Aces High Bulletin Board

General Forums => Hardware and Software => Topic started by: atlau on December 22, 2018, 11:46:08 PM

Title: CPU usage 100% = bottleneck?
Post by: atlau on December 22, 2018, 11:46:08 PM
I looked at my CPU usage and it appears that when playing AH one of the cores is pegged at 100% usage while the GPU is not hitting its performance limit according to GPU-Z. The other  Does that mean the CPU is limiting performance? (i5-6600).

thanks in advance
Title: Re: CPU usage 100% = bottleneck?
Post by: Skuzzy on December 23, 2018, 07:27:28 AM
That is a fair assumption, but not always true.  It depends on how the software is measuring the load.
Title: Re: CPU usage 100% = bottleneck?
Post by: atlau on December 23, 2018, 08:42:47 AM
I was simply using the task manager/resource manager
  Do you have a different program that youd recommend?

Also if i upgrade to a new cpu, is there a good reason to get one with more than 6-8 cores for AH?
Title: Re: CPU usage 100% = bottleneck?
Post by: Bizman on December 23, 2018, 08:57:46 AM
HWmonitor tells the current, max and min values: https://www.cpuid.com/softwares/hwmonitor.html (https://www.cpuid.com/softwares/hwmonitor.html)
Title: Re: CPU usage 100% = bottleneck?
Post by: atlau on December 23, 2018, 09:07:18 AM
Tried that one too. It does max to 100% for core 0 so i assume a stronger processor might be beneficial. Just not sure if price/benefit ratio is worthwhile!
Title: Re: CPU usage 100% = bottleneck?
Post by: Bizman on December 23, 2018, 09:44:55 AM
For AH I have understood that anything past 3 GHz should be golden.

For the record I just made a short test in the Mission arena. Both of the cores of my Core2Duo E8500 overclocked to 3.8 GHz reached the top. #0 started to climb faster, but when it topped #1 was also at 98%. According to a couple of benchmarks yours should be over twice as fast in single core operations and even faster in multi-core tasks. Yet I have no issues to feed the GTX970.

That said, even with the latest flagship processor you might still see the 100% usage. Your processor has 250% the performance of mine, an i7-9600k is only ~30% faster than yours. I wonder if that little extra power would make any difference in AH.
Title: Re: CPU usage 100% = bottleneck?
Post by: atlau on December 23, 2018, 09:54:14 AM
Yep i dont think an upgrade at this point makes sense financially for such a small relative gain.
Title: Re: CPU usage 100% = bottleneck?
Post by: DaddyAce on December 23, 2018, 02:15:09 PM
I use MSI Afterburner to monitor hardware usage.  I also use it to set a more aggressive cooling curve for my GPU fans.  I initially used and i3 6100 cpu in this puter, but discovered that there were times when it seemed to bottleneck my puter, like when close to the ground or when I was in a big FSO fight.  I upgraded the cpu to an i5 7600K a while back and have not been noticing any issues since.  I've been happy enough with the i5 - 3GB 1060 combo running my Rift, and have felt no real need to upgrade yet.  I'll try to remember to to review the Afterburner output when I fly AH next and report back.
Title: Re: CPU usage 100% = bottleneck?
Post by: atlau on December 23, 2018, 02:28:12 PM
Id be tempted to upgrade cpu. But id have to replace the motherboard too along with a cooler for a k processor and dont have energy to do that :)
Title: Re: CPU usage 100% = bottleneck?
Post by: DaddyAce on December 23, 2018, 03:09:57 PM
I went with the K version in part because it was on sale at the time ($199.89), and it also gives significantly better performance than the non-K version without overclocking, and I think it was also less $ at the time.    You can see a comparison here:  https://cpu.userbenchmark.com/Compare/Intel-Core-i5-7600K-vs-Intel-Core-i5-7600/3885vs3890

