cv2702 Posted July 27, 2017 Share Posted July 27, 2017 Same environment (Radeon HD 7850, i7-4770, 16GB, SSD), same plan. During rendering of a walkthrough: X8: CPU 30% GPU 15-20% X9: CPU 15% GPU 90-95% Rendering a walkthrough for that plan under X9 takes 8-9 times what it took under X8. Granted, overall 3D performance has greatly improved with X9, but this situation forces me to do renderings while I'm busy on something else, then you loose that work flow . I also understand that there are some new parameters (bloom, occlusion...) which make performance comparisons more difficult. Given the augmented usage of the GPU under X9, I know my aging video card is begging for an upgrade. None the less, I'm curious if anybody else has similar or dissimilar experiences to share about walkthrough rendering under X9. Link to comment Share on other sites More sharing options...
Chopsaw Posted July 28, 2017 Share Posted July 28, 2017 Are you using the same codec in X8 and X9 ? Link to comment Share on other sites More sharing options...
Richard_Morrison Posted July 28, 2017 Share Posted July 28, 2017 This sounds suspiciously like a laptop with an auxiliary graphics card. (i.e. the Radeon). You need to explicitly set the laptop to use the Radeon card with X9, because otherwise it will use the onboard graphics processor, which can be very slow. Link to comment Share on other sites More sharing options...
cv2702 Posted July 29, 2017 Author Share Posted July 29, 2017 22 hours ago, Chopsaw said: Are you using the same codec in X8 and X9 ? First thing I checked. x264vfw in both cases. Link to comment Share on other sites More sharing options...
cv2702 Posted July 29, 2017 Author Share Posted July 29, 2017 6 hours ago, Richard_Morrison said: This sounds suspiciously like a laptop with an auxiliary graphics card. (i.e. the Radeon). You need to explicitly set the laptop to use the Radeon card with X9, because otherwise it will use the onboard graphics processor, which can be very slow. Preferences/render/Video Card Status: shows the Radeon HD 7800 series (HD 7850) being used in both cases. When I'm monitoring the render, TaskMgr shows activity against the PCI Radeon, not the embedded Intel HD 4600. Windows boots with this video card as primary. I do see something suspicious, as per attached. X9 shows a lower level Open GL version: 3.3.13491 vs 4.5.13491. Notice the common trailing '13491',but different prefixes? Even worst, when I query my video card directly, it reports 6.14.10.13491. Can anybody make sense of this? Better yet, has anyone done some benchmarks on walkthroughs between X8 and X9? Link to comment Share on other sites More sharing options...
Chopsaw Posted July 29, 2017 Share Posted July 29, 2017 (edited) 2 hours ago, cv2702 said: Better yet, has anyone done some benchmarks on walkthroughs between X8 and X9? Running one for you now X9 5400 frames No shadows No Bloom 8:56, X8 5400 frames No Shadows 8:00 My Preferences also list Open GL 4.5 in X8 and 3.3 in X9 The one thing I have not done yet is update my library for X9 because I was waiting to install my new SSD so that is the next thing on my to do list now that my drive seems to be running well. Curious what kind of quality you are getting with x264vfw and have you tried Xvid. Edited July 29, 2017 by Chopsaw Got to thinking maybe I did some edits in the X9 file so I brought forward a fresh copy and retested getting a faster result running the X8 file in X9. Sorry I just noticed that the frame sizes were not the same so I ran it again and it's faster in X8. Link to comment Share on other sites More sharing options...
cv2702 Posted July 30, 2017 Author Share Posted July 30, 2017 Just tried with Xvid: same slowness under X9. Thanks, any pointer is appreciated. As for walk-through rendering parameters, I also have no shadows No Bloom. Your (much appreciated) test shows that you are also getting slower performance with walk-through rendering under X9 versus X8, albeit with less impact than my test. I'm generating 1653 frames at 30fps. I don't think the number of frames to produce has any impact; the complexity of the frames may impact rendering times. But, back to my original post, same plan and same everything else, ratio of 8-9 (X9 -vs- X8), this is abnormal. BTW, I'm running 19.2.0.39. Where are you at? Link to comment Share on other sites More sharing options...
Chopsaw Posted July 30, 2017 Share Posted July 30, 2017 I am updated to 19.3.1.7 and my tests were both run at 15fps but I do usually shoot a final view with shadows at 30Ffps and upload to YouTube for 720p quality. I would expect that the Ambient Occlusion takes extra rendering time in X9 and it was set to 100% where I usually have that dialed down quite a bit. I gave the x264vfw a try last night and ended up with a file that was more than 2.5 x larger than with Xvid codec. Are you able to record a video at better than 720p with x264 ? If so would you mind posting your x264 settings. Yes I think you are having some unusual slowness that needs to be figured out. Link to comment Share on other sites More sharing options...
