Who can run Metro 2033 maxed?

[Suarez7]

Yup, thats what Toms Hardware concluded, seems that still hasn't changed with Kepler.

Your running a GTX 680? am i right in thinking the less CUDA cores you have the lower Physx performance will be?

I'm thinking of those poor buggers running the run of the mill ~£200 GTX 660.

they would probably be better off running Physx off the CPU, provided they have a Sandy / Ivy Bridge K, or a Phenom II x6.

Yeah, using a 680. I don't think the extra cores help with physx much in any game. It doesn't need many cores to be able to run it at the highest setting in games without lag.

I'm going to try the WHQL drivers as I've seen people achieving better scores with these than the new beta's.

 

[humbug]

Yeah, using a 680. I don't think the extra cores help with physx much in any game. It doesn't need many cores to be able to run it at the highest setting in games without lag.

I don't quite understand what your saying there, are you saying the amount of CUDA Cores you have makes no difference to your Physx performance?

Would a GTX 660 get the same Physx performance in Metro2033 as a GTX 680?

 

[Suarez7]

I don't quite understand what your saying there, are you saying the amount of CUDA Cores you have makes no difference to your Physx performance?

Would a GTX 660 get the same Physx performance in Metro2033 as a GTX 680?

Yeah that's what I was saying. It's probably hard to actually measure properly but I don't believe the extra cores on the 680 would give it better physx performance. People use really low end nvidia gpu's as physx cards and it lets them run physx at the highest settings without any lag.

 

[weehamish]

Yeah that's what I was saying. It's probably hard to actually measure properly but I don't believe the extra cores on the 680 would give it better physx performance. People use really low end nvidia gpu's as physx cards and it lets them run physx at the highest settings without any lag.

This is true, its just Nvidia software screwing the AMD cards really.

I had a 660 in a Q9450 @ 3.7Ghz it lagged with max physx. Same card in my 4.6GHz i5 same settings no lagg.

So they still rely on CPU most part.

 

[humbug]

That ^^^^ maybe true for todays Physx, but at the time of this game Physx was nicly threaded, a 6 core AMD did well in this Physx SKD.

I think CPU's in general just became to powerful, they cought up with Nvidia GPU Physx, so perhaps Nvidia did something about it, not to hit AMD. but to hit all CPU's, Intel included.

Yeah that's what I was saying. It's probably hard to actually measure properly but I don't believe the extra cores on the 680 would give it better physx performance. People use really low end nvidia gpu's as physx cards and it lets them run physx at the highest settings without any lag.

Physx Hybrid, like running a dedicated Nvidia GPU on an AMD rig as a Physx slave

Does that not matter because thats all the dedicated Physx card is doing?

In other words your able to dedicate the whole GPU to physx rendering instead of having to share Graphics and Physx, in that sense a budget card would do the job.
Whats more the higher end the dedicated Physx card is the higher Physx performance you will get.

Which is why i ask, if the GTX 680 is not getting much higher Physx performance in this game than the CPU, then i would ask what is that performance? as i can't engine its all that high unless you have an Intel 6 core or perhaps an i7 3770K overclocked.

So if there is a performance hit on GPU Physx on a lower end card, i can't imagine it would take much to push it under the performance of a good CPU.

 

[Suarez7]

All I can say is Metro is the only game I've noticed running it on the CPU doesn't take much of a hit. For every other game the highest physx setting usually has AMD users crawling to a halt.

 

[humbug]

All I can say is Metro is the only game I've noticed running it on the CPU doesn't take much of a hit. For every other game the highest physx setting usually has AMD users crawling to a halt.

Not going to get a straight answer, am i?

It does for Boarder Lands 2, just as it does on an Intel.

 

[weehamish]

Yes even at medium on my 7970 it dies in BL 2 :\.

 

[Suarez7]

Not going to get a straight answer, am i?

It does for Boarder Lands 2, just as it does on an Intel.

I really don't know what you're asking. You aren't making much sense. I told you what I thought. The extra cores between a 680 and a 660 make little to no difference in metro or any other title for physx performance, otherwise people would use separate dedicated physx cards, and as you know hardly anybody uses them now.

 

[Nuclear Fusion]

I think I could last time I played it at 1080p on my 7950 at 1000/1400.

 

[T345]

900p, Radeon 5770
Highest settings, but lower AA (I think?)
I usually get around 40 fps

 

[humbug]

I really don't know what you're asking. You aren't making much sense. I told you what I thought. The extra cores between a 680 and a 660 make little to no difference in metro or any other title for physx performance, otherwise people would use separate dedicated physx cards, and as you know hardly anybody uses them now.

