Aginor Posted July 6, 2016 Posted July 6, 2016 We just discussed that recently in another thread. It is not as easy as it may seem. DCSW weapons cheat sheet speed cheat sheet
Bluelight Posted July 7, 2016 Posted July 7, 2016 Yes it typically is, but Windows doesnt just swap memory to virtual memory, it as well compress the memory when in use and then swap it. (Swapping means it is moved from RAM to disk). And that compression causes a situation where DCS process can just drop from a 30-40% of CPU consumption to 10-20% without any logic by first sight. As you look the DCS process and it isn't eating CPU or GPU so you think that everything is OK and there is a problem with settings. While you can't even easily see the NT operating system own resource consumptions, why any of third party process managers can't show it either easily as NT hides it from them. So the problem is that DCS starts caching lots of RAM for textures and models from towns etc. But it doesn't efficiently use RAM as NT starts compressing DCS process memory even when you have like 16GiB free from 32GiB. And then it doesn't always swap it, just compress it. And that eats I/O and processing as DCS is still just designed for a single core. So DCS process CPU process can stay low and it looks like CPU is idling for no reason, while NT is seriously slowing caching and that causes serious framedrops or even jittery and pauses as the process just is waiting data that NT is processing and denying DCS to have as single core design can't expect to be waiting that RAM. I had a lot of problems even when system resources are way over needed recommendations. And simple way (of not only one) disable RAM compression is to disable paging file. I believe that it can be disabled per process(/executable) but i don't care as 32GiB is enough for everything and this doesn't affect for sleep function. If i disable all extra "useless" startup process etc (reset/reinstall windows), i get boot time now to 7-10 second. Same but without disabling paging file i can't get it under one minute by any means. This NT memory compression is just a problem and DCS isn't only one that gets affected from it. Sadly this didn't seem to have any affect on my situation. Thanks for the suggestion though. HTC Vive, Saitek X52 Pro, i7-950 Overclocked to 4ghz, Noctua NH-D14 cooler, ASRock x58 Extreme MB, EVGA GTX 970 FTW, 24GB G Skill Sniper DDR3 1600 RAM, EVGA 650-GQ (650 watts) PSU, Windows 10 Home
BiBa Posted July 7, 2016 Posted July 7, 2016 (edited) We just discussed that recently in another thread. It is not as easy as it may seem.There are a lot of threads on this topic... Just having had that theme again futilely discussed / ruminated, doesn't entitle its incontinence any validity, because 99% of them cause only diversion away from the Programmer's incompetence. Please cite only the link that is REALLY NOT delusory. Edited July 7, 2016 by Biba 1 BiBa...............BigBang WIN 10-64 Pro. MoBo: ASUS ROG STRIX Z390-F GAMING. 1TB Samsung SSD 960 PRO M.2 + 4TB SSD LEXAR 790. CPU: Intel Core i9-9900K. GPU: MANLI RTX 2080 Ti. RAM: HyperX 3200 MHz 64 GB. Monitor: ASUS 4K 28"/VR: Pimax 4K/TrackIR-5/SAITEK X55 Rhino HOTAS-Yoke-Rudder-Trim Wheel-THRUSTMASTER TCA Quadrant.
Fri13 Posted July 7, 2016 Posted July 7, 2016 Even the old FSX dinosaur has Addons like FSPS - Xtreme FSX PC, which can read the Hardware and can activate all processor cores! If the DCS programmers are too lazy / or not been payed well enough to fulfill this task, why don't DCS open the doors to theird party specialists like FSPS and beams uns eventually into the 21st ?! Because in fact a single-core CPU is faster than multi-core for a single process. This is reason why a Intel wins AMD in single process as Intel has larger CPU cache that speeds radically processing. But start running multiple processes and AMD wins. That is reason why i would take a CPU that has even a 512MiB L1 cache for single core CPU than a 8 core with 32MiB L1 and 32MiB L2 for each core, even when clock speed would be same. But for a servers where you run multiple processes instances etc... multi core system is better. And ED made correct decision for design to single core systems, but question is how well they have threaded the process so when one thread is waiting resources, other thread is undisturbed from it And that is where problems start. A easy way is to separate video and sound, but to split video to multiple ones becomes very tricky quickly. And this even if software is designed for multi- or single-core systems. But we don't run DCS as single process on computer, and have OS for that so we can have multiple processes running same time as OS task is to give each process the required time on CPU and specific amount of RAM or control of the different inputs or outputs. And it just makes easy for many to have a multi-core CPU as OS can assign each core to each process with better control, even when all are single-core designed (Like a music player, a word processor and a image editor.) So DCS that needs to start caching textures and models from drawing distance and further that distance is, more it needs it to do in same time, in inverse square factor (double the distance and you quadruple amount). And of the caching causes other graphical processing to wait... you get the idea. And now if the caching is slowed down by compressions by trying to compress RAM on the fly, it will slow down all. Why the RAM compression can cause problems but not always. i7-8700k, 32GB 2666Mhz DDR4, 2x 2080S SLI 8GB, Oculus Rift S. i7-8700k, 16GB 2666Mhz DDR4, 1080Ti 11GB, 27" 4K, 65" HDR 4K.
