ATi responses challenge of NV30 with R350

[Hellbinder]

Besides, the 9700 (despite seriously embarrassingly bad drivers), is faster than nVidia's current offering. Thats fine. But remember that the 8500 was supposed to be their answer to the GF3 Ti. heh, they [ATI, in case you were wondering) failed.

Then came the R300 when nVidia was, of course, ramping up for the NV30

All in all, the R300 was designed to beat the NV25 - which it did. And since the NV30 - by all accounts - is gearing up to give the R300 a righteous spanking, ATI are trying to - again - leapfrog one generation by prematurely (IMO) releasing the R350

Typical.. Completely Nvidia Fanboi overstaded popycock. Nothing unusual from you.

The Radeon 8500 was released to compete with the GF3, not the TI. And even now the Ti and 8500 are on pretty even ground. Your statement is completely off base and reveals just how *in the dark* you really are. I dont care what developer relations you have.

The R300 is not premature. It was released just under one year from the 8500. Where do you even get off posting something like that? There is no indication that the Nv30 will be extremely faster. Its all on paper. thats it.

You have a completely *Twisted* one sided view. why is it ok for Nvidia to release not one.. but Two rehashed producs in the same damn year.. But Ati is acting prematurely with every product they release? Give me a damn break.

One last thing. I play at least 10 games regularly. I do not have a significant driver issue with any of them.

 

[Evildeus]

Actually, it says, "when Nvidia’s 0.13-micron NV30 chips hit the market in volume after November." Note "chips" and "after". That sounds like volume production of *chips* (which still need to become cards) in December to me, tho there is possibly some wiggle room there if the author was being imprecise in his language.

What's the market? For me it's on shelves, so to be in volume on shelves in december in volumes means to be produced in volume in november. But you are right could be february and then ....[/b]

 

[Ichneumon]

]

The Radeon 8500 was released to compete with the GF3, not the TI. And even now the Ti and 8500 are on pretty even ground. Your statement is completely off base and reveals just how *in the dark* you really are. I dont care what developer relations you have.

The R300 is not premature. It was released just under one year from the 8500. Where do you even get off posting something like that? There is no indication that the Nv30 will be extremely faster. Its all on paper. thats it.

Stated exactly correct. the 8500 was to compete with the GF3 line... the TI series was Nvidia's trump card on the 8500 which they played exceedingly well. The 8500 came a long way though in its competition to the Ti series, and it always did so at a significantly cheaper price/performance ratio.

It is quite amusing to hear some bonehead try to say the r300 was designed to beat the nv25. Complete idiocy. Of COURSE it was supposed to beat the nv25 series... The r300 was desiged to go head to head with the nv30. The nv25 and r300 are not even close to being in the same class of tech.

Then to have it said that the r300 is "premature" because the nv30 isn't here yet is even more silly... the nv30 is late the r300 was not early.

The NV30 will extremely likely be faster than the R300... it bloody well BETTER be if Nvidia is to avoid being seriously laughed at... given that they're not releasing the NV30 until MONTHS after the R300 is in peoples hands, they'd better figure out how to make it faster than the R300 if they haven't yet.

 

[V3]

The NV30 will extremely likely be faster than the R300...

That's no guarantee is it ? It could be roughly the same speed, but it just seems to have better caunt here and there, though with those big shaders, will NV30 be fast enough to execute them all ?

 

[Evildeus]

The NV30 will extremely likely be faster than the R300... it bloody well BETTER be if Nvidia is to avoid being seriously laughed at... given that they're not releasing the NV30 until MONTHS after the R300 is in peoples hands, they'd better figure out how to make it faster than the R300 if they haven't yet.

I don't see why. I don't remind people (well perhaps some fanatics of Nvidia brand) laughing at Ati at the launch of the 8500, and it was released MONTHS ( ) after the GF3, and wasn't "extremely faster".

The only problem i see for Nvidia, if it's the case, is the 3-4 months advance of ATI in the released of its futur products...

 

[Derek Smart [3000AD]]

]
Typical.. Completely Nvidia Fanboi overstaded popycock. Nothing unusual from you.

You mean as opposed to the usual fanATIc load o' bollocks that you and your lady friends are used to typing?

