Will 3DMark03 ever become popular?

[Myrmecophagavir]

Originally posted by K.I.L.E.R
2k3 is 2300. :)2k3 = 2 * 1000 * 3 = 6000. Hence we should say UT2k+3, but why bother since that's the same number of chars as UT2003... :cool:

[StealthHawk]

Originally posted by DaveBaumann
Really, the only time a DX developer would use Cg is if they are porting across to OpenGL as well, if you are doing DirectX only development, why would you use Cg?

if Cg produced code faster than the DX9 HLSL for all video cards. even if Cg produced faster code than DX9 HLSL for only nvidia cards, but the code produced for other vendors was just as faster as the code produced by the DX9 HLSL, i don't see why Cg shouldn't be used.

basically, as long as Cg doesn't hurt other vendors with respect to the DX9 HLSL i see no problem with using Cg, or any other compiler any company decides to make.

i could see in the case of a benchmark that you would want to be "fair" and thus not use a first party compiler. but for retail games, optimizations will be made, by hand if need be, as that is the reality right now. so compilers like Cg might as well be used if there is anything to gain.

[Lezmaka]

Originally posted by K.I.L.E.R
2k3 is 2300. :)

:)

oooooooook, riiiiiiiiiiiiiiiight

maybe you better tell the peeps at infogrames that their website is messed up :rolleyes:

2.3k would be 2300, not 2k3

2k = 2000
3 = 3
2k3 = 2003

it's called a shortcut

[DaveBaumann]

Originally posted by StealthHawk
if Cg produced code faster than the DX9 HLSL for all video cards. even if Cg produced faster code than DX9 HLSL for only nvidia cards, but the code produced for other vendors was just as faster as the code produced by the DX9 HLSL, i don't see why Cg shouldn't be used.

But this is still a DirectX benchmark - what is the point of using a non-Microsoft/DirectX specific HLSL when DirectX9 comes with its own? DX9 HLSL is optimised - its optimised for DirectX, which is exactly what Futuremark are looking for.

[Cool Barn]

Originally posted by Admiral Horror
When does the geforceFX Warp25 Omega Turbo come out? Do u have any info on the specs?
It should be out just before NV30 ;)

[Kruno]

Originally posted by Lezmaka
oooooooook, riiiiiiiiiiiiiiiight

maybe you better tell the peeps at infogrames that their website is messed up :rolleyes:

2.3k would be 2300, not 2k3

2k = 2000
3 = 3
2k3 = 2003

it's called a shortcut

There was someone at Beyond3d who pointed out that what you say is wrong. :)


"Also... 2k is engineering notation for 2000 -- so 2k2 means 2200, and 2k3 means 2300...the 'k' is a multiplier, in this case 1000, which goes where the decimal place would be. So it should be 2k003. Which isn't any savings at all and doesn't look nearly so cool. Sorry... pet peeve :)"

http://www.google.com.au/search?q=cache:qdobYHcszzkC:www.dev0n.com/archives/003735.html+2k3+%3D+2300&hl=en&ie=UTF-8

That's also not including my electronics and physics class. :)

Don't argue with me. :)
Argue with my teachers and those people. :)
:rolleyes:

[jbirney]

Originally posted by StealthHawk
if Cg produced code faster than the DX9 HLSL for all video cards.

Well since CG still does not support PS1.4 why should it have been used in FurtureMark again? :) I also dont see how CG code could ever be faster then HLSL code. Just depends on the compilers/backend?

[stncttr908]

Originally posted by K.I.L.E.R
2k3 is 2300. :)

:)
Really? 2k=2000, 3 = 3. He said 2k3, not 2k3hundred.

Two thousand 3 = 2003. :afro:

[Hanners]

Originally posted by scott123
This effort by Futuremark is like amature hour. The graphics look terrible, and the benchmarks are based on conjecture not fact. The 1999/2000/2001 efforts were great, but 2003 IS a disaster.

I'll agree with you on one point here, 3DMark 2003s visuals just don't seem to have the same 'wow' effect on me as previous versions have (Although I imagine watching it run at 1FPS on my 8500 doesn't help :rolleyes: ;) ), and I can't quite put my finger on why. However, I think you almost have to ignore that fact and look what the tests are doing 'under the bonnet' as it were - The important thing is what is being tested, and to be honest if you want a benchmark for DirectX9 video cards, I can't see anything better to use right now.

[Kruno]

Originally posted by stncttr908
Really? 2k=2000, 3 = 3. He said 2k3, not 2k3hundred.

Two thousand 3 = 2003. :afro:

Like I said, don't bug me about it. I trust what my teachers say. :)
Bug them about it. :)

Also bug the guy who told me at Beyond3D. :)
Was it Mufu? Got to find that thread.

[creedamd]

Originally posted by K.I.L.E.R
Like I said, don't bug me about it. I trust what my teachers say. :)
Bug them about it. :)

Also bug the guy who told me at Beyond3D. :)
Was it Mufu? Got to find that thread.

2k3 = 2003 period, you are high.

[abb]

Originally posted by scott123
This effort by Futuremark is like amature hour. The graphics look terrible, and the benchmarks are based on conjecture not fact. The 1999/2000/2001 efforts were great, but 2003 IS a disaster.

Futuremark, to be a valid benchmark needs ALL the big players on board. Right now the 3dmark03 benchmark has been deminished, and is not really the benchmark of choise anymore.

