300fps no reduce buffering vs display based cap reduce buffering

I can’t add you so add me. I might be able to help.

Apologies, have a full friends list almost all the time because of the small limit.

Hi, I have an 8700k @ 4.9ghz and RAM xmp’d to 3200 mhz as well, but my computer can barely stay at 240fps and drops during team fights. I use GTX 2080, do you know what could be the issue? Thanks!

Make sure your settings are on all low, not high/ultra, and that the render scale is at 100% or lower. I prefer frames over the high settings.

Background processes will also do it, make sure all other things are closed when trying to max frames.

Disabling HPET may also fix the issue- google it, and try it, it can’t hurt and is reversible.

Hey, this might be a bit of a dumb question, but I assume you have 16GB or ram running in dual channel?

I also want you to try something just for an experiment: Go into your bios and disable Hyperthreading. Since you got a six-core it will get maxed out by the game engine anyway.
Secondly, I want you to go into nvidia settings-> Manage 3D settings -> Global settings -> scroll down and disable ‘Threaded optimization’
I can’t link a picture of where to find it in here but it is easy to Google if you fail to do so.

Please try these two things independently from another. So first, try disabling Hyperthreading, look at the results. Then enable it again and try the nVidia Settings, look at the results.

I use a 6v6 Zarya bot match on Illios Well map to benchmark. But Quickplay should be fine, too.
It is useful to have the MSI Afterburner Overlay enabled so you can see your CPU/GPU saturation respectively, but the ingame counter is fine for starters. Also make sure to unlock your fps (slider at 300) while doing this.

Would be really interested to hear back from you and what results you got. These 2 options are really quick to change if anything unusual happens, so no worries. Just don’t queue for ranked while doing this :joy: Good luck.

Thanks for your reply, I already have everything set low and render scale is set to 75%. I made sure nothing is running in the background. My motherboard doesn’t have the option to disable HPET so I disabled HPET from device manager only. Right now I can manage to play at 230-240fps with occasionally drops to 200ish but 85% of the time it stays around 240 so I am ok with it. Just want to see if I can get any higher.

Hi, thanks for your reply, yes, I have 16gb dual channel rams. I will try disable hyperthreading tonight and test it. Can you explain what’s the difference between both experiments? I assume disable hyperthreading would make sure OW is utilize all 6 real cores instead of virtual ones? Not sure what the “treaded optimization” does though.

As far as testing, I used 6v6 Zarya on Lijiang tower and I was able to stay 280-300 using in game counters but if I used 6v6 Mei, then when there were ults throwing around the fps fluctuated from 200 to 300. These are the tests I did before, I will test again tonight after I change the settings as you suggested.

Another thing I found interesting was when I tested/played in QP or arcade, my fps was stable at much higher rate but if I played Rank then the fps fluctuated noticeably. I assume because there were more real team fights with different effects in Rank play and might also depended on maps?

1 Like

Hey,

thanks for responding!
You are right, for me it is really interesting to investigate how the game engine handles 6 real cores vs. having to ‘manage’ additional 6 threads. Before 6C/12HT or even 6C became mainstream this wasn’t really possible.

I noticed back when I had a 4C/8T Intel, disabling nVidia Threaded Optimization gave me a lot more stable (minimum) fps, because I was dealing with annoying fps drops back then, too. This wasn’t linked to anything graphical demanding but rather physics, yet it is hard to confirm. Leading to my next point and you already mentioned that ranked games kind of ‘feel’ different in terms of performance. It just might be a heightened attention because of additional ranked pressure but I noticed something similiar over the years. Btw disabling HPET always gave me horrible mini freezes, but I also can’t disable it in my BIOS ( screw you, Asus! :stuck_out_tongue: )

Maybe the engine handles/prioritizes certain assets differently in ranked mode. The cpu has to calculate things off-screen all the time, especially with unlocked fps the burden is huge. That’s also why a lot of OW streamers swear by their 2 PC setups.
The Mei case is interesting but makes sense if you look at my point above (physics) because it is a pretty big AOE ultimate compared to grav with a lot more interactions (when does sb get in the radius, when does he get out etc.) all this takes additional processing power.

Looking forward to see your results! Btw I forgot to add you should test HT off + Threaded Optimization off, too.

DO NOT disable threaded optimization; that isn’t hyperthreading in nvidia options, it’s simply allowing the game to use more than one core. It’s totally unrelated. Edit: You can disable threaded optimization globally and then enable it solely for Overwatch, which may also improve performance.

However, you can try reducing overclocks slightly or cleaning out PC also as options. With HPET, you need to make sure you run the cmd lines as well.

Then how come MSI Afterburner showed me the utilization of all cores in OW even when it was disabled? This is a common fix for Battlefield 1 as well.
If your assumption were true, why would you disable it globally then?

I did my test last night and along with some researching on what should be on and off. For other ppl, seems like toggling HT or threaded optimization have helped them. For me, I think turn off Hyperthreading was definitely the worst. I used 6v6 mei testing and got result of 240-300 fps. Having HT on with or without threaded optimization didn’t really affect my performance much as I was able to stay in the 270-300 range and 285+ for the most part.

So I left HT and threaded opt on and tuned down my gpu oc slightly and got 290+ stable fps with Mei testing. I thought it was great then I played one Rank game to test and got about 280+ fps. However, my fps starts to dip after first game. I thought might be the map but I played a couple more games and fps fluctuated too much so I had to limit my fps to about 245 and then it became more stable.

