I'm wrapping up my reviews and testing of the Radeon HD 6000-series (HD 6770 will be out soon), and news of the Radeon HD 7000 is starting to hit the web. I'm upgrading my test rig from an i7-920 to an i7-2600k. So, with this I need to re-baseline my cards (only going to carry over the 6000 series). Additionally, I have an AMD FX 8150 that I need to test and compare to both Intel chips. AND, come hell or high water I'm going to start testing NV cards. So, it's time to review what I'm testing and the time invested.
I want to get feedback from the community on what you'd like to see, and where you see value.
Games
Aliens vs. Predator - I like this one. The big cards still aren't hitting 60fps in Eyefinity, and single cards aren't even hitting 30fps. Widescreen is still a challenge for mid-range cards, and single high-end cards still aren't hitting 60fps.
Battle Forge - This title still punishes single cards in Eyefinity, and CFX sets are less than 60fps. Widescreen isn't much of a challenge any more. Even the low-end cards are hitting 30fps in widescreen. However, does anyone play this title. Does knowing how a card performs in Battle Forge help you? I'm looking to drop this title.
Dirt 2 - Widescreen isn't a challenge for any card. Most mid-range cards hit 60fps in widescreen. Almost everything clears 30fps in Eyefinity, and CFX setups cross 60fps. However Dirt 3 is out, and I'm planning to simply update titles.
Dirt 3 - More demanding, and a good replacement for Dirt 2.
F1 2010 - I hate testing F1. It takes forever to load into the title. The game won't run at 3x1080p Eyefinity on a 1GB card, and I don't see any eye candy to justify the limitation. Planning to drop and replace with Dirt 3.
Far Cry 2 - Easy to run, but getting old. Even mid-range cards hit 60fps at max settings. Those same cards will hit 30fps in Eyefinity, and higher cards will hit 60fps. Is Far Cry 2 still relevant to today's titles and gamers? If so, I'll continue with it, but if not it may be time to retire.
HAWX - This title doesn't test any but the lowest cards in widescreen. The title does pose some stress for cards in Eyefinity, but HAWX 2 is now out. Will probably retire for HAWX 2.
HAWX 2 - Updates the HAWX series to DX11, and is a bit more strenuous on cards. Will replace HAWX with HAWX2. However, is HAWX still a relevant benchmark overall? I do remember some stink about the demo/benchmark tool being biased.
Heaven v2 (DX9, DX10 and DX11) - I started testing Heaven v1, and have updated to v2. Version 2.5 is out now. DX9 version is no challenge for any card in widescreen, though it poses some stress in Eyefinity. DX10 is a bit more strenuous, though only in Eyefinity. The DX10 version looks a lot nicer, and only puts a slight hit on the performance. DX11 is where the real test is. So, what about updating to v2.5, but only running DX11? That would cut the time on this title by 2/3.
Just Cause 2 - Like many titles, widescreen is no real test. Additionally, many cards hit 30fps in Eyefinity, and the CFX configs hit 60fps. It can still show the differences in cards, but is it a good reference for current gamers?
Mafia II - Widescreen isn't tough on most cards. It really hammers the cards in Eyefinity, and it's certainly a keeper.
Metro 2033 - A hell of a benchmark. Even widescreen can still stress cards, and Eyefinity won't even run on a 1GB card. A keeper, and a good replacement for Far Cry 2.
STALKER: Call of Pripyat - Widescreen and Eyefinity are both a challenge. It's a long benchmark to run, but easy to do. Is it still relevant for most gamers?
Total War: Shogun 2 - A big challenge for cards and a good keeper.
Warhammer 40k: Dawn of War II - Looks to be a pretty good test, but possibly more CPU bound and not CFX friendly. On my stock i7-920, hit 55fps on a HD 6970 2GB at 1080p. It actually lost 1 frame on a HD 6990 4GB. I got the same results on a 6770 CFX setup, and even a single 6790 was right at 50fps. There is more variation in Eyefinity, but I'm not sure it's worth the time.
Thoughts - So, in writing all of this I'm looking to scale down to AVP, Dirt3, HAWX 2, Heaven 2.5 DX11, Mafia II, Metro 2033, STALKER and Shogun 2. This would bring me from 16 tests down to 8 tests. I left Just Cause 2 and DOW2 as maybes. Considering the other titles I'm keeping, do those add any value? Are any of the titles I've chosen redundant or no longer relevant/popular? Are any not relevant because they are simply too demanding?
I know there are some other titles out there that can be tested. Lost Planet 2 has a benchmark, but I think I remember having trouble getting it to run at one point. I used to do World in Conflict, but got graphics corruptions issues at some point and couldn't fix it with a reinstall of the game or the GPU driver. Also, I know Civ5 gets benchmarked, but I'm not sure if it has a canned test.
I need a canned test for consistency and time management, so unless someone can show me how to get a canned timedemo in a COD or BF game, I hard pressed to add them to the routine.
Settings
Currently I'm testing each game at four resolutions: 1600x900, 1920x1080, 3x1600x900 and 3x1920x1080. Are these still good settings? Most other sites are just looking at 1920x1200 or 1920x1080. HardOCP picks up 2560x1600, but only for SLI/CFX. Hardware Heaven sticks with 1080p and 3x1080p, but they test widescreen in 8xAA and Eyefinity in 4xAA.
Should I keep up all four resolutions? Some games simply won't do 3x1080p with 4xAA on a 1GB card. Is the answer to show the 3x1600x900, or do 3x1080p at 2xAA. When a game runs at all resolutions, the performance between the resolutions is often relatively close (10fps or so). Is knowing the 1080p and 3x1080p good enough? Would you rather know what settings are needed to hit 30fps or 60fps?
|