My Mobo is an MSI B-150 Gaming M3, and does not support overclocking.   But it does support all kinds of USB connections without having to go to an external hub.     :cheers:
Title: Re: CPU usage 100% = bottleneck?
Post by: DaddyAce on December 23, 2018, 06:48:53 PM
Took a quick look at my graphs after flying a bit.  Frame rates on my Rift were generally between 45 & 90.  Both CPU 1 and GPU usage were up and down and spiked often approaching 100%. but generally were not that high, although GPU spikes closely approached 100% more often.    My conclusion is this cpu/gpu combo is balanced about right for AH3 running the Rift, at least for the in-game conditions I encountered earlier.
Title: Re: CPU usage 100% = bottleneck?
Post by: atlau on December 23, 2018, 07:07:24 PM
Do you dial back thr graphics settings to run smoothly?

Im.guessing my.slowdowns near busy fields with lots of ack and objects being generated is being caused by my cpu. When flying away from any action at higher alt im getting 120+ fps

How much ram do you have? Ive got 8 and as far as i can tell its never fully used.
Title: Re: CPU usage 100% = bottleneck?
Post by: DaddyAce on December 23, 2018, 10:20:45 PM
I did not dial back the graphics.  I recently noticed that I had the option to turn on reflections and shadows, and I enjoy the more immersive feeling of having those, and especially enjoy the reflections on the canopy, and off the wings, so if anything I goosed the graphics settings up from how they were set more than year ago.  My Rift was a surprise gift from one of my sons: it was unplanned for me.  I initially had mixed feelings because I was holding off waiting for the technology to mature more, and had committed to an upgrade without planning to add VR near-term.   I expected that I would need more CPU & GPU to run the Rift well.  But as it played out my system seems to run it fine, for my taste.  I have 8 GB 2400 DDR 4.  Looking at my Afterburner output it looks to me like I'm using about half of that while playing AH3, and about 2/3 of my 3 GB VRAM.
Title: Re: CPU usage 100% = bottleneck?
Post by: atlau on December 23, 2018, 10:58:39 PM
been playing with the graphical settings trying to figure out if certain ones impact the CPU more than GPU. Like detail range perhaps?
Title: Re: CPU usage 100% = bottleneck?
Post by: DaddyAce on December 24, 2018, 02:10:31 PM
been playing with the graphical settings trying to figure out if certain ones impact the CPU more than GPU. Like detail range perhaps?

Sorry, don't have a sense for that atlau.

I did go in and check my current settings:

I have Pilot mesh & Anti-aliasing both off, all else is on.

Environment mapping is 0.  I may try to bump this one up.

Sliders:
Object Detail  is in  the middle.
Terrain detail is set to max.
Both  ground and tree detail is set about 20% up from minimum.

My sense again is that my system is pretty well-balanced, but it appears I need more processing power for full graphics detail, best guess is that a better GPU would help the most.   Bottom line though is that I'm happy with what I have and am not sure it would be worth the $ to upgrade at this point, or that I'd get a significant gain in visual quality, especially given current VR technology.  If I were to upgrade I think I'd go with a 1070 & perhaps a more powerful CPU, but suspect I'd get the most from the GPU upgrade.
Title: Re: CPU usage 100% = bottleneck?
Post by: Bizman on December 25, 2018, 01:59:05 AM
Environment mapping is the most effective frame rate killer.
Title: Re: CPU usage 100% = bottleneck?
Post by: atlau on December 25, 2018, 08:38:37 AM
What exactly does environment mapping do? I see no difference visually when i change that setting
Title: Re: CPU usage 100% = bottleneck?
Post by: Bizman on December 25, 2018, 09:15:26 AM
For what I've understood it mirrors the surrounding of shiny objects like a bright aluminium P51. The higher the value the more real time effect you get.

As you say, the difference may be hard to see since you rarely just stare at a mirror shiny plane closely enough to see any details. In flight you mostly see it in the cockpit gauge glasses and windows, a lower value mirroring the grass beside the runway at the very worst. That negative feature has maybe been fixed since I don't remember having seen it in a long time.
Title: Re: CPU usage 100% = bottleneck?
Post by: DaddyAce on December 26, 2018, 08:11:25 PM
I just tried adjusting up EM while climbing out in a jug with a shiny skin.  The effect was pretty striking on the canopy and the wings, and I realized it was something I liked when I was using a monitor (pre-VR), and I like the effect.   The lower settings were not nearly as satisfying as having it maxed out;  I think my preference would be all or none.  The only trouble, as you pointed out Bizman, was my framerates dropped to slow enough to be distracting.    Now I've got thoughts of upgrading my GPU, and am thinking hard about a 1070 or 10670 ti...........
Title: Re: CPU usage 100% = bottleneck?
Post by: Bizman on December 27, 2018, 01:24:45 AM
DaddyAce, as EM is the most demanding single adjustment in the game (in any game) you'd need the best video card you can afford. At one point Skuzzy said that no video card on the market could use it on full without a punishment but that was before the 10## series. However, if you're planning to get a better monitor i.e. 144 Hz or 1440p or both in the near future, even the 1070Ti might be too weak for full EM.
Title: Re: CPU usage 100% = bottleneck?
Post by: DaddyAce on December 27, 2018, 10:51:31 AM
DaddyAce, as EM is the most demanding single adjustment in the game (in any game) you'd need the best video card you can afford. At one point Skuzzy said that no video card on the market could use it on full without a punishment but that was before the 10## series. However, if you're planning to get a better monitor i.e. 144 Hz or 1440p or both in the near future, even the 1070Ti might be too weak for full EM.

Yes, you're now the third person to suggest the 1080 in the last half-day.  The prices seem really high right now.....maybe missed all the recent sales already.  Although a quick look it seems like a RTX 2070 may be a better deal....will likely do some more research, and likely waiting before I jump in to upgrade my GPU....but I think I have the bug now.   Thank you for the input Bizman.   :salute
Title: Re: CPU usage 100% = bottleneck?
Post by: atlau on December 27, 2018, 06:45:17 PM
Wait for the next ebay 10% sale and 2070s will be about 450. Still not cheap. But maybe more reasonable
Title: Re: CPU usage 100% = bottleneck?
Post by: DaddyAce on December 28, 2018, 05:49:50 PM
....caught a eBay 15% off sale today, and sprung for a used    Gigabyte AORUS GeForce GTX 1080 for $340.   Hopefully it'll last a while......      :airplane:
Title: Re: CPU usage 100% = bottleneck?
Post by: atlau on December 31, 2018, 10:29:48 AM
So anyway back on target...

Sometimes my framerate drops but when I check my GPU usage, CPU(including individual cores), or RAM usage none of them are maxed out. Trying to figure out the culprit and if it's something I need to upgrade.
Title: Re: CPU usage 100% = bottleneck?
Post by: DaddyAce on December 31, 2018, 10:34:48 AM
So anyway back on target...

Sometimes my framerate drops but when I check my GPU usage, CPU(including individual cores), or RAM usage none of them are maxed out. Trying to figure out the culprit and if it's something I need to upgrade.

Sorry atlau, no good ideas.  Hopefully some of the much more knowledgeable guys can help.
Title: Re: CPU usage 100% = bottleneck?
Post by: Skuzzy on December 31, 2018, 11:04:20 AM
There are, literally, hundreds of reasons why frame rates will fluctuate.  Load measurements are not effective in catching most of them as resource related issues can happen even when the CPU/GPU are at 10% load.

Measuring resource usage is almost impossible without some profiling tools and a version of the executable built to support them.  Even then it is difficult without those same debugging aids being built into the video card driver, audio device driver, network interface driver, and input device driver.

All those devices are constantly grabbing and releasing resources.
Title: Re: CPU usage 100% = bottleneck?
Post by: Shuffler on December 31, 2018, 12:18:49 PM
My frame rates stay very steady. I use a system that takes a given rate over a given time and averages it all almost instantaneously. It works on all machines.