cv2702 Posted July 31, 2017 Author Share Posted July 31, 2017 On 7/29/2017 at 9:25 PM, Chopsaw said: ... Are you able to record a video at better than 720p with x264 ? If so would you mind posting you x264 settings. I couldn't find any CA parameter around H264 settings, so below is Media Info and x264vfw settings: ID : 0 Format : AVC Format profile : High@L4 Format settings, CABAC : Yes Format settings, RefFrames : 4 frames Codec ID : H264 Duration : 55 s 100 ms Bit rate : 6 988 kb/s Width : 2 114 pixels Height : 830 pixels Display aspect ratio : 2.547 Frame rate : 30.000 FPS Color space : YUV Chroma subsampling : 4:2:0 Bit depth : 8 bits Scan type : Progressive Bits/(Pixel*Frame) : 0.133 Stream size : 45.9 MiB (100%) Link to comment Share on other sites More sharing options...
Chopsaw Posted July 31, 2017 Share Posted July 31, 2017 Ok so it appears that you are using the default x264vfw settings. Thank you for the Media Info, it is very extensive... Much more than windows provides. What is your source if I may ask. When I use the default x264vfw settings and then uploading to Youtube it will only register as 720p quality. Are you able to achieve 1080p somehow without post processing? Have you run this problem past tech support or did you come here first. I may have to check to see how much my video card is loading up as you did in your first post as mine is fairly new and may handle things better in X9. Link to comment Share on other sites More sharing options...
cv2702 Posted July 31, 2017 Author Share Posted July 31, 2017 34 minutes ago, Chopsaw said: Ok so it appears that you are using the default x264vfw settings. Thank you for the Media Info, it is very extensive... Much more than windows provides. What is your source if I may ask. When I use the default x264vfw settings and then uploading to Youtube it will only register as 720p quality. Are you able to achieve 1080p somehow without post processing?. I use 2 utilities: MediaInfo to get specs from the output video (Open Source) Codec Tweak Tool to read or set codec parameters (part of K-Lite Codec Pack found here) I uploaded my test video tu Youtube. I can only play it at maximum 720P. I normally don't use Youtube for my clients, I link them to cloud storage. Link to comment Share on other sites More sharing options...
Chopsaw Posted August 1, 2017 Share Posted August 1, 2017 I Think I have some bad news for you as I reran an X9 test with No Shadows, No Bloom, and Ambient Occlusion set to 0% and the same 5400 frames in 5:55. So X9 is faster and seems to create a smaller file. CPU averaged about 22% and GPU about 26%. Or maybe if you do not care for Ambient Occlusion this may solve your problem. Give it a try. My X8 numbers were CPU 24% average and GPU 18% average. My X9 numbers were CPU 24% average and GPU 36% average with A.O. @ 100% Link to comment Share on other sites More sharing options...
cv2702 Posted August 1, 2017 Author Share Posted August 1, 2017 Thanks @Chopsaw, I'll be running some more tests; it does appear the issue is with my setup. Link to comment Share on other sites More sharing options...
cv2702 Posted November 7, 2017 Author Share Posted November 7, 2017 I'm doing more walkthroughs as this project gets to an end and it's painful. There is an increased reliance on the graphics card with X9 compared to X8, as both @Chopsaw's test and mine indicate. Depending on which GPU benchmark, a GTX1070 runs 3 to 5 times faster than my HD7850. The HD7850 runs the walkthroughs at closer to 100% rather than my initial reporting of 90-95. So I suspect the CPU is actually regularly idling, waiting on the HD7850 to complete. That could explain the lower CPU consumption in my test. So, I have two options. It would be nice if CA would allow users to specify the load spread between CPU and GPU, but I suspect that would be difficult to implement. The other option is to try out a GTX1070 which I plan to do shortly. I'll report when I get there. Link to comment Share on other sites More sharing options...
Chopsaw Posted November 7, 2017 Share Posted November 7, 2017 They just brought out a new 1070 Ti if you want to check it out or there are still faster models if they are within your budget. https://www.evga.com/articles/01161/evga-geforce-gtx-1070-ti/ Link to comment Share on other sites More sharing options...
cv2702 Posted November 7, 2017 Author Share Posted November 7, 2017 A 1080 would be nice, but not at those prices I was looking at the 1070ti, but few online stores (Canada) have stock. Then moving away from AMD, I need to decide amongst Asus, EVGA, MSI, Gigabyte, Zotac... I assume you have an EVGA. What made you decide on that brand? Link to comment Share on other sites More sharing options...
Chopsaw Posted November 8, 2017 Share Posted November 8, 2017 6 hours ago, cv2702 said: I assume you have an EVGA. What made you decide on that brand? I think that it was a matter of price and availability as well as not being able to find many bad reports on EVGA. Also Got the Founders Edition with the nearly silent fan which I really like. I ended getting mine from Amazon. Link to comment Share on other sites More sharing options...
Recommended Posts
Please sign in to comment
You will be able to leave a comment after signing in
Sign In Now