Like I said before, 240 fps is really all I need so I am okay with whatever I can hit now., but with my setup, I thought it could hit higher fps. My friend mentioned that might be b/c of my motherboard is limiting the performance since I bought the PC directly from bestbuy so they might used cheaper boards but idk.

One last thing, when I tried to do the Mei testing, I kept getting unexpected server errors so I could only test for like 2 mins each time. And to really get the accurate FPS for me is to do it in Rank since my fps only fluctuates when I play rank which is also weird.

1 Like

Are you using reduce buffering? Make sure you are toggling it if so, otherwise those fps problems only get worse.

Turning it off globally makes your background processes use less cores in general, so the main game you are playing with it on should be using more cores and the background processes will be trying to split load less.

Frankly, the nvidia CP is pretty bad and it needs a workover very bad. Inspector is slightly better.

Threaded optimization splits gpu load unto multiple cores instead of one. Modern games do this reguardless. Threaded optimization is for old pre-multi-core era games.

Disabling and enabling hyper threading really depends on how well a program can take advantage of the extra cores.

There is so much talk about hpet on vs hpet off that I’ve been unable to find good information on it. I use a program called windows system timer tool which sets the timer (idk which one(s)) to the fastest speed available so that it never changes. The system timer will commonly change speed which is known to cause microstutter. This program resolved my microstutter issues. There is also a utility called dpc latency checker to see if your timers are congested, but it can be difficult and time consuming to resolve all of the issues, but imo worth it.

I know a trick benchmarkers use to get an extra 5% out of their nvidia video cards performance. This is HIGHLY RISKY. Create a system restore point and make sure you know how to launch your computer in safe mode. Please read guides on how to do this and PLEASE create a system restore point and know how to launch your computer in safe mode. Use MSI utility v2 to switch your video card from IRQ (interrupt request) to MSI (message singal interrupt [not related to the company]). DO NOT SWITCH ANYTHING ELSE TO MSI. DO NOT SWITCH ANYTHING ELSE TO MSI. IF YOU DO NOT SEE ANYTHING CHECKED AND ALREADY USING MSI THEN DO NOT SWITCH YOUR VIDEO CARD TO MSI MODE as your motherboard may not support it.

(IRQ)Your gpu communicates to your cpu by sending your cpu an interrupt request but this is done on one core.
(MSI)Your gpu communicates to your cpu by modifying a location they decided to use in ram. This can alert any core.

Nvidia graphics cards are MSI compatible but default to IRQ to be compatible for virtual machines. Not all motherboard are MSI compatible.

This is known to reduce input latency and was noticeable when I switched. It was also hilarious when one benchmarker was 5% ahead of the best benchmarkers and they were scratching their heads.

EDIT: Also it is very common that some or all of your usb ports are running on the same IRQ channel as your video card. Tests have shown unstable mouse performance (fluctuating poll rates) which were resolved by using USB ports that were not on the same IRQ as the video card (test results showed stable 1000hz). Since windows 7 you have not been able to manually assign IRQ channels. This leaves 2 options, used the right ports or switch your video card to MSI mode(RISKY)

We have been doing these things, if you send me a friend invite to my alt tabinotochuu#11176 (Which should have open friend slots) I can give you more solid information on this and what to do with those tools, as we’ve come to some extensive research on it, though nothing absolute has been determined that works for every user, and placebo is common. I use the MSI tool myself as well; just make sure you don’t set SATA to MSI mode, or you will brick windows.

By the way; ENABLE HPET in BIOS. Disable in Windows device manager with the proper commands. It’s been a mistake all along to keep it disabled in the bios if you have the option to do so. If you don’t have it in the BIOS, then it is enabled by default, and windows timer tester will give you .5 ms; if it’s disabled in bios, it will give a strange .498 timer, instead of the true minimum, which causes frame issues.

Recent discovery and very relevant.

Hey, do you mind sharing your windows setup? i have a smiliar setup with i7 7700k @ 5ghz, 3200mhz ram and a GTX 1080 and i hit like 250 fps in normal games.
would love to do some improvments i havent already done :slight_smile:

I must rescind my previous post and say that enabling HPET in BIOS is really not that big of a deal, and only causes a slight OCD reaction to the .5 timer. What matters is the relevant edits in bcdedit.

Sorry for the late reply my friend, but I have a
8700k @ 4.9
1070 sometimes OC’d not really tho
CL15 3200mhz RAM at CR1; I can get this higher/lower CL but don’t see a purpose to
960 EVO
And a very stripped OS. Stripping the OS and making it fresh, as well as playing without too much going on, helped the most. I don’t really play anymore, but I still find interest in fixing these issues.

I agree, as well on the above, that threaded optimization (the option in nvidia) is widely misrepresented. People seem to think it means you won’t use more cores on your CPU. Basically saw no difference.

I must also bump this thread because reduced buffering is still a problem.

Blizzard, please figure something out about the reduce buffering bug. Collaborate with someone, Nvidia, whoever, about it. It’s one of the worst bugs OW has dealt with and has been clearly an issue for everyone using the feature. I worry about how many people are trying to play this game with sudden drops and horrible stutter because they leave it on and don’t realize their game just slowly gets worse.

Fix the game’s performance issues, UI stutter, and consistency and you might be able to fight Project A. Good luck.

[Edited by Moderator]
This thread has been inactive for some time. Please be mindful of resurrecting older threads. A new or more recent thread can be created to continue the discussion. Locking this thread.

2 Likes