Most want to see 1000+ units once you go above a certain unit amount in strat games they crawl due to lack of CPU grunt 8 player sins of solar empire battlestar galactica mod is a prime example. ( Cylon caps come with 100+ fighters per cap x20 per player game just crawls due to the amount of CPU calculations required.
A vote for metro 2033 and probably last light when it comes out. Some older games like borderlands would be nice
Yes, I think this is a good suggestion. Perhaps pick demanding titles where possible across each of the main engines. Maybe a demanding Dx9 and 10 title if they aren't already covered.
Great minds n' all Not sure how fair a test MP3 would be as I think the more Vram you have, the more objects it puts into the game. At least I think that's how it works.
If it's going live to the public it needs to include the biggest games that people play along with the biggest (recent) releases from a bigger variety of genres (can't have them all as RPG's for instance). BF3 Witcher 2 Skyrim WoW (Such a big game that a lot of people will base their decision on something like this... despite it being heavily CPU dependent - something that only those 'in the know' understand) Starcraft II Arkham City Dirt 3/Showdown
If you're going to be testing with WoW, can I recommend a full 25-man raid of Black Temple? There was something about the instances around that area that was really hard work for my computer.
Glad to see that BT are testing with demanding games. Saw another site test against COD MW3 and Diablo 3. Now if I'm not mistaken, both of those, whilst recent, aren't exactly known for pushing hardware.
I also think Batman AC and Max Payne 3 would be good inclusions. They are both reasonably demanding, both recent and popular.
Unigine Heaven would be the most obvious choice. Fix the settings to something a large proportion of the user base can run and you have the perfect comprison tool. Personally I would prefer a focus on actual gameplay performance rather than simply "benchmarking". Whilst you lose some of the direct repeatability you gain in the fact that the results you are seeing is what you can actually expect when you play the game.
So for our new benchmarks, I've run BF3 @ ultra Crysis 2 w.DX11 and High res Witcher 2 w/o UberSampling Skyrim w.Highres And unigine @ 1920, so you can run a benchmark at home and compare your performance. Might add Sleeping Dogs/ F1 2012 too if I have the time in the next few months.
/vote shogun 2. Shogun 2 is a good choice. Have to be careful to override the function that automatically lowers settings when it detects low VRAM (think this still effects benchmark mode). Also, when I played it (when it first came out) it was horribly CPU limited due to shitty multi thread support (that may have been improved by one of the many patches though). Shogun 2 also has dedicated bench marks that people can run at home, plus many options for anti aliasing and direct x 11 effects. It also uses a lot of VRAM at high settings, so should show up cards with low video memory / weak memory performance. Sleeping dogs is too variable to be a good choice. It uses low VRAM even with high texture pack and has variable fps depending upon where you are (fps drop in daytime). I addition it will run on a chocolate teapot until you whack up AA and SSAO (when performance tanks disproportionally). There is also micro stuttering problems which rumour has it will be patched. Possible forthcoming patches and driver updates could effect performance which is also a reason not to include it until it is more mature. I think that BF3 and Crysis 2 are very good choices as they represent two of the most up to date and prevalent engines. Many future games will use these. For example Command and Conquer will use Frostbite while Mechwarrior is using Cry Engine.