Half-Life 2 DirectX 9 Performance
Posted by: Newsfactory on: 08/28/2003 01:51 PM [ Print | 14 comment(s) ] · 6333 views
Here's a story I posted at Driver Heaven earlier today! As I'm pushed for time, I'll post it as is.
I recently posted an article from Beyond 3D called ATI vs. Nvidia Tomb Raider: Angel Of Darkness DX9 Benchmarks! The importance of this benchmark is that it is the first to full utilise PS2.0 DX9 functionality within a gaming environment.
What most people seem to have missed was they weren't looking at the game which in our Great leaders words "Zardon: I agree totally the game is utter garbage" but were looking at how well ATI's and Nvidia's High end and Midrange cards compared when the DX9 Pixel Shader 2.0 functionality is enable! Here's a rip from the thread.
Looking at the benchmarks , Nvidia is getting a spanking! also worth noting is that in all cases Cg compiled shaders were enabled for the NVIDIA boards and disabled for the ATI boards. There is no difference in the output of the shaders compiled by Cg, however this should represent the the best case for the NVIDIA boards.
WaltC: I think if you read the article you'll see that the game runs much less "sh*tty" on the Radeons...
Come on, now, after all the huff & puff coming out of nVidia about using "real 3d games" you surely can't object to using a real DX9 game as a test...? It's not a benchmark--it's a real 3d game. Lots of "real 3d games" run far less optimally than benchmarks, and other 3d games. Doesn't mean they aren't "real 3d games," however. What's interesting to me about this game is the fact that it's probably the first real DX9 game to hit the market, and looking at nVidia's DX9-feature support scores in this game it's not hard to see why the company quit FutureMark last year... If anything, nVidia's DX9 feature performance is even worse in this "real 3d game" than it is in 3dMk03.... It's an eye-opener in that regard, IMO...
and I said this at the time WaltC is right with what he says and just wait and see what happens when Half-Life 2 is released? I bet you wished you'd had bought an ATI card .
Now over to my good freind Matt Burris at 3DGPU
This isn't confirmed as 100% official, so keep that in mind. On the HalfLife2.net forums, a gamer emailed Gabe Newell of Valve Software, and asked him a question in regards to GeForce FX cards dismal performance in Tomb Raider using PS 2.0 brought up by the article on Beyond3D (see this post). Here's the question he asked Gabe:
Is a ATi 9800pro card really alot better for HL2 then Nvidia's FX5900? Or is the difference not that big (quality & fps wise)?
For the answer he got: head on over to 3DGPU
I recently posted an article from Beyond 3D called ATI vs. Nvidia Tomb Raider: Angel Of Darkness DX9 Benchmarks! The importance of this benchmark is that it is the first to full utilise PS2.0 DX9 functionality within a gaming environment.
What most people seem to have missed was they weren't looking at the game which in our Great leaders words "Zardon: I agree totally the game is utter garbage" but were looking at how well ATI's and Nvidia's High end and Midrange cards compared when the DX9 Pixel Shader 2.0 functionality is enable! Here's a rip from the thread.
Looking at the benchmarks , Nvidia is getting a spanking! also worth noting is that in all cases Cg compiled shaders were enabled for the NVIDIA boards and disabled for the ATI boards. There is no difference in the output of the shaders compiled by Cg, however this should represent the the best case for the NVIDIA boards.
WaltC: I think if you read the article you'll see that the game runs much less "sh*tty" on the Radeons...
Come on, now, after all the huff & puff coming out of nVidia about using "real 3d games" you surely can't object to using a real DX9 game as a test...? It's not a benchmark--it's a real 3d game. Lots of "real 3d games" run far less optimally than benchmarks, and other 3d games. Doesn't mean they aren't "real 3d games," however. What's interesting to me about this game is the fact that it's probably the first real DX9 game to hit the market, and looking at nVidia's DX9-feature support scores in this game it's not hard to see why the company quit FutureMark last year... If anything, nVidia's DX9 feature performance is even worse in this "real 3d game" than it is in 3dMk03.... It's an eye-opener in that regard, IMO...
and I said this at the time WaltC is right with what he says and just wait and see what happens when Half-Life 2 is released? I bet you wished you'd had bought an ATI card .
Now over to my good freind Matt Burris at 3DGPU
This isn't confirmed as 100% official, so keep that in mind. On the HalfLife2.net forums, a gamer emailed Gabe Newell of Valve Software, and asked him a question in regards to GeForce FX cards dismal performance in Tomb Raider using PS 2.0 brought up by the article on Beyond3D (see this post). Here's the question he asked Gabe:
Is a ATi 9800pro card really alot better for HL2 then Nvidia's FX5900? Or is the difference not that big (quality & fps wise)?