Bluelight Posted July 8, 2016 Posted July 8, 2016 Because in fact a single-core CPU is faster than multi-core for a single process. This is reason why a Intel wins AMD in single process as Intel has larger CPU cache that speeds radically processing. But start running multiple processes and AMD wins. That is reason why i would take a CPU that has even a 512MiB L1 cache for single core CPU than a 8 core with 32MiB L1 and 32MiB L2 for each core, even when clock speed would be same. But for a servers where you run multiple processes instances etc... multi core system is better. And ED made correct decision for design to single core systems, but question is how well they have threaded the process so when one thread is waiting resources, other thread is undisturbed from it And that is where problems start. A easy way is to separate video and sound, but to split video to multiple ones becomes very tricky quickly. And this even if software is designed for multi- or single-core systems. But we don't run DCS as single process on computer, and have OS for that so we can have multiple processes running same time as OS task is to give each process the required time on CPU and specific amount of RAM or control of the different inputs or outputs. And it just makes easy for many to have a multi-core CPU as OS can assign each core to each process with better control, even when all are single-core designed (Like a music player, a word processor and a image editor.) So DCS that needs to start caching textures and models from drawing distance and further that distance is, more it needs it to do in same time, in inverse square factor (double the distance and you quadruple amount). And of the caching causes other graphical processing to wait... you get the idea. And now if the caching is slowed down by compressions by trying to compress RAM on the fly, it will slow down all. Why the RAM compression can cause problems but not always. U said a single core is better.. if I disabled all my cores except one or 2 in my BIOS, would that make a difference? Even so, virtual reality requires 4 cores so if they want to make those of us who plan to play this game only in VR happy (money dropping) customers, they must fix these issues somehow. HTC Vive, Saitek X52 Pro, i7-950 Overclocked to 4ghz, Noctua NH-D14 cooler, ASRock x58 Extreme MB, EVGA GTX 970 FTW, 24GB G Skill Sniper DDR3 1600 RAM, EVGA 650-GQ (650 watts) PSU, Windows 10 Home
Fri13 Posted July 8, 2016 Posted July 8, 2016 U said a single core is better.. if I disabled all my cores except one or 2 in my BIOS, would that make a difference? Even so, virtual reality requires 4 cores so if they want to make those of us who plan to play this game only in VR happy (money dropping) customers, they must fix these issues somehow. It wouldn't as it is CPU design that is required to be changed so. So specific CPU design with lots of L1 cache etc would be a different efficiently. And the is no need to disable any cores via BIOS as you can use OS features to limit how many cores process can use. NT problem is that you can't tie process to a specific core like Linux allows you to do, as NT will allow process execution to hop between cores that is assigned, just keeping the amount as ordered. So if DCS would run on Linux or BSD, we could really manage what resources we give it, like we could command it to use cores 3 and 6 from 8 cores, use a RAM area between 16GB and 28GB from 32GB and then have a highest execution privilege after OS (=kernel) and even a assign a totally own swap for a own fast drive if wanted, or make virtual one to RAM (like from 28GB to 32GB). Then we could assign VR to have own resources that are separated from DCS and they never need to share any of the resources and that way being really running parallel without affecting each others if we wouldn't throttle them down. But as you get it, we would be doing part of the OS tasks then. But on Windows i would like to see would it be possible to fit a DCS to virtual drive in RAM so you don't slow down from even SSD. The whole DCS wouldn't be needed to be mapped to RAM, just textures and mapping and models so they are cached as quickly as RAM can read and write. That of course if the I/O of the drives is bottleneck for performance in a first place. There are many ways to try boost games performance, especially those that has huge resource requirements. i7-8700k, 32GB 2666Mhz DDR4, 2x 2080S SLI 8GB, Oculus Rift S. i7-8700k, 16GB 2666Mhz DDR4, 1080Ti 11GB, 27" 4K, 65" HDR 4K.
