Borderlands 2 GPU/CPU benchmarks [TechSpot/HardOCP/Others]

[blackened23]

That is borderline lying.

MOST people don't have YOUR problem...aka they experinece fine gameplay.

Cart before horse.

Have you checked your system for DCP?
Or do you just want to whine up and do no searching for the cause?!

 

[KingFatty]

Here's a pretty mild run through caustic canyon or whatever it was called. The x is seconds, and the y FPS. I had the frames limited to 60 or there would have been jumps up to 200 most likely. This was only with 1 and then 2 other people. (1 joined somewhere in the timeframe)

I still need some more data from 4 person because single/2player "feels" smoother then with the 4 person chaotic battles.

Can you clarify what this graph shows? Is the first X seconds showing single-player, and then the next seconds X to Y are two-player?

Also, it seems nobody else is willing to go through the effort you did to make this chart. Can you share the steps you went through to create the chart? Perhaps quantify what time and effort is involved? maybe if others see it as easier, they won't be so resistant to back up their claims and show that it's very possible to avoid dips to 30 FPS when running borderlands 2 on a GTX690 (even better, avoiding 30 FPS dips on a GTX680, 670, or 660ti). I'm happy to hear about other's eyeballs telling them this is the case, but graphs are just nicer to see.

 

[blackened23]

Yep, it's time to put him on the witness stand. I expect 17 eyewitness reports and credibility statements regarding his character - and a youtube video of the last 5 hours he spent playing it. Additionally his hard drives will need to be confiscated (in addition to his computer) so that we can get to the bottom of this mystery. Lastly, I will need a polygraph for all statements related to physx and BL2. /jokes

 

[wand3r3r]

Can you clarify what this graph shows? Is the first X seconds showing single-player, and then the next seconds X to Y are two-player?

Also, it seems nobody else is willing to go through the effort you did to make this chart. Can you share the steps you went through to create the chart? Perhaps quantify what time and effort is involved? maybe if others see it as easier, they won't be so resistant to back up their claims and show that it's very possible to avoid dips to 30 FPS when running borderlands 2 on a GTX690 (even better, avoiding 30 FPS dips on a GTX680, 670, or 660ti). I'm happy to hear about other's eyeballs telling them this is the case, but graphs are just nicer to see.

I asked about benchmarking and someone offered the suggestion to use http://www.fraps.com/download.php which saves (default) to c:\FRAPS\Benchmarks in 3 *.csv files. There are the min, max, and average fps in one file, frame times in another, and then every second and the FPS for that second in a third.

You just take the fps file and open it in Excel and insert - line chart. The default came like the chart I posted. The FPS are the Y axis and that chart I posted above was 1200 seconds so 20 minutes of playing. The frames were capped at 60, thus no higher, but the drops show as spikes below 60.


Here is the latest gameplay.

Lonbjerg close your eyes, more actual data incoming

Now I just ran through the part where bloodwing was (map wildlife refugee(?), where Slag is) and here are the results of playing with one other person (2 people). In my experience 1 player is a lot less intense, 2 is still not bad, 4 is chaotic especially using sirens freeze thing, the guys turret, elemental weapons etc.

 

[DimmyK]

This is quick 8-min solo run (I have no friends, foreveralone.jpg lol) on a single 680 @ 1325/3312. 1920x1080, all settings on max, fps capped @ 120, tundra express, closest outpost to the spawn spot, 20-25 bandits, 5 helicoptery flying things. I was using corrosive grenade mod that spawns 5 child grenades + corrosive pistol + few barrels of goo around = pools of liquid everywhere.

If someone can point me to more demanding spot (where I can just jump in and clean out the outpost or whatnot without doing specific missions), I'd be happy to bench it too.

Here is 30 minutes of slaying crystalisks in Fridge, same settings as above. It starts with bunch of bandits at the entrance of the map, lots of explosions/corrosion liquid spills.