For the answer he got: head on over to 3DGPU
« Exclusive: Symantec Product Activation Only For North America · Half-Life 2 DirectX 9 Performance
· Opera 7.20 Beta 7 - Mozilla 1.5 Beta »
Comment
|
digitalwonderer Unregistered |
#46707 Posted on: 08/28/2003 06:49 PM
WaltC ain't wrong a whole lot, at least I can't recall the last time. Zardon, on the other hand..... |
Comment
|
incoherent Unregistered |
#46710 Posted on: 08/28/2003 07:21 PM
Good thing I moved all my friends to ATi boards (BBA...none of this oem stuff, they don’t know computers very well which means ill be getting calls about fried video cards because they thought over clocking "was a good thing")to 9600 's and up.... Slowly persuading those who don’t that Half-life 2 powered games are going to suck on nv boards. props to the Nforce 2. Much love. |
Comment
|
shaderboy Unregistered |
#46714 Posted on: 08/28/2003 11:26 PM
Its down hill all the way for the FX. The games im working on all show the same behaviour. It looks like only DOOM 3 will be different, but thats an engine and not a game. Dont bank on it being a particularly good game ! One game is not enough to make the FX worth while, especially when you consider that the R3x0 wont be as far behind in DOOM as the FX is in every other game. |
Comment
|
Runner Unregistered |
#46715 Posted on: 08/28/2003 11:47 PM
I believe that this will end up being a non-issue. The 50.xx drivers will fix it, one way or another. |
Comment
|
shaderboy Unregistered |
#46716 Posted on: 08/28/2003 11:54 PM
No they wont... sorry but they wont. |
Comment
|
thomas997 Unregistered Posts: 11 Joined: 2003-04-21 |
#46721 Posted on: 08/29/2003 03:28 AM
Drivers arent everything, you have to have hardware to go along with them.. |
Comment
|
Runner Unregistered |
#46722 Posted on: 08/29/2003 03:45 AM
You don't seem to understand that they have the hardware...just watch. Shit like this is rediculous, they WILL fix it, one way or another. Nvidia knows what they're doing, and you can bet the 50's will be used to rain on ATi's parade, probably around the 9900 launch, as they've done before. Bank on it. |
Comment
|
shaderboy Unregistered |
#46729 Posted on: 08/29/2003 10:39 AM
This undying faith in nvidia is laughable, i cant believe people still believe it will be ok. The main reason why NV30/35 is slower than R3X0 in PS2.0 effects is the hardware configuration. The NV30/35 has 4 PS2.0 units that must also do some texture lookup work and thus can't calculate shader programs at full speed. Moreover, most NV30/35 assembly instructions take 1 clock cycle, but some of them need 2, which slows down the processing even more. R300 has 8 PS2.0 units that are reserved completely for shader execution, and they can even do 2 instructions in one cycle when the slowest are 1 clock cycle per instruction. This is theoretically 3 to 5 times faster than NV30/35. The only advantage the FX line has is higher clockspeed, so theoretically R3X0 should be 2 to 3 times faster in PS2.0 shading, which seems to be the case in reality. No amount of driver optimization will close that kind of gap, UNLESS cheating is involved and the reduction of image quality is high. Yes they did close the gap in 3Dmark03 PS2.0 test, but this was due to re-writing the code. They cant do this for every game, the drivers would be massive and the cost high. The chances are they will just slowly forget the NV3x and move on to better things. |
Comment
|
hinkle Unregistered |
#46730 Posted on: 08/29/2003 10:45 AM
hm, maybe HL2 doesn't need 8 shader units to run flawlessly... Did that come up in your mind? I would doubt if more than 25% of the calculations for HL2 are PS2.0 calculations. Mostly it's |
Comment
|
hinkle Unregistered |
#46731 Posted on: 08/29/2003 10:46 AM
... mostly it's |
Comment
|
shaderboy Unregistered |
#46733 Posted on: 08/29/2003 11:05 AM
HL2 scales down very well, so it will run well on lesser cards. It doesnt scale down on a high spec 9800 machine, it does on a high spec 5900 machine. Sorry but he FX isnt a patch on the 9800 in DX9 games, it has to run with less effects. HL2 doesnt need all the power of the R3x0 but it needs more than the FX has (to run flat out). I can say this because i have been loaned a copy of the game by a friend who works for valve. Im thinking of changing jobs after seeing it in the flesh, its fantastic. |
Comment
|
hinkle Unregistered |
#46735 Posted on: 08/29/2003 11:19 AM
I already played the beta version of HL3 on a nv45 and it worked like a charm... whatever |
Comment
|
shaderboy Unregistered |
#46736 Posted on: 08/29/2003 11:25 AM
Maybe you should accept that there are people out there who A) know more than you. B) work in the industry and therefore have access to things you do not. C) arent blinded by fanboyism. |
Comment
|
Runner Unregistered |
#46786 Posted on: 08/30/2003 07:12 AM
Undying faith? Hardly, I detest nvidia as a company, but there isn't a chance in hell this won't be fixed by drivers. Call it whatever you like, but they will get the performance up to par, one way or another. Im about as far from a fanboy for nvidia as you can get, btw. But personally I think you're underestimating the lengths this particular company will go to in order to keep the performance crown they don't believe they have lost(they actually have, they just won't admit it). Image is everything, and im not talking about the graphics that get displayed on the screen either, we all know that tired story. |