Goldsmack Posted July 8, 2016 Posted July 8, 2016 I have the EXACT same problem as well. It ran perfectly fine until I updated to 1.5.4. The town drop my FPS to 40 where they use to be 80s-90s. None of my settings have changed. I have a FPS test mission and I have def lost frames but only when looking at towns. I've tried damn near everything. I also checked other games and there all fine. Nevada map seems to run ok especially after the building draw distance was pulled back its only in the caucuses map I think. I can't be sure about Nevada more testing is required. Sorry just thought I'd chime in here I was going to make a new thread then I saw this one. Windows 10 Pro 64, I5 4690k @4.6GHz with CAPTIAN 240EX AOI, Samsung 850 EVO ,G Skill Ripjaws 16G RAM, Nvidia GTX 970 STRIX, MSI Z97 GAMING 5, WD Blue 1TB HDD, Seasonic M12 II EVO psu, Track IR 5, Pro Flight X-55 Rhino H.O.T.A.S
Bluelight Posted July 8, 2016 Posted July 8, 2016 I have the EXACT same problem as well. It ran perfectly fine until I updated to 1.5.4. The town drop my FPS to 40 where they use to be 80s-90s. None of my settings have changed. I have a FPS test mission and I have def lost frames but only when looking at towns. I've tried damn near everything. I also checked other games and there all fine. Nevada map seems to run ok especially after the building draw distance was pulled back its only in the caucuses map I think. I can't be sure about Nevada more testing is required. Sorry just thought I'd chime in here I was going to make a new thread then I saw this one. Well this sucks to hear are you saying you just updated to 1.5.4 today and it hurt performance?? Or are you saying the beta is what did it? Cause everyone was saying the same thing about going from 1.5.2 to 1.5.3. I am on 1.5.3 right now should I not update to 1.5.4 official release??? I would hate to think every version of this game that is updated, gets worse. HTC Vive, Saitek X52 Pro, i7-950 Overclocked to 4ghz, Noctua NH-D14 cooler, ASRock x58 Extreme MB, EVGA GTX 970 FTW, 24GB G Skill Sniper DDR3 1600 RAM, EVGA 650-GQ (650 watts) PSU, Windows 10 Home
EAKMotorsports Posted July 8, 2016 Posted July 8, 2016 stick to the 1.5.3 no update, the 1.5.4 still buggy and heavy Intel® Core™ i5-2500k CPU@4.20GHz 64 bit operation System Windows 10+ Pro NVIDIA GeForce GTX 1060 - Memory: 16.0 GB - 500gb ssd samsung - Samsung 27"SyncMaster TA550 monitors [SIZE=1][B]- [/B][/SIZE][FONT=Arial][SIZE=2]TM Hotas Warthog[/SIZE][/FONT] Trackir4 - TM Rudder Pedals.
