Author Topic: CPU core count important?  (Read 4456 times)

Offline DaddyAce

  • Silver Member
  • ****
  • Posts: 1248
Re: CPU core count important?
« Reply #30 on: November 20, 2016, 11:08:43 PM »
Wow Pudgie, you're really into this, I like it.  You're into it much deeper than I am; but I am into trying to understand my system better, and I think this exercise this will help me decide what to upgrade if I want to increase performance after I spring for a nice 1440p monitor, hopefully next week.

I was using AB to record frame rate (mine is currently limited to 60 by my monitor), % GPU usage, % CPU usage (overall) , and % CPU1 usage, and % CPU 2 usage.  I've attached 2 snips of the graphs (I like the snipping tool, thank you for that tip).  Both are offline runs on the dueling terrain where I was often flying close to the trees, etc.  One graph was running the DX 9 version, and the other the DX 11 version.  I was not very careful to exactly reproduce my flights, so my summary impression is somewhat anecdotal.  Looking them over my initial impression is that  DX 11 makes better use of both CPU and GPU resources, as you indicated.  I seemed to get more frame rate drops below 60 in DX9.  Frame rate drops in both versions corresponded to the GPU being maxed out at 100% .... where FR drops below 60 you see the GPU curve flattened at 100%.    My CPU did not seem similarly maxed out and did not reflect the strong pattern seen in the GPU curve.  I now think more that my system is GPU limited (GTX 1060)  and not CPU limited by the i3 6100.  My son just emailed me that it's possible that the CPU may be more limiting when playing online because "....lots of information from the server needs to be processed in serial."  I'll have to record more online runs to verify the CPU is not limiting for online usage, although I suspect this is not the case.

Note, on both graphs the top line is Frame rate; I cut the label off on the DX11 graph....

 :salute
« Last Edit: November 20, 2016, 11:38:08 PM by DaddyAce »

Offline Pudgie

  • Silver Member
  • ****
  • Posts: 1280
Re: CPU core count important?
« Reply #31 on: November 21, 2016, 12:06:02 AM »
Thanks for posting those graphs DaddyAce!

Looking at both graphs of your I3 6100 2-core CPU's individual core usage I'm seeing the exact same CPU core usage pattern as I'm seeing on mine (I just have 4 more cores to add into the mix than you do) when AHIII is run under Dx9 vs Dx11.

Win 10 is utilizing both CPU cores 1 and 2 when AHIII is run under Dx9 but only CPU core 1 mainly when AHIII is run under Dx11 so your I3 6100 2-Core CPU is being more utilized under Dx9 vs Dx11, just as my I7 5820K 6-core CPU is.

This is pushing more towards something instructing the OS under Dx9 to use CPU Core1 and Core2 running AHIII then offloads the CPU cores when running AHIII under Dx11.....what this is that is doing this is beyond me at this time as the game client is exactly the same coding being used under both OS's (Win 7 SP1 on my box and Win 10 on yours) Dx API's which should be the same in both OS's and the game client isn't coded to request any CPU core affinity that I know of but the OS's appear to be doing this running AHIII under Dx9 vs Dx11.

This is good stuff as I love tinkering w\ computers (does suck though when it comes to playing AHIII using Dx11 on my box having to deal w\ the occasional screen freeze\pause)..........but not tinkering w\ them for a living though.

Curious as to your son's reasoning that the CPU may be limited due to serial processing........?

The only serial paths left that I know of would be the ancient PS\2 ports for input devices (mouse\keyboard) or the DMI\QPI link between the Intel chipset to CPU.......

A 2560 x 1440 res monitor will work your GPU some more vs the current 1920 x 1080 res unit you're currently using but that card you got will do better than you may think it will @ 2560 x 1440 res................

Anyway, thanks again!

 :salute
« Last Edit: November 21, 2016, 12:09:54 AM by Pudgie »
Win 10 Home 64, AMD Ryzen 9 3900X, MSI MPG X570 Gaming Plus, GSkill FlareX 32Gb DDR4 3200 4x8Gb, XFX Radeon RX 6900X 16Gb, Samsung 950 Pro 512Gb NVMe PCI-E SSD (boot), Samsung 850 Pro 128Gb SATA SSD (pagefile), Creative SoundBlaster X7 DAC-AMP, Intel LAN, SeaSonic PRIME Gold 850W, all CLWC'd

Offline Pudgie

  • Silver Member
  • ****
  • Posts: 1280
Re: CPU core count important?
« Reply #32 on: November 21, 2016, 12:16:36 AM »
Just now looked again at your graphs....................... ..

