RussianSensation
Elite Member
That's fair if those outliers are either showing ambiguous results that don't match up with the rest of the review sites (Wolfenstein NWO) or if the game was blatantly coded for a particular AIB with total disregard for competing AIB's products (Project CARS). The point I was making is that you cannot correlate the improvements in AMD's cards relative to NV's only due to newer drivers because those 2 games were heavily influencing the results given how much of an outlier they were.
Complaints about AMD drivers are more often than not coming from NV users who haven't used AMD drivers in years. Based on the driver support we've seen since R9 280X/290/290X, AMD deserves an A for its support. What about Kepler? Look at 780, OG Titan, 680/770. Ya, not much more needs to be said. Those cards have bombed.
Yes, so if I plan to play TW3/FO4, Metal Gear Solid V, GTA V and SW BF for most of 2015/1H of 2016, I wouldn't care to look at the averages of any review since those are irrelevant to me. That means if I am upgrading to a new card, and plan on spending 100s of hours on those games, I would only use those games as what primarily drives my GPU choice. I couldn't care less what the average is in 20-30 games I won't play. Since we can't read people's mind, when recommending GPUs, AT we compare videocards based on averages since it's impossible to quantify who plays what games, what fraction of their time they spend playing game XYZ, etc. If Gamer A spends the majority of his/her time playing Project CARS, WoW or games that highly favour NV hardware, it's self explanatory what videocard Gamer is going to buy. This changes nothing about what is a fair method of comparing videocards, not games. Also, to keep things consistent, the same sites that removed games that highly favoured AMD (Dirt Showdown) are now blatantly including games like Project CARS that highly favour NV. To be professional, there has to be some consistency to the methodology.
This is the key difference between TPU/Computerbase/Sweclockers/AT and HardOCP. The purpose of the former is to provide as wide range as possible for representing many modern games, and their underlying engines (by including many diverse games/engines, this allows these reviewers to capture trends of performance when making GPU recommendations), that a gamer might play in the immediate or near future and try to gauge where various GPU products land on that basis. The purpose of HardOCP's GPU reviews is not to review videocards, but to review specific game experiences by looking at IQ and performance of that specific game provided by a specific videocard. What that means is a site like HardOCP might review 3, 5, or 7 games but they review all of these games individually. In theory, this is an excellent approach and is unique in the industry. The flaw in HardOCP's intended methodology is that they claim to review games per their owners/editors, but then draw conclusions on videocards at the end. This is completely contradictory to their ideology. What they should do instead is compile a chart of games and what videocards provide the best gaming experiences in those games based on their testing. This way someone can easily select 10-15 games from their database and quickly compare which videocards provide what level of an experience without looking at %s or actual FPS data.
OTOH, sites like TPU/Computerbase/AT/Sweclockers/PCGamesHardware/TechSpot are looking to compare videocards, not games. That's why these reviewers need to have a fair representation of how well a videocard might perform in 50, 100, 150 other games. That's the whole point of using 15-20 games in reviews to try to get a more accurate average.
If you don't like the idea of averages, not a problem but since we can't ask 1000+ members on AT what games they play, since all the data is available for them, as mentioned earlier they are free to buy an NV card for WoW, Project CARS, etc. There have been quite a few posters on AT that bought a 750Ti/960 over R9 290 for WoW and no one says anything because their intentions are genuine and well understood.
Last edited:
