PC Lets Talk Performance

[unholyjoe]

One thing is that we are still in the experimental phase of their development. TFP's opinion is that EA means they do their experiments as freely as possible and just make sure that they get a somewhat working version out from time to time for players to inspect.
While they are changing systems around, chances are that many optimizations would be made obsolete by later changes or even counter-productive (as optimizations usually make assumptions about the data that may change later and produce hard to find bugs). Optimization usually is done in the beta or lets call it polishing phase. I would guess they have to really start entering beta in the summer if they want to release 1.0 at the end of the year.

Another thing is that a voxel world put on top of an engine (i.e. Unity) that is optimized for rendering polygons is not optimal. Voxels always have been resource-intensive and were always behind in FPS to conventional rendering engines. While Unity was probably necessary so they don't have to built up everything from ground up, they will have a hard time optimizing anything inside the unity engine. There is a limit to the optimizations they can achieve down there.

My guess is that the amount of voxels that have to be juggled every second is simply bringing the data busses and CPUs on the PCs to their limits. Not much left for all the other necessary tasks like path finding and animation, all of which also need lots of data to be moved.

TFP said they are looking at optimizations but in the development of A17 it seems many of the optimization wins were countered by features that cost cpu and cpu cycles again. Now they might have a hard time finding further worthwile improvements. I'm sure they will be able to still improve the engine in the beta phase but the times where 7D2D would run on a potato are probably over.

thank you and others for an opened minded discussion. you are correct in most of this (too advanced and pointless for me to go further), but as you noted we are in development and we (entire pimps staff), really need support from the player base because things WILL change periodically and the extensive work is always something to keep in check.

 

[Kubikus]

Another thing is that a voxel world put on top of an engine (i.e. Unity) that is optimized for rendering polygons is not optimal. Voxels always have been resource-intensive and were always behind in FPS to conventional rendering engines. While Unity was probably necessary so they don't have to built up everything from ground up, they will have a hard time optimizing anything inside the unity engine. There is a limit to the optimizations they can achieve down there.
My guess is that the amount of voxels that have to be juggled every second is simply bringing the data busses and CPUs on the PCs to their limits. Not much left for all the other necessary tasks like path finding and animation, all of which also need lots of data to be moved.

The voxel argument always comes up when the "improvable" performance is discussed. Since I have been playing older alphas a lot recently, particularly A12 and A14, I'm no longer buying that one. A12 should have as many voxels as A17 and it runs just wonderful. It also runs just wonderful with 100 zombies around. A17 drops the framerate down to slideshow-speed when I enter a prefab, with, what?, 10 zombies around.
Voxels still might add to it, but the degree, in my uneducated opinion, should be negligible. Something else was done to decrease performance from where it was in the older alphas, the permanent stuttering/frequent micro-freeze was practically absent in A12. I'm assuming it's graphical improvements, such as the zombie textures that were updated in A13. A17 has some additional graphical ... "features" that hit the framerate really hard for me. It's pretty much unplayable at times, as I said, particularly inside prefabs.

 

[Maguvas]

thank you and others for an opened minded discussion. you are correct in most of this (too advanced and pointless for me to go further), but as you noted we are in development and we (entire pimps staff), really need support from the player base because things WILL change periodically and the extensive work is always something to keep in check.

Woooow a TFP man! Hi man and really I may disagree on the direction the game is taking ( more action dependent, less adventure oriented) but I cannot stop from thanking you to made a Linux native game.

Thank you.

Thank you.

 

[Khulkhuum]

Okay, let's talk performance...

Not gonna talk much about my PC (would need to check my settings, requiring a bit of time), but lets say with my current game options i'm sporting around a stable 30-40 fps.

I started with spawning regular Zs in an open field. 2nd batch of 25 didn't drop my frames, but next 25 made my fps drop to around 25. Another 25 Zs dropped fps to around 15.

Killall to get rid of zeds. Spawning a mix of regulars and rads got similar effects. Killall. Spawning only rads proved to lower fps just a couple points lower. Killall.

Spawning regular cops dropped fps at 50 of them (second 25 spawned, so quicker) and third spawning reduced fps similarly as the fourth regulars. This is most likely due to cops puking.

Killall and i entered into a skyscraper POI (for sleeper numbers). Awaking to below 50 walking/running Zs lowered fps briefly as they spawned, but as the numbers grew closer and above 60, the fps dropped more and more.

Further tests pending.

 

[Maguvas]

Okay, let's talk performance...
[...]

Further tests pending.

Your post gave me an idea: wouldn't be nice to have a wonderful 'do benchmark' command as startup parameter or as menu option?

While the GPU war between ati and nvidia raged in the past (now they're politely splitting the market ) I remember a lot of games with that feature.

I do not know if it would be costly for TFP doing something like that.

 

[Minandreas]

I really hope optimization of the game will see a huge difference in the long run... I have my concerns. I've seen a lot of games evolve and optimization has always seemed to go the same way. "Optimizations will come after alpha!" And then "It's only beta. The game's not finished. It will be better on release!" And then release comes and its hardly better than it was in alpha...

Really hope TFP will break the mold and show that those past experiences were just with lazy crap devs.