Boy I messed that up!

Your CPU usage shows to be opposite of what I'm seeing w\ mine (got confused w\ the graph line above CPU Core1 usage which is the total CPU usage graph line on the Dx9 graph you attached).

So your CPU is being utilized more under Dx11 than under Dx9........which is what I would expect to see.

Time for me to go to bed..........................

 :old:

 :salute
Win 10 Home 64, AMD Ryzen 9 3900X, MSI MPG X570 Gaming Plus, GSkill FlareX 32Gb DDR4 3200 4x8Gb, XFX Radeon RX 6900X 16Gb, Samsung 950 Pro 512Gb NVMe PCI-E SSD (boot), Samsung 850 Pro 128Gb SATA SSD (pagefile), Creative SoundBlaster X7 DAC-AMP, Intel LAN, SeaSonic PRIME Gold 850W, all CLWC'd

Offline DaddyAce

  • Silver Member
  • ****
  • Posts: 1248
Re: CPU core count important?
« Reply #33 on: November 21, 2016, 11:02:22 PM »
Okay Pudgie, I ran some more data, this time online.  I organized the graphs a bit better I think.  I get 5 CPU reads, a general CPU usage, then usage for CPU 1, 2, 3, & 4.  Since I only have 2 cores, perhaps one for each of 4  threads?  My take is that for my setup again, I get better performance and with GPU maxed less often, and also less load on the of CPU with DX11.  There is also a different pattern in how the threads (assuming that is what is being measured) are used between DX9 & DX11.

Offline Pudgie

  • Silver Member
  • ****
  • Posts: 1280
Re: CPU core count important?
« Reply #34 on: November 22, 2016, 10:38:49 AM »
Hi DaddyAce,

I've looked at your graphs and I'm seeing the same pattern of CPU usage % as I'm seeing w\ mine....................

It looks like when AHIII is being run under Dx11 the CPU cores are being under utilized much more than under Dx9.
This CPU core under utilization is what I'm thinking is contributing to the petite freezes, screen freezes and screen lockups being witnessed when running the game under Dx11 as this is definately indicating that there is a lot more CPU core wait time occurring as even Core1 isn't being fully utilized much less the rest of them. Intel CPU's operate better when the cores that are being used are being utilized as close to 100% load as possible as this keeps the Intel SpeedStep CPU clock\power saving control at bay (stops SpeedStep from underclocking\underpowering CPU cores due to being under utilized which can cause all kinds of mayhem when running games).

To see the CPU core usage dropping off while running AHIII Dx11 vs AHIII Dx9 can't be a good thing IMHO.........quite the opposite. I would rather see the CPU core loads go up even more................

The later Intel CPU's (which most of us are using now....starting at either Intel SB or IB....I definately noted this w\ my X79 series I7 4820K CPU which is IB-E) have this basic control built in at the hardware level now and it can't be disabled. The Enhanced Intel SpeedStep can be disabled in the BIOS\UEFI (I have this disabled on my box due to me discovering that this will prevent the Intel TurboCache from boosting the CPU clock speeds when playing games) but the CPU will still speedstep if the CPU core loads are below a certain threshold due to the hardware level SpeedStep (I have witnessed this going on w\ my I7 5820K CPU thru my Gigabyte SIV while my box was at desktop w\ the Enhanced SpeedStep disabled in the UEFI).

I've never noted any screen freezing\pausing w\ AHIII Dx11....either during Beta or Release....until after the VR coding was added to the game client for Dx11. This is NOT to say that there is an issue w\ the coding itself but I do wonder if the VR coding IS causing a byproduct effect to users running AHIII Dx11 w\o a VR headset attached that is causing the CPU core loads to drop off due to the client not seeing a VR headset.

To date I haven't heard\read posting of any VR user playing AHIII under Dx11 make any complaints of seeing screen freezes\pauses thru their headsets......only from users that aren't using a VR headset. This is what has ran up a red flag for me and got me to start checking into CPU\GPU usage and to discover what you're seeing yourself now.

Yesterday I had a clear graph of my CPU usage dropping on Core1 while playing AHIII under Dx11 and recording a screen freeze once the Core1 usage dropped below 34% w\ the other 5 cores usage below 10% then immediately after this the Core1 usage went back up to over 85% but I didn't snip it..........wished I had done so now.

