Want to see this type of testing on another map? let me know in the comments. Sorry for the slight shift in the text from 3:17 and onwards, I tried to fix this after rendering the video many times but could not solve it. Join the discord server ( discord.gg/6Utxzsa9wc ) that is run by RUclipsr Klemintime @klemintime1452 (I highly recommend you check out his channel!), there is a big community active that is always willing to help you out with any pc or performance related issues. Personally I'm very active in this discord so join up if you need any help with Tarkov or other related PC issues.
I think lower FOV acts like higher LOD somehow, hence the fps differences. Borderless is I think because of windows 11? not sure though, borderless has always been better for me but might be different for every system. Tarkov is indeed a mistery
@@p8gaming669 hags and vrr are on, nothing else. Disable Fullscreen Optimization check and unchecked for tarkov.exe resulted in the same fps ever since moving to windows 11.
@@tomfurstyfield I did a quick retest of some of the settings in a more open place within factory and the results didn't change. I agree that Factory isn't the greatest test for settings like LOD as an example but people still play Factory, so these results are still relevant for this map. Lighthouse will be up this weekend!
Doing the testing you did, in such a liner confined space was nearly pointless, as seen by the fact that most of the settings made no difference. Goto light house. Stand on dead tree rock and do it there. That way, draw distance and almost everything else (grass shadows etc) will be taken into account.
Lighthouse is in the works, plan is to have it up next sunday! I don't see why Reflex would gain you any FPS, maybe on+boost because it will force your GPU to clock higher constantly. It's also important to know how you came by those numbers so you can recreate it. Tarkov is hard to get solid benchmarks numbers from, every raid performs different, even offline.
One of the major reasons for GPU bound testing is because the majority of the settings within tarkov are GPU related/dependent (not sure if dependent is the right word lol). The second reason is: having a stable baseline to work with. Testing in a GPU bound scenario is very easy and consistent to work with, so you get to see very clean scaling of options within a setting without any other interference. CPU bound testing is EXTREMELY hard to get right in my opinion for Tarkov, when you do CPU bound testing either online or offline with AI, you will find out really fast how inconsistent testing becomes, especially online raids. Online raids are notorious difficult to test in because there are so many moving parts like: players, AI, amount of AI, weather, time of day and the list goes on. Offline testing even with AI does NOT perform the same every raid since AI spawn locations seem to be based off where you spawn. From my experience I would say 70% of the player base is CPU bound, so yes you're completely right in saying that most people are CPU bound. CPU bound testing should show that the majority of settings will have no effect on your FPS, and keep in mind that you need A LOT of GPU headroom for a pure CPU bound test, which most don't have. I would say most people have their GPU at like 60-80% usage, at 80% usage you turning up HBAO to the max setting will put you at a GPU bound scenario instantly. So it's easy to be GPU bound (easier then most people think) and therefor testing in a GPU bound scenario will give you better data if you will. I will of course repeat this testing on another map while being GPU bound and I will try my best to do all of these tests in a CPU bound situation.
How is the aspect ratio testing being done? It's not gonna be the same resolution so how are they comparable? Did you go for the ones that were closest in pixel count? Otherwise it was a nice video even if Factory probably isn't the best map to test things out on.
No settings were changed, just swapping between the aspect ratios. You're right, no they won't result in the same resolution but no one that switches between aspect ratio's is going to account for that, people often pick 4:3 because that's 'best' for example. This was my first time testing every setting so I figured factory was a good place to start because it's the only map that doesn't have time of day and weather changes constantly happen. I just got done recording the same tests on lighthouse, which was a major pain to do because the weather changes alone lol.
For GPU usage I'm running the game in 4K with medium ish settings as the baseline, that will put most GPUs at 100% usage. For CPU usage, I did not have that included in the overlay but my CPU usage doesn't go above 50% normally. If your GPU usage is normally around 50% than you mostlikely have the following going on: 1. your CPU is not fast enough 2. your GPU is way overkill compared to the CPU 3. your system is not running optimally 4. your settings are set 'to low' for what your GPU can handle (I'll have to know your specs & settings to know which one) Low CPU usage is normally for Tarkov. Tarkov tends to use 1 core at 85% and the other cores at 20/30%, and that results in you reading an overall low usage %.