I know that, like most here, you consider yourself to be an authoritative figure because you've been here longer, have more posts etc. You really do need to think before you type - because I really don't give a toss WHO you are - you type crap to me, I'm going to take you up on it. Fool.

The Radeon 8500 was released to compete with the GF3, not the TI.

According to whom? You clueless fanbois? Gimme a break. The GF3 and GF3 Ti cards were out at a relatively close time frame (don't make me go look!) and then the 7xxx and 8xxx cards came out. IIRC, it didn't even had anything to do with the GF3 or the GF3 Ti, it just seemed that way. ATI simply released their suite of cards and hoped for the best. Its kind like throwing crap on the wall and seeing how much of it sticks. In their case, they were lucky, the 8xxx stuck - but thats no thanks to the driver developers because almost 99% of the reviews said the same thing Great card. Piss Poor drivers - when will they learn?

And even now the Ti and 8500 are on pretty even ground.

Rubbish. Apart from that, the 8500 drivers can't even compete with the GF3 nor Ti drivers

Your statement is completely off base and reveals just how *in the dark* you really are. I dont care what developer relations you have.

Yeah, I'm in the dark, but at least I have night vision on and can pretty much make out what I'm talking about, even if I don't have the facts. You and your friends take speculation, hearsay and outright rumors as if they were the gospel. And you have the nerve to point the finger at others.

The R300 is not premature. It was released just under one year from the 8500. Where do you even get off posting something like that? There is no indication that the Nv30 will be extremely faster. Its all on paper. thats it.

Oh look!! You never fail to disappoint. WHERE did I say in THIS thread, that the R300 was premature. Go quote the phrase or take your dunce cap and head for the nearest corner.

Taking posts out of context and posting flat out lies is your thing, but one would think that after being called on it so many times, that you'd actually quit doing that.

You have a completely *Twisted* one sided view. why is it ok for Nvidia to release not one.. but Two rehashed producs in the same damn year.. But Ati is acting prematurely with every product they release? Give me a damn break.

See above. And take your meds when you're done

One last thing. I play at least 10 games regularly. I do not have a significant driver issue with any of them.

I don't give a shit how many games you are playing or whether you're playing with yourself. You can post any crap you want. In fact, I play over fifteen games and have no significant ATI driver problems with any of them. *snicker*.

Hint: Golly Bull was a moron

It is quite amusing to hear some bonehead try to say the r300 was designed to beat the nv25.

You mean as opposed to hearing some frock wearing fanATIc dumbass posting unfounded speculatory nonsense? Yeah, its amusing alright - hilarious even.

 

[MrFloopy]

Could someone point me to a clear and precise list of driver issues with 7xxx, 8xxx, 9xxx drivers. Sorry to be a pain, but every link I've found to a previous thread seems to point to the same thread (in content anyway).

Am looking for actual bug descriptions and conditions in which they occur.

know this is OT but is most current thread about this point.

cheers.

 

[Derek Smart [3000AD]]

Could someone point me to a clear and precise list of driver issues with 7xxx, 8xxx, 9xxx drivers. Sorry to be a pain, but every link I've found to a previous thread seems to point to the same thread (in content anyway).

Am looking for actual bug descriptions and conditions in which they occur.

cheers.

Go to the www.rage3d.com forums and take your pick

 

[MrFloopy]

Go to the www.rage3d.com forums and take your pick

Was looking for bugs from a developers perspective if possible.

Having worked closely with 3dfx, quantum3d and nvidia (Mostly through Quantum3D) over the past 7 years, I am familiar with a lot of the driver bugs with these companies, but not ati so any specific pointers would be helpfull.

 

[rubank]

 

[sumdumyunguy]

Re: Hmmm

Ok well being in the IT world and Programming world I can say that there is the total possability of any one overclocking a Radeon 9700 Pro past 400 Mhz. I have talked to many people and they have showed me how they have overclocked it to insantity (One guy lately told me he overclocked it to 450Mhz I think). Now for the 400Mhz question with a .15m I would have to say that if they use the new fad (which is replacing all copper on the board with fiber (aka network wise fiberoptics)) they would be able to increase performance around about 3000% if they can get the stupid hardware to be fast enough to handle it.

From what I can calculate if ATI would use fiber they would be able to boast up there memory to about 1500Mhz to 2500Mhz. I have seen some network components using fiber instead of copper and those units were handling about 3.5 terabytes a sec (if I remember correctly). Plus there are some companies in which I have talked to some people in those projects that they are creating memory that uses pure fiber.

