![]() ![]() If the closure lit refactor is the root cause fixing this would slow down other platforms as that was the goal of the task. Older platforms can become slower as their drivers don't receive the optimizations we are targeting (responsibility of the OS driver developer). In blender we try to optimize for newer platforms. Sometimes tradeoff have to be made in terms between maintainability and platform performance. To be frank, the only thing that I miss in 2.83 is the Asset Browser, except that I could live with old version until v3.2 (as I checked that's when EEVEE would be rewritten) or there's a chance I'd upgrade my setup by then. I just stuck for now with mojave OS and I can't really help it. Is there anything else for now that I can do? I totally understand the approach the dev team has, that's fair. The good news is that #93220 (EEVEE rewrite) will be improved compilation times. I suspect D10390: EEVEE: Refactor closure_lit_lib.glsl. Also you can find a blend file that was use to perform the test. I attached a video, it's realtime screen recording (the only editing within the video is blurring the mess on my desktop). I tried also 2.93.8 version (LTS tab on ) and the result is the same. I know my machine is not a great performance beast and my OS is pretty outdated (but I can't upgrade because of other apps that I use), but you can clearly tell that 3.0.1 works much slower. Wait until material is fully visible in the viewport **Exact steps for others to reproduce the error**ģ. performance-test.blendīroken: (example: 3.0.1, 2.93.8 edbf15d3c044, master,, as found on the splash screen) Wait until material is fully visible in the viewport.Switching to "Material Preview" view mode (or Rendered) and any manipulation of a material (like adding new node) takes significantly longer than in previous version.Įxact steps for others to reproduce the error Graphics card: Intel Iris Graphics 550 1536 MBīroken: (example: 3.0.1, 2.93.8 edbf15d3c0, master,, as found on the splash screen) Thanks.Operating system: MacOS Mojave 10.14.6 (18G9323) So I am trying once again to raise this issue. Working around this is time-consuming, but even with this issue PDFPenPro has been an indispensable tool for my work. ![]() But I may be ignorant of some obvious reason why adding this functionality is difficult.) (Indeed this is what one does with Acrobat Pro. (This happens with Acrobat Pro.) If that is the case, a solution would be to implement an option of running OCR on a specified range of pages, so that a user can OCR a document not in one go but in smaller steps. My (technically illiterate) suspicion has been that, with larger files the memory runs out and causes the OCR process to halt. I can upload an example document which gives me grief, as well as a verbose log taken during an attempt at OCR-ing it. ![]() But this is unpredictable, and I often have to try a number of times. Sometimes, one of the two “PDFpenOCR” dies and the other keeps going, and OCR is completed. The application (PDFpenPro) itself thereby hangs, and I have to force quit it. What happens (which one can see from Activity Monitor is that one or two processes called 'PDFpenOCR' start when I run the “OCR document” command, and after a while these processes slow down, and eventually halt. I am hoping that the new support team at Nitro might offer further help. I have raised this issue a couple of times with the Smile support team. This has been a problem for me since before version 10. PDFpenPro's OCR process often stalls when being run on a relatively long document, whether it is run with the document open, or from the “OCR files…” command. I am using PDFPenPro 13.1 for Mac on MacBook Pro (13-inch, 2016, Four Thunderbolt 3 Ports), running macOS Mojave 10.14.6 (18G9323). ![]()
0 Comments
Leave a Reply. |