Bluelight Posted July 8, 2016 Posted July 8, 2016 stick to the 1.5.3 no update, the 1.5.4 still buggy and heavy ok .. thats sux though. Didn't they release it as a non beta official update today? HTC Vive, Saitek X52 Pro, i7-950 Overclocked to 4ghz, Noctua NH-D14 cooler, ASRock x58 Extreme MB, EVGA GTX 970 FTW, 24GB G Skill Sniper DDR3 1600 RAM, EVGA 650-GQ (650 watts) PSU, Windows 10 Home
Goldsmack Posted July 8, 2016 Posted July 8, 2016 To be honest I never used 1.5.3 allot and never did a benchmark because I almost exclusively ran 2.0 NTTR. but it seemed to run ok for me but allot of users have/are reporting problems with 1.5.3. I only noticed the problems after updating to 1.5.4 So all I can say for certain is 1.5.2 ran better. This appears to be a bug with DCS so we might just have to wait till another update/hot fix. I'm sure ED is now aware of the issue. Windows 10 Pro 64, I5 4690k @4.6GHz with CAPTIAN 240EX AOI, Samsung 850 EVO ,G Skill Ripjaws 16G RAM, Nvidia GTX 970 STRIX, MSI Z97 GAMING 5, WD Blue 1TB HDD, Seasonic M12 II EVO psu, Track IR 5, Pro Flight X-55 Rhino H.O.T.A.S
wolle Posted July 8, 2016 Posted July 8, 2016 For all it's worth: I have the same problem. Low FPS when looking at towns. Seems to be related to low graphics card usage when looking at towns. My graphics card usage for 1.5.4 is typically 60%, but when I look at town it is only 30%, with a proportional drop in FPS. For comparison, 2.0 runs at 100% GPU usage. [sIGPIC][/sIGPIC] Intel Core I7 4820K @4.3 GHz, Asus P9X79 motherboard, 16 GB RAM @ 933 MHz, NVidia GTX 1070 with 8 GB VRAM, Windows 10 Pro
adrianstealth Posted July 8, 2016 Posted July 8, 2016 be nice if we could just switch towns off, eg leave airports and run ways but an option to exclude the towns , they are not optimised enough to be there
SkateZilla Posted July 8, 2016 Posted July 8, 2016 (edited) ^ That Reply Made No Since. I'll Explain: 1. A Single Core Will Process a Serial (Single Threaded) Task Faster than Multiple Cores, (Ie Taking A Serial task and Spreading it across 4+ Cores), as every Calculation depends on another, and the others will have to wait to be processed until it receives the data from the other cores, latency will cause the thread to process slower. DCS's Engine is a Serial Process, It was designed to run on a Single Thread and have All the Information readily available to the sim at any given time in that single thread. 2. Disabling Other CPU Cores wont increase Performance, it will make it worse, Why? At Any Given Time, your PC is Processing 100s-1000s of Threads/Processes, If you Drop your CPU to 2 Cores, all of those Processes for Windows and other applications will be compacted to those 2 cores, taking CPU Processing time away from the DCS Process, the More cores you have, the more windows can spread those other tasks around. 3. DCS is not Required to be 4 Cores Minimum, There's nothing to "Fix" 4. The Issue is Draw Call Overhead on the CPU Process for DirectX, The Easiest way to Fix this, is to limit how many objects are asked to be drawn, Which is done by reducing the Visibility Range of objects etc etc. Edited July 8, 2016 by SkateZilla Windows 10 Pro, Ryzen 2700X @ 4.6Ghz, 32GB DDR4-3200 GSkill (F4-3200C16D-16GTZR x2), ASRock X470 Taichi Ultimate, XFX RX6800XT Merc 310 (RX-68XTALFD9) 3x ASUS VS248HP + Oculus HMD, Thrustmaster Warthog HOTAS + MFDs
wolle Posted July 9, 2016 Posted July 9, 2016 ^ That Reply Made No Since. 4. The Issue is Draw Call Overhead on the CPU Process for DirectX, The Easiest way to Fix this, is to limit how many objects are asked to be drawn, Which is done by reducing the Visibility Range of objects etc etc. Makes no difference. I reduced visible range from extreme to high, only a slight improvement, if any (the only thing I notice in a decrease in latency when I move the head around). As soon as I look at (even a nearby) town, FPS and GPU usage drops by a factor of 2 at least. [sIGPIC][/sIGPIC] Intel Core I7 4820K @4.3 GHz, Asus P9X79 motherboard, 16 GB RAM @ 933 MHz, NVidia GTX 1070 with 8 GB VRAM, Windows 10 Pro
