Metro Last Light early benchmarks on gtx 670

[tviceman]

http://www.guru3d.com/articles_pages/asus_geforce_gtx_670_mini_review,18.html

Ouch I hope nvidia has some drivers in store to improve performance.

 

[MrK6]

I'm not sure why you're rolling your eyes, because there really isn't anything incorrect about toyota's statement. MSAA has a big performance hit in Metro 2033, and DOF creates stuttering in many areas. Many people in this thread have stated that DOF makes the game feel very much less smooth and less fluid, and it's a true statement. I'd say the vast majority of players that have played it through fully did so without DOF. There's something very wrong with the DOF implementation in Metro 2033.

I agree with both of you, and my point was that if you turn off the broken garbage of course the game is going to "feel" better. As long as we keep using crap/unoptimized code to benchmark our hardware it will be difficult to be satisfied.

 

[toyota]

I agree with both of you, and my point was that if you turn off the broken garbage of course the game is going to "feel" better. As long as we keep using crap/unoptimized code to benchmark our hardware it will be difficult to be satisfied.

then why were you rolling your eyes? I already pointed out that its not the average that makes the game sluggish. again its mostly dof and msaa that will simply just make the game too inconsistent at least with a card like 680 and below. so poohbear and anyone else claiming the game is "smooth" on full max settings with a gtx670 do not have a clue what smooth is.

 

[f1sherman]

@tviceman

my fault... Last REFUGE(!)

 

[Enigmoid]

Ouch I hope nvidia has some drivers in store to improve performance.

Very likely, The 650, 660 are quite competetive (or even better than usual) when compared to the Amd cards. Its only until the 660 ti + that you see poor scaling or fps vs power.

 

[toyota]

Ouch I hope nvidia has some drivers in store to improve performance.

and that is only using AAA and no physx

 

[SirPauly]

Would like to see that bench without DOF though!

 

[bystander36]

I was thinking the same. DoF utilizes DirectCompute heavily, which is Nvidia's biggest short coming. At least this series. That is, of course, assuming they are using advanced DoF with that benchmark.

 

[toyota]

without dof or 4x msaa the framerate is nearly doubled.

 

[Headfoot]

Eloquence vs nerd rage.

There's no screenshot to take, i never said it runs @ 60fps nor do i think thats necessary in this game for smooth gameplay. The game ran smooth at whatever fps i was getting, thats all i said & then the nerd rage exploded like a cat on crack.

We have no data to talk about without a screenshot of the benchmark or a frame rate graph from a run through a level. If you want to have any sort of real discussion you need hard evidence. Otherwise, it's just words

 

[2is]

We have no data to talk about without a screenshot of the benchmark or a frame rate graph from a run through a level. If you want to have any sort of real discussion you need hard evidence. Otherwise, it's just words

And "just words" is just fine. It's really very simple. He said the game ran smoothly for him. Toyota posts TPU charts. I looked at them and thought "you know, I remember my in-game performance being better than those charts" and replied with that. He's not out to "prove" anything nor am I...

What proceded that was sudden nerd rage by two members, one of them driven to near insanity. The nerd rage also included insults getting tossed around, and when those two special members got insulted back, they got offended and raged some more. One of them clealy being emotionally compromised that a stranger in the Internet classified this game as an RPG. The icing was when Poohbear insinuated he has a better social life. The number of times his words were cynically repeated is amusing and telling at the same time.

I've never seen such emotional outbursts over something so trivial

 

[toyota]

And "just words" is just fine. It's really very simple. He said the game ran smoothly for him. Toyota posts TPU charts. I looked at them and thought "you know, I remember my in-game performance being better than those charts" and replied with that. He's not out to "prove" anything nor am I...

What preceded that was sudden nerd rage by two members, one of them driven to near insanity. The nerd rage also included insults getting tossed around, and when those two special members got insulted back, they got offended and raged some more. One of them clealy being emotionally compromised that a stranger in the Internet classified this game as an RPG. The icing was when Poohbear insinuated he has a better social life. The number of times his words were cynically repeated is amusing and telling at the same time.

I've never seen such emotional outbursts over something so trivial

I talked about REAL game performance too. and you also claimed to get better benchmark results than those shown on tpu.

 

[2is]

I talked about REAL game performance too. and you also claimed to get better benchmark results than those shown on tpu.

Congratulations.

Your first sentance is completely irrelavent to anything I said in my previous post.

As to your second sentance, I specifically stated I've never ran the benchmark.

 

[toyota]

Congratulations.

Your first sentance is completely irrelavent to anything I said in my previous post.

As to your second sentance, I specifically stated I've never ran the benchmark.

no its relevant because you keep acting like I am only talking about the benchmark.

and I misread your earlier comment in this thread as I thought you had also referred to your benchmark being higher so my mistake.

 

[2is]

no its relevant because you keep acting like I am only talking about the benchmark.

and I misread your earlier comment in this thread as I thought you had also referred to your benchmark being higher so my mistake.