@@wildgg here my specs : Rtx 4070 I5-11600k 32gb 3600mhz Game on nvme pcie 4.0 Maybe 2 wipe ago I disabled the desktop optimisation in property of escapefromtarkov.exe and my GPU/CPU usage jump to 100% and I double my fps, but it doesn’t work anymore :/ And yes my game settings are on low/off, but even with that, street is unplayable with 60 fps and a lot of shutter
@@wildgg Hi, here my spec : RTX 4070 I5-11600k 32GB DDR4 3600MHZ Game on NVME PCIE 4.0 few wipe ago i disabled the full screen optimisation in the escapefromtarkov.exe property and my GPU/CPU usage jump to 100% but it don't work anymore now it use like 30/40%
I could honestly not tell you but I have some ideas of why that is. Some games are bound by drawcells and in those games higher FOV hurts FPS. I don't think Tarkov is bound by that so therefor lower FOV with more 'zoom' has to render the objects closest in more detail, hence the lower fps. That's my theory on it anyways, it could very well be something that happens in certain scenarios and maps and where you are on the map etc. Tarkov is notorious hard to benchmark accurately so it could be anything to be honest with you. I'm currently re-doing all of these tests on Lighthouse, maybe FOV will give a different results there. (also keep in mind this is a GPU bound test, maybe in a CPU bound test FOV has a bigger impact)
The Manage 3D Settings should be used more as tools rather than tweaks. Last time I looked at them it didn't change the fps in any meaningful way. At some point I will test every settings in there but that's not high up on the list currently.
It would be great do this with STREETS. Bc it's the most fps dropping map. I have a i9 14900 en RTX 4090 32 gb mem and still on STREETS only 60 FPS. It's Russian spagetti code.
With those specs I would say use what ever really if you're on 1080 or 1440p. If you're GPU bound you basically want to turn down the settings that have the biggest impact like shown in this video. I'm planning on uploading a CPU bound version of this video by next sunday. I would check out the channel Klemintime, his latest video explains every setting really well.
@@Ikuorai Great question, something I probably should've included in the video. GPU bound basically means bottlenecked by the GPU, or the GPU is the limiter. bound by the GPU. CPU bound basically means bottlenecked by the CPU, or the CPU is the limiter. bound by the CPU. When someone says "I am GPU bound" it means their GPU is running at 100% and the GPU can't output more FPS. When someone says "I am CPU bound" it means their CPU can't deliver more 'work' to the GPU. A simple example would be: Both CPU and GPU are at 50% usage, this means you're CPU bound. CPU is at 50% and the GPU at 100%, this means you're GPU bound. I say simple example because of course ram plays a role and other things.
Unfortunately without 1% lows/.01% lows this basically tells us nothing, as well as the fact that you’re standing still looking at a wall. The fps differences are negligible and are no indication of what setting does what. You should benchmark by walking the same path looking in the same direction. At the very least do a 360 pan so you get those drops in fps and the graphic settings will then determine how low the drops in fps are.
Want to see this type of testing on another map? let me know in the comments.
Sorry for the slight shift in the text from 3:17 and onwards, I tried to fix this after rendering the video many times but could not solve it.
Join the discord server ( discord.gg/6Utxzsa9wc ) that is run by RUclipsr Klemintime @klemintime1452 (I highly recommend you check out his channel!), there is a big community active that is always willing to help you out with any pc or performance related issues. Personally I'm very active in this discord so join up if you need any help with Tarkov or other related PC issues.
>increasing fov increases fps
>borderless increases fps
>literally the opposite in every other game ever made
The way Tarkov works is indeed a mistery
I think lower FOV acts like higher LOD somehow, hence the fps differences.
Borderless is I think because of windows 11? not sure though, borderless has always been better for me but might be different for every system.
Tarkov is indeed a mistery
@@wildgg w11 has both borderless and full screen optimizations, do you have both activated?
@@p8gaming669 hags and vrr are on, nothing else.
Disable Fullscreen Optimization check and unchecked for tarkov.exe resulted in the same fps ever since moving to windows 11.