wolle Posted July 9, 2016 Posted July 9, 2016 I'm sure ED is now aware of the issue. Wouldn't count on it. Several users have applauded 1.5.4 for an increase in FPS, so unless we keep this thread alive, ED will think everything is great. [sIGPIC][/sIGPIC] Intel Core I7 4820K @4.3 GHz, Asus P9X79 motherboard, 16 GB RAM @ 933 MHz, NVidia GTX 1070 with 8 GB VRAM, Windows 10 Pro
Goldsmack Posted July 9, 2016 Posted July 9, 2016 True. I also noticed SkateZilla made a post so atleaste the moderators skim through here but no Idea if/how that info is passed on. As for the FPS issue, I have no Idea whats causing it. Maybe its an Nvidia thing? Anyone using AMD? or have rolled back there drivers? Iv tried a clean driver install up to 368.69 restoring all my Nvidia profiles, DCS repair. Adjusting some setting like shadows helps "A BIT" but If it was working great before???. Nevada seems to run fine maybe 5fps less in some spots but nothing like the caucuses map. It almost looks like building draw dissidence is different in NTTR from Caucuses? since one of the last updates maybe a different way of doing it which would make sense seeing as how the Nevada strip is congested with buildings, but my FPS have gone UP over Vegas since the last update and I noticed the building draw distance has been reduced. I have tried dropping pre load radius but nothing....I'm stumped:doh: I have never seen this problem before. If posting some kind of log would help I'd be more then happy to share. Windows 10 Pro 64, I5 4690k @4.6GHz with CAPTIAN 240EX AOI, Samsung 850 EVO ,G Skill Ripjaws 16G RAM, Nvidia GTX 970 STRIX, MSI Z97 GAMING 5, WD Blue 1TB HDD, Seasonic M12 II EVO psu, Track IR 5, Pro Flight X-55 Rhino H.O.T.A.S
Fri13 Posted July 9, 2016 Posted July 9, 2016 Makes no difference. I reduced visible range from extreme to high, only a slight improvement, if any (the only thing I notice in a decrease in latency when I move the head around). As soon as I look at (even a nearby) town, FPS and GPU usage drops by a factor of 2 at least. That was the key why i found the RAM compression to be a problem as if it would be in DCS problem, the resource uses would go up and not down. And when you look towns a DCS will start cache data to RAM and then it will trigger windows to compress it. And it will take all resources from other processes like DCS, and slow them down. Why you see that DCS CPU and GPU consumption goes down and not up. It was clue that something else is eating the resources so DCS has been throttled down below requirements. As the resource managers didn't show any other process using more CPU or GPU but they were "idling". So Check the disk I/O. You can see how if it will jump up and down and being a bottleneck for CPU and GPU. DCS the days does better than it did with 1.2.x version when it comes to CPU and GPU consumption, but windows doesn't like it's RAM usage so much. And the odd part was that DCS wasn't recaching towns, it was just jittery. And that can as well be something where VRAM and RAM can be causing as bottleneck as VRAM is full (3GB and 4 GB here on me) and it gets cached to RAM, but OS want to compress it to virtual memory (paging file) even when RAM was half empty (i don't remember anymore does new windows show the disk cache in RAM in used or not.... so problem can have been that DCS caches all to disks and out gets filled to RAM, that then needs to fill DCS itself for second time, then VRAM gets cached to RAM as well, that runs out (32 GB in this case) and it gets heavy compression by windows and this makes cached towns in RAM unavailable for a time needed and CPU with GPU is throttled down as they don't have resources to consume and their consumption goes down as those are idling when memory compression is taking place and slow all down. But looks it doesn't help everyone. I had as well compression disabled to speed other low end computer to run DCS perfectly (16gb RAM, G750ti, 4-core AMD and HDD) like main computer as it suffered from compression, no matter what were the DCS setting (all low at minimal resolution or all full at max resolution). i7-8700k, 32GB 2666Mhz DDR4, 2x 2080S SLI 8GB, Oculus Rift S. i7-8700k, 16GB 2666Mhz DDR4, 1080Ti 11GB, 27" 4K, 65" HDR 4K.
