I wouldn't really trust hwmonitor IMO. I use hwinfo all the time. It shows a lot more information of your pc.. You can configure it a lot more too. For example I can see how much power my gpu.... I hit around 267watts with my 970 as I bios modded it and overclocked it like crazy. It will peak at 290watts at full load. I'm kinda surprised how well my games run at 4k with "nvidia dsr" on a 21:9 screen. Nearly all of them run on max settings without much of a issue except for using double the power compared to stock lol. If I haven't overclocked/overvolted/unlocked the power, I wouldn't have no where close to the playable frame-rates with an 970 at "4k" lol. I am surprised I haven't seen many people modding their 970s.. A lot of performance to be unlocked with these cards. I lock my fps at 100, as I don't need 100+fps for my games.
Finally giving my PC the reset zap, had enough of applications crashing, wifi not working, crazy lag on some things, and just other madness. Like apps failing to u-install.
@SixSixSevenSeven This is with i7-6700 and gtx 1080, CPU 5 is limiting, that is one where graphics were running (DX11) on this run, you can't see it hitting 100% at all, yet it really is, very fast spiking kind of load which causes delay and GPU waits for CPU: 43% GPU load, 44% GPU power usage SSAO on, dynamic reflections 60m and lowest faces per update, high details. I'm quite curious if you can keep up fps at 60 with 3 or 4 cars there, which should not be challenging at all in terms of physics or in terms of GPU power. Settings used: If it would not be single core limited, there would be 60fps.
Youve introduced an additional bottleneck artificially through running windowed. Windowed mode causes the GPU to render a framebuffer, only to then perform a CPU based copy of that framebuffer into a drawable bitmap on a window, which is then blitted to the display in a second later operation, introduce large performance slowdowns. Literally the difference between 80 and 130+ fps for many games.
You can see clock on screenshot, non K model, I did set XMP on at BIOS and don't touch anything else. 161MBeams/s over 40MBeams/s with 1 car, so for non K this should be pretty fast as I have seen worse from 6700K. On that scene setup there just is so many drawcalls that DX11 murders almost any CPU, so it really does not matter much which GPU you have when doing that test. 1050Ti can do around 35-38% of gtx1080 so I think if you do same test setup you should have close to same framerate with your 1060, maybe bit faster if your CPU can do more than 40MBeams/s with single car in banana bench, I have no idea how your CPU fares on banana bench really. I did not even have additional lights running, which would murder CPU even more, thanks to DX11 running on single thread, which has to process stuff for GPU, shadows eat CPU because how they work and add lights, add more shadows, less room for more cars etc. However greatest misconception here seems to be that not seeing 100% single core usage would not be single core limited, while it can perfectly be, it is enough if your single core is 5ms at 100%, it does not show up, but is hitting fps already. It just is how thing works. @SixSixSevenSeven here is fullscreen version of that re-created so it is bit different, but still you can see same effect, I'm very curious if in your system there is no this kind of effect?: --- Post updated --- I haven't observed such effect with my setup, it is pretty much same, windowed, borderless, full screen, but windowed is using slightly smaller resolution.
But is that GPU or CPU that needs to work harder? I don't know too much of programmer side of this, but I know when drawcalls are incresed CPU needs to work harder on this game and that work happens on single core because of DX11, but for me it appears to be that windowed mode is not causing impact to performance, which could mean that it is GPU that has to work harder and when I have over 50% reserve there, such does not cause any impact on my system. Almost 6000 drawcalls, this CPU can do about 4000, varies a bit, but around there is maximum, that is DX11 and also how shadows use drawcalls I guess, JC3 with 6000 drawcalls is not yet hitting single core limit like BeamNG is, but reasons for me are unknown, but I know BeamNG is very single core limited when using higher graphics currently, which means Threadripper guys are not getting performance they hope for. If someone has Beamng running without hitting single core limits like this, I would like to know how because I want to run Beam so well too, with 8086k it is only 2 more cars really, despite it has twice the multicore power. So that is why I bring this up, I'm curious if there would be something there that could affect this, how this test works out on your system?
It is CPU that needs to work harder for windowed. Beamng issue isn't dx11. Never has been. It's crap render code. Nothing more to it than that
I wonder why there is no really much difference between windowed and full screen for me in CPU limited situation then, maybe that is not causing higher load on same CPU core that is gfx limiting in example situation? Whatever it is, it is causing single core load being the limit at least on this end, crap render code causing heavier DX11 CPU load than needed? All I can tell is that fastest single core performance works best for current version of BeamNG and that BeamNG does not like from high number of textures at all, performance increases hugely with joined UV maps and combined textures Keeping SSAO off and dynamic reflections off reduces drawcalls a lot and allows much more vehicles to be used in example situation, so until they get Vulkan version out, with those settings one can improve performance a lot, but I'm always interested to hear if there is way to get around this limit, so that I could actually get all cores used for fully while keeping 60fps, I doubt that being possible with current version though.
ceasing this massive argument over BeamNG performance, i bought a monitor for £40, 1080p and it's a VA panel, and wow IPS panels are wayy over-rated. The VA's increased contrast is really good. Played FH3 in the dark and wow it's like i should get popcorn Also 1920x1080 is such a massive upgrade over 1280x1024 productivity wise. It's soooo nice. I can have a 720p video on, and facebook or whatever vertically optimized site on the side.