Wild and Klemintime are producing 10/10 content for the Tarkov community, thank you
Pyurology is goated too
How is this 10/10?! There's barely any difference in the FPS because it's in a terrible place for testing
@@tomfurstyfield I did a quick retest of some of the settings in a more open place within factory and the results didn't change.
I agree that Factory isn't the greatest test for settings like LOD as an example but people still play Factory, so these results are still relevant for this map.
Lighthouse will be up this weekend!
@@tomfurstyfield factory is the most stable enviroment tho. weather and time never changes.
Sweet man. Thanks!
Doing the testing you did, in such a liner confined space was nearly pointless, as seen by the fact that most of the settings made no difference. Goto light house. Stand on dead tree rock and do it there. That way, draw distance and almost everything else (grass shadows etc) will be taken into account.
Agree with you 100%, I also noted this in the beginning of the video.
Will definitely do other maps!
Appreciate the format, but agree that you should test other maps. Nvidia reflex more than doubles my FPS where it showed almost no difference for you.
Lighthouse is in the works, plan is to have it up next sunday!
I don't see why Reflex would gain you any FPS, maybe on+boost because it will force your GPU to clock higher constantly.
It's also important to know how you came by those numbers so you can recreate it. Tarkov is hard to get solid benchmarks numbers from, every raid performs different, even offline.
@@wildggI think it doubles OC’s fps because it removes the fps cap. Without the v-sync trick game is fps capped
@@themarketgardener Good point! I forgot about that
Nice job
Thank you!
do this on lighthouse on the main road and on rogues
What is your philosophy behind gpu bound testing when most players are probably(atleast i think) cpu bound?
One of the major reasons for GPU bound testing is because the majority of the settings within tarkov are GPU related/dependent (not sure if dependent is the right word lol).
The second reason is: having a stable baseline to work with. Testing in a GPU bound scenario is very easy and consistent to work with, so you get to see very clean scaling of options within a setting without any other interference.
CPU bound testing is EXTREMELY hard to get right in my opinion for Tarkov, when you do CPU bound testing either online or offline with AI, you will find out really fast how inconsistent testing becomes, especially online raids. Online raids are notorious difficult to test in because there are so many moving parts like: players, AI, amount of AI, weather, time of day and the list goes on. Offline testing even with AI does NOT perform the same every raid since AI spawn locations seem to be based off where you spawn.
From my experience I would say 70% of the player base is CPU bound, so yes you're completely right in saying that most people are CPU bound.
CPU bound testing should show that the majority of settings will have no effect on your FPS, and keep in mind that you need A LOT of GPU headroom for a pure CPU bound test, which most don't have. I would say most people have their GPU at like 60-80% usage, at 80% usage you turning up HBAO to the max setting will put you at a GPU bound scenario instantly.
So it's easy to be GPU bound (easier then most people think) and therefor testing in a GPU bound scenario will give you better data if you will.
I will of course repeat this testing on another map while being GPU bound and I will try my best to do all of these tests in a CPU bound situation.
How is the aspect ratio testing being done? It's not gonna be the same resolution so how are they comparable? Did you go for the ones that were closest in pixel count? Otherwise it was a nice video even if Factory probably isn't the best map to test things out on.
No settings were changed, just swapping between the aspect ratios. You're right, no they won't result in the same resolution but no one that switches between aspect ratio's is going to account for that, people often pick 4:3 because that's 'best' for example.
This was my first time testing every setting so I figured factory was a good place to start because it's the only map that doesn't have time of day and weather changes constantly happen.
I just got done recording the same tests on lighthouse, which was a major pain to do because the weather changes alone lol.
How do your CPU/GPU are used at 100% ? mine never go over 50%
For GPU usage I'm running the game in 4K with medium ish settings as the baseline, that will put most GPUs at 100% usage.
For CPU usage, I did not have that included in the overlay but my CPU usage doesn't go above 50% normally.
If your GPU usage is normally around 50% than you mostlikely have the following going on:
1. your CPU is not fast enough
2. your GPU is way overkill compared to the CPU
3. your system is not running optimally
4. your settings are set 'to low' for what your GPU can handle
(I'll have to know your specs & settings to know which one)
Low CPU usage is normally for Tarkov. Tarkov tends to use 1 core at 85% and the other cores at 20/30%, and that results in you reading an overall low usage %.