IMHO it looks like for us users that aren't using a VR headset the only choice for us at this time is to use the Dx9 version of AHIII to get away from these petite freezes, screen freezes\pauses or lockups as it is clearly showing that under Dx9 the game client will load the CPU cores much more heavily which is IMHO a contributor to the game client running and not exhibiting these petite freezes, screen freezes\pauses and lockups.

I have pretty much exhausted any\every thing else on my box at my end and have eliminated all else as a contributor..........except the CPU core usage patterns exposed when running the game client under Dx9 vs Dx11.

I even went into my box's UEFI and reset the CPU Upgrade (Gigabyte's UEFI quick CPU OC) from I7 5820K 4.0 (sets the Intel Turbo Cache boost limit to 4.0 Ghz) to Auto (sets this back to default @ 3.5 Ghz) then ran AHIII again under both Dx9 and Dx11 and saw no change in CPU core usage patterns.

 :salute
« Last Edit: November 22, 2016, 10:48:34 AM by Pudgie »
Win 10 Home 64, AMD Ryzen 9 3900X, MSI MPG X570 Gaming Plus, GSkill FlareX 32Gb DDR4 3200 4x8Gb, XFX Radeon RX 6900X 16Gb, Samsung 950 Pro 512Gb NVMe PCI-E SSD (boot), Samsung 850 Pro 128Gb SATA SSD (pagefile), Creative SoundBlaster X7 DAC-AMP, Intel LAN, SeaSonic PRIME Gold 850W, all CLWC'd

Offline MADe

  • Silver Member
  • ****
  • Posts: 1117
Re: CPU core count important?
« Reply #35 on: November 23, 2016, 12:42:32 AM »
pudge the idea of under used cpu cores as issue does not jive.
ah3 was redone to use more gpu and less cpu..................to say the cpu is not maxed is to say its handling the load, yes..................

AH3 added DX11 years after the fact, dX9 is native so I think here is the rub. Why did not HT use PhysX, to much difference in coding .............

I have been testing slowly and well the micro freezes, stutters, CTD's............... I just get micro stutters, very limited, so I assume the all my new hardware is handling it all fine.
The internet will cause stuttering, pc not set up right to game will cause stutters..................... ...

this DX11 issue should be broken down to
-new machine, latest hardware
-old, quality machine with new VID CARD added
-middle of the road pc...
-laptop
-NVidia based
-AMD based

 :salute
ASROCK X99 Taichi, INTEL i7 6850@4.5GHz, GIGABYTE GTX 1070G1, Kingston HyperX 3000MHz DDR4, OCZ 256GB RD400, Seasonic 750W PSU, SONY BRAVIA 48W600B, Windows 10 Pro /64

Offline Skuzzy

  • Support Member
  • Administrator
  • *****
  • Posts: 31462
      • HiTech Creations Home Page
Re: CPU core count important?
« Reply #36 on: November 23, 2016, 06:06:14 AM »
<snip>
AH3 added DX11 years after the fact, dX9 is native so I think here is the rub. Why did not HT use PhysX, to much difference in coding .............<snip>

We are not big about supporting proprietary API's. Never have been.
Roy "Skuzzy" Neese
support@hitechcreations.com

Offline Pudgie

  • Silver Member
  • ****
  • Posts: 1280
Re: CPU core count important?
« Reply #37 on: November 23, 2016, 01:15:27 PM »
I just had a thought and so I set up and ran a test on my box to kinda simulate the Internet being in use then having MSI Afterburner open to record the CPU core usage while Firefox is running and also having Gigabyte's SIV open and running in the background to monitor the CPU core speed as well.

Then after this ran for a while I then took my mouse pointer and started moving it across the MSI Afterburner graph to move the white status line that runs vertically over the entire graph to monitor CPU core usage changes.

Here is a snippet of the results for your viewing provided below.
(Note: I do wish I knew the in's\out's of making a YouTube movie of this as it would have backed up what I'm about to post)

You can clearly see that when I started moving my mouse pointer in the fashion described above the CPU core usage moved to approximate the percentage of CPU core usage as recorded\shown in earlier posting w\ AHIII Dx11 running, not necessarily using the same CPU core affinity (the OS made these CPU core affinity choices to assign the threads to). But what you don't see in this snip that I witnessed is the Gigabyte SIV showing the CPU core speeds fluctuating from as low as 1200 MHz all the way up to the Turbo Boost speeds of 4003 MHz while I was moving the mouse pointer. Note this is operating w\ Intel Enhanced SpeedStep disabled in my box's UEFI (which does NOT disable the hardware level SpeedStep now built into all Intel consumer CPU's). Note also during this test the vid card is hardly being used (mostly due to Aero Glass which will use some 3D rendering but 2D rendering is being used the most) but the GPU is clocking at 300 MHz (lowest speed it will run) w\ mem clocked at 500 MHz (this is on this graph but can't all be seen on screen thru snip).

