Framerate Problems Alpha 19

[SylenThunder]

Not sure why this game is so demanding on hardware when it doesn't even look that great.  I get 100+ fps on warzone with graphics set on high.

It's a 3D voxel game with structural integrity and a complex AI. Warzone is a 2D game with static maps and a simple AI. Add to that the fact that the game is still in Alpha development, where Warzone has already finished beta, optimizations, and has been fully released. You're comparing apples to oranges. Just like every one comparing this game to GTA-V.

Your specs are more than enough to get a solid 60 FPS at 1080p. Assuming you aren't using any un-optimized graphic features like reflections. You'll also want to have your save data on an SSD as well.

 

[jdgreenfriend]

I'll agree that the game is complex and has a lot of depth to it that other games do not have, but using the alpha excuse is straight crap.  It's been out for years and years and cost money.  They can call it alpha all they want but that's just a lame excuse.  You can put syrup on @%$# but it doesnt make it pancakes.

 

[SylenThunder]

I'll agree that the game is complex and has a lot of depth to it that other games do not have, but using the alpha excuse is straight crap.  It's been out for years and years and cost money.  They can call it alpha all they want but that's just a lame excuse.  You can put syrup on @%$# but it doesnt make it pancakes.

Sorry but when core game features are still being developed and implemented, that IS the definition of Alpha development.  You have to create the batter before you can have pancakes.

98% of games that exist, you never saw Alpha development for because it was all internal.  Here you get to participate. If you can't deal with that, then don't participate in Early Access.  It's not like Steam didn't throw the definition of it, and the risks, in your face before your purchase.

 

[The_Great_Sephiroth]

CPU matters in this game. The world is voxel-based. Each block must be calculated as connected or not connected to one next to it, how it is connected, it's shape, etc and then all of that data goes to the GPU for rendering. You can easily choke on a weak CPU even if you have a massive video card. You may also have other apps accessing your disk while you play, which can KILL framerate because chunks are taking longer to load in, process, and render. Each system is unique and presents its own issues and advantages. 7 Days requires a lot of CPU and GPU, unlike most games, which only require a good video card. RAM helps too, and disk can be important here.

When posting your specs, post your specs completely. Example:

Windows 7 Pro 64bit OEM

Core i7-6950X (3.0-3.5GHz)

128GB DDR4 3200 (PC4-25600)

RTX 2080 Ti 11GB

Creative Ae5

1 x WD Blue SATA 6Gbps 1TB (OS only)

4 x WD Black SATA 6Gbps 1TB single-platter disks in BTRFS RAID10 (Most games)

1 x M.2 NVME 3,500Gbps SSD (Ark, 7 Days, etc)

That explains how you are actually set up. Tackling just the massive difference a disk can have, if you have a single-platter disk it only has to read and write one platter and is WAY faster than the more common three or four-platter disks, which have to load and save data across many more platters. The type of SSD is important also. a SATA SSD is capped around 6Gbps, but an M.2 NVME SSD can go up to around 4,000Gbps with the right model.

 

[-Holo-]

I'd like to add a bit of what I believe is Unity flakiness too,

I have seen this on many (all?) sandbox unity games, (and to lesser extent UE games) across several platforms and over many years.

Regardless of how good of a system it seems the engine itself chokes on something. (complexity?)

I have a logitech  keyboard with the display, that MSI afterburner is displaying all kinds of info in realtime for me,

CPU usage per core, temps, Mem usage, Vram usage, GPU usage and temp, disk usage. Everything useful

And even if no part of the system is even close to max, the game dips and runs at 30-50 fps in some cases. (Dishong tower especially nasty for this)

No videosettings changes affect the framerate with more then a few fps, so it really has to be something funky going on behind the curtains in the engine.

 

[Vampirenostra]

I'd like to add a bit of what I believe is Unity flakiness too,

I have seen this on many (all?) sandbox unity games, (and to lesser extent UE games) across several platforms and over many years.

Regardless of how good of a system it seems the engine itself chokes on something. (complexity?)

I have a logitech  keyboard with the display, that MSI afterburner is displaying all kinds of info in realtime for me,

CPU usage per core, temps, Mem usage, Vram usage, GPU usage and temp, disk usage. Everything useful

And even if no part of the system is even close to max, the game dips and runs at 30-50 fps in some cases. (Dishong tower especially nasty for this)

No videosettings changes affect the framerate with more then a few fps, so it really has to be something funky going on behind the curtains in the engine.

May that be the requirement to do way to many calculations per cycle? More cores required to do the trick, but multithreading is set to use like 4 cores max as I've heard?

 

[-Holo-]

May that be the requirement to do way to many calculations per cycle? More cores required to do the trick, but multithreading is set to use like 4 cores max as I've heard?

