Benching, games, optumizations, synthetics and you…

The_Neon_Cowboy · Mar 11, 2005, 10:10 AM

I find my self arguing more and more about a point which I find a good topic for discussion. See what you views and opinions on this topic are… I will throw up a couple of basic questions / points and my views and see what you have to say... then we will get to my actual point is twads the end but this is a primer...

[color=blue]Think, How do you feel about game optimizations in general?[/color]

I personally think their fine assuming they try to support what they can obviously they can’t support everything.

[color=blue]Think, How do you feel about game advanced optimizations? [/color]
[color=blue](As for affecting the performance of 1 brands cards only)[/color]

This is where the problems lye to me when a developer is in bed with a company and they add only there extra futures or tune the game to run better on only their brand cards. They can even make the game run quite differently. Separate paths, textures, bit depths methods of doing things... almost like 2 very different games in 1. Being different not really / fairly comparable…

[color=blue]Think, How do you feel about any preferential treatment by game developers?[/color][color=blue](As in thorough compatibility and bug checking for only one brand)[/color]

Again another big problem for example a “Meant to be played” game thoroughly optimized tweaked and tested on a range of NVIDIA hardware with little to no ATI testing or feedback. Notorious for having bugs on ATI cards be it the drivers or games fault. It seems the developer doesn’t care about how well and proper their game runs on all hardware but long as it run well on their partner brands hardware… I personally thing the graphics companies and developers should cut it out and work together more closely together regardless of sponsorship. Game makers should worry about their games not about it promoting a given brand of hardware. That really not their business or is it? Any more it’s hard to tell.

[color=blue]Think, How do you feel about people basing their futures purchases on the[/color] [color=blue]results of 1 or more optimized games? For example ATI for only HL2 benchmark scores or NVIDIA for only doom III benchmark scores?[/color]

Personally I think that nuts unless you plan to play one game on your pc for the life of the card (Which isn’t going to happen). You should look at a lot of games and synthetic benchmarks. Several reviews users feed backs etc… It seems games have taken on a role as marketing tools but appears they really aren’t seen that way by average the consumer.

[color=blue]Think, How do you feel about reviews comparing say an ATI and NVIDIA brand card using a large grouping of mostly “Meant to be played” or “Get in the game” [/color][color=blue]games?[/color]

Personally I think it’s pretty unfair to compare for example an ATI and NVIDIA brand card in a lopsided pool of tests. Like 10 meant to be played games and 1 get in the game. Or like 8 get in the game, 2 unendorsed OpenGL, and maybe 1 get in the game. I mean if you want to portray an image you just need to fish for the right games and you can make a card look superior or inferior buy the games you choose as as tests. For example there a dozen game benchmarks that make a X800XT look faster then a dual 6800U setup and then again a bunch more that shot the SLI setup much faster. That will never change but that’s not my point. That why you should read several reviews...

This is something I want people to realize….
It’s not the reviewers at all I’m picking on here.

[color=blue]Think, Who polices games for driver optimizations? Who polices games optimizations added by the developer for any given brand...?[/color]

ANSWER: for the most part NO ONE

You wouldn’t trust a benchmark put out by a given brand and base your purchases on it would you. One thats optimized for heavily for their cards and their driver highly optimized for the benchmark. Would you? No that would be obvious and silly… yes the ability to turn of all optimizations via the driver and game could help but those options are not really available for whas built in the game…

But that’s what we are accepting! But in another form games! Games! We rely more on games benchmarks any more. Most sites have dropped synthetic testing all together!!! And users give them less and less weight… When games are more and more becoming marketing pawns for hardware…not much we can do about that ...

But those same synthetics such as 3dmark that’s the only ones that are policed to any extent and are fairer are the ones we turning away from. Go figure! Thats why I say we still need good synthetics

Discuss....

PoopyTheJ · Mar 13, 2005, 02:54 AM

well, optimizations for D3, HL2, CSS, farcry are included in the DHzp drivers because their big titles, we obviously want people to have the best IQ and performance we can get and we plan on bringing those same opts over for ATI cards. Are we biased? it's a rhetorical question mostly as sites like DH who use a very wide range of games to benchmark and test out compatibility issues are a far better read, and more reliable than OMFG 3D05 ownzers on the Nvidia 7800000000XT PEFX!

