note: for 4870x2 cf test numbers scroll down to 10th post
note: for gtx295 sli test numbers scroll down to 16th post
just got my 5870 order in to day. thought i'd share some test results at 25x16. i decided to test with crysis warhead, dark athena, batman arkham asylum, and resident evil 5. crysis and riddick are fraps recorded actual play tests, while batman and re5 results are from the in game benchmarks. at 25x16 it's kind of necessary to have as much aniso on as possible so i forced 16x quality from catalyst. i also forced super sampled anti aliasing. my test system specs are as follows:
i7 920 @ 3.6ghz
msi x58 platinum
6gb corsair cas8 1600mhz ram @ 2t 8-8-8-24
thermal take toughpower 1000w psu
vista ultimate 64
crysis warhead:
all settings at enthusiast but no aa. fraps recorded play test of the 'hells heart' stage beginning on the train in the tunnel and ending as the train makes its first stop. the same settings were used for both the single card and crossfire tests. as you can see form the fraps log above, the first recording in single card mode returned 7 min, 33 max, and 20 avg. hardly performance i would call playable. once the second card is enabled, we get returns of 8 min, 47 max, and 28 avg. don't let the 1 fps improvement in minimum frame rate fool you, warhead became very playable, maintaining very consistent frame rate throughout the test.
dark athena:
in game settings at max with no aa, but ambient occlusion set to low in single card mode and high in crossfire. fraps recorded play test of the prologue beginning with riddick waking up on the beach, and ending after the run down the darkening tunnel. single card mode returned 22 min, 239 max, and 33 avg. with crossfire turned on and maximum ambient occlusions applied, we get 22 min, 603 max, and 40 avg. the extraordinarily high max numbers are anomalous recordings attributed to the times when the screen goes black during the first in game cinematic. i also noticed some slight micro stuttering during the cinematics but not during the actual game.
batman arkham asylum:
i wasn't able to play test this game because it was recently purchased, and haven't had time to play much of it. the game has a built in benchmark that is no more than a fly by of some of the locations in the game. i used fraps to record the fly by as there doesn't seem to be anything keeping track of frame rates during the test. i used maximum in game settings with physx disable or course, and 2x aa forced in catalyst for the single card test with 4x aa used for the dual card test. the benchmark returned 26 min, 53 max, and 41 avg. the crossfire test returned 16 min, 65 max, and 45 avg. the game appears much cleaner with 4x aa enabled but the 10 frame loss in mins is a bit troublesome. certainly subsequent driver releases will improve minimum frame rate performance.
re5:
this is another game i wasn't able to play much, but at least it comes with a couple of benchmarking tools. the first available benchmark is an ai simulated play test of an area in the first stage. the second benchmark is a fly by of a crowd of infected in the first. the variable test is predictably the most useful. i used maximum in game settings including 8x aa from the settings menu. i didn't use fraps to record frame rates, but the test concludes with a helpful chart of the peaks and valleys.
that's all for now. i'll try to update with some 4870x2 cf comparisons tomorrow, and maybe gtx295 sli tests if i manage to get my other x58 buils stable at 3.6ghz.
note: for gtx295 sli test numbers scroll down to 16th post
just got my 5870 order in to day. thought i'd share some test results at 25x16. i decided to test with crysis warhead, dark athena, batman arkham asylum, and resident evil 5. crysis and riddick are fraps recorded actual play tests, while batman and re5 results are from the in game benchmarks. at 25x16 it's kind of necessary to have as much aniso on as possible so i forced 16x quality from catalyst. i also forced super sampled anti aliasing. my test system specs are as follows:
i7 920 @ 3.6ghz
msi x58 platinum
6gb corsair cas8 1600mhz ram @ 2t 8-8-8-24
thermal take toughpower 1000w psu
vista ultimate 64
crysis warhead:
all settings at enthusiast but no aa. fraps recorded play test of the 'hells heart' stage beginning on the train in the tunnel and ending as the train makes its first stop. the same settings were used for both the single card and crossfire tests. as you can see form the fraps log above, the first recording in single card mode returned 7 min, 33 max, and 20 avg. hardly performance i would call playable. once the second card is enabled, we get returns of 8 min, 47 max, and 28 avg. don't let the 1 fps improvement in minimum frame rate fool you, warhead became very playable, maintaining very consistent frame rate throughout the test.
dark athena:
in game settings at max with no aa, but ambient occlusion set to low in single card mode and high in crossfire. fraps recorded play test of the prologue beginning with riddick waking up on the beach, and ending after the run down the darkening tunnel. single card mode returned 22 min, 239 max, and 33 avg. with crossfire turned on and maximum ambient occlusions applied, we get 22 min, 603 max, and 40 avg. the extraordinarily high max numbers are anomalous recordings attributed to the times when the screen goes black during the first in game cinematic. i also noticed some slight micro stuttering during the cinematics but not during the actual game.
batman arkham asylum:
i wasn't able to play test this game because it was recently purchased, and haven't had time to play much of it. the game has a built in benchmark that is no more than a fly by of some of the locations in the game. i used fraps to record the fly by as there doesn't seem to be anything keeping track of frame rates during the test. i used maximum in game settings with physx disable or course, and 2x aa forced in catalyst for the single card test with 4x aa used for the dual card test. the benchmark returned 26 min, 53 max, and 41 avg. the crossfire test returned 16 min, 65 max, and 45 avg. the game appears much cleaner with 4x aa enabled but the 10 frame loss in mins is a bit troublesome. certainly subsequent driver releases will improve minimum frame rate performance.
re5:
this is another game i wasn't able to play much, but at least it comes with a couple of benchmarking tools. the first available benchmark is an ai simulated play test of an area in the first stage. the second benchmark is a fly by of a crowd of infected in the first. the variable test is predictably the most useful. i used maximum in game settings including 8x aa from the settings menu. i didn't use fraps to record frame rates, but the test concludes with a helpful chart of the peaks and valleys.
that's all for now. i'll try to update with some 4870x2 cf comparisons tomorrow, and maybe gtx295 sli tests if i manage to get my other x58 buils stable at 3.6ghz.
Last edited: