Metro 2033 Performance

Rofflay

Rofflay

Rofflay

Permabanned
Joined
28 Nov 2006
Posts
5,750
Location
N Ireland
Ive had a search through the forum and suprisingly cannot find any offical threads on performance on this game.

Im building a new rig soon(hopefully 6 core cpu + GTX480) and im hearing bad things about this game.Id of thought with the above kind of pc i could pretty much max out crysis and metro2033 and have a platform able to produce almost photo realistic games to last me 3 years into the future and through direct x 11's lifetime.


Max settings direct x 11 4xMSAA+16XAF

Core i7 3.8 6gb ddr3 and a 5970 cannot pull of 60fps and is more like 20-30fps.
from what ive seen this is an even bigger hog than crysis...


what gives? is this game cpu based or graphics based? or would the fact thier optimal specs mentioned above 8gb of ram and most people have 6gb be hampering performance?
 
I had a ****load of problems running this on a decent rig. But it runs smooth as silk now. My problem was that I had Data Execution Prevention enabled on EVERYTHING which was slowing down access to the game files. Once I provided Metro 2033 as an exception, it runs in 1920 res, High settings and DX10.
Check out my posts in the main Metro 2033 thread for specs and details of my frustrating journey.
 
Rofflay said:
Ive had a search through the forum and suprisingly cannot find any offical threads on performance on this game.

Im building a new rig soon(hopefully 6 core cpu + GTX480) and im hearing bad things about this game.Id of thought with the above kind of pc i could pretty much max out crysis and metro2033 and have a platform able to produce almost photo realistic games to last me 3 years into the future and through direct x 11's lifetime.


Max settings direct x 11 4xMSAA+16XAF

Core i7 3.8 6gb ddr3 and a 5970 cannot pull of 60fps and is more like 20-30fps.
from what ive seen this is an even bigger hog than crysis...


what gives? is this game cpu based or graphics based? or would the fact thier optimal specs mentioned above 8gb of ram and most people have 6gb be hampering performance?
Click to expand...

If you wanting to build a pc thats future proof your gonna need to probably look at 12gb of ram and a PSU capable of perhaps adding in a second 480 at a later date.

Metro 2033 runs like cack if you enable DOF and tesslation and there are no benchmarks or performance information regarding the new fermi based nvidia cards saying they can do both smoothly either just yet. Crysis runs really well on current top end hardware without any slowdowns unless you game at extreme resolutions + AA at that res :P
 
psu wise for a 2nd graphics card down the line look for someting like the 850w corsair modular... but it all comes down to what cards youre looking to get and system setup...

as for performance in metro 2033, its obviosuly a poorly coded, unoptimised game so i wouldnt worry about a super computer to run it at max
 
phil_c_ said:
psu wise for a 2nd graphics card down the line look for someting like the 850w corsair modular... but it all comes down to what cards youre looking to get and system setup...

as for performance in metro 2033, its obviosuly a poorly coded, unoptimised game so i wouldnt worry about a super computer to run it at max
Click to expand...
I wouldn't say poorly coded, it could probably be better but for the graphical quality i'm quite happy with how it runs on my rig anyway.
 
Runs really badly under DX11 with 2 5770's in Xfire for me.

Had to use DX10 very high settings runs much better.

Very nice looking game from what I've played so far which is not much atm.

All the other DX11 games I've played so far are fine but the poor frame rate make this unplayable for me under DX11.

Might have to wait for a driver hotfix or updated application profile hopefully that will fix it unless the game is badly optimized for DX11.
 
Another comparison.

At DX10 Very High, two 5870s pretty much cane the game; nearly 100% Crossfire scaling there. 77 fps average at 1680, 54 fps average at 1920x1200, Catalyst 10.3 preview driver.

A few other things stand out: the game only uses two processor cores, CPU overclocking with an i7/Phenom II makes literally zero difference to framerate, and TWIMTBP or not, Nvidia cards get pummelled in both of this page's posted comparisons, hmm...

Time for me to install those two 5870s I swindled back in December for a combined £430 and have never actually opened. Adieu 260s, you've served your very venal master well.
 
Rapido! said:
Another comparison.

At DX10 Very High, two 5870s pretty much cane the game; nearly 100% Crossfire scaling there. 77 fps average at 1680, 54 fps average at 1920x1200, Catalyst 10.3 preview driver.

A few other things stand out: the game only uses two processor cores, CPU overclocking with an i7/Phenom II makes literally zero difference to framerate, and TWIMTBP or not, Nvidia cards get pummelled in both of this page's posted comparisons, hmm...

Time for me to install those two 5870s I swindled back in December for a combined £430 and have never actually opened. Adieu 260s, you've served your very venal master well.
Click to expand...

Very interesting read actually, especially the dual core part! For a brand new game of this graphical level not using quads is abit of a joke to be honest.
 
Run in dx11 or dx10, if dx11 turn of tesselation and DOF, it makes almost zero difference. For me it looks BETTER without DOF on, Tesselation appears to make little to no difference.

The fact its a "launch" title for Fermi and in certain artificial situations tesselation can be faster on the architecture I wouldn't be surprised if they implemented tesselation in a way purely designed to add a specific type of gpu load that makes Fermi look great.

Problem is the 50% performance hit for using it would be fine, IF it made a noticeable increase in quality, it doesn't, its simply wasted performance, either very bad use of a new effect which the developers are known for, or Nvidia playing games.

Whats odd is that Stalker had a horrible time with TWIMTBP, released two buggy as crap titles, AMD took over supporting the second game and added dx11 and helped make it more stable with the increases in quality/speed, and much of the same team making Metro decided to go with physx/Nvidia combo, barely any worthwhile physics in the game, a few cloth effects very early on that weren't done well, and then, well, nothing to be honest.

Disable both those effects and it will run very well on a 5850 or above at 1920x1200, run both the effects and it will run like crap.
 
drunkenmaster said:
Run in dx11 or dx10, if dx11 turn of tesselation and DOF, it makes almost zero difference. For me it looks BETTER without DOF on, Tesselation appears to make little to no difference.

The fact its a "launch" title for Fermi and in certain artificial situations tesselation can be faster on the architecture I wouldn't be surprised if they implemented tesselation in a way purely designed to add a specific type of gpu load that makes Fermi look great.

Problem is the 50% performance hit for using it would be fine, IF it made a noticeable increase in quality, it doesn't, its simply wasted performance, either very bad use of a new effect which the developers are known for, or Nvidia playing games.

Whats odd is that Stalker had a horrible time with TWIMTBP, released two buggy as crap titles, AMD took over supporting the second game and added dx11 and helped make it more stable with the increases in quality/speed, and much of the same team making Metro decided to go with physx/Nvidia combo, barely any worthwhile physics in the game, a few cloth effects very early on that weren't done well, and then, well, nothing to be honest.

Disable both those effects and it will run very well on a 5850 or above at 1920x1200, run both the effects and it will run like crap.
Click to expand...

Same thing noticed on my rig as well. Tessellation and DOF truly destroy the FPS. I can't see a difference in tess on this but to be fair I also could bearly see it on AVP either at the highest settings with tess on maximum lol. Problem is when stuff moves so quickly anyway you dont notice the finer detail.

Its kinda like watching a blu ray film on a 1080p tv but the film has really shoddy camera work!
 
what does everyone makes of this article then? there are reports that 2 cores are enough and that adding in another2 made zero frame rate difference?!?!

http://games.on.net/article/7836/Metro_2033_-_Technical_QA


Quote from the dev's "If you are not bottlenecked by the video card, we have linear scaling – so double the performance. This goes all the way up to eight and sixteen cores"

so why are people just not seeing this or are they? im confused
 
I remember one or two forum members mentioning that when the cpu was overclocked the fps improved substantially.

I was playing with my cpu at 3.4 and my video card at stock and apart from the occasional slow down, and some weren't that bad to be fair, it ran quite well. The frames ranged from (absolute worst) around 10 to about 150. The usual fps I got was around 30 - 90.
 
You must log in or register to reply here.
Back
Top Bottom