I don't know if Futuremark will ever come out with another valid benchmark, but I wish them luck because it's back to the drawing board.

Scott Actually, the graphics look GREAT! if they do not, then it must be your card. The Ti4600 is a good card (I know, because I have one in my sons computer- but not in my main one). It is just not meant for this benchmark, just like the ATI Rage was not meant for 3DMark 2K1 when it first came out. It is a good benchmark. It is just not for Nvidia cards untill they up their technology and start putting out cards that can compete in todays future market. Nvidia had it with the NV30, but its final result ended with a card that is extremely hot, extremely loud and not that great of a performer compared to the hype that this card had. It is time for NVIDIA to go back to the drawing board. If NVIDIA had a board that would perform extremely well in this benchmark, then 3dmark 03 would never be in question.
Abb

[Chalnoth]

Originally posted by DaveBaumann
But this is still a DirectX benchmark - what is the point of using a non-Microsoft/DirectX specific HLSL when DirectX9 comes with its own? DX9 HLSL is optimised - its optimised for DirectX, which is exactly what Futuremark are looking for.
Only 3DMark03 is marketted as a gamers' benchmark, not as a DirecX benchmark. I think it would have been awesome if Futuremark put used Cg, if only to compare Cg vs. HLSL, or Cg vs. hand-optimized assembly (using the lowest PS vesrion possible).

[creedamd]

Originally posted by abb
/snip

. If NVIDIA had a board that would perform extremely well in this benchmark, then 3dmark 03 would never be in question.
Abb

bingo

[DaveBaumann]

Originally posted by Chalnoth
Only 3DMark03 is marketted as a gamers' benchmark, not as a DirecX benchmark.

Oh, don't be daft.

http://www.futuremark.com/companyinfo/3dmark03_whitepaper.pdf

3D Platforms. The dominant 3D platform on PCs is Microsoft DirectX. 3DMark releases have coincided with major DirectX versions - 3DMark 2000 with DirectX7, 3DMark2001 with DirectX8, and 3DMark2001 SE with DirectX 8.1. Accordingly, 3DMark03 targets DirectX® 9 features and continues the tradition being a forward-looking benchmark. We hope to give the user a view into state-of-the-art 3D graphics not only today, but also up to one and half years into the future. OpenGL is another popular 3D platform, especially in CAD and
scientific applications. As Direct3D has higher usage in games and more uniform driver support, we only support Direct3D. However, we actively monitor the progress of OpenGL adoption.

No doubt you've already read the comments fron Worm/Neeyik/AJ at B3D in what they are targetting - it just obtuse to say that its marketted as a 'Gamers benchmark' and not a DirectX benchmark. It is margetted as both.

[Kruno]

Originally posted by creedamd
2k3 = 2003 period, you are high.

No, I just listen to teachers rather than strangers. :)

Just asking: Is this an American way of doing it?

[digitalwanderer]

Originally posted by K.I.L.E.R
2k3 is 2300. :)

:)

? No, it's 2003...isn't it?

What's 2003 then?

EDITED BITS: I posted this before I noticed there was a second page where this was already thoroughly discussed. :rolleyes: I gotta stop posting before my first pot of coffee.

Can we just agree to disagree on this one? I'm gonna stick with the 3dm2k3 labelling convention just 'cause it seems to work with the way I referred to their past attempts as 3dm2k & 3dm2k1se. :)

[kyleb]

... if you are doing DirectX only development, why would you use Cg?

because that is what you got payed to do. ;)

[Cotita]

Originally posted by DaveBaumann
Thats exactly as I wrote in our Postmotem article. However, under no circumstances does it make sense for them to use Cg when there is a perfectly good, native DirectX, vendor unspecific, HLSL in DirectX9.

Really, the only time a DX developer would use Cg is if they are porting across to OpenGL as well, if you are doing DirectX only development, why would you use Cg?

There are many other reasons why a developer would use CG for directx, but there are already several threads discussing this, so I won't go into detail.

[Cotita]

Originally posted by Myrmecophagavir
2k3 = 2 * 1000 * 3 = 6000. Hence we should say UT2k+3, but why bother since that's the same number of chars as UT2003... :cool:

I understand this "equation" but where does the 2300 figure comes from?

Hmmm, maybe math is different across the world :confused:

[kyleb]

k = 1000

2k = 2000

2k3 =2003


thats math on my part of the planet :D

[jbirney]

Originally posted by Cotita
There are many other reasons why a developer would use CG for directx, but there are already several threads discussing this, so I won't go into detail.

Its been show that currently CG only supports features found in NV hardware (no support for Turform or PS1.4). If a developer has all the tools he needs to write HLSL which is a standard for all hardware, why not write just in HLSL?

[Myrmecophagavir]

Originally posted by Cotita
I understand this "equation" but where does the 2300 figure comes from?I have no idea, ask K.I.L.E.R... I refuse to say 2k3 meaning anything other than 6000 ;)

[Chalnoth]

Originally posted by jbirney
Its been show that currently CG only supports features found in NV hardware (no support for Turform or PS1.4). If a developer has all the tools he needs to write HLSL which is a standard for all hardware, why not write just in HLSL?
Cg has full support for PS 2.0 and the ARB extension fragment/vertex programs (Meaning it has full support for the R300 architecture). Truform needs no support in shaders.

[UDawg]

I can't shoot anything or interact with anything in 3DMark 2003 so I have no use for it. In other words, it isn't a game.