The higher range the GPU the better Physx performance is, so what was at first your speculating, has now turned into factual statements, and is wrong.

There will be a difference between a GTX 460 and a GTX 480, just as there is a difference between a GTX 660 and a GTX 680 in Physx abilities.

If a GTX 680 is already at the fringes of playable (not being much faster than your CPU) then losing a bunch of CUDA cores (GTX 660) you may have to use the CPU in Metro2033.

This unless your GTX 680 Physx performance is, lets say 45 FPS and your CPU 40 FPS then a GTX 660 at 35 FPS you would be better off with the CPU.

So what i'm asking is, what is your GPU physx performance? because your Metro2033 performance slide does not look much better than mine with you being on your GTX 680 GPU Physx than me on AMD Phenom II x6 1090T CPU Physx.

You see what i'm saying, in this game Nvidia cant afford to lose any Physx performance, but i think you know that.

 

[Suarez7]

The higher range the GPU the better Physx performance is, so what was at first your speculating, has now turned into factual statements, and is wrong.

There will be a difference between a GTX 460 and a GTX 480, just as there is a difference between a GTX 660 and a GTX 680 in Physx abilities.

If a GTX 680 is already at the fringes of playable (not being much faster than your CPU) then losing a bunch of CUDA cores (GTX 660) you may have to use the CPU in Metro2033.

This unless your GTX 680 Physx performance is, lets say 45 FPS and your CPU 40 FPS then a GTX 660 at 35 FPS you would be better off with the CPU.

So what i'm asking is, what is your GPU physx performance? because your Metro2033 performance slide does not look much better than mine with you being on your GTX 680 GPU Physx than me on AMD Phenom II x6 1090T CPU Physx.

You see what i'm saying, in this game Nvidia cant afford to lose any Physx performance, but i think you know that.

There's an average of 10fps between mine and your slides. You're going on about factual performance of physx between a 660 and 680 yet I haven't seen you give any hard evidence either. A 660 and a 680 will both run borderlands maxed out no problem. The 680 will have higher fps with it just being a more powerful card, not because it handles physx any better than the 660. That's my opinion of it anyway.

Also I've already explained to you that metro has little to no benefit between a physx gpu or running it on the cpu so I don't expect better physx performance over you.

 

[humbug]

There's an average of 10fps between mine and your slides. You're going on about factual performance of physx between a 660 and 680 yet I haven't seen you give any hard evidence either. A 660 and a 680 will both run borderlands maxed out no problem. The 680 will have higher fps with it just being a more powerful card, not because it handles physx any better than the 660. That's my opinion of it anyway.

Also I've already explained to you that metro has little to no benefit between a physx gpu or running it on the cpu.

With DOF also enabled yes, DOF is another Nvidia feature, tho i'm not sure if there is any GPU performance hit Nvidia vs AMD, certainly it runs well on my 7870 with high, but not very high settings.

Average game FPS is nothing to do with Physx, i'm looking where the Physx are in the bench on the slides, my FPS only drop about 3/5 FPS lower than your GTX 680 GPU Physx.

The evidence is you need a higher end Nvidia GPU to get good GPU dedicated Physx. for example, a GTX 4/550 does not get you good GPU Physx, a GTX 4/560 is whats known as a good all rounder for price, a GTX 4/570 is better.

I'm pretty sure GPU Physx performance depends on how many CUDA cores you have, a lower end Nvidia GPU has less CUDA cores, hence the performance difference on dedicated GPU Physx.

From that i would suggest a with a GTX 680 not getting you great Physx performance in this game, a GTX 660 would be certainly less so.

 

[Suarez7]

Average game FPS is nothing to do with Physx, i'm looking where the Physx are in the bench on the slides, my FPS only drop about 3/5 FPS lower than your GTX 680 GPU Physx.

The evidence is you need a higher end Nvidia GPU to get good GPU dedicated Physx. for example, a GTX 4/550 does not get you good GPU Physx, a GTX 4/560 is whats known as a good all rounder for price, a GTX 4/570 is better.

I'm pretty sure GPU Physx performance depends on how many CUDA cores you have, a lower end Nvidia GPU has less CUDA cores, hence the performance difference on dedicated GPU Physx.

From that i would suggest a with a GTX 680 not getting you great Physx performance in this game a GTX 660 would be certainly less so.

