djdallmann
What's the actual impact on latency and 0.1%/1% lows in games?
I was looking at this chart and it seem that we are playing with such a narrow window for optimization (mostly driver level) with all of these tweaks (even the ones who are actually not bullshit):
Have you done tests, both on latency and performance? I have went through tons of resources on this, including valleyofdoom's guides, just out of curiosity, but there are 0 benchmarks that I could've found covering the non-bullshit part, at least in a well controlled manner.
Overall I am confused, seems like we are majoring on the minor and there is no one benchmarking the results of it. This is not a critic, but I am genuinely looking for opinions/proofs and since you and valleyofdoom are probably one of the best resources on the topic on the internet, which is not bullshit (following MS recommendations and beyond), I figured out that it's great place to open this discussion.
Appreciate the discussion, there has been some benchmarks of this repo against other preconfigured ISOs or guides etc. This guide didn’t give the best results in those tests but that could also be said because there’s lots of information I have chosen not to include, or research I never made time to include (some is on discord). People should start with some baseline (this or similar guide), then further optimize for the game and situation.
If the community works together and starts from a same/similar point then it’s much easier to learn, share and compare. Did X show measurable differences on most systems when added to the baseline, if yes/no track when it did, why, and why not. Even a bad outcome is something learned and can be shared.
Hardware/bios/uefi configuration/peripherals play a huge role in out of box experience even when there’s bad code, or less than ideal configuration. (Operating system, drivers, software). Some GPU perform significantly worst with some games, and you may not learn that without having researched it first.
If you compare this guide to many others it’s much safer than others in terms of absolute optimization, and has shown ways to challenge and investigate things in new ways.
Thanks. I absolutely agree that this guide and less than a dozen of others, including @valleyofdoom's, are the only few in existence that try to cover system latency properly, referencing mostly great sources and a few other goodies outside system latency. However, none of these guides include game benchmarks that properly display the impact all the optimizations have with the various titles out there, and from the chart above, it does seem we are playing in a very narrow range.
Without those real-world benchmarks, and with the theoretical limit we have upon how far these optimization can go, it's frightening for me to say, if all of this matters at all, and that's coming from someone who's spend a good amount of time going through them all and have over 60-70k word notes on them.
I do understand the theoretical interest behind all of the research, it's a great resources for people working with music where such system latency can matter a whole a lot, and it's generally interesting to learn how some parts of the computer work. Yet, it's quite possible that the impact on gaming is barely anything more than a rounding error.