The_Neon_Cowboy · Mar 13, 2005, 03:07 AM

Quote:

Originally Posted by PoopyTheJ

well, optimizations for D3, HL2, CSS, farcry are included in the DHzp drivers because their big titles, we obviously want people to have the best IQ and performance we can get and we plan on bringing those same opts over for ATI cards. Are we biased? it's a rhetorical question mostly as sites like DH who use a very wide range of games to benchmark and test out compatibility issues are a far better read, and more reliable than OMFG 3D05 ownzers on the Nvidia 7800000000XT PEFX!

I'm not talking about modded drivers but those direct from the companies...

Also Yes DH does and many other di run a wide array, that why I said i'm not really poicking on reviewers but other sites I've seen some pretty lop sided testing done... kinda likeThe scales where tipped before the 1st test was ever run...

What I want to see is a wide range of synthectics brought back into the mix. I would also like to see verry strict policeing by the companies that put these out as to look for any specfic bechmark optumiztions. Kinda like what 3dmark is and has been doing...

PoopyTheJ · Mar 13, 2005, 03:24 AM

Frankly 3dmark blows, their shader programming is deranged at best, and no other games use engines even reasonably close to the futuremark engine that I've seen. Is it a good idea to look for cheats in benchmarks? probably, however the whole don't render the scene if you can't see it thing nvidia did a long time ago would be great if you could manage to do that kind of z/w occlusion culling in practice....

The_Neon_Cowboy · Mar 13, 2005, 04:01 AM

Quote:

Originally Posted by PoopyTheJ

Frankly 3dmark blows, their shader programming is deranged at best, and no other games use engines even reasonably close to the futuremark engine that I've seen. Is it a good idea to look for cheats in benchmarks? probably, however the whole don't render the scene if you can't see it thing nvidia did a long time ago would be great if you could manage to do that kind of z/w occlusion culling in practice....

3dmark03 is a joke becouse it runs diffrently on diffent cards.
Not running unsuported test isn't what i'm talking about either...

but I'd rely alot of 01 and 05.. and actually the scores and results i see in 3dmark01se can tell me leaps a bound about a card, a system, drivers.... it also can show me the tinyist varriation in my setting weather it improves on hampers my performance

But the point is if it done in every applicaion thats fine but when it does it just for the bechmark thats when the problem shows up. Becouse thier running diffrently just for the bechark to get a higher score. even algoritoms can be used to tell when it thinks it being bechmarked and it can use diffrent settings to up the FPS artifically to get a better score...

What nv did a while back they where rendering the bullets and alot of effects fo a verry long time. Hince it was a massive chnage when I switched over to ati in just about all my games... effects and IQ that I had been missing.. no doubt by now they've got it beter perfected but still nothing I approve of...

poxmem · Mar 13, 2005, 05:52 AM

[COLOR=DarkOrange]How do you feel about game optimizations in general?[/COLOR]

Optimization is one of the most important aspects for any program not just games and it's often neglected (or sometimes there isn't time or the team isn't capable) just like people neglect to buy a good PSU which is the most important component.

[COLOR=DarkOrange]How do you feel about game advanced optimizations?
(As for affecting the performance of 1 brands cards only)[/COLOR]

IMO a game should be designed and optimized first to run on all major without specific optimization, second it should be a priority to optimize it for the brand that has the best hardware&driver (currently ATI), next it should be optimized for other brands if it's worth it (there's time, money, competence available) and NVIDIA has a decent Shader Model 3 among other things but ATI's technologies including 3Dc still have the advantage. To be fair a game would have to be optimized for all graphics cards without specific optimizations, although I'm no NVIDIA fan HL2 should have had problems with NVIDIA cards but maybe NVIDIA is partly responsible here.

[COLOR=DarkOrange]How do you feel about any preferential treatment by game developers?(As in thorough compatibility and bug checking for only one brand)[/COLOR]

A game should run without problems on all cards that meet the minimum system requirements, to test it mostly on a single brand it not recommended. The game should detect the graphics card model and use the best settings (from it's database) for that card's performance.
“Meant to be played” and "Get in the game" are bad strategies, a developer shouldn't promote one brand or the other. There are games that are "meant to be played" on NVIDIA that run and look better on ATI.

[COLOR=DarkOrange]How do you feel about people basing their futures purchases on the results of 1 or more optimized games? For example ATI for only HL2 benchmark scores or NVIDIA for only doom III benchmark scores?[/COLOR]

They're free to do what they want, its their wasted money, they could even pay 600$ for a 6800 GT for all I care.

