weird issues running Metro 2033 bench

toyota

Lifer
Apr 15, 2001
12,957
1
0
I can not figure this out and have never seen anything like it. when I run the bench I get completely different results on each run even with the same settings. I have ran the bench 10 times and gotten everything from 19 fps to 30 fps on the same settings.

what is odd is if I lower the resolution, I can get consistent results. I thought maybe vram might be an issue on the higher resolution but its using the same amount on each run so it makes no sense why its so inconsistent. plus its only hitting 800mb max anyway.
 
Last edited:

Keromyaou

Member
Sep 14, 2012
49
0
66
If I do the Metro2033 benchmark with 4xMSAA three times consecutively (even 1680x1050), the first run shows more fps than the second and third runs (the second and the third runs show more or less the same fps). However if i use AAA setting, fps is almost the same for all three runs (even with 1920x1080). Since 4xMSAA setting should use more v-ram than AAA setting, I always think that this is caused by the vRam limit (My card has 1GB vRam). But I don't know for sure.
 

toyota

Lifer
Apr 15, 2001
12,957
1
0
well at 1280x720 it is 58 fps every time I run it. at 1920x1080, I have gotten 19 fps, 25 fps, 22 fps, 31 fps, 22 fps, 30 fps, 21 fps, and then 31 fps. o_O
 

Jaydip

Diamond Member
Mar 29, 2010
3,691
21
81
well at 1280x720 it is 58 fps every time I run it. at 1920x1080, I have gotten 19 fps, 25 fps, 22 fps, 31 fps, 22 fps, 30 fps, 21 fps, and then 31 fps. o_O

Hey T is your hard disk properly defragmented ?
 

toyota

Lifer
Apr 15, 2001
12,957
1
0
Hey T is your hard disk properly defragmented ?
yeah its fine plus again it only happens when running at 1920x1080 so I dont see how the drive would matter in this scenario. this is probably the oddest thing I have seen though as I can run the bench and its fairly smooth and then the next run its choppy and slows way down and then run it again and its fine. lower the res and its fine the whole time.
 

3DVagabond

Lifer
Aug 10, 2009
11,951
204
106
yeah its fine plus again it only happens when running at 1920x1080 so I dont see how the drive would matter in this scenario. this is probably the oddest thing I have seen though as I can run the bench and its fairly smooth and then the next run its choppy and slows way down and then run it again and its fine. lower the res and its fine the whole time.

Have you tried monitoring your GPU clock while running it?
 

Lonbjerg

Diamond Member
Dec 6, 2009
4,419
0
0
why what?

Why do you whine over a benchmark-tool?
Before you even tried the game...aka play the fracking game!
(Read: Play the fracking games and ignore the useless benchmark tools already)

I am just slighty amused over your "adventures"...and worried what you will do when the game run fine...but the benchmark tools is wonky...

But more amused that "frack" is okay to write...while "fu*k" isn't...even if the meaning is the same...don't mind me....got paided vacation until 31st Jan 2013...so I am mentally high ^^
 

toyota

Lifer
Apr 15, 2001
12,957
1
0
Why do you whine over a benchmark-tool?
Before you even tried the game...aka play the fracking game!
(Read: Play the fracking games and ignore the useless benchmark tools already)

I am just slighty amused over your "adventures"...and worried what you will do when the game run fine...but the benchmark tools is wonky...

But more amused that "frack" is okay to write...while "fu*k" isn't...even if the meaning is the same...don't mind me....got paided vacation until 31st Jan 2013...so I am mentally high ^^
lol, go away if that is all you have to add.
 

Lonbjerg

Diamond Member
Dec 6, 2009
4,419
0
0
lol, go away if that is all you have to add.

I say:
Try the game.

Way better than trying to figure out why a synthetic benchmark won't give you any usefull information...and much more fun.

Who cares if the benchmark is borked...the game rules....*sigh*
 

toyota

Lifer
Apr 15, 2001
12,957
1
0
I say:
Try the game.

Way better than trying to figure out why a synthetic benchmark won't give you any usefull information...and much more fun.

Who cares if the benchmark is borked...the game rules....*sigh*
I have already played the actual game though. I even normally use the game for running my own benchmarks. I simply used the built in benchmark today to directly compare results in a review that also used the benchmark. unfortunately it is not running properly at the settings I needed to run it at.
 

Alusan

Member
Mar 5, 2010
33
0
66
Were you running with MSAA at 1920x1080 or AAA? With the lower results, are the dips in certain parts or consistently? Same problem after restarting? I thought maybe the memory usage fluctuated between runs but I just did a few and according to Afterburner the memory usage was constant after a few seconds into the first runthrough.
 

toyota

Lifer
Apr 15, 2001
12,957
1
0
1920x1080
high DX11
no physx
no DOF
16x AF
4X MSAA

those are the same settings tomshardware used in for the built in benchmark in their gtx660 review. when I tried the bench it was really really slow as I knew I should get around low 30s since the gtx560 ti got just over 40 fps. the next time I ran it I got normal performance. I can run it over and over and the performance is all over the place. there is no certain place where it just slows down and if the results are low then its slower throughout.

again though simply lowering the resolution will let me get consistent results. I got 58 fps on all three runs at 1280x720.
 

toyota

Lifer
Apr 15, 2001
12,957
1
0
yeah must be a driver issue or somethings specif about those settings.

I just tried the actual game on those settings and it is inconsistent too. I started it up three different times and gotten different results in same exact spots. one time I can be at 20-25 fps then next time at 10 to 12 fps in same spot. its like a different game each time I start it up.
 

toyota

Lifer
Apr 15, 2001
12,957
1
0
I updated to the new Nvidia driver and no more issues. I ran the bench 5 times in a row and its consistent now at 32 fps. that was a pretty odd bug to just cause an issue on a certain resolution.