It's quite possible since it generally happens 'mid-lategame' when bases starts to become advanced, (in all games it happens in)

I'm fairly sure atleast 7days uses all threads available, albeit not much, most of my threads are at 20-30% on both client and the server (both ryzen 12 thread systems) with  two threads being used more at around 50-60%

Wish they'd had some sort to check for issues like this, it's so annoying to have 30 fps, when not a single part of the system is used more then 50%

 

[kptb]

I am having frame rate issues since 19.2 stable came out. Yesterday the game was running fine, today the frame rates dropped to the single digits and the game is unplayable as the rate drop occurs with movement.

 

[Ponasity]

My game is unplayable at low settings. I have 250 hours played with no issues at medium settings, pre A19. I havent played since March so I thought Id load up my A18 save. Nope. Tried to start a new save and its literally unplayable. I play every single other game I own on Ultra, getting at least 40 FPS. Not gonna bother posting specs as every person who has posted specs in this thread was met with aggression. Just letting Fun Pimps see yet another customer who is unable to play their game.

 

[Vampirenostra]

My game is unplayable at low settings. I have 250 hours played with no issues at medium settings, pre A19. I havent played since March so I thought Id load up my A18 save. Nope. Tried to start a new save and its literally unplayable. I play every single other game I own on Ultra, getting at least 40 FPS. Not gonna bother posting specs as every person who has posted specs in this thread was met with aggression. Just letting Fun Pimps see yet another customer who is unable to play their game.

Oh you are so wrong...

Why are you here on this forum?  You actually came with your first post to blame product for poor quality, and throw @%$# all around,  you're not here to attempt to solve your problem if you do really have one what I strongly doubt.

 

[Liesel Weppen]

May that be the requirement to do way to many calculations per cycle? More cores required to do the trick, but multithreading is set to use like 4 cores max as I've heard?

Hmm, for me 7d2d during normal gameplay uses all of my 8 cores/16 threads equally and then has an overall usage of like 20-25%. So i can't confirm that there is a limit of using only 4 cores in 7d2d.

I had massive framedrops during last bloodmoon with darkness falls, but there was so much action, i didn't look at the system load.

Maybe i catch a look during the next BM.

Iirc structural integrity also uses more threads. You can easily test that. Start a game, turn on dm and cm, look for a high building with little floor space or little support blocks. Use dynamite to blow up the support blocks. The construction site works nice. One of the skyscrapers is also easy to make collapse. Then watch your system load while the POI collapses.

 

[Tjb]

Oh you are so wrong...

Why are you here on this forum?  You actually came with your first post to blame product for poor quality, and throw @%$# all around,  you're not here to attempt to solve your problem if you do really have one what I strongly doubt.

All I can add is try Alpha 19.x maybe its better with your configured system.

 

[Daemonjax]

This game pushes my 1070ti past its limit during dusk and dawn hours @ 1920x1200 resolution.  My cpu is an i5 3570k (very old cpu with a mild OC), and it stays around 25-30% total cpu usage on singleplayer (but I didn't stress test it by blowing up a whole building or something like that).  I had to tone down some settings to keep the framerate playable during those dusk and dawn hours, but not that much -- still looks close enough to everything high/ultra.  Outside of those hours, gpu usage is around 50% iirc.  You'd figure this is the exact problem that the Dynamic Resolution setting is supposed to address, but that setting seems broken for me -- it doesn't do anything at all on my machine (static "works", but that's not what I need or want).  Anyways... Going from memory, I needed to lower the following (starting from the Ultra preset) to keep the fps drop during those hours within the playable range: 1) Object Detail to High; 2) Anti-aliasing Low; 3) Motion Blur Off. 

The OP's laptop's gpu and cpu don't seem too bad to me, although they are both weaker than my aging PC.  Although, the 2.2Ghz clock rate on the 8750h is concerning.  I imagine he's playing at a lower resolution than me, so that'll help out his GPU.   Maybe start from High settings, and use that as the starting point before lowering the above 3 things I mentioned (with Object Detail set to medium)?

I'd move the game from the HDD to SSD.  No reason not to.  I'm also assuming he's not accidentally using the integrated gpu.

 

[Tjb]

I on the other hand will add this my 2 cents.

Shut down your system and open it up. Usually some screws at the back of the case.

Take it out side and blow with compressed air if you can the heat sinks and fans. Get the dust and crap out of the thing. Off the heat sinks and fans.

Don't go ballistic just easy and gently blow or otherwise clean all the crap out.

 

[Vampirenostra]

Hmm, for me 7d2d during normal gameplay uses all of my 8 cores/16 threads equally and then has an overall usage of like 20-25%. So i can't confirm that there is a limit of using only 4 cores in 7d2d.

I had massive framedrops during last bloodmoon with darkness falls, but there was so much action, i didn't look at the system load.