[COLOR=DarkOrange]How do you feel about reviews comparing say an ATI and NVIDIA brand card using a large grouping of mostly “Meant to be played” or “Get in the game” games?[/COLOR]

I don't like the “Meant to be played” or “Get in the game” but I think that by testing with different games people can see the card's performance under different situations, one game may have dynamic shadows while another may have better A.I. and character animation. Every card should be tested with 40+ games and all benchmarks but that is not possible due to time constraints.

I give 3dmark less and less weight because it favors NVIDIA cards or Forceware is optimized for 3dmark, it's ridiculous to see a NVIDIA card better than an ATI card in 3dmark when it's inferior in all the current games. What does 3dmark05 give me ? Performance in future games ?

Mar 11, 2005, 10:10 AM	#1
The_Neon_Cowboy HardwareHeaven Extreme Member Join Date: Dec 2002 Location: U.S.A. Posts: 16,009 Rep Power: 92 System Specs	Benching, games, optumizations, synthetics and you… I find my self arguing more and more about a point which I find a good topic for discussion. See what you views and opinions on this topic are… I will throw up a couple of basic questions / points and my views and see what you have to say... then we will get to my actual point is twads the end but this is a primer... [color=blue]Think, How do you feel about game optimizations in general?[/color] I personally think their fine assuming they try to support what they can obviously they can’t support everything. [color=blue]Think, How do you feel about game advanced optimizations? [/color] [color=blue](As for affecting the performance of 1 brands cards only)[/color] This is where the problems lye to me when a developer is in bed with a company and they add only there extra futures or tune the game to run better on only their brand cards. They can even make the game run quite differently. Separate paths, textures, bit depths methods of doing things... almost like 2 very different games in 1. Being different not really / fairly comparable… [color=blue]Think, How do you feel about any preferential treatment by game developers?[/color][color=blue](As in thorough compatibility and bug checking for only one brand)[/color] Again another big problem for example a “Meant to be played” game thoroughly optimized tweaked and tested on a range of NVIDIA hardware with little to no ATI testing or feedback. Notorious for having bugs on ATI cards be it the drivers or games fault. It seems the developer doesn’t care about how well and proper their game runs on all hardware but long as it run well on their partner brands hardware… I personally thing the graphics companies and developers should cut it out and work together more closely together regardless of sponsorship. Game makers should worry about their games not about it promoting a given brand of hardware. That really not their business or is it? Any more it’s hard to tell. [color=blue]Think, How do you feel about people basing their futures purchases on the[/color] [color=blue]results of 1 or more optimized games? For example ATI for only HL2 benchmark scores or NVIDIA for only doom III benchmark scores?[/color] Personally I think that nuts unless you plan to play one game on your pc for the life of the card (Which isn’t going to happen). You should look at a lot of games and synthetic benchmarks. Several reviews users feed backs etc… It seems games have taken on a role as marketing tools but appears they really aren’t seen that way by average the consumer. [color=blue]Think, How do you feel about reviews comparing say an ATI and NVIDIA brand card using a large grouping of mostly “Meant to be played” or “Get in the game” [/color][color=blue]games?[/color] Personally I think it’s pretty unfair to compare for example an ATI and NVIDIA brand card in a lopsided pool of tests. Like 10 meant to be played games and 1 get in the game. Or like 8 get in the game, 2 unendorsed OpenGL, and maybe 1 get in the game. I mean if you want to portray an image you just need to fish for the right games and you can make a card look superior or inferior buy the games you choose as as tests. For example there a dozen game benchmarks that make a X800XT look faster then a dual 6800U setup and then again a bunch more that shot the SLI setup much faster. That will never change but that’s not my point. That why you should read several reviews... This is something I want people to realize…. It’s not the reviewers at all I’m picking on here. [color=blue]Think, Who polices games for driver optimizations? Who polices games optimizations added by the developer for any given brand...?[/color] ANSWER: for the most part NO ONE You wouldn’t trust a benchmark put out by a given brand and base your purchases on it would you. One thats optimized for heavily for their cards and their driver highly optimized for the benchmark. Would you? No that would be obvious and silly… yes the ability to turn of all optimizations via the driver and game could help but those options are not really available for whas built in the game… But that’s what we are accepting! But in another form games! Games! We rely more on games benchmarks any more. Most sites have dropped synthetic testing all together!!! And users give them less and less weight… When games are more and more becoming marketing pawns for hardware…not much we can do about that ... But those same synthetics such as 3dmark that’s the only ones that are policed to any extent and are fairer are the ones we turning away from. Go figure! Thats why I say we still need good synthetics Discuss.... __________________

