JF_Aidan_Pryde
Regular
It seems the majority is not pleased with the Tomshardware/Anandtech reviews of graphics cards. The question is what would constitute a "good" way to measure performance and IQ fairly.
This is something I've been thinking about.
First, depending on the game, we decide on a Minimum Frames Per Second number. This is important as speed requirements for 3D Games are vastly different.
While Rev's comment on 60FPS for ALL games is in all good intentions, I disagree as *VERY* highly competitive games, in the hands of serious players need much more than that. I and many others can easily feel the difference between 60 and 120 FPS in Quake. It does impact your performance, especially in tight situations. This is why different Minimum FPS acceptance levels are needed for different games. Finally I think it should be the *player* or buyer of the card to decide what's the minimum for him/her, not the reviewer. This may be why so many reviews just cover maximum FPS, just for the hell of it.
The idea I've been thinking is, instead of measuring Maximum FPS for all games tested, it should be testing the best image quality achievable for a given FPS. For example, if I decide that 80FPS (personal preference) is just right for Quake3, then we crank up the IQ features on a card until it is lowed down to 80. It may look like:
82 FPS - CardA: 1600x1200 with (-)xAA and (-)xAniso
79 FPS - CardB: 1280x1024 with (~)xAA and (~)xAniso
At this setting, the speed should average at 80FPS and not drop below 60FPS. As said before, this is a personal preference but I feel 80FPS in Quake3 is very reasonable.
Two cards should be benched such that they both reach an 80FPS average and their image quality compared. The one with the better IQ is the one which manage to produce the BEST image quality under the pre-decided, desired FPS.
This is just to test one game. To make sure the review actually represents a real gamer's experience, games of different generes and engines should be tested.
Warcraft 3 may only need 30 FPS. If that's the case, what's the point of testing Maximum FPS? IQ features should be pushed until it reaches the magic FPS and then a meaningful test can be conducted.
In Flight and Mech sims, the frame rate needed is easliy limited by the vehicle mechanics. A Jet or Mech can only turn at a fixed speed per second while it's possible to turn 180 Degrees in Quake3 in less than a second with the flick of a mouse. 40-50FPS hence would be quite comfortable for game vehicle limited genres (car, mech, plane).
Sorry for the long post, just had to get it out of my head.
JF_Aidan_Pryde
This is something I've been thinking about.
First, depending on the game, we decide on a Minimum Frames Per Second number. This is important as speed requirements for 3D Games are vastly different.
While Rev's comment on 60FPS for ALL games is in all good intentions, I disagree as *VERY* highly competitive games, in the hands of serious players need much more than that. I and many others can easily feel the difference between 60 and 120 FPS in Quake. It does impact your performance, especially in tight situations. This is why different Minimum FPS acceptance levels are needed for different games. Finally I think it should be the *player* or buyer of the card to decide what's the minimum for him/her, not the reviewer. This may be why so many reviews just cover maximum FPS, just for the hell of it.
The idea I've been thinking is, instead of measuring Maximum FPS for all games tested, it should be testing the best image quality achievable for a given FPS. For example, if I decide that 80FPS (personal preference) is just right for Quake3, then we crank up the IQ features on a card until it is lowed down to 80. It may look like:
82 FPS - CardA: 1600x1200 with (-)xAA and (-)xAniso
79 FPS - CardB: 1280x1024 with (~)xAA and (~)xAniso
At this setting, the speed should average at 80FPS and not drop below 60FPS. As said before, this is a personal preference but I feel 80FPS in Quake3 is very reasonable.
Two cards should be benched such that they both reach an 80FPS average and their image quality compared. The one with the better IQ is the one which manage to produce the BEST image quality under the pre-decided, desired FPS.
This is just to test one game. To make sure the review actually represents a real gamer's experience, games of different generes and engines should be tested.
Warcraft 3 may only need 30 FPS. If that's the case, what's the point of testing Maximum FPS? IQ features should be pushed until it reaches the magic FPS and then a meaningful test can be conducted.
In Flight and Mech sims, the frame rate needed is easliy limited by the vehicle mechanics. A Jet or Mech can only turn at a fixed speed per second while it's possible to turn 180 Degrees in Quake3 in less than a second with the flick of a mouse. 40-50FPS hence would be quite comfortable for game vehicle limited genres (car, mech, plane).
Sorry for the long post, just had to get it out of my head.
JF_Aidan_Pryde