So I DO see this as a very potential issue when running AHIII Dx11 on my box showing this low of CPU utilization and seeing screen freeze\pause that in contrast I do NOT see happening running AHIII Dx9 on my box which is shown to more heavily load the CPU core utilization and NOT see ANY screen freeze\pause using the exact same game client software using the exact same client settings using the exact same driver settings on the exact same computer configuration using the exact same Internet routing system w\ the same variables upping\playing from the exact same area of the game map across both Dx API's.....all this should be far more of a usage variance issue as far as the GPU usage\utilization is concerned but IMHO NOT the CPU usage\utilization side of this.....shouldn't be this large of a variance in CPU usage between 2 MS Dx API's being used by the exact same OS w\ the exact same game client software running on it.

I'm open to any understanding of how MS Directx API's affect the CPU usage side of a game to this extent when being run that I don't already know about from the reading that I have already done to help me to see\understand the wide CPU usage variance that I'm seeing currently between a single game client being used under 2 different Dx API's that I've witnessed issues from 1 but not the other...........

As far as PLAYING the game goes, this is an easy choice to make....just stop using the Dx11 version and go w\ the Dx9 version of AHIII if a player doesn't have\is using a VR headset (also kinda what HTC has alluded to do in a roundabout way for the time being while they're looking into this issue). My box runs AHIII Dx9 damn well as it ALWAYS has, but this doesn't address the issue at hand as the game is currently coded as for optimum GPU game performance is concerned all testing on my end has shown that the AHIII Dx11 version IS the 1 to be using....and now we're at the crux of this issue, at least for me but I discern that I'm not the only 1 here......including the nice folks at HTC as well.

 :salute
Win 10 Home 64, AMD Ryzen 9 3900X, MSI MPG X570 Gaming Plus, GSkill FlareX 32Gb DDR4 3200 4x8Gb, XFX Radeon RX 6900X 16Gb, Samsung 950 Pro 512Gb NVMe PCI-E SSD (boot), Samsung 850 Pro 128Gb SATA SSD (pagefile), Creative SoundBlaster X7 DAC-AMP, Intel LAN, SeaSonic PRIME Gold 850W, all CLWC'd

Offline MADe

  • Silver Member
  • ****
  • Posts: 1117
Re: CPU core count important?
« Reply #38 on: November 23, 2016, 05:22:37 PM »
I have disabled speedstep, turboboost, I have cpu at 4ghz and it's locked in.
I would disable all the crap, then test.
You can also disable core parking in w7, do this as well!
« Last Edit: November 23, 2016, 05:24:39 PM by MADe »
ASROCK X99 Taichi, INTEL i7 6850@4.5GHz, GIGABYTE GTX 1070G1, Kingston HyperX 3000MHz DDR4, OCZ 256GB RD400, Seasonic 750W PSU, SONY BRAVIA 48W600B, Windows 10 Pro /64

Offline MADe

  • Silver Member
  • ****
  • Posts: 1117
Re: CPU core count important?
« Reply #39 on: November 23, 2016, 10:30:00 PM »
allowing the pc to make choices and operate on its own will inevitably bite you.
if its green its disabled
if its autofyed its disabled
any optimization is scewed for performance


all this new hardware is changing things as well, GTX 1000 series cards are incredible really.
ASROCK X99 Taichi, INTEL i7 6850@4.5GHz, GIGABYTE GTX 1070G1, Kingston HyperX 3000MHz DDR4, OCZ 256GB RD400, Seasonic 750W PSU, SONY BRAVIA 48W600B, Windows 10 Pro /64

Offline Pudgie

  • Silver Member
  • ****
  • Posts: 1280
Re: CPU core count important?
« Reply #40 on: November 24, 2016, 10:05:04 AM »
I have disabled speedstep, turboboost, I have cpu at 4ghz and it's locked in.
I would disable all the crap, then test.
You can also disable core parking in w7, do this as well!

