Radeon 9800 se -> 9800

[Bjorn]

I agree with Joe. Both ATi and NV are misleading in their naming.

Your average computer user knows absolutely nothing about graphics cards. If they see GeForce4 Ti 4600 and GeForceFX 5200 Ultra they will always think that the higher number with the 'Ultra' added is the faster chip, especially if they have heard that NV's latest and greatest is called the GeForceFX.
..

The problem here as i see it is that some of the new cards might be faster under certain circumstances and slower under other compared to the previous generation. With FSAA, AF f.e. You also have the problem with drivers that might hamper a new gen card in the beginning. And yet another problem is, what do you use to measure performance ?
Which games, syntethic benchmarks ?

Maybe a new card needs a new game to show it's true performance. And you might not see that game until a couple of months after the game is released. Or 1 year or more for that matter.

Bottomline is, it's not that i don't think that it would be a good idea with the name telling you speed/features and everything, but the problem is, can it be done in a good way ?

 

[Mariner]

Ah - it was so much easier in the olden days.

You knew that a Voodoo2 was faster than a Voodoo 1 in all situations. Ditto with a TNT2 over a TNT, a G400 over G200 or a Radeon over a Rage Fury.

Which was the first card to start these trends? Was the low-end Voodoo 3 slower than Voodoo 2 SLI? I seem to remember that (in some situations) GeForce was slower than TNT2 Ultra and then GeForce 3 was slightly slower than GF2 Ultra in the odd game.

The problems occur these days because the IHVs don't just bring out a single chip, they tend to bring out a whole new family of chips at the same time all of vastly differing performance.

Must be difficult for the various marketing departments of these companies to come up with some new names. That said, what's the betting that the NV40 is named "GeForce FX2" or something similar! Radeon 10K, anyone? 8)

 

[demalion]

demalion,

your posts in this thread show that you simply can't accept other people having a different opinion on a matter than you do.

Actually, this directly contradicts what I actually said to you, without even the beginning of an attempting to show why my stated reasons for the assertion are incorrect. All sorts of meaningless statements can be made this way, or by confusing references to one opinion with all other opinions you present.

However, it does seem to accurately describe your position. Did it occur to you to consider your characterization in the context of your own stance?

This only reinforces my opinion that I stated in my replies to you.

Well, this only reinforces that your opinions making sense are less important to you than your opinions being yours, as I've outlined.

I regret that I actually asked a question in my first reply, even though I thought it was needed for clarification, it apparently gave you the opportunity to pull me into a debate that focussed more around your definition of certain words or terms than the issue at hand.

It isn't just "my definition" of certain words, it is me establishing that my definition makes sense and your saying your definition is the only one that does.

Any further reply to your posts was a mistake. This whole thread, probably a few thousand words in total, has led nowhere and I'm not really surprised about that outcome.

Nor should you be when you fail to pay attention to what the other person says to you.

But hey, all this is subjective, everyone is right in their opinion. Well, except when you subjectively decide the other guy doesn't make sense, of course.

 

[KimB]

Ah - it was so much easier in the olden days.

You knew that a Voodoo2 was faster than a Voodoo 1 in all situations. Ditto with a TNT2 over a TNT, a G400 over G200 or a Radeon over a Rage Fury.

There were prolems then, too.

For example, around the time of the release of the TNT2, nVidia released the TNT2 M64, which had higher clock speeds than the original TNT, but a 64-bit memory bus. Prior to that was the failed Vanta...

And then the TNT2 vs. Voodoo3 debates were quite vehement.

Then that fall, there was a big showdown between the GeForce SDR/DDR, the Savage 2000, and the Rage Fury MAXX. There were "golden sample" versions of the TNT2 that came close to 200MHz for core and memory.

And then came the release of the GeForce2 GTS and the GeForce2 MX.

Anyway, I guess I'm just saying that it's been confusing for a long time.

 

[demalion]

Are you losing track of the fact that these names are associated with products? If the GF 4 MX product offered vertex and pixel shader functionality, ATI's naming of the 9000 being "at least as bad" would no longer be an unreasonable statement. Do you think this doesn't make sense?
If you haven't read my link to where I discussed all of this already, please do so as soon as possible.