 I call it "Frozen Margaritas"
Title: Re: CPU usage 100% = bottleneck?
Post by: 68falcon on December 31, 2018, 03:45:53 PM
 :rofl :rofl :rofl
Title: Re: CPU usage 100% = bottleneck?
Post by: atlau on December 31, 2018, 03:59:55 PM
There are, literally, hundreds of reasons why frame rates will fluctuate.  Load measurements are not effective in catching most of them as resource related issues can happen even when the CPU/GPU are at 10% load.

Measuring resource usage is almost impossible without some profiling tools and a version of the executable built to support them.  Even then it is difficult without those same debugging aids being built into the video card driver, audio device driver, network interface driver, and input device driver.

All those devices are constantly grabbing and releasing resources.

Dang Skuzzy! I wish there was an EZ button! Its mostly frustrating because those darn clouds will cut it form 90 to 45 hz and it feels choppy when it does so!
Title: Re: CPU usage 100% = bottleneck?
Post by: Brooke on January 04, 2019, 04:55:55 AM
Your setup is probably totally fine for Aces High (assuming your screen isn't 4k, or you aren't running 2 monitors or VR -- and assuming you've got a graphics card that costs about $140).

What's your graphics card and what's your monitor resolution?

I run AH on a Pentium g4400 and a GTX 1050 graphics card on a 1080p monitor and get 60 fps (including in heavy clouds) as long as I turn off reflections and shadows and set environment slider to none.  It's a $275 computer system plus a $140 graphics card.

Look at CPU usage without AH running and make sure CPU usage is usually 0-1% with only some blips higher.  If not, you've got some program running that is hogging some CPU that you should hunt down and turn off.

Then try turning off reflections, shadows, and setting environment slider to "none".  If you still don't get a steady 60 fps, you can also try setting (from the starting AH screen) video settings, max texture size to 2048 or even 1024.
Title: Re: CPU usage 100% = bottleneck?
Post by: atlau on January 05, 2019, 09:39:25 AM
Sorry I should have added the this is in VR so essentially 2 monitors as well. I did a clean install of windows recently but haven't checked the background applications.

Single core usage on a i5 6600 often approaches 90-100%
Title: Re: CPU usage 100% = bottleneck?
Post by: Bizman on January 05, 2019, 12:49:51 PM
Did you also install the motherboard drivers?
Title: Re: CPU usage 100% = bottleneck?
Post by: atlau on January 05, 2019, 06:39:39 PM
Umm.. not sure. I'll check. I assumed windows installed the appropriate drivers for the devices.
Title: Re: CPU usage 100% = bottleneck?
Post by: Bizman on January 06, 2019, 02:14:10 AM
Umm.. not sure. I'll check. I assumed windows installed the appropriate drivers for the devices.
It doesn't. It installs generic drivers for limited functionality so you can download and install the real drivers from within Windows. Most likely there's a bunch of Windows drivers from 2006 in the System Devices of your Device Manager. The mothervoard manufacturer should have the latest drivers for your board on their website. -How many components in your system were even on the drawing board back in 2006?

Just to clarify, some of the drivers still will remain as those generic versions so don't try to update each and every device driver you find having an ancient date. That's perfectly OK, they're either for old devices like ISA slots, floppy drives etc. or they're for devices that can install their own drivers when plugged in like many USB devices including memory sticks and game controllers.
Title: Re: CPU usage 100% = bottleneck?
Post by: atlau on January 06, 2019, 08:24:31 AM
I downloaded the ASrock [mb manufacturer] software and updated the drivers thru their app but none of them really seemed relevant.

