![]() ![]() Today I've tried again, unplugged TrackIR and didn't have the Software started. ![]() Sorry for my late reply but I've had no time to play BoS last week. Switch off TrackIR and see if the horizon still stutters when you roll. I'm lost with this, running out or ideas what to try next. GPU usage is at it's limit, but it has been like that on my old system on 3.011 as well, without these stutters. I've also tried lowering settings, like shadows high->medium, landscape filter blurred->off, landscape distance 130->70km, grass ultra->distant, SSAO off, all of them together - no difference concerning these stutters.FPS is rock solid at 90 as you can see, that's because I've limited it to that value (that was for testing purpose, I've had it uncapped before with the same result). Meanwhile I've also tried to uninstall the Intel Rapid Storage Driver (using Microsoft Standard AHCI instead) and manually setting the PCIe slot version to 3.0, each to no avail. This is me flying through a furball with 13 planes involved: 2 Player fighters, 8 AI fighters, 3 AI bombers (Peshkas). Yesterday we've been flying online again, stuttering occured at the same occasions as before (crowded airfield, furball) but this time I have it in the recording as well. The proper ordering of these settings from least to most interrupt requests are:Įxtreme, High, Medium, Adaptive (Also dependent on load), Low, Minimal, Off/DisabledSo I stand corrected concerning the network thing. The gap between Off and the next least restrictive setting (minimal) is very significant DPC latency in general is consistent between runs NDIS dpc latency spread (across cores) isn't always equally balanced between runsīut DPC latency performance does not change regardless Interrupt Moderation Disabled = same as with Enabled but Off ![]() Off (but Interrupt Moderation Enabled) = 2,650,000 ![]() RSS Enabled, 2 RSS on cores 3 & 4, NUMAStaticģ0sec simulation +900Mbps, TCP, 100MB transfers, no fragmentation PCI-E Network Adapter (Intel Gigabit Desktop CT) To note some of these results will likely change based on several factors, RSS, RSS affinity, rx/tx buffers, and timer resolution. Below are some test I ran to see the difference between interrupt moderation settings during identical loaded network simulations and by comparing the differences between ndis dpc latency and the number of interrupt requests. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. Archives
January 2023
Categories |