Phenom II 940 + 5850 crossfire mini review

I am installing games as we speak and setting things up as i go

Phenom II 940be at 3ghz 3.2ghz 3.4ghz and higher depending on how high i can overclock
4gigs 2x2gbs Gskill ddr2 pc6400 at 5 5 5 15
1 terabyte samsung F3 system drive
1 terabyte samsung F1 game drive
2 ati 5850s in crossfire
Vista 64 running with all extras turned off fresh install fully updated

cant remember what cat im using i got it from amd's website today so whatever lol driver string from gpu z read as follows 8.14.10.0700 anyway..

games to be tested

Crysis
Crysis Warhead
Oblivion
Fallout 3
Fear
Neverwinter Nights 2
Resident Evil 5
The Last Remnant
Devil May Cry 4

some other 15 games + give or take ill update the list as i go

granted we all know what 2 5850s can do with a heavily overclocked i7 now lets see how they do with a simple Phenom and some DDR2 so the Core2 and Phenom users can get a better idea of what the cards offer

dont kill me highest res i can go is 1680x1050 which is a COMMON screensize anyway down to buisness

also note each set of posts is split by game genre

set 1 is FPS

set 2 is RPG

set 3 is RTS




Overclock results 1 will be posted soon
overclock test run is as follows

Phenom II 940 at 3.42ghz NB at 2200mhz
http://valid.canardpc.com/show_oc.php?id=774432 link to show initial OC test note i WILL overclock further this is so ppl can compare the 940 to a 965 anyway stay tuned testing almost complete


as of this moment overclock testing is on hold currently the drivers for 5850 crossfire are junk even with 400mhz overclock prime 95 stable for 2hrs and the games show no improvement or minimal improvement at best. I will update this thread at a later date when a new set of drivers are released



and for a friend out there in the internet wastes yes Aydin im THAT guy

Crysis 32bit exe DX10



first lvl of game tested from parachute in till u reach your dead comrade

settings 1680x1050 4xaa Very high (even motion blur etc)

run min / avg / max

test1 26 / 38 / 66

test2 22/ 39 / 69

test3 25/ 35 / 64

seems the high numbers are a bit low but everything was fluid drivers need more work as there is some stuttering in a few spots but there was no MICRO stuttering after reducing motion blur 99% of the stutter was gone and fps naturally improved but the nature of the test was to see what MAX settings would do and on a non overclocked cpu ram and 2 gpus all at stock these results are better then expected the game is at this point 100% enjoyable areas where the stuttering was noticed tended to be when the game had sudden actions taken or dramatic shifts in view (example the crash into the water after parachute event frame rate is 29-32fps but it feels sluggish) so far tho when in control of the character gameplay is silky smooth otherwise. ill update this post with overclocked results after finishing the other games (images coming soon )

another note reducing motion blur to medium resulted in a the minimum frame rate jumping to 30fps ie the magic number


Crysis Warhead DX10



settings 1680x1050 4xaa Enthusiast (even motion blur etc)

run min / avg / max

test1 22 / 36 / 56

test2 21/ 35 / 53

test3 22 / 37 / 52

crysis warhead runs far worse then crysis BUT there was no stuttering issues i suspect improved drivers will help improve frame rates im gonna run the test a few more times to be safe updates soon

** updated frame rates for warhead seems windows was having an issue restarting gave improved frame rates still as far as the crysis series goes oringial crysis gives better frame rates BUT warhead FEELS smoother


FEAR

tested at 1680x1050 4xaa max settings with volumetric lights and softshadows on if it can be turned on its on



Fear in game benchmark

run min / avg / max

test1 92 / 259/ 761

test2 80 / 245 / 758

test3 86 / 253 / 752

I am completely 100% cpu limited the Phenom at stock clocks is just unable to supply the gpus fast enough


Resident Evil 5



max settings 1680x1050 DX10 8xAA note i used the fixed benchmark as its FAR more demanding using the variable benchmark resulted in 20fps MORE on average some scenes increasing by 40 fps on average.

i believe the game is 100% cpu limited so while the 2 5850s allow max settings ATI's architecture shows its more cpu limited then Nvidias architecture even so the game is 100% stable and runs flawlessly

tested the game with 5 runs all 5 runs averaged within 1 frame of each other (tested as i though there might be an error)
stay tuned for Overclocked results



Devil May Cry 4

Max settings DX 9 8xaa


Max settings DX 9 8xAA at 3400mhz cpu 2200mhz NB


Max settings DX9 8xAA at 3400mhz cpu 2200nb **update** at bottom of this post


Max settings DX10 8xaa


Max Settings DX10 8xaa at 3400mhz cpu 2200mhz NB


As you can see the Phenom II 940 at 3ghz isnt enough to push the 2 ati 5850s to there full potential we will should see some serious performance gains with and overclocked Cpu in this test later on

the overclocked results are a mixed bag or so it seems at the moment theres a slight improvement with the 400mhz oc to the CPU and the NB but not much of one to write home about so far

** UPDATE **
after uninstalling and reinstalling drivers i decided to try something different after it made no difference i installed RC6 and then installed RC7 OVER RC6 and u can obviously see what happened in the DX9 8xaa Devil may cry 4 bench ill try RE5 next to confirm and YES crossfire was working in the previous tests i tried it buy removing a gpu and testing again showing a significant loss in fps so crossfire was working eitherway be it fluke or not crossfire is scalling ridiculously well right now ill rerun the crysis benchmarks asap

Oblvion GOTY

max in game settings including shadows 8xAA with bloom at 1680x1050

ive done NO modification to the config file it is well known that by increasing certain memory limits in the config file u can get massively improved performance i have neglected to do this as the average gamer wont be using said tweaks.



test was done as follows from new character creation to running around the imperial city to the bridge between the town and the Imperial city said bridge is the most demanding area for the oblivion engine to render and to increase graphics strain i did the test during a rainstorm.