Goldsmack Posted July 9, 2016 Posted July 9, 2016 No mater what happens in life...know that theirs always someone worse off then you. Software update causes $286 million Japanese satellite to break apart in orbit http://www.geek.com/science/software-update-causes-286-million-japanese-satellite-to-break-apart-in-orbit-1654675/ Windows 10 Pro 64, I5 4690k @4.6GHz with CAPTIAN 240EX AOI, Samsung 850 EVO ,G Skill Ripjaws 16G RAM, Nvidia GTX 970 STRIX, MSI Z97 GAMING 5, WD Blue 1TB HDD, Seasonic M12 II EVO psu, Track IR 5, Pro Flight X-55 Rhino H.O.T.A.S
Goldsmack Posted July 9, 2016 Posted July 9, 2016 After some more testing in the Caucuses map with my standard test mission here are my results. I can confirm the GPU usage drops significantly when looking at towns/building areas. Using ASUS GPU Tweak 2 for monitoring GPU .............Average ....... Looking At Towns VRAM %: 50% down to 10% GPU Usage %: 85% down to 45% Total Power %: 100% down to 60% Average FPS : 170 FPS down to 39 FPS (No Mirrors, F-15E) CPU and RAM average are the same even when looking at towns Windows 10 Pro 64, I5 4690k @4.6GHz with CAPTIAN 240EX AOI, Samsung 850 EVO ,G Skill Ripjaws 16G RAM, Nvidia GTX 970 STRIX, MSI Z97 GAMING 5, WD Blue 1TB HDD, Seasonic M12 II EVO psu, Track IR 5, Pro Flight X-55 Rhino H.O.T.A.S
wilbur81 Posted July 9, 2016 Posted July 9, 2016 After some more testing in the Caucuses map with my standard test mission here are my results. I can confirm the GPU usage drops significantly when looking at towns/building areas. Using ASUS GPU Tweak 2 for monitoring GPU .............Average ....... Looking At Towns VRAM %: 50% down to 10% GPU Usage %: 85% down to 45% Total Power %: 100% down to 60% Average FPS : 170 FPS down to 39 FPS (No Mirrors, F-15E) CPU and RAM average are the same even when looking at towns I'm getting significantly better FPS directly over Nevada (both at low and high alt.) in 2.0 than I'm now getting with the new 1.5.4 update.... especially with mirrors on in the F-15. My settings in both are the same. i7 8700K @ Stock - Win10 64 - 32 RAM - RTX 3080 12gb OC - 55 inch 4k Display
wolle Posted July 9, 2016 Posted July 9, 2016 That was the key why i found the RAM compression to be a problem as if it would be in DCS problem, the resource uses would go up and not down. And when you look towns a DCS will start cache data to RAM and then it will trigger windows to compress it. And it will take all resources from other processes like DCS, and slow them down. Why you see that DCS CPU and GPU consumption goes down and not up. It was clue that something else is eating the resources so DCS has been throttled down below requirements. As the resource managers didn't show any other process using more CPU or GPU but they were "idling". So Check the disk I/O. You can see how if it will jump up and down and being a bottleneck for CPU and GPU. DCS the days does better than it did with 1.2.x version when it comes to CPU and GPU consumption, but windows doesn't like it's RAM usage so much. And the odd part was that DCS wasn't recaching towns, it was just jittery. And that can as well be something where VRAM and RAM can be causing as bottleneck as VRAM is full (3GB and 4 GB here on me) and it gets cached to RAM, but OS want to compress it to virtual memory (paging file) even when RAM was half empty (i don't remember anymore does new windows show the disk cache in RAM in used or not.... so problem can have been that DCS caches all to disks and out gets filled to RAM, that then needs to fill DCS itself for second time, then VRAM gets cached to RAM as well, that runs out (32 GB in this case) and it gets heavy compression by windows and this makes cached towns in RAM unavailable for a time needed and CPU with GPU is throttled down as they don't have resources to consume and their consumption goes down as those are idling when memory compression is taking place and slow all down. But looks it doesn't help everyone. I had as well compression disabled to speed other low end computer to run DCS perfectly (16gb RAM, G750ti, 4-core AMD and HDD) like main computer as it suffered from compression, no matter what were the DCS setting (all low at minimal resolution or all full at max resolution). Hi, Thanks for the explanations. I have a question: how do I check disk I/O (I have been looking at it using the performance monitor accessed via CTRL-ALT-DELETE (I have disk C SSD, disk D HD, disk F SSD and the highest usage ever is 6%). My RAM and VRAM are nowhere near full (I have been testing it with a single flight mission). So do you think the disk I/O problem could apply to me, probably not, right? [sIGPIC][/sIGPIC] Intel Core I7 4820K @4.3 GHz, Asus P9X79 motherboard, 16 GB RAM @ 933 MHz, NVidia GTX 1070 with 8 GB VRAM, Windows 10 Pro
Fri13 Posted July 9, 2016 Posted July 9, 2016 Hi, Thanks for the explanations. I have a question: how do I check disk I/O (I have been looking at it using the performance monitor accessed via CTRL-ALT-DELETE (I have disk C SSD, disk D HD, disk F SSD and the highest usage ever is 6%). My RAM and VRAM are nowhere near full (I have been testing it with a single flight mission). So do you think the disk I/O problem could apply to me, probably not, right? Basically check the physical LED on computer about disk operations. As NT operating system doesn't reveal the RAM compression I/O performance and you just need to try disabling the paging files for all drives, reboot and check DCS performance. If it doesn't help, enable them back and reboot. Worth the two boots time ;-) i7-8700k, 32GB 2666Mhz DDR4, 2x 2080S SLI 8GB, Oculus Rift S. i7-8700k, 16GB 2666Mhz DDR4, 1080Ti 11GB, 27" 4K, 65" HDR 4K.