Till later
Raystream

P.S. this is no crap ther are some articles that have just been released about it so do some searching and you will find them. Right now I can not get any of you links (Stupid HOMEWORK). Sry

Ok Raystream, I'll bite. Working for a large telecommunications concern, & myselg being primarily concernd with SDH/SONET on a day to day basis let make take a stab

Are you saying that the interconnects on the pcb should be fiber instead of copper? True this would make transit times virtually nothing, however what about the EOE (electrical-optical-eletrical) conversion?

In modern ADM's (ADD-DROP Muxes) the conversion process (all the switching is done in the electrical domain; the holy grail is true optical switching which has yet to be realized as a practical matter) is what takes all the time. When you say things like 3.5 teraybytes on the backplane this specifically does not mean the the processing is done at that speed or anywhere near that. Actually that really refers to capacity (ports if you will in our ADM example) but you still have electrical components that are still "working" as slow (relatively speaking) as they always did. Furthermore all those EOE transducers would push one single R400 retail product would cost as much as a fiber repeater kit, say, a few hundred thousand dollars!

If you are aware of some breakthrough optical technology, then by all means man, post it!!

 

[Humus]

Go to the www.rage3d.com forums and take your pick

Was looking for bugs from a developers perspective if possible.

Having worked closely with 3dfx, quantum3d and nvidia (Mostly through Quantum3D) over the past 7 years, I am familiar with a lot of the driver bugs with these companies, but not ati so any specific pointers would be helpfull.

I can contribute with two I've encountered myself. With later drivers, dependent texture reads on a render texture will produce black. Also, using GL_SGIS_generate_mipmap on a render cubemap will cause a reboot.

 

[MrFloopy]

I can contribute with two I've encountered myself. With later drivers, dependent texture reads on a render texture will produce black. Also, using GL_SGIS_generate_mipmap on a render cubemap will cause a reboot.

Will check these out tomorrow.

On the last one, definitely sounds bad as nothing should do that.

I assume you are using the described method found in the ARB_render_texture specification with regards to cube maps and mipmap generation?

 

[Geo]

What's the market? For me it's on shelves, so to be in volume on shelves in december in volumes means to be produced in volume in november. But you are right could be february and then ....[/b]

Certainly that's a defensible interpretation. OTOH, who is NVidia's "market"? Not you and me. It's Leadtek, Gainward, etc. So that is a defensible interpretation too. Personally, I'd like to see them get it on the streets rights after Comdex as some are suggesting will be the case. I like competition and I don't want NV to get too far behind ATI.

 

[Randell]

re: ATI driver bugs - the problem with looking for bugs at www.rage3d.com are twofold:

- Lots of issues posted there, a lot are PEBCAK. Here are a couple of examples;

According to received canon in some quarters my system combo never works properly (8KHA+/XP1600/Radeon8500) although its always been fine with me (unless I do something like overclocking the fsb too much).

DAoC doesnt render guild emblems on cloaks correctly - again no problems on my system.

- 2nd a lot of real driver bugs that affect games devleopment will never be posted there as consumers dont see them and go through the developer support channels. e.g. DS's issues with w-buffer and MT didnt show up there.

 

[Typedef Enum]

As long as this thing has turned into a topic about drivers, and the fact that I haven't really had too much time of late to post...

I really do believe that the 9700, though (once again) a very good piece of hardware, is (ahem...once again) really tarnished by the drivers. I mean, look...I fully realize you cannot go to a forum like Rage3D and expect to find no issues...this is what the forum is designed for. Or the fact that one doesn't necessarily generalize about the state of a product based on what is seen @ forums...

But if you look @ the problems people are experiencing with this product, there are a couple of things that jump out...

1. Power related: This isn't one of these deals that you can point the finger directly @ ATI...BUT, it's really starting to look like the issues that impact some folks could be directly related to the massive power requirements...which is indirectly ATI's fault (if you're interested in placing blame).

2. General driver bugs: It seems to be better than the 8500...But in many respects, still much worse than nVidia. There will be those that will debate this issue until their blue in the face...about how ATI has come a long way, and there's a very slim difference between the 2 camps...I don't believe that at all. I still think nVidia has a very massive lead on ATI, when it comes to driver/software developers.