I have already disabled the Enhanced Intel SpeedStep in the UEFI....been doing this since my X79 box when I discovered that it defeated Intel TurboBoost. I also have disabled Hyperthreading in the UEFI so this took care of Windows parking the logical CPU cores....Windows only sees the 6 physical CPU cores which Windows will not park so my CPU cores are running 1 thread at a time per CPU core now instead of trying to execute 2 threads per CPU core. Made these changes after doing some reading up on how Windows processes threads across CPU cores and on how Hyperthreading actually works.....realized that w\ a mulit-core CPU being used w\ a consumer level usage load that hyperthreading is pretty much useless if a CPU has a minimum of 2 physical cores but definately when a CPU has 4 or more physical cores on die.....unless a consumer is doing work that can actually saturate all the physical CPU cores, but at a hexacore\octacore or more level unless you're running server related types of usage I just can't see a need for hyperthreading at all. This also simplifies thread assignment\management for Windows which could net some slight CPU efficiency\performance gains as well under a consumer usage load w\ Hyperthreading disabled on a multi-core CPU.

Now I haven't disabled Intel TurboBoost as I kinda liked the advent of free CPU OC'ing since my understanding of how this works (doesn't suppose to kick in until after the CPU is running at the base clock speeds as set in the UEFI (w\ my I7 5820K this is 3.3 GHz) then if the CPU operating power\temp levels are sufficiently low enough at base clock speeds then the CPU will be OC'd to the turbo boost clock speeds as set in the UEFI (default is 3.5 GHz)...IOW's free OC'ing w\o the hassle of actually setting this up). I have set the CPU Upgrade setting in my Gigabyte UEFI (which are presets that set the upper TB clock limits) to I7 5820K 4.0 GHz and have checked this to actually work as advertised....once you get the Enhanced Intel SpeedStep out of the way (also have set up in Windows Power Management to High Performance plan which should shut down any Windows controlled power management schemes....which Enhanced Intel SpeedStep was 1 of them).

Might look into trying my box w\ TB disabled to see if there's a difference.

The only CPU power\clock control left that can't be disabled on Intel CPU's is the hardware level 1st gen Intel SpeedStep which is looking at CPU core load to determine when to reduce power and clock speeds w\ the intent of determining a desktop type CPU load usage vs heavy app usage (such as a game being run) to save power at very low CPU load usage. This is coded in at the actual CPU die level on chip now.....used to be accessible thru the BIOS back in the day but since the Pentium days this has been installed at the chip level.

This is why it's important for a game client software to load an Intel CPU (or an AMD CPU as they use the PowerNow! equivalent of Intel's SpeedStep) above a certain load threshold and maintain the CPU usage above it to prevent the CPU from trying to go into speed stepping due to the CPU load falling below the set load threshold level to trigger this.

This is what I'm trying to prove is what is triggering the screen pauses\freezes and whatnot that is going on w\ running AHIII Dx11. If I can get the CPU core load usage to climb high enough to stay above the 50% usage mark across at least 2 CPU cores as I'm seeing being done under AHIII Dx9 I believe this stuff will cease to occur. From all the data that I've gathered and the items that I've addressed\witnessed on my end this is the only logical conclusion that I can come to to explain what is causing these issues under the Dx11 version but not under the Dx9 version.

So far, everything I've tried\done I can't get the CPU core usage when running AHIII Dx11 on my box to stay up above the 40% level across any CPU core and every screen pause I've seen since I've set up MSI AB to record the individual CPU core usage on my CPU has shown the CPU core usage on any CPU core to be below 35% or lower at the time when they do occur. I believe that the CPU is trying to speed step the cores at this low of CPU load level causing the issues to appear but I can't prove any of this unless I can get the CPU core usage up consistently higher then monitor the game from there.

I believe all the other items, though important to find, were only really symptoms that the real cause amplified.......and why it's so hard to isolate.

I have learned thru my soon to be 34-yr career in the petroleum industry to not assume if possible but to test, document and verify even if you are sure of a source of an issue and to not discount any potential cause until it is proven w\ verifiable documentation to be\not be the cause, no matter how silly or mundane it may sound to others.

