the bricks and stones that make up the staircase. In the scene featuring the Buddha, the full image comes to life with the use of SM 3.0, with the technique applied to multiple objects."

 

[Alkali]

Quote from an unnamed developer:

ps 3.0 == ps 2.0 + more registers and dynamic branching

- anything you can do with ps3.0 you can do with ps2.0, it just might require more passes
- dynamic ps3.0 branching in upcoming hardware is going to be very slow, and barely worth using
- therefore, ps 3.0 == ps 2.0

any comparisons supposedly showing the power of ps3.0 over ps 2.0 are ridiculous... the two instruciton sets are virtually identical. not to mention that the ps stuff is just a *programming language*... its like showing videos comparing games made in C++ vs games made in C, and claiming that one is better than the other.

i'll be more excited when we get much faster ps3.0 hardware, and/or ps4.0 hardware

Btw, here you can get fully educated about PS3.0

PS3.0 Full Technology Review

 

[UlricT]

AIWGuru... 1.x shaders are from DirectX 8!
I think the argument here is DX9b vs. DX9c!!!

 

[AIWGuru]

Originally posted by: UlricT
AIWGuru... 1.x shaders are from DirectX 8!
I think the argument here is DX9b vs. DX9c!!!

Yes, but what we're talking about is PS 3.0 being able to do the same ammount of work in 1/10th the ammount of time as PS 2.0.
Saying PS 2.0 = PS 3.0 just because it's capable of producing the same visuals EVENTUALLY (with much more work) is stupid.
It can't be PRACTICALLY done because it's MUCH MUCH slower.
That's like saying that a pentium 1 and a pentium 4 are the same since given the same SETI unit, they'll both come up with the same result even though one takes longer.

Obviously, PS 3.0 DOES allow for more advanced visual effects because the horsepower is not present to perform them under 2.0, not because it's not technically possible.

 

[AIWGuru]

Quick question:
If PS 3.0= PS 2.0 why are the developers of Farcry sinking all of this money into the game Post Release to update its support for PS 3.0?