Surprising that you had such resolution so long. I used to have 19" flat panel with 27" and before that two CRT screens, I still haven't used to single screen, despite it is years since I had two. While I can have two HDMI sources on my current monitor, it takes over 15 seconds to this cheap monitor to switch between inputs, it takes some blinded button clicking to actually be able to keep changed input as confirmation box might display for 1 second or so after monitor manages to do a switch. So buy another and you are doomed to have large table space etc. forever, there really is no going back from dual monitor setup.
I had a 1920x1200 monitor (HP ZR24w-professional IPS monitor, really good colours etc. but nothing fancy and was CCFL+scaler didn't work on 1080p (silly really) ), upgraded to a 1920x1080 144Hz monitor, but that's by my Logitech G27 wheel, my main monitor would still be the old Sony monitor. I had an Acer monitorbefore that (cheap one from maybe 2007), but was VGA and shite, the Sony monitor had DVI-D so no fuzzy text (yay!). 1280x1024 was manageable tbh. But single tasking only. And the minute I tried video editing on the 1080p screen i have now, i have never been so happy. It's true, video editing on a 4:3 (or 5:4) screen will drive you up the wall! Fun fact, the sony that i had appeared in an early 2004 PCWorld article. https://www.pcworld.com/article/114576/article.html It had a shocking response time (literally shocking), where 30FPS video was like 60FPS since the motion blur it induced made it smooth. It overclocked to 78Hz, but that disabled the on screen display (it just disappeared until you went back to 60Hz).
Just saying. In the ultra low voltage lines. Most of the intel CPUs of risk core count perform the same regardless of clock speed. Artificial throttling at a TDP limit yo
Yeah, 1080p is a huge upgrade over 1280x1024.. I have a ultra wide LG monitor 25" I got for around $200 from Canada computers. Its a great monitor. I overclocked it to 75hz, and enabled NVIDIA DSR -its stunning.. If you have a powerful enough video card you might be able to try 1440p using NVIDIA DSR, or an AMD equivalent. I'm running my games at 3840x1620 at max graphics settings and its beautiful. I'm not going back to 1080p for a while haha. It sure does push my 970 to the max though. It still pushes 40-60fps even at max settings. Some games will go above 100fps. My gpu is not stock in any way, its overclocked like crazy.
This is very much true, chips might be able to do better, but are held back. For example disabling 2 cores of 6 core CPU it is possible to gain performance, thing puts out same amount of wattage and heat as with 6 core, but computes lot more per core. Intel's heat issues are quite big with 6 cores, TIM is not transferring heat fast enough, even though it is said that 8086K got better TIM used in workstation CPUs, it is still poor, cooler really does not matter much when heat is not transferring from chip to heat spreader. With 6 cores turbo drops off really fast, having only 4 cores active turbo stays up and then there actually is some good performance, but 6700K and 8086K stock are not that different in Bananabench with 1 car. For beamNG 8 threads vs 12 threads is also not so big difference, much better to have faster per core speed while avoiding hitting TDP limits, so for most I would think 8th series with only 4 cores active would be better in BeamNG, that bigger cache does help a bit. Now with 6700 non K, there is very low TDP, 65W I guess it was, I'm hitting 74-65W at max with GPU installed, without GPU it is hitting 64-65W, it does not come issue with the BeamNG most of the time though as BeamNG with single vehicle is using something less than 30W, doing 3 vehicles etc it might get even up to 50W on CPU heavy map like WCUSA, but it is still not very bad, even turbo stays up quite well, hitting power limits is easy though. Coming AMD offerings are something very interesting at least on paper, hopefully those would at least pave way out of these limiters of performance, chips would do nicely, if thermal etc. limits would allow them to. 8 core intel chips are something that probably are much worse than 6 core ones, even though they say it would have soldered heat spreaders, haven't seen too much from those though. One warning though, at least 8th gen motherboards are much worse than 6th gen, had to RMA mine, will take weeks from them to deliver new one. For the same money that one could buy quality Z170 board when they were new, it seems one can get only rubbish level Z370 boards. So if anyone is going to get 8th gen, spend a lot to motherboard, really a lot something like 200 or more seems to give similar level of quality as one could get with 150 for 6th gen. --- Post updated --- DSR can be nice, but it works only when full screen and I don't usually ever use full screen as for me there is no performance difference really. It is bit odd though, some textures remain blurry mess, some that without it look even worse, might look much better.
It's a far bigger issue in their low voltage ultrabook parts where most won't hit base clock on all 4 cores let alone boost for more than a 20 second burst. They've got this TDP(down), TDP, TDP(up) stuff going on that hear throttles the max power consumption of the CPU even when within thermal limits
So i recently played the new forza it is awesome and my pc can take max settings mostly but sometimes i have a bit of freezing but not always at the same spots and not only in forza i also sometimes have problems with my second screen turning black i had it overclocked to 4.2 but i already got it back to stock clocks and it does happen less but still sometimes
