June 15, 20179 yr It seems p3d v4 is nicely optimized to take advantage of multiple cpu cores. I'm running an i5-6600k at 4.7ghz and p3d will max out all four cores. Are those of you with 6 or 8 core cpus seeing a similar result ? How many cores is p3d coded to fully utilize at this point? Lian Li 011 Air Mini | AMD 9800X3D | Asus ROG STRIX B650E-F | Arctic Cooling Liquid Freezer II 280mm RGB | 2x32GB G.Skill DDR5-6000 | ASUS TUF RTX 5090 | Seasonic Prime Platinum 1000W | Pimax Crystal Light
June 15, 20179 yr There seems to be a tipping point, where scenery loading; split up amongst too many threads has no benefit or could actually slow things down. There is no hard fast number as it is also highly dependant on clock speed of the cpu, IPC of your CPU and subsystems etc etc. You'd really have to ask someone who has say gone from for instance a 4 C , 8 T or straight 4 core to a HEDT CPU with 8+ cores. I expect it drops off at or after 6 core 12 thread but it's just supposition until the community tests it further. Edit; If you have 10 people trying to screw in a lightbulb does it actually get done faster? Same thing applies here. Steve McNitt
June 15, 20179 yr In my case its using all 6 actual and the 12 virtual cores very nicely. I have been told the 2nd virtual core is usually the least loaded because many windows services run on the 1st actual core 0 and many of those dont hyperthread to Core 1. Dont know if it is true but the process explorer evidence I see suuports that. I also see it using Handbrake for video conversions it will max out any CPU you use and I mean 100% but the second virtual core will always be less than the rest ( on my 17-3930K anyways) just like P3D v4. It is always a test of how many other things you have hooked into the sim too. For example TrackIR I have seen run anywhere between 2 and 10% cpu load depending on what you are doing with it. Level flying over ORBX scenery with almost everything to do with scenery maxxed (except speed trees) at FL100 and viewing the scenery loads my CPU at about 50 to 60% I am GPU bound on my 4K screen the GTX980Ti hits 99% when the screen opens on the tarmac. A.Chryss - near YSCB 1. ASUS TUFF X570E wifi, Ryzen 5950X EK 420x45 RAD, D5 Pump, EK monoblok, 32GBs GSkill Ram, Gig Aorus RTX 3080TiTi, SSung 980 Pro 2TB & 1Tb. XBox controller and Stream Deck XL for - camera and sim control (non AC) 2. ASUS Tuff Z690 i7-14700K, Zotac RTX 4080Super, 32MHz ram 2x 970EVO 1TBNVME 3x ssd etc. MSFS2020 & DCS. 49" Predator 240Hz OLED monitor & tablets. Warthog controllers, Honeycomb Yoke, Thrustmaster TPR pedals, TrackIR & KVM.
June 15, 20179 yr 43 minutes ago, Slayer said: There seems to be a tipping point, where scenery loading; split up amongst too many threads has no benefit or could actually slow things down. There is no hard fast number as it is also highly dependant on clock speed of the cpu, IPC of your CPU and subsystems etc etc. You'd really have to ask someone who has say gone from for instance a 4 C , 8 T or straight 4 core to a HEDT CPU with 8+ cores. I expect it drops off at or after 6 core 12 thread but it's just supposition until the community tests it further. Edit; If you have 10 people trying to screw in a lightbulb does it actually get done faster? Same thing applies here. Nope.. bad analogy.
June 15, 20179 yr I've been using a 4820K @4.5GHz for about 4 years now paired with a Titan X (Maxwell). I've been very happy with P3D v3 performance. When v4 dropped I was able to find a 4960X on eBay for $200. The 4960X is the "fastest" or "Best" CPU I can plug into my socket 2011 motherboard. I was able to overclock it to 4.7Ghz with a corsiar H100i v2 water cooler. I definitely noticed a performance increase with the 2 additional cores. I find that I can increase autogen draw distance\scenery complexity without a noticeable hit to FPS. I found an interesting post from Beau on the LM forums. Take a look at his post here: http://www.prepar3d.com/forum/viewtopic.php?f=6315&t=124932 Here's a quote from Beau: "The speed of the primary core will be the determining factor FPS assuming your settings have you CPU-bound. For raw FPS, less cores at a higher clock will yield better results. On the other hand, more cores will improve the speed at which new terrain textures, and autogen data load in." Interesting... No? Benjamin Nash AMD Ryzen 9800X3D, 64GB DDR5 RAM CL30, Asus ROG Strix 4090, Asus ROG x870E Hero Motherboard, Samsung 9100 Pro 2TB PCIE5 NVME SSD, LG OLED C1 48"
June 15, 20179 yr Thanks for the link to Beau's response, it is valuable to me for all the other info he provided about how P3D works now. Given I haven't seen my CPU run at 100% on all cores (and I dont want to) I also found Beau's comment about more cores and scenery telling in the context of me having a bound GTX980Ti. 64 bit and multi threading in flight sims has finally arrived in time to watch a giant battle of cores about to commence between the two makers of HEDT CPUs and we can only benefit. A.Chryss - near YSCB 1. ASUS TUFF X570E wifi, Ryzen 5950X EK 420x45 RAD, D5 Pump, EK monoblok, 32GBs GSkill Ram, Gig Aorus RTX 3080TiTi, SSung 980 Pro 2TB & 1Tb. XBox controller and Stream Deck XL for - camera and sim control (non AC) 2. ASUS Tuff Z690 i7-14700K, Zotac RTX 4080Super, 32MHz ram 2x 970EVO 1TBNVME 3x ssd etc. MSFS2020 & DCS. 49" Predator 240Hz OLED monitor & tablets. Warthog controllers, Honeycomb Yoke, Thrustmaster TPR pedals, TrackIR & KVM.
Archived
This topic is now archived and is closed to further replies.