No, it doesn't make sense since you continue to compare the 9000 to the generation before (8000+, DX8) instead of the 9500+, DX9 that you should compare it according to it's name, "technology generation".

OK, this is different than the last time I addressed this exact sentiment, how?

Bjorn, consider this example of your argument technique:

"It doesn't make sense to say that the carpet is white, because when I look at it, it looked red to me."

Now, if you happen to have a red light in the room, and turn it on, a white carpet looks red to you, and you have a valid reason for saying that it is red as far as just that information.

However, you can close your eyes or turn your head when someone turns on a white light, and continue to believe the carpet is red, even if it is then revealed not to be the case to someone who actually looks at the evidence provided.

Talking with you strikes me as you trying to hold a contest to see whether you get tired of squeezing your eyes close before I get tired of turning on the white light, where I can't even get you to recognize that I'm turning on a switch or changed the bulb from red to white.

I'm not interested in continuing to watch you contort yourself turning your head and squeezing your eyes shut any more, but it was "interesting" to watch for a while.

 

[Mariner]

And then the TNT2 vs. Voodoo3 debates were quite vehement.

It's really more the cards from the same companies that I'm thinking of.

You are quite right about the M64 version of the TNT2 - this was probably the first of the new chips which played on the name of the high-end version for OEM purposes. It obviously worked as this was NVidia's best selling chip at the time, I believe!

Thinking about it, however, I suppose the Voodoo Rush and Voodoo Banshee played on the Voodoo name as well, despite offering lower performance than the Voodoo itself. The Rush was a hack, however in an attempt to get the combined 2D/3D started for 3DFX and the Banshee was their first 'proper' combined 2D/3D accelerator and both were utter failures, really.

Ultimately, it is the OEM market which requires these bastardised low-end chips and I suppose marketing has to give them similar names to the high end versions to attract the OEM deals. It would be nice if a scheme could be found which didn't mislead about the capabilities or performance of a budget chip, but I think there are just too many variables for this to happen.

 

[Bjorn]

Bjorn, consider this example of your argument technique:

"It doesn't make sense to say that the carpet is white, because when I look at it, it looked red to me."
......

I must say that i haven't had this much fun in a long time.

Anyway, i'm of the opinion that i made it rather clear as for why i think you're wrong so i think it's time that we end this discussion.

 

[RussSchultz]

What!? You've only gone 5 pages.

Wimp.

 

[Bjorn]

What!? You've only gone 5 pages.

Wimp.

Yep, and i'm not ashamed to admit it either

 

[Althornin]

The problem here as i see it is that some of the new cards might be faster under certain circumstances and slower under other compared to the previous generation.

Well, I'd have to say i dont find the 9500/9600 naming to be very confusing at all:
in DX9, the 9600 is gonna pwn the 9500, according to the B3D article dealing with TR:AOD.
In fact, i bet that the 9000 beats the 8500 in it. The only naming problems ATI has are the 9100 (never shoulda rebadged the same card with a higher number) and the 9800SE.
people like chalnoth hyperventilatiing over the 9600 and 9000 thing are missing the boat.

 

[KimB]

Well, I'd have to say i dont find the 9500/9600 naming to be very confusing at all:
in DX9, the 9600 is gonna pwn the 9500, according to the B3D article dealing with TR:AOD.

TR: AOD isn't representative of every "DX9" game. The simple fact is that in the majority of programs, the 9600 is behind. I see no reason why the 9600 would necessarily be any better in those games that make significant use of 3rd-gen features.

In fact, i bet that the 9000 beats the 8500 in it.

I highly doubt that! The 9000 is still little more than an 8500 with a 4x1 architecture.

The only naming problems ATI has are the 9100 (never shoulda rebadged the same card with a higher number) and the 9800SE.

Um, the 9100 was renamed because of the 9000. How can the 9000's name possibly not be problematic if the 9100's is?