So this concept is what I'm bringing to the table to work this issue to try to be of help to HTC w\ this game because it's 1 of my favorite pastimes and I can't stand it when something isn't working as it should so I always try to eliminate my side of this 1st before going any further and at this time I've pretty much eliminated everything on my side that I'm aware of.....just got done to finally reroute a phone line yesterday to get the wife's all-in-one printer's fax modem off our dedicated ADSL line (found out why the filter was missing earlier after finding this from noticing my new Netgear Nighthawk modem\router was dropping out approx every 2-3 days....she had did it earlier due to a fax not going thru and didn't tell me about it.....now all will go thru the Centurylink installed industrial-grade ADSL filter they installed in the phone service box which gives me a fully dedicated, unhindered ADSL line ALL THE TIME now to the modem\router w\o needing to install any of the cheap ADSL line filters anymore......beautiful!).

 :salute
Win 10 Home 64, AMD Ryzen 9 3900X, MSI MPG X570 Gaming Plus, GSkill FlareX 32Gb DDR4 3200 4x8Gb, XFX Radeon RX 6900X 16Gb, Samsung 950 Pro 512Gb NVMe PCI-E SSD (boot), Samsung 850 Pro 128Gb SATA SSD (pagefile), Creative SoundBlaster X7 DAC-AMP, Intel LAN, SeaSonic PRIME Gold 850W, all CLWC'd

Offline Pudgie

  • Silver Member
  • ****
  • Posts: 1280
Re: CPU core count important?
« Reply #41 on: November 25, 2016, 08:50:19 AM »
I have disabled speedstep, turboboost, I have cpu at 4ghz and it's locked in.
I would disable all the crap, then test.
You can also disable core parking in w7, do this as well!

Hi MADe,

I got another hunch this morning and went into the UEFI in my box and looked around under the MIT then under Advanced CPU Core Settings....then I kept tabbing the selections down and I found a setting that I thought wasn't there......called CPU EIST Function (which is where this CPU power\frequency saving control can be enabled, disabled or allow the UEFI to determine it's usage at the hardware level).

So I quickly disabled this, saved and rebooted then checked the CPU thru Gigabyte SIV.......voila! My I7 5820K CPU is now locked into the Intel TurboBoost settings showing CPU clocks staying mostly at the 4003 MHz clocks but occasionally will dip to 3942 MHz so I believe she won't clock down lower than the base clock speed of 3300 MHZ going forward.

Gonna run some tests w\ AHIII Dx11 now.......................... .

FYI.......................... ........

 :salute
Win 10 Home 64, AMD Ryzen 9 3900X, MSI MPG X570 Gaming Plus, GSkill FlareX 32Gb DDR4 3200 4x8Gb, XFX Radeon RX 6900X 16Gb, Samsung 950 Pro 512Gb NVMe PCI-E SSD (boot), Samsung 850 Pro 128Gb SATA SSD (pagefile), Creative SoundBlaster X7 DAC-AMP, Intel LAN, SeaSonic PRIME Gold 850W, all CLWC'd

Offline MADe

  • Silver Member
  • ****
  • Posts: 1117
Re: CPU core count important?
« Reply #42 on: November 25, 2016, 10:48:21 PM »
CIEST, except for thermal protections I disable all. I believe here is where the cpu throttling is controlled. Basically any feature that was added for laptops batt longevity is useless in a desktop machine.

msconfig>boot>advanced, here you will find an option to tell machine to use all cpu cores for boot process. Now I believe this also reports to OS about # of cores and can screw with things like cpu-z and OS resource monitor...................... ......in case you were unaware :salute

I have had HT disabled because of my cpu OC. I had cores parked, so I disabled core parking on my W7 i7 machine. The new machine reports cores differently, You never see HT cores in resource monitor altho they are enabled. Right now I use a BCLK of 101 and a multiplier of 40. CPU does not vary except for basic fluctuations. Theres no throttling and all cores get same all time as far as freq and voltage, even at idle.

I also played with my displays settings, set for game-original, and switch to aspect scaling with gpu, from no scaling. Game did load slightly different......

I decided to test my new rig as well with DSR, gonna up my res and crank it up with v sync disabled. I have already set it up just have not got into online arena yet, offline it looks phenominal with steady frame rate at 120.
« Last Edit: November 25, 2016, 10:58:38 PM by MADe »
ASROCK X99 Taichi, INTEL i7 6850@4.5GHz, GIGABYTE GTX 1070G1, Kingston HyperX 3000MHz DDR4, OCZ 256GB RD400, Seasonic 750W PSU, SONY BRAVIA 48W600B, Windows 10 Pro /64