I did try to look at when my frame rates dropping (usually during heavy maneuvering with lots of other action around). I think vr just takes a heavier toll on the system. Maybe due to the required software running in the background?
Title: Re: CPU usage 100% = bottleneck?
Post by: Bizman on January 06, 2019, 09:12:59 AM
Depending on the motherboard maker, model and age for Win10 the most essential drivers are called Chipset drivers. You may also have them separated to several other Intel drivers and utilities. The Asrock app may do a decent job but since your system already is a few years old, AsRock may not have updated the drivers to the latest ones. Some manufacturers even leave the drivers to be updated by Windows, in which case you should go through the Device Manager item by item and manually check for updates for each (right click-update driver). If Windows can find a newer one it will update it. I've found most of the updateable drivers under the System Devices using this method but often also under other titles so starting from the first one is not overkill. It's not the way I'd like to update drivers but sometimes it's the only method available. 

I found your DxDiag output from last July. If you ran a new one now you could compare the driver data side by side. That might reveal some issues.

Basically your system should be capable of running VR pretty well, but you may need to decrease some settings. Environment mapping, post lighting, shadows are the worst resource hogs and least needed.
Title: Re: CPU usage 100% = bottleneck?
Post by: atlau on January 06, 2019, 10:28:47 PM
I've turned down the shadows and reflections as I found they made the game visuals worse.

I'm gonna try and see if there's maybe background programs that might be making the issue worse. I think the windows mixed reality does add a lot of overhead tho
Title: Re: CPU usage 100% = bottleneck?
Post by: Skuzzy on January 07, 2019, 07:12:02 AM
Many older systems will run Windows 10, but only have the generic Microsoft drivers available, which will impede the system from reaching its full performance potential.  In many cases stability will be affected as well, particularly with the audio drivers as the Microsoft supplied generic driver is not stable when asked to do multi-threading, which we do.
Title: Re: CPU usage 100% = bottleneck?
Post by: Bizman on January 07, 2019, 10:26:54 AM
Umm, Skuzzy, while your information is valid per se, I don't think his system is that old. The i5-6600 and the Z170 chipset were launched at the same time as the initial launch of Windows 10 so most likely Intel had already thoroughly studied the Preview Builds of the new OS to keep the customers happy. The drivers I found for the chipset in Microsoft Update Catalog were from 2016, March and August.
Title: Re: CPU usage 100% = bottleneck?
Post by: atlau on January 07, 2019, 09:21:36 PM
So I went thru the chipset drivers and they seem up to date. When I run various benchmarks such as Firestrike or Timespy the scores are in line with other similar systems.

I'm leaning towards the possibility that having to run the windows mixed reality + steam + AH3 VR is significantly more taxing than when I simply ran AH on my monitor even though the total pixel count isn't that much higher in VR.

Too bad Intel keeps changing the motherboard/cpu compatibility every other generation otherwise I'd consider upgrading to a 9600K.
Title: Re: CPU usage 100% = bottleneck?
Post by: FESS67 on January 07, 2019, 11:38:40 PM
Many years ago i had a system where the cpu spiked badly to the point that gaming was impossible.

Tracked it down to a usb joystick.  Powered usb hub solved it for me.  Is it worth adding usb hardware selectively in different usb sockets to see if there is a tipping point?
Title: Re: CPU usage 100% = bottleneck?
Post by: Bizman on January 08, 2019, 02:30:54 AM
FESS, in the Device Manager>USB controllers you can find Host Controllers whose Advanced tab will tell how much bus width each one is reserving. Under USB Root Hubs/Advanced you'll find which controllers you have plugged into that hub and how much power it uses.

For some more information see https://superuser.com/questions/690074/what-is-the-power-output-of-a-usb-port (https://superuser.com/questions/690074/what-is-the-power-output-of-a-usb-port)
Title: Re: CPU usage 100% = bottleneck?
Post by: FESS67 on January 08, 2019, 05:20:57 AM
Yeah thanks I have a computer networking degree I was offering an avenue for Atlau to consider exploring not seeking assistance for myself.
Title: Re: CPU usage 100% = bottleneck?
Post by: Bizman on January 08, 2019, 06:03:36 AM
Heh, OK. The question at the end of your post fooled me, I didn't figure out it was meant as a suggestion.   :cheers:
Title: Re: CPU usage 100% = bottleneck?
Post by: atlau on January 08, 2019, 07:42:03 AM
Hmmm so I have something plugged into just about every USB output. I'll try to figure out what you guys are talking about there to investigate:) I did turn the power settings to max if that means anything
Title: Re: CPU usage 100% = bottleneck?
Post by: Skuzzy on January 08, 2019, 10:16:23 AM
If the power draw from the USB ports comes close to the limits of the built-in hubs, then a brown-out condition can occur where the USB hub sends millions of interrupts to the CPU causing a heavy spike in CPU usage.

