benchmarking Performance Impact Of Nvidia Shadowplay Instant Replay Testing

[LeviBW]

This is testing I have done over the last few days on the impact of Nvidia Shadowplay Instant Replay Recording on gaming preformance. These tests were run on my main gaming PC (Specs Link https://pcpartpicker.com/b/cvjypg) with no applications running other than the benchmarking software. The benchmarking software was restarted after every run of testing.

All testing was run using Nvidia WHQL Driver 384.76, and Geforce Experience Version 3.7.0.81  

 

The setting for Nvidia Shadowplay Instant Replay that were used:

Resolution: 1080p

Frame Rate: 60 FPS

Bit rate: 30 Mbps

Replay Length: 5 Min.

 

Link to testing Data and Analysis: https://1drv.ms/x/s!AnPE2UQhg2sqiR3E0pVshAQTh3hq

 

If you have any input on how I can improve my spreadsheet or suggestions for other tests I should run, all input is welcome.

 

[Zando_]

So you only lose about 1fps?

[Shimejii]

The biggest dip in FPS for Instant replay is when you actually hit Save. Otherwise it does not affect performance that much. If you do manual recording then you will see a much bigger difference

[Bustapalapeno]

Pretty cool! It would be neet if you tested @Shimejii's claim of manual recording with shadowplay as well. 

[Shimejii]

2 minutes ago, Bustapalapeno said:

Pretty cool! It would be neet if you tested @Shimejii's claim of manual recording with shadowplay as well. 

I use it all the time for my gaming youtube channel. You see a large dip in performance when you start and end the video with manual, with the inbetween being about the same if not slightly less (or in games like ARK 10-15 FPS less). 

[LeviBW]

15 minutes ago, Shimejii said:

The biggest dip in FPS for Instant replay is when you actually hit Save. Otherwise it does not affect performance that much. If you do manual recording then you will see a much bigger difference

Alright. I don't know that, I may re do all of these tests while recording with the same settings instead.

 

Thanks for the Idea 

[LeviBW]

On 7/24/2017 at 2:46 PM, Shimejii said:

I use it all the time for my gaming youtube channel. You see a large dip in performance when you start and end the video with manual, with the inbetween being about the same if not slightly less (or in games like ARK 10-15 FPS less). 

@Bustapalapeno @Shimejii @Zando Bob

I started testing the performance impact of recording and figured out what was going on. Basically the fps while recording is exactly the same as in the data I have above. You do see an FPS dip when you hit the Start and Stop button if the save location is on the same drive as the game you are playing is running off of. 

 

As long as you change your temp file location and save location to a drive that is different from the one you are playing your games off of, you will see the exact same performance as I have in the spread sheet. 

[Shimejii]

22 minutes ago, LeviBW said:

@Bustapalapeno @Shimejii @Zando Bob

I started testing the performance impact of recording and figured out what was going on. Basically the fps while recording is exactly the same as in the data I have above. You do see an FPS dip when you hit the Start and Stop button if the save location is on the same drive as the game you are playing is running off of. 

 

As long as you change your temp file location and save location to a drive that is different from the one you are playing your games off of, you will see the exact same performance as I have in the spread sheet. 

Mine are, i have 3 SSD's, 500 GB M.2 for boot drive and games, 250 GB SSD for Recording footage, and a 750 GB ssd for temp files. I still see performance dips when starting and ending clips, but the inbetween like i said should be near the same if not a bit less depending on the system and other variables.

[alphaproject]

Yes during recording I don't really see much of a dip....just when I hit record. I only use manual settings.

[LeviBW]

1 hour ago, Shimejii said:

Mine are, i have 3 SSD's, 500 GB M.2 for boot drive and games, 250 GB SSD for Recording footage, and a 750 GB ssd for temp files. I still see performance dips when starting and ending clips, but the inbetween like i said should be near the same if not a bit less depending on the system and other variables.

That's interesting, In my testing I have the game running on my 2TB ssd and then then both the temp folder and save folder are on my 500GB samsung 960 evo and I saw 0 fps dip when starting and stopping the recordings in that configuration

[Curufinwe_wins]

That's pretty cool. I've just had some stability issues and other gameplay failures related to shadowplay being enabled (I honestly don't remember anymore, but I do remmeber looking them up and people being like "turn off shadowplay" and then everything working again.)

 

Otherwise I'd totally be using it myself.

[SkilledRebuilds]

The first two seconds of my capture (make my games drop a few fps and then it quickly recovers)

The capture itself does not have this initial slowdown and my games only lose 1-2fps when using 1080p-1440p but when I am actually playing at 1600p-2160p then record, I can lose up to 10% of ingame performance on average. (I just make sure I have enough performance before this)

(Recording to Raid0 SSD) (In-GameRes/60FPS@100Mbps)

*Also never had any of these 'issues' that plague people..

GFE/Shadowplay has always worked whenever I needed it... but I know thats not the case for everyone.

[Curufinwe_wins]

3 hours ago, SkilledRebuilds said:

The first two seconds of my capture (make my games drop a few fps and then it quickly recovers)

The capture itself does not have this initial slowdown and my games only lose 1-2fps when using 1080p-1440p but when I am actually playing at 1600p-2160p then record, I can lose up to 10% of ingame performance on average. (I just make sure I have enough performance before this)

(Recording to Raid0 SSD) (In-GameRes/60FPS@100Mbps)

*Also never had any of these 'issues' that plague people..

GFE/Shadowplay has always worked whenever I needed it... but I know thats not the case for everyone.

Yeah seems like SLI+multimonitor+shadowplay+high refresh rate is the litany of asking for trouble.