 

[TheUnk]

Try these settings to help with physx. Probably want some lower settings for the heap size. I have 3Gb cards and mem usage went up to 2Gb. Turning off light shafts seemed to help a good amount for me. Not sure what the cache does exactly or if it helped.

bAllowLightShafts=False
PhysXGpuHeapSize=1024
PhysXMeshCacheSize=256

 

[wand3r3r]

I had a run through jacks lair whatever it was called. There were 4 players but it wasn't as intense as the close quarter fights in other locations as it's a wide open space, and everyone was spread out vs. side by side fighting the same guy.

gtx 690
1920x1200
I updated the physx driver to 9.12.0613 as someone said that helped them.
~20 mins

Here's a capped at 60fps run in single player. It was where you need to blow up the furnace. It felt pretty mild and wasn't anywhere near the intensity of multiplayer. (this was before updating physx) In the wide open spaces in single player it doesn't even feel like it's very intense even with elemental weapons etc.

I tried to run across from the teleport station straight to another exit in the desert, shooting a few skags along the way along with a couple birds. This is as non-scientific as it gets, but the scenery is pretty empty with about a dozen enemy kills and only 1-2 minute runs.

I noticed when I look across the map where you can see a distance the frames were almost half then when I turned mostly facing a massive pump which had tremors when pumping.

 

[toyota]

screw physx even being on medium on me as the spots that tank with it on high tank just as bad on medium too. I get 80-100 fps on low physx but just 20-40 fps in the same spots on medium or high. this is on the part I mentioned earlier where you kill the assassins. all those silly little chunks that we have seen a million times just sit there on the floor tanking my framerate. its only in one section but it ruins the game there as its not even playable at times in specific spots.

 

[Durvelle27]

what i got in this

AVG ~ 62

MIN ~ 1

MAX ~ 64

1920x1080 everything set to the highest except physx which was medium

specs in sig

 

[supremor]

I'm running with FPS capped to 60 and it's usually pegged to 60 with occasional dips to ~45. Settings are maxed out except I have DOF off because I can't stand it, resolution is 5990x1200 bezel corrected and Physx is set to high.

This is on a 3770K at a modest 4Ghz and 2 ASUS DCII 670's oc'd to 1000/6300 and ingame boost of 1150-1200 according to afterburner.

Noticed no FPS variance between 2 player and 3 player games, no 4 player results.

 

[railven]

Stopped by the article in the OP, and sad to see they still haven't fixed their mistake.

So much for credibility.

 

[f1sherman]

Stopped by the article in the OP, and sad to see they still haven't fixed their mistake.

Well Physxinfo have updated their original article. Twice.

3 different runs. 2nd being the worst case scenario, and 3rd is upcoming ingame benchmark.

http://physxinfo.com/news/9425/borderlands-2-is-cpu-capable-of-handling-the-physx-effects/

 

[railven]

Well Physxinfo have updated their original article. Twice.

3 different runs. 2nd being the worst case scenario, and 3rd is upcoming ingame benchmark.

http://physxinfo.com/news/9425/borderlands-2-is-cpu-capable-of-handling-the-physx-effects/

Thanks, that's a good follow-up by PhysX Info. Kudos to them for doing what TS doesn't seem to want to do - journalism.

 

[biostud]

Is there a difference between gtx 670/680 and 570/580 when you compare their compute performance in physX. Does the 570/580 handels physX better than the 670/680?

 

[Jaydip]

No 670/680 will handle it better.

 

[SirPauly]

Thanks, that's a good follow-up by PhysX Info. Kudos to them for doing what TS doesn't seem to want to do - journalism.

Zogrim is impressive and scores some nice interviews. Here is another with the developers of FunCom about PhysX:

Funcom plans to further expand PhysX integration into their games

PhysXInfo.com: Are you planning to use any client-side physics in The Secret World or future projects? We mean anything from trivial ragdolls to complex effects, like destructible environments or character clothing simulation via NVIDIA APEX.

Rui Casais: Yes, we will be adding more and more PhysX and APEX support. We have some experiments going on internally and we hope that soon we’ll have them ready for the players on The Secret World.

PhysXInfo.com: Will you add any types of hardware accelerated PhysX effects to The Secret World?