Bluelight Posted July 9, 2016 Posted July 9, 2016 (edited) ^ That Reply Made No Since. I'll Explain: 1. A Single Core Will Process a Serial (Single Threaded) Task Faster than Multiple Cores, (Ie Taking A Serial task and Spreading it across 4+ Cores), as every Calculation depends on another, and the others will have to wait to be processed until it receives the data from the other cores, latency will cause the thread to process slower. DCS's Engine is a Serial Process, It was designed to run on a Single Thread and have All the Information readily available to the sim at any given time in that single thread. 2. Disabling Other CPU Cores wont increase Performance, it will make it worse, Why? At Any Given Time, your PC is Processing 100s-1000s of Threads/Processes, If you Drop your CPU to 2 Cores, all of those Processes for Windows and other applications will be compacted to those 2 cores, taking CPU Processing time away from the DCS Process, the More cores you have, the more windows can spread those other tasks around. 3. DCS is not Required to be 4 Cores Minimum, There's nothing to "Fix" 4. The Issue is Draw Call Overhead on the CPU Process for DirectX, The Easiest way to Fix this, is to limit how many objects are asked to be drawn, Which is done by reducing the Visibility Range of objects etc etc. For God sakes Skatezilla... where do I begin: 1. Yes this all makes sense but you are making excuses for POOR PERFORMANCE in game. This is a repetitive thing here that we, the paying customers, need to stop accepting. How many single core gaming PC's exist in 2016!? Probably none unless someone is nostalgic and likes that sort of thing. 4 cores is the standard for gaming now... I realize changing things to work differently is hard.. but what you are saying is "our engine is old and out of date and doesn't perform well with current CPUs etc, deal with it!" Not good enough! If you have an old car and you keep putting fresh paint jobs, accessories, new tires, and spinner rims on it but you never rebuild the engine... you still have a piece of crap car that you are trying to pass off as something more. You can never have a car, or game, that can run well in current times without rebuilding or upgrading the engine. 2. Yes, makes sense but still an excuse for why you must use a 4 core machine but the game won't run correctly with it. 3. Are you serious? No it is designed for a single core or 2 core machine but that isn't the standard for gaming anymore and this game is supposed to be VR compatible!! The standard for VR is 4 cores! That is the minimum... therefore, your game MUST perform correctly with 4 cores or atleast utilize the resources it is meant to efficiently, to get a high frame rate enough for VR. 4. Yes, this is great that you know what the problem is but turning the draw distance down DOES NOT fix it. I have my visibility down to low and my tree draw distance on the minimum and yet I can look to the left of my heli in one mission and see over 2000 objects being loaded on screen with over 270,000 triangles. The GPU actually slows down from a whopping 40% to 30% sometimes and goes from boost clock to normal... all the while the fps drops like crazy because nothing is being utilized to help maintain the fps. I have invited you to my thread about this before and thusfar you have not stopped by to help there with the specifics and watch the videos with the numbers showing what I am talking about... proving that draw distance doesn't fix the problem. If there was a way to select how many objects were drawn per town or trees, or more control over it.. it may help but these are all bandaids for the problem of this engine not utilizing currently hardware in a way thats acceptable for VR. Again, please stop by this thread so you can show the irritated people there that someone is paying attention, like you did here, and lets discuss the specifics of this further. http://forums.eagle.ru/showthread.php?t=169738 Edited July 10, 2016 by Bluelight 1 HTC Vive, Saitek X52 Pro, i7-950 Overclocked to 4ghz, Noctua NH-D14 cooler, ASRock x58 Extreme MB, EVGA GTX 970 FTW, 24GB G Skill Sniper DDR3 1600 RAM, EVGA 650-GQ (650 watts) PSU, Windows 10 Home
Recommended Posts