Indoor
min avg max
test 1 45 / 175 / 300
test 2 43 / 180 / 323
test 3 39 / 177 / 289

Outdoors
min avg max
test 1 43 80 176
test 2 42 77 154
test 3 44 82 164

there you have it i do believe these settings are cpu limited altho i ran into a wierd glitch (as far as im concerned) with my character and monsters frame rate remains high when npcs are on screen frame rate seems to stick to 55-63ish untill i turn away from them eitherway the game is entirely playable at max settings with some tweaks the game would be limited by cpu only i surmise that a faster clocked dualcore would be preferable in this game most likely an e8400 or 8500 around 4ghz or more as oblvion dosent seem to even use more then 1 core on my phenom II at this point (ill test further a bit later)

note for posterity and for those that can tweak oblivions configs via notepad it should be known that doing the tweaks for cell loading and framebuffers allowed me to raise the minimum frame rate by 20fps AND eliminate all stuttering at max settings

following the tweaks here http://www.tweakguides.com/Oblivion_9.html resulted in the following

Outdoors
min avg max
test 1 56 80 176
test 2 66 77 154
test 3 59 82 164

may not seem like much but the random fps spikes were eliminated overall performance is limited by cpu clock but still i nearly 20fps increase on the minimum is rather remarkable soon ill overclock and see what improvements are made till then stay tuned


Fallout 3

Max in game settings vsync off 1680x1050 8xaa HDR u name it its maxed out



Test was done from start of new character to the run to megaton

min avg max
test 1 33 56 63
test 2 37 57 63
test 3 34 56 63

Even with vsync off and FORCED off via CCC the game still maxes out at 63fps not that i mind as the game is 100% playable but as usual there is nearly no difference with crossfire on or off (i tested in this title the scores remain the same exactly) 1 5850 is more then enough for the common man to enjoy fallout 3 with all eyecandy im sure with some tweaks like in my oblivion review the games performance will increase but at this time i dont see the reason as the game is smooth as glass to play. I also note here that i may also be 100% cpu limited. This is a game where i see a dual core clocked at ridiculous speeds paying huge dividends. in performance. I will say the dip in frame rate is the same with oblivion as the game loads new cells it results in a very quick drop. If u remove the lowest fps average from the loading spikes and focus only on frame dip when in tense situations then the frame rate never dropped below 50 fps expect for said spikes when loading

to improve performance in fallout 3 http://www.tweakguides.com/Fallout3_10.html follow the memory and multi-threading guide




Neverwinter Nights 2



max settings 1680x1050 4xaa all in game option enabled

I test the region where you first run into Neeshka the Tiefling rouge as it has a fairly densely populated fort and a good open region to test draw distances overall the gameplay was smooth but this game even after being patched fully still shows just how unoptimized the game engine really is.

5 minute play test using fraps

Run min avg max
test1 39 73 83
test2 38 75 81
test3 41 69 84

The Last Remnant

1680x1050 max in game settings except shadows which are 1 slot down from max due to the issue with ATI gpus and how the game renders shadows



when running around exploring the FPS averages between 220-300fps depending on location which isnt really worth benchmarking so lets take a look at the battle results

Battle vs 3 enemy squads with 3 squads of my own

fight min / avg / max
test1 77 127 169
test2 66 120 158
test3 71 125 162

reserved for benchmarks 3

crysis results up for non overclocked reference machine

well ill keep at it i know this isnt exactly the worlds greatest review but i figure for those of us on a gaming budget this review will come in handy and again the first benchs are all stock so no cpu overclock no gpu overclock no ram overclock no nothing this is just my machine slapped together and finally running stable game performance will only get better as i start tweaking with the overclocked results

crysis warhead added

i dont have vantage or 3dmark besides those scores while nice show nothing of real world performance which is why im doing this review eventually i might download the trial version of 3dmark and give them a go but currently w1zzards review shows what the cards can do in 3dmark so hes covered that so far (just use the gpu score to get an idea)

reran tests still lower but otherwise warhead FEELS smoother crysis gives better frame rates take or leave it i think its the phenom II stock clock causing it as in im bottlenecked i do believe eitherway we will test that theory later on

i have added oblvion tests to the mix next will be fallout 3 and fear

updated with FEAR benchmarks

btw WOOT WOOT google brought up my review lol sweet (never had that happen before cut me some slack )

ill do more benchmarks tomorrow this is still a MAJOR wip -_-

ive got

Assassins Creed. Call of Jaurez Bound in Blood, Drakensaang , Fallout3, The Last Remnant, Sacred 2, World In Conflict, Medieval 2 total war +expansion, Empire total war, Neverwinter Nights 2, Devil May Cry 4, Dead Space, Tomb Raider Underworld, Doom3 , The Witcher, among many more so if u see something above u really absolutely want me to test asap well say so lol and ill start on that first thing tomorrow right now its 1am im damn tired gonna sleep

sure ill try it out granted it will be just the starting area at best but then agian when i bought the game in 06 it was so buggy the first area was a crippler lol i remember my old 8800gts 640 CHUGGING at 19fps at my native res -_- so id like to see how these cards handle it as well i also MIGHT extend my review to be both single card AND crossfire to see if im getting a decent boost or not. so stay tuned