Rui Casais: Yes, some of the effects are GPU accelerated.

http://physxinfo.com/news/9586/funcom-plans-to-further-expand-physx-integration-into-their-games/

 

[railven]

Zogrim is impressive and scores some nice interviews. Here is another with the developers of FunCom about PhysX:

Funcom plans to further expand PhysX integration into their games




http://physxinfo.com/news/9586/funcom-plans-to-further-expand-physx-integration-into-their-games/

Is that even a good idea? TSW pretty much flopped, I don't think spending more money on an API that will offer little to the game is a good use of their already limited resources. The game at highest settings runs awful on both nVidia and AMD hardware. Adding more effects on top will result in <30 FPS for most people.

 

[SirPauly]

PhysX is not an API but a middleware. Funcom has been using PhysX for some time on the server side.

 

[railven]

PhysX is not an API but a middleware. Funcom has been using PhysX for some time on the server side.

Have they optimized the game since their free 3-day trial? I remember having to turn off Tessellation on both our rigs to get decent >45 FPS. I remember in London just walking through the streets due to the cobblestone roads FPS would tank into the 30s for me and 40s for her, and when we came around the corner to where players were oh man did it become a jittery mess.

If they made some strides on the performance, then kudos, but from what I remember, adding a layer of PhysX is going to hurt performance, unless that performance hit was always there and now we just get to see the swirlies.

After playing Guild Wars 2 and Panderia, TSW is going to have an uphill battle. SW:TOR is going F2P too.

I guess my opinion doesn't matter, just curious more so how much more performance players are willing to give up for the eye candy. That engine felt so bloated for the little IQ you gained. My opinion, of course.

 

[SirPauly]

And if the performance is too much of a hit for a specific gamer for their GPU, they may lower the setting or trade-off a setting if they desire. Just a choice for a gamer to consider that may improve immersion -- no one is forcing anyone to use it.

 

[Termie]

For a light-hearted break from the intense discussion of PhysX and bottlenecks, check out TechReport's article on the highly addictive nature of BL2: http://techreport.com/blog/23649/strung-out-on-borderlands-2

 

[railven]

And if the performance is too much of a hit for a specific gamer for their GPU, they may lower the setting or trade-off a setting if they desire. Just a choice for a gamer to consider that may improve immersion -- no one is forcing anyone to use it.

I'm not even arguing any of that, hell I already had to lower the settings on our rigs to get satisfactory results. My question is more - is this even worth their efforts considering the game flopped and their budget/workforce was reduced because the game flopped.

You don't seem to counter what I remember so I will assume this game still has a poorly optimized engine, and adding fluff that I'm not even sure how many will enjoy seems counterproductive to fixing the issues the game had (I stopped playing because I ran into a bug where a monster I had to kill wouldn't spawn. I guess this was so well known a lot of people moved on from the zone. Have they fixed these issues or is adding PhysX more important?)

 

[Termie]

New HardOCP article on performance and IQ:

http://www.hardocp.com/article/2012/10/01/borderlands_2_gameplay_performance_iq_review

I've changed the title of this thread to reflect that the OP now lists multiple articles on Borderlands 2 performance.

For better or for worse, HardOCP provides no benchmarks of AMD vs. nVidia on equivalent settings, as they run all the nVidia cards with PhysX. This may simply be a philosophical choice - PhysX seems to add to the graphical quality of this game, so they run it with that option turned on if it's available.

The most interesting comparisons, therefore, are within products lines. The (admittedly OC'd 660Ti) absolutely kills the 660, which is pretty unexpected. Also, not surprisingly, the 7970 GHz Edition distances itself from the 7950 and 7870, which is reflective of the current pricing of the cards.

 

[SirPauly]

I'm not even arguing any of that, hell I already had to lower the settings on our rigs to get satisfactory results. My question is more - is this even worth their efforts considering the game flopped and their budget/workforce was reduced because the game flopped.

You don't seem to counter what I remember so I will assume this game still has a poorly optimized engine, and adding fluff that I'm not even sure how many will enjoy seems counterproductive to fixing the issues the game had (I stopped playing because I ran into a bug where a monster I had to kill wouldn't spawn. I guess this was so well known a lot of people moved on from the zone. Have they fixed these issues or is adding PhysX more important?)

What's the point about already complaining when the feature isn't officially implemented or investigated yet?

 

[NIGELG]

What's the point about already complaining when the feature isn't officially implemented or investigated yet?

What's the point of already posting when the feature isn't officially implemented or investigated yet?