The problem isn't my GPU getting good physx performance, it's because the game doesn't take advantage of GPU physx as much as other titles do. It's pretty obvious to see that when you compare different games.

 

[humbug]

The problem isn't my GPU getting good physx performance, it's because the game doesn't take advantage of GPU physx as much as other titles do. It's pretty obvious to see that when you compare different games.

Oh its the games fault....

Compare what games? if you compare any Physx game from the time or before i would be willing to bet a powerful modern CPU can run Physx in those games just as well as your GPU and one from the time.

I don't think its got anything to do with the game using your GPU properly, or the GTX 480 from the time.

I think its entirely possible CPU's got to powerful and Nvidia changed the Physx engine to strangle all CPU's.

But i will say this, it is the only Physx game where i think the Physx in it are truly something to talk about, it doesn't have blobular water. or other blobular things rolling around on the ground.

What this has is massive amounts of fine particle physx, a lot of incredibly detailed atmospheric dust and mist thats very much 3D and moves about as you would expect. it looks stunning and its everywhere all the time.

Its the only Physx game i have ever seen that in my opinion they had actually got it right.
No, i don't think BL2 looks much good.

What, if anything did they change, will we ever see anything like that again?
Certainly whats in Metro2033, to me, requires a lot more processing power than what you see in BL2.

 

[weehamish]

Oh its the games fault....

Compare what games? if you compare any Physx game from the time or before i would be willing to bet a powerful modern CPU can run Physx in those games just as well as your GPU and one from the time.

I don't think its got anything to do with the game using your GPU properly, or the GTX 480 from the time.

I think its entirely possible CPU's got to powerful and Nvidia changed the Physx engine to strangle all CPU's.

But i will say this, it is the only Physx game where i think the Physx in it are truly something to talk about, it doesn't have blobular water. or other blobular things rolling around on the ground.

What this has is massive amounts of fine particle physx, a lot of incredibly detailed atmospheric dust and mist thats very much 3D and moves about as you would expect. it looks stunning and its everywhere all the time.

Its the only Physx game i have ever seen that in my opinion they had actually got it right.
No, i don't think BL2 looks much good.

What, if anything did they change, will we ever see anything like that again?
Certainly whats in Metro2033, to me, requires a lot more processing power than what you see in BL2.

I agree with this as i have said i ran the 660 with 2 different CPUs same settings high physx on BL2 and the difference was massive.

People say the card does all the work, does it crap, the CPU does it just Nvidia telling the computer to run good because theres an Nvidia card inside!

BL2 is nothing on BF3 and such so it doesnt stress the CPU that much and for graphics my 7970 goes to about 80% tops.

 

[Suarez7]

Oh its the games fault....

Compare what games? if you compare any Physx game from the time or before i would be willing to bet a powerful modern CPU can run Physx in those games just as well as your GPU and one from the time.

I don't think its got anything to do with the game using your GPU properly, or the GTX 480 from the time.

I think its entirely possible CPU's got to powerful and Nvidia changed the Physx engine to strangle all CPU's.

But i will say this, it is the only Physx game where i think the Physx in it are truly something to talk about, it doesn't have blobular water. or other blobular things rolling around on the ground.

What this has is massive amounts of fine particle physx, a lot of incredibly detailed atmospheric dust and mist thats very much 3D and moves about as you would expect. it looks stunning and its everywhere all the time.

Its the only Physx game i have ever seen that in my opinion they had actually got it right.
No, i don't think BL2 looks much good.

What, if anything did they change, will we ever see anything like that again?
Certainly whats in Metro2033, to me, requires a lot more processing power than what you see in BL2.

Oh so I'm wrong for suggesting that it's this game that has the problem taking advantage of physx dedicated GPU's yet it's ok for you to suggest it's nvidia sabotaging the code to strangle it working on CPU's. It's obvious your hatred for nvidia here and I will not be continuing this discussion as it will spiral into a fanboy war.

So easy to see that you're a fanboy when somehow the only game that runs physx well for you is the only one you like the physx effects in. Many people here will tell you physx is fantastic in games like batman and mirrors edge.

 

[V3teran]

A dedicated Physx card makes a lot of difference in Borderlands 2, Mafia 2, Batman etc etc.

There are only a handful of games that use Physx it but the ones that do make a huge difference in immersion.

 

[weehamish]

In Batman its good, but in Borderlands 2 its nothing special.

You have to see though ACE it is Nvidia gimping it.