It's not relavent at all, you're just being dense. I was not even talking about the game or the benchmark. I was paraphrasing the series of events that took place and the douchebaggery that ensued.

You've tested the game, inside and outside the benchmark. I get it now and I got it the first, second and third time you said it. Didnt change anything then, doesn't change anything now.

 

[poohbear]

I feel like the counter arguments are being presented by people who don't own the game and haven't played it. Maybe they played it for 10 minutes and they think they have an idea of what's going on. I don't know. But, this thread has already veered so far from the original topic that it's just tiring now. The actual Last Light game is due in a few days and hopefully we can get some real benchmarks then.

Finally i can agree with you on something. The problem is your posts were posted in an inflammatory manner, and hence why you get 3-4 pages of people arguing back and forth about nonsensical details. We insist it ran fine for US, and you insist it didn't, which is annoying as hell cause your standards are NOT the same as ours and you think you can perceive our enjoyment of the game based on YOUR standards. How can you argue it didn't run fine for us?

Next time, to avoid hijacking a thread and starting a flame war, don't post things in an asinine manner.

 

[sontin]

PCGH has new numbers from the review version:
http://www.pcgameshardware.de/Metro...ests/Metro-Last-Light-Benchmark-Test-1069286/

A GTX680 is 34% faster than the 7970GHz (Cat 13.5). Titan (876MHz) is nearly 32% faster than the GTX680.

 

[bystander36]

PCGH has new numbers from the review version:
http://www.pcgameshardware.de/Metro...ests/Metro-Last-Light-Benchmark-Test-1069286/

A GTX680 is 34% faster than the 7970GHz (Cat 13.5). Titan (876MHz) is nearly 32% faster than the GTX680.

I bet the other one was benched with something like advanced DoF on, and these don't. So depending on what you settings you want, it could go either way.

 

[Grooveriding]

I bet the other one was benched with something like advanced DoF on, and these don't. So depending on what you settings you want, it could go either way.

There is no DOF setting in Metro Last Light, these are the only options available to customize.

Resolution
Quality (Low, Normal, High, Very High)
SSAA (Auto, Off, 0.5x, 2x, 3x, 4x)
Texture Filtering (AF 4x, 16x)
Motion Blur (Low, Normal)
Tesselation (Off, Normal, High, Very High)
VSYNC (On, off)
Advanced Physx (On, off)


Basically from looking at the old and new benches, the game is unplayable with SSAA and gpu physx unless you are running multiple Titans or 680s/670s.

Turn those off though and it looks to play quite well. This is similar to 2033 and the MSAA/DOF settings. Same situation, different system killer settings :whiste:

 

[blackened23]

There is no DOF setting in Metro Last Light, these are the only options available to customize.

Well it looks like they took the most worthless option from Metro 2033 and removed it from last light. I won't miss it. Apparently, the 4A engine has the worst DOF implementation known to man.

Actually, DOF seems to vary wildly across games - Witcher 2's DOF implementation REALLY impressed me - and there isn't a performance anomaly for using it. But it's very variable across different games. Anyway, good riddance DOF in metro last light. How I won't miss you.

 

[sontin]

DoF is not an dedicated option in Last Light. It's under the Image Post-Processing tab.

 

[Grooveriding]

Well it looks like they took the most worthless option from Metro 2033 and removed it from last light. I won't miss it. Apparently, the 4A engine has the worst DOF implementation known to man.

Actually, DOF seems to vary wildly across games - Witcher 2's DOF implementation REALLY impressed me - and there isn't a performance anomaly for using it. But it's very variable across different games. Anyway, good riddance DOF in metro last light. How I won't miss you.

Total Biscuit's review : http://www.youtube.com/watch?v=DWimz-duun8&feature=player_embedded

He's running SLI Titans and said he had to turn off SSAA.. lol. Also no FOV controls and it has a narrow FOV like the last one. In 2033 you could adjust it in an .ini config, in Last Light there is not such config... crappy.

These are the only options available:

 

[bystander36]

The reason DoF was so heavy of a performance hit on Nvidia cards in Metro 2033, is how DoF was implemented to use DirectCompute. It meant it could do more, but Nvidia doesn't do DirectCompute so well, which is why AMD wins with DoF on.

 

[sontin]

Total Biscuit's review : http://www.youtube.com/watch?v=DWimz-duun8&feature=player_embedded

He's running SLI Titans and said he had to turn off SSAA.. lol.

I don't get it. SSAA is the most demanding option out there. What are people expecting? Doing 4x the processing of pixel should be result in a low performance impact?

I would make more sense if people start to educate themself what each option does.

 

[Elfear]

I don't get it. SSAA is the most demanding option out there. What are people expecting? Doing 4x the processing of pixel should be result in a low performance impact?

I would make more sense if people start to educate themself what each option does.

Maybe people are hoping that $2000 worth of GPUs would allow for the best eye-candy?