@@wildgg here my specs :
Rtx 4070
I5-11600k
32gb 3600mhz
Game on nvme pcie 4.0
Maybe 2 wipe ago I disabled the desktop optimisation in property of escapefromtarkov.exe and my GPU/CPU usage jump to 100% and I double my fps, but it doesn’t work anymore :/
And yes my game settings are on low/off, but even with that, street is unplayable with 60 fps and a lot of shutter
@@wildgg Hi, here my spec :
RTX 4070
I5-11600k
32GB DDR4 3600MHZ
Game on NVME PCIE 4.0
few wipe ago i disabled the full screen optimisation in the escapefromtarkov.exe property and my GPU/CPU usage jump to 100% but it don't work anymore now it use like 30/40%
Thx for your work! i dont get why FOV 55 has lower FPS than 75. So seeing more Details (55) drains more fps than seeing a wider image (75)?
I could honestly not tell you but I have some ideas of why that is.
Some games are bound by drawcells and in those games higher FOV hurts FPS. I don't think Tarkov is bound by that so therefor lower FOV with more 'zoom' has to render the objects closest in more detail, hence the lower fps.
That's my theory on it anyways, it could very well be something that happens in certain scenarios and maps and where you are on the map etc. Tarkov is notorious hard to benchmark accurately so it could be anything to be honest with you.
I'm currently re-doing all of these tests on Lighthouse, maybe FOV will give a different results there.
(also keep in mind this is a GPU bound test, maybe in a CPU bound test FOV has a bigger impact)
what are your thoughts on "Manage 3D Settings" in nvidia control panel? i feel like changing settings there dont affect ur fps at all
The Manage 3D Settings should be used more as tools rather than tweaks. Last time I looked at them it didn't change the fps in any meaningful way.
At some point I will test every settings in there but that's not high up on the list currently.
I still don't get it why is borderless better then on fullscreen?
I have no idea either, this is one of those settings you'll have to test for yourself to see which is better.
It would be great do this with STREETS. Bc it's the most fps dropping map. I have a i9 14900 en RTX 4090 32 gb mem and still on STREETS only 60 FPS. It's Russian spagetti code.
Streets is planned but it's a mess to test, soon hopefully.
If you get 60 on streets with those specs you're doing something wrong.
I tuned up my LoD from 2.5 to 4 on woods and easily lost like 30 fps for that.
@0:55
@@wildgg Yup. I know. Didn't forget about that. Just thought I'll mention it.
please do streets of tarkov
what are the best settings for squeezing out the most fps with a 7800x3d and a 4090
With those specs I would say use what ever really if you're on 1080 or 1440p. If you're GPU bound you basically want to turn down the settings that have the biggest impact like shown in this video. I'm planning on uploading a CPU bound version of this video by next sunday.
I would check out the channel Klemintime, his latest video explains every setting really well.
@@wildgg thanks man
@@wildgg what do you mean by "gpu bound" vs "cpu bound"
@@Ikuorai Great question, something I probably should've included in the video.
GPU bound basically means bottlenecked by the GPU, or the GPU is the limiter. bound by the GPU.
CPU bound basically means bottlenecked by the CPU, or the CPU is the limiter. bound by the CPU.
When someone says "I am GPU bound" it means their GPU is running at 100% and the GPU can't output more FPS.
When someone says "I am CPU bound" it means their CPU can't deliver more 'work' to the GPU.
A simple example would be: Both CPU and GPU are at 50% usage, this means you're CPU bound.
CPU is at 50% and the GPU at 100%, this means you're GPU bound.
I say simple example because of course ram plays a role and other things.
Unfortunately without 1% lows/.01% lows this basically tells us nothing, as well as the fact that you’re standing still looking at a wall. The fps differences are negligible and are no indication of what setting does what. You should benchmark by walking the same path looking in the same direction. At the very least do a 360 pan so you get those drops in fps and the graphic settings will then determine how low the drops in fps are.
Thank you for your feedback, I will await your video.
I think you should have tested this on another map
Lighthouse was uploaded yesterday!
bro testing on factory lol
You're free to do it better, lol