In Windows 7, it was a critical condition which could cause the system to crash.  In Windows 8, and later, Microsoft instituted a hack/fix, of sorts, which shuts down the USB hubs ability to generate an interrupt and then they switch to polling mode.  It is less efficient and uses more CPU time, but it keeps the system from crashing.

To my knowledge, there has never been a motherboard which had full support for the power draw limits specified for the USB bus.


Oh Bizman, my early response about the older chip sets was more of an FYI thing that may or may not have anything to do with the system in question.
Title: Re: CPU usage 100% = bottleneck?
Post by: Bizman on January 08, 2019, 10:46:14 AM
Oh Bizman, my early response about the older chip sets was more of an FYI thing that may or may not have anything to do with the system in question.
Thought so.  :cheers:
Title: Re: CPU usage 100% = bottleneck?
Post by: FESS67 on January 08, 2019, 03:08:11 PM
Atlau, are you using any powered usb hubs or is everything straight into the PC?  Often a powered hub is useful.  I have an 8 port powered hub that handles my trackir, pedals, stick, throttle and some other peripherals leaving the pc ports for yhe vr sensors and headset.
Title: Re: CPU usage 100% = bottleneck?
Post by: atlau on January 08, 2019, 03:56:35 PM
Everything is plugged into my computer directly. Mouse, keyboard, printer, usb wifi, throttle, joystick, vr headset. Is that too many?
Title: Re: CPU usage 100% = bottleneck?
Post by: The Fugitive on January 08, 2019, 05:01:47 PM
Everything is plugged into my computer directly. Mouse, keyboard, printer, usb wifi, throttle, joystick, vr headset. Is that too many?

That could be. The power for all those comes from your power supply. If you dont have a robust one in your computer it may give you trouble. Powered USB hubs dont have the greatest power supplies, but it does take some of the load off of the computers power supply.

In my case, flight controls are all on one USB hub, peripherals on a second, and VR stuff on the computer.
Title: Re: CPU usage 100% = bottleneck?
Post by: Bizman on January 09, 2019, 01:49:36 AM
---
In my case, flight controls are all on one USB hub, peripherals on a second, and VR stuff on the computer.
That sounds like a perfect solution since a desktop usually has a few USB lines. Spreading the load to all of them and adding extra power seems more reasonable than using just one mega hub.
Title: Re: CPU usage 100% = bottleneck?
Post by: Skuzzy on January 09, 2019, 06:20:02 AM
That could be. The power for all those comes from your power supply. If you dont have a robust one in your computer it may give you trouble. Powered USB hubs dont have the greatest power supplies, but it does take some of the load off of the computers power supply.

In my case, flight controls are all on one USB hub, peripherals on a second, and VR stuff on the computer.

The amount of power which can be supplied to the USB peripheral is not regulated by the power supply of the computer.  It is regulated by how much current the USB chipset can allow.  While the specification for USB 1.x and 2.x allows for a maximum of 500mA, and 900mA for USB 3.x, the amount of current is usually far from that in the actual implementation.

The maximum per connected USB device was set at 100mA for USB 1.x and 2.x, and 150mA for USB 3.x.  In reality there have been many devices which violated that current draw.


A computer has X number of USB hubs in it, with Y number of USB ports attached to each hub.  No manufacturer, I am aware of, has ever put external labeling on the ports to tell you which hub they may be connected to leading to easily overloading any give hub.