Kisvakond

Hi, I get the point. Thanks for the info.

Hey all, Anyone knows of any plans/rumors for a future SteamOS (Linux+Vulkan) support? DCS V1.5 brought indeed some improvement in game performance, even for AMD machines, but since dx12 is planned to be win10-exclusive, the next step is to decide if going dx12 or not (=Vulkan). BR

Hello all, I'm new to this forum and the DCS series. I've read among the new EDGE features that the CPU multi-threading is not in the scope of development. I have an old AMD config (phenom II X4 955 BE, HD6850) and I struggle with the frame rates (even worse with Crossfire). 2 CPU cores near peak load, two near idle. Just the well-known stuff, no hard feelings on that, really not a topic. :D Now I have the following experience: * When my aircraft suffers damage or when I'm hit by a missile and I just flat spinning towards the ground, my fps literally skyrockets (from 20-35 fps jumps beyond 60)! The same happens if I'm dead and I just cycle through the other units. The scenery remains the same (e.g I see my plane (or the wreckage) going down 70 fps, while it was 20 fps when I could fly it), so it doesn't look like GPU bottleneck for me. AI carries on with its stuff, continues dogfighting with wingmen, so I assume it's not the AI or 3D World portion, either. * My fps won't increase if I just pass out after a hard turn, but the aircraft remains intact. (sometimes I just pass out after a flame-out inducing negative push while trying to counter the ever-pitching habit of the SU-27, and for some reason I never re-gain consciousness. Ok, it's a flat spin but does not seem that bad, maybe 3 Gs. Why? :O) * fps is varying with the different aircraft being flown. A SU-27 or the MIG-29 has a much lower frame rate than the SU-25T or the A-10A. Interestingly, clickable birds like the A-10C or the Ka-50 are in the middle in terms of fps drop. All this tells me that the simulation of on-board systems of the player aircraft is a CPU hotspot and not the 3D world or the AI. Maybe a false observation but aircraft equipped with radar and high-end avionics tend to give less frame rate than those with no radar or without MFDs. A-10C is the odd man out, having no radar but a lot of clickable avionics to simulate up to the smallest details. :) Now, I can undestand that there are tasks that you can't easily distribute between CPU cores as they shall exchange much information among them in a timely fashion. This would present sync problems that might cancel out the benefit of parallel execution. But (and please correct me if I'm too naive or incorrect at any of my assumptions), the player aircraft simulation (avionics-electronics, physics, cockpit etc.) exchanges a low amount of data with the 3D world simulation (player aircraft simulation output eventually boils down to aircraft position and speed vector, projectiles fired, damage status etc.). As far as I read it in the SU-27 manual, there is a fairly complex simulation of the fly-by-wire system with only a few inputs and outputs. Such mathematical models (even when simplified) are prefect subjects to be calculated asynchronously. So my (purely theoretical) question: Is there something against a CPU core dedicated to the player aircraft simulation (assuming that the level-x caches of the CPU is enough to solve inter-core data exchange)? That would be one for the sound, one for the player aircraft and one for the world being simulated. Hopefully no one finds this topic boring or offensive. It's just a slow quad-core CPU owner wondering. I'm very curious about the new engine, hopefully it brings a lot of improvement. At least I would be happy with a consistent minimum of 35 fps. :) Best Regards