Maybe i catch a look during the next BM.

Iirc structural integrity also uses more threads. You can easily test that. Start a game, turn on dm and cm, look for a high building with little floor space or little support blocks. Use dynamite to blow up the support blocks. The construction site works nice. One of the skyscrapers is also easy to make collapse. Then watch your system load while the POI collapses.

That is strange... I can clearly see not all cores are used in my MBP...

 

[linewalker]

That is strange... I can clearly see not all cores are used in my MBP...

View attachment 17107

it appears to be using the physical cores and not the fake hyperthreaded cores. which is pretty typical for a lot of earlier chipsets. especially if they are amd.

 

[Liesel Weppen]

That is strange... I can clearly see not all cores are used in my MBP...

But you can also see that it uses more than 4.

I didn't look that close, i just saw that it all cores are used. Some might indeed have lower usage anyway (or maybe even used for other background processes).

it appears to be using the physical cores and not the fake hyperthreaded cores. which is pretty typical for a lot of earlier chipsets. especially if they are amd.

It makes sense anyway. The hyperthreaded ones are not real, full cores. Especially if the real core ist at 100%, the hyperthread one can not bring the same performance as another real core on top. However, even if a real core is shown at 100% it doesn't mean all of his ALUs/FPUs are really used. Those unused units are still available to use for the hyperthreaded core, but that may not be enough to do further heavy calculations with them. That's why hyperthreading was invented.

From what i've seen in tests (and tested my self like 20 years ago on the first P4HT) Hyperthreading brings "only" 0-30% additional performance. How much it really is, hardly depends on the type of workload (and type of the cpu). Back then some specific workloads ran even faster if HT was turned off, but i guess that is solved nowadays, because the OS-scheduleres are now aware of how hyperthreading really works, what cpu is used exactly and how the workload looks like. And the result of that optimizations should look exactly like the screenshot above.

 

[meganoth]

I turned off hyperthreading in A17 and it got me a few fps at the time. Sounds like I should try turning it on again, the negative impact might be gone.

 

[napalm00]

I'm also facing horrible performance in Alpha 19.

No matter what I do with the video settings, vsync, dynamic resolution, the game still dips from like 120-200fps down to 20 for a few seconds, then back again.

This happens randomly and also when loading parts of the map that have already been visited. Both single and multi player.

PC Specs:

AMD Ryzen 9 5950X

NVIDIA RTX 3090 24GB

2x 16GB G.Skill 3600MHz

2x Sabrent Rocket 4.0 1TB

2 monitors (4K main and 1080p secondary)

The game has no reason to run this bad on such a rig and yet it does. It's the only game I've encountered that has such bad performance.

The funny thing is Alpha 18 used to run fine on my older rig (i7 4770k (?), 32GB G.Skill 1333MHz, EVGA GTX 980).

 

[Vampirenostra]

it appears to be using the physical cores and not the fake hyperthreaded cores. which is pretty typical for a lot of earlier chipsets. especially if they are amd.

This is an 15" MBP late 2018, There can't be any amd. it is an I-7 8750H 6 cores, 12 threads. I assume it is strange cause I can fully load the stone building my java projects, it starts throttling heavily under load, but keeps distributing the load upon all threads. And I know Mac is a horrible place to play games)), but it is more than sufficient and comfy for my work, especially if you get it as a tool from your employer.

But you can also see that it uses more than 4.

I didn't look that close, i just saw that it all cores are used. Some might indeed have lower usage anyway (or maybe even used for other background processes).

It makes sense anyway. The hyperthreaded ones are not real, full cores. Especially if the real core ist at 100%, the hyperthread one can not bring the same performance as another real core on top. However, even if a real core is shown at 100% it doesn't mean all of his ALUs/FPUs are really used. Those unused units are still available to use for the hyperthreaded core, but that may not be enough to do further heavy calculations with them. That's why hyperthreading was invented.

From what i've seen in tests (and tested my self like 20 years ago on the first P4HT) Hyperthreading brings "only" 0-30% additional performance. How much it really is, hardly depends on the type of workload (and type of the cpu). Back then some specific workloads ran even faster if HT was turned off, but i guess that is solved nowadays, because the OS-scheduleres are now aware of how hyperthreading really works, what cpu is used exactly and how the workload looks like. And the result of that optimizations should look exactly like the screenshot above.

Yes you're very right. More than 4 for sure. As we can see real cores are also far from 100% loading, here the problem is more on GPU side, as can be seeing from another chart a bit higher on the right poor Radeon Pro 555X 4Gb struggles to run the game even with lowest settings. 

BTW speaking of frame drops while collapsing large buildings - I've noticed it stutters more when looking at  blocks hitting the ground and not the ones that fall from the building getting destroyed.