Mar 13, 2005, 02:54 AM	#2
PoopyTheJ DriverHeaven Extreme Member Join Date: Jan 2003 Location: Ohio Posts: 1,588 Rep Power: 65	well, optimizations for D3, HL2, CSS, farcry are included in the DHzp drivers because their big titles, we obviously want people to have the best IQ and performance we can get and we plan on bringing those same opts over for ATI cards. Are we biased? it's a rhetorical question mostly as sites like DH who use a very wide range of games to benchmark and test out compatibility issues are a far better read, and more reliable than OMFG 3D05 ownzers on the Nvidia 7800000000XT PEFX! __________________ [COLOR=Blue]“Alright boys, let’s start her up and see why she doesn’t work.” - John Fritz "I aim to misbehave" Malcolm Reynolds ATI Catalyst Beta Tester[/COLOR]

Mar 13, 2005, 03:24 AM	#4
PoopyTheJ DriverHeaven Extreme Member Join Date: Jan 2003 Location: Ohio Posts: 1,588 Rep Power: 65	Frankly 3dmark blows, their shader programming is deranged at best, and no other games use engines even reasonably close to the futuremark engine that I've seen. Is it a good idea to look for cheats in benchmarks? probably, however the whole don't render the scene if you can't see it thing nvidia did a long time ago would be great if you could manage to do that kind of z/w occlusion culling in practice.... __________________ [COLOR=Blue]“Alright boys, let’s start her up and see why she doesn’t work.” - John Fritz "I aim to misbehave" Malcolm Reynolds ATI Catalyst Beta Tester[/COLOR]

Mar 13, 2005, 05:52 AM	#6
poxmem DriverHeaven Junior Member Join Date: Mar 2005 Posts: 25 Rep Power: 0	[COLOR=DarkOrange]How do you feel about game optimizations in general?[/COLOR] Optimization is one of the most important aspects for any program not just games and it's often neglected (or sometimes there isn't time or the team isn't capable) just like people neglect to buy a good PSU which is the most important component. [COLOR=DarkOrange]How do you feel about game advanced optimizations? (As for affecting the performance of 1 brands cards only)[/COLOR] IMO a game should be designed and optimized first to run on all major without specific optimization, second it should be a priority to optimize it for the brand that has the best hardware&driver (currently ATI), next it should be optimized for other brands if it's worth it (there's time, money, competence available) and NVIDIA has a decent Shader Model 3 among other things but ATI's technologies including 3Dc still have the advantage. To be fair a game would have to be optimized for all graphics cards without specific optimizations, although I'm no NVIDIA fan HL2 should have had problems with NVIDIA cards but maybe NVIDIA is partly responsible here. [COLOR=DarkOrange]How do you feel about any preferential treatment by game developers?(As in thorough compatibility and bug checking for only one brand)[/COLOR] A game should run without problems on all cards that meet the minimum system requirements, to test it mostly on a single brand it not recommended. The game should detect the graphics card model and use the best settings (from it's database) for that card's performance. “Meant to be played” and "Get in the game" are bad strategies, a developer shouldn't promote one brand or the other. There are games that are "meant to be played" on NVIDIA that run and look better on ATI. [COLOR=DarkOrange]How do you feel about people basing their futures purchases on the results of 1 or more optimized games? For example ATI for only HL2 benchmark scores or NVIDIA for only doom III benchmark scores?[/COLOR] They're free to do what they want, its their wasted money, they could even pay 600$ for a 6800 GT for all I care. [COLOR=DarkOrange]How do you feel about reviews comparing say an ATI and NVIDIA brand card using a large grouping of mostly “Meant to be played” or “Get in the game” games?[/COLOR] I don't like the “Meant to be played” or “Get in the game” but I think that by testing with different games people can see the card's performance under different situations, one game may have dynamic shadows while another may have better A.I. and character animation. Every card should be tested with 40+ games and all benchmarks but that is not possible due to time constraints. I give 3dmark less and less weight because it favors NVIDIA cards or Forceware is optimized for 3dmark, it's ridiculous to see a NVIDIA card better than an ATI card in 3dmark when it's inferior in all the current games. What does 3dmark05 give me ? Performance in future games ?