Let me give you some perspective on the topic of drivers. I know a lot of people laughed @ Derek when he originally stated that ATI had crappy drivers, and listed Matrox as one of the best (OK, there really are only 3 players left)...

Many were like, "WTF is he talking about? Matrox?"

Having used nVidia stuff for as long as I have, I can back this statement up 100%. I have been using this Matrox board now for about....hmmm....I guess going on 4 weeks now. Not ONCE...no kidding...not one single BSOD. I haven't had any significant bugs, with respect to the product.

I have been in close contact with Matrox on some specific bugs, most of which are very tame...a small flashing texture here, another there...and they have been very receptive to looking into these issues. But on the whole, their drivers are exceptional, with respect to quality.

It's really hard to place blame, and I don't generally like doing it...But let's just, for a second, pretend that ATI didn't have this massive history of crappy drivers...Let's pretend that the slate is clean, and this was ATI's first offering. I really believe that they would have been better served to get their drivers in a more polished state, rather than release the thing as early as they did. Hell, they knew NV30 was going to REALLY be late, so any additional time they didn't count on could now be better put to use on improving their drivers.

That's just my opinion on the matter. On the whole, I think ATI is slowly clawing its way back, as far as perceptions go...but they have a ways to go.

 

[Humus]

I assume you are using the described method found in the ARB_render_texture specification with regards to cube maps and mipmap generation?

No, the described method in ARB_render_texture doesn't use GL_SGIS_generate_mipmap. The method described in there is hardly practical when dealing with a complete mipmap chain.

 

[Nebuchadnezzar]

What's so hard to understand???

R9700 = R300 / we all know
R9500 = RV300 / 4 Pipe, 1TMU, 128 Bit bus, 0.15µ, it's just a low-end R300 / Q4 2002
R350   = All same as R300 + DDR2, MAYBE 2 TMU per Pipe, Pixel & Vertex 2.0+ (But sureley not 3.0), 0.15µ / Q1 2003
RV350 = Some low-end version of R350, 0.13µ, Probally same performance as a R9700 / Q2-3 2003
R400   = Super-hyper DX9 Ultra card with unified P&V Shaders 3.0+ / Q4 2003

 

[T2k]

What's so hard to understand???

R9700 = R300 / we all know
R9500 = RV300 / 4 Pipe, 1TMU, 128 Bit bus, 0.15µ, it's just a low-end R300 / Q4 2002
R350   = All same as R300 + DDR2, MAYBE 2 TMU per Pipe, Pixel & Vertex 2.0+ (But sureley not 3.0), 0.15µ / Q1 2003
RV350 = Some low-end version of R350, 0.13µ, Probally same performance as a R9700 / Q2-3 2003
R400   = Super-hyper DX9 Ultra card with unified P&V Shaders 3.0+ / Q4 2003

Based on latest ATI's product cycle, R400 should be announced before Q3 2003.

 

[FUDie]

]

]
Typical.. Completely Nvidia Fanboi overstaded popycock. Nothing unusual from you.

You mean as opposed to the usual fanATIc load o' bollocks that you and your lady friends are used to typing?

At least Hellbinder has friends, I doubt you have any.

I know that, like most here, you consider yourself to be an authoritative figure because you've been here longer, have more posts etc. You really do need to think before you type - because I really don't give a toss WHO you are - you type crap to me, I'm going to take you up on it. Fool.

I know that you consider yourself to be an authoritative figure because you made a couple of crappy games and have released more patches for your games than all other developers combined, but you really do need to think before you type - because no one really gives a toss WHO you are.

And even now the Ti and 8500 are on pretty even ground.

Rubbish. Apart from that, the 8500 drivers can't even compete with the GF3 nor Ti drivers

Rubbish. Maybe you should go check out some benchmarks sometime.

Yeah, I'm in the dark, but at least I have night vision on and can pretty much make out what I'm talking about, even if I don't have the facts. You and your friends take speculation, hearsay and outright rumors as if they were the gospel. And you have the nerve to point the finger at others.

You have no facts, you said it yourself. Take your useless speculation to a forum that gives a shit what you say.

Derek, you provide no facts to back what you say and resort to personal attacks whenever possible. Maybe you should grow up.