 

[OpenGL guy]

Well, I'd have to say i dont find the 9500/9600 naming to be very confusing at all:
in DX9, the 9600 is gonna pwn the 9500, according to the B3D article dealing with TR:AOD.

TR: AOD isn't representative of every "DX9" game. The simple fact is that in the majority of programs, the 9600 is behind. I see no reason why the 9600 would necessarily be any better in those games that make significant use of 3rd-gen features.

Higher engine clock = more shading power. Pretty simple, really. If an application is using lots of long shaders, then the engine clock can become very important.

 

[Althornin]

TR: AOD isn't representative of every "DX9" game. The simple fact is that in the majority of programs, the 9600 is behind. I see no reason why the 9600 would necessarily be any better in those games that make significant use of 3rd-gen features.

Oh, so you know what is representative of every DX9 game?
The simple fact is, as OpenGl guy says, the 9600 will perform better in DX9 shader intensive games. This is backed up by TR:AOD. Where is your evidence to the contrary?

I highly doubt that! The 9000 is still little more than an 8500 with a 4x1 architecture.

I was under the impression that some shader improvements were made?

EDIT:

What's interesting here is that the performance of Radeon 9000 PRO in the raw Pixel Shader test is actually a little above that the Radeon 8500 in most cases, maintaining about a 10% difference across the middle resolutions. The lack of the extra texture unit is not going to harm the performance much in the raw Pixel Shader performance and given the clock speed equality between 9000 PRO and 8500 it may be that ATI have optimised the pipeline a little as well.

Thats from B3D's Review of the 9000. I was under the impression that it was larger than that, but still, it might be enough to matter.

Um, the 9100 was renamed because of the 9000. How can the 9000's name possibly not be problematic if the 9100's is?

Um, what?
The 9100 was renamed to get rid of a surplus of R200 cores....
Bigger number=sell more....
Has jack all to do with 9000
EDIT2: Besides, you cant bash ATI for BOTH of them! If you think the 9000 name was a mistake, then the 9100 name FIXES the mistake... (numbers relative to performance). If you dont, then the 9100 was the mistake.

 

[Bjorn]

Um, what?
The 9100 was renamed to get rid of a surplus of R200 cores....
Bigger number=sell more....
Has jack all to do with 9000

If the only thing they needed was a bigger nr, why not call it 8600 ?
Yes, the reason for the renaming was the surplus, but what was the reason for that specific name ?

I would say because the 9100 was a bit faster (although that wasn't always the case) then the 9000 and thus, it has a lot to do with the 9000.

EDIT2: Besides, you cant bash ATI for BOTH of them! If you think the 9000 name was a mistake, then the 9100 name FIXES the mistake... (numbers relative to performance). If you dont, then the 9100 was the mistake.

The full ATi statement was "first digit = technology generation, second relative to performance". So you are free to bash Ati for both names if you have the same opinion as i do about technology generation. If not , then we disagree

Well, I'd have to say i dont find the 9500/9600 naming to be very confusing at all:
in DX9, the 9600 is gonna pwn the 9500, according to the B3D article dealing with TR:AOD.
In fact, i bet that the 9000 beats the 8500 in it. The only naming problems ATI has are the 9100 (never shoulda rebadged the same card with a higher number) and the 9800SE.

And here we see the problem that i was talking about with regards to numbers indicating performance. I looked at an old review of the then freshly released 9100 and the 9000 and 9100 swapped places quite a lot.

The 9100 was faster in 3D Mark 2001, UT2003 and slower in Serious Sam, Castle Wolfenstein

And if we do the same with the 9600, slower in UT2003, Aquanox 2 and 3D Mark 2003 and faster in Splinter Cell, TR:AOD

And enable 8X AF in f.e UT2003, 1600*1200 and you'll get 15.7 fps on a R9600 PRO vs 37.4 on a 9500 PRO. Enable 4X FSAA to even it out a bit, although the 9500 is still quite a bit faster. 15.7 FPS (same as with only 8X AF btw) for the 9600 PRO and 26.8 on the 9500 PRO. The difference is smaller in lower resolutions though. Numbers taken from Tom's hardware.