Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion. Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules. |
14th March 2011, 02:33 | #5981 | Link | |
Registered User
Join Date: Feb 2006
Posts: 293
|
Quote:
__________________
Spec: Intel Core i5-3570K, 8g ram, Intel HD4000, Samsung U28D590 4k monitor+1080p Projector, Windows 10. |
|
14th March 2011, 10:30 | #5982 | Link |
Registered User
Join Date: Apr 2010
Posts: 163
|
Code:
D3D9 Surface speed test: NV12: upload 458 fps, download 27 fps, trick download failed YV12: upload 278 fps, download 26 fps, trick download failed |
14th March 2011, 11:18 | #5984 | Link |
Registered User
Join Date: Apr 2010
Posts: 163
|
@Madhsi - thanks for that, it is quite interesting.
On another note regarding my delayed frames on XP, the card is maxed at 100% performance and was still dropping/delaying frames after running a while, however i have managed to stop this by configuring the NVidia 3D settings to "Let the 3D application decide". It would seem the more things you leave to the card driver to decide, the more things go wrong. |
14th March 2011, 11:57 | #5986 | Link |
Registered User
Join Date: Apr 2010
Posts: 163
|
I don't tend to force anything and previously just left the 3D settings as the default NVidia setup, but this caused playback problems after about an hour.
Playback performance is better if I let the 3D application decide and constant reboots are not necessary. I don't think my particular situation is helped by using an NForce 630a motherboard which has very poor NVidia driver support/compatibility. The main thing is playback is now sorted and I can enjoy superior PQ |
14th March 2011, 12:14 | #5987 | Link |
Registered Developer
Join Date: Sep 2006
Posts: 9,140
|
Does your GPU support CUDA? If so, you can get newest drivers by using the CUDA development drivers which seem to support all GPUs that can do CUDA. I was able to update to 270.32 drivers this way with my NVidia 9400 integrated mainboard.
|
14th March 2011, 13:17 | #5988 | Link |
Registered User
Join Date: Apr 2010
Posts: 163
|
Yes it is a 2gb GT430 and has 96 CUDA cores. I am currently running the XP 267.24 BETA drivers purely because I was having issues with 266.58 WHQL and thought things might be sorted with newer drivers.
When I get home tonight, I will PM you the madNV12Test results, I have 7600GS on Win7 32bit at work for which I have sent you results. 270.32 drivers on download now thx. |
14th March 2011, 13:55 | #5989 | Link |
Registered User
Join Date: Sep 2006
Posts: 2,197
|
would it be possible to create a kind of instruction for ati catalyst and nvidia forceware settings which are recommended to be (de)activated for madvr in terms to reach best quality (and performance)? because I havent much a clue which settings could hurt here.
|
14th March 2011, 14:01 | #5990 | Link | |
Registered User
Join Date: Apr 2008
Posts: 1,106
|
Quote:
All I can think of is if madVR could set the preferred settings, and then after the 3d device / surface is created, check to see if its settings match what madVR asked for (ie if the drivers are forcing anything other than that or not) and perhaps provide some sort of warning, or info in the ctrl-j stats that settings seem to be getting forced by the drivers? At least then the user could know to go and create a custom profile for the app, say mpc-hc, to take settings from the app rather than forcing specific settings itself.. Just a thought... Last edited by mark0077; 14th March 2011 at 14:05. |
|
14th March 2011, 14:18 | #5993 | Link |
Registered User
Join Date: Apr 2008
Posts: 1,106
|
Thing is, for people with some of the higher powered gfx cards at least, many like myself are used to setting a global setting like, 16 anisotropic, 4xaa or higher as I said. I'm just thinking alot of those users, when they use madVR might not cop on to the fact that this might negatively effect madVR (I assume it does..)... me included until I read the last few posts When madvr creates the 3d device, does it always just get the device in return or is it easy for madVR to know if it got exactly what it asked for, or some variation of it with forced settings like we speak about. I assume it would have to be a manual check on madVR's part if it was to do this check. If its not alot of work, and will improve quality / performance having specific settings, it might be a future improvement for madVR to perform this check and inform the user somehow?
Last edited by mark0077; 14th March 2011 at 14:38. |
14th March 2011, 15:13 | #5994 | Link |
Registered User
Join Date: Sep 2006
Posts: 2,197
|
but sometimes I also play older games like wc3 and in game, I cannot set options like aliasing and such. therefore, Id like to have as much enabled as default as possible, with the expections of those settings which should better be disabled for madvr. whichever these should be.
|
14th March 2011, 15:18 | #5995 | Link |
Registered Developer
Join Date: Sep 2006
Posts: 9,140
|
Well, what can I say. I don't know which settings affect madVR and which not. And I don't have the time to find out. But it's easy enough to find out for anyone who wants to know: Just compare the madVR performance (e.g. windowed mode rendering times, dropped frames etc) with various settings set to either "let the application decide" or to "forced enabled". Also you could take screenshots to check whether forcing any options results in a hit in image quality. If you turn off dithering in the madVR settings, screenshots should be bit-for-bit identical. If they're not, then obviously forcing some settings on via GPU control panel harms image quality.
|
14th March 2011, 20:29 | #6000 | Link |
Registered User
Join Date: Apr 2008
Posts: 1,106
|
madshi, one of the mp4 files I'm having issues with frame drops seems to have this issue according to mkvmerge when I was trying to make an mkv from it. No remuxing so far has allowed me to get it to play without evr-cp showing jittery graphs, or madVR dropping frames. ffdshow says its 23.974fps, my display is about 23.962hz~ (nvidia).
"The AVC video track is missing the 'CTTS' atom for frame timecode offsets. However, AVC/h.264 allows frames to have more than the traditional one (for P frames) or two (for B frames) references to other frames. The timecodes for such frames will be out-of-order, and the 'CTTS' atom is needed for getting the timecodes right. As it is missing the timecodes for this track might be wrong. You should watch the resulting file and make sure that it looks like you expected it to." I assume this is why the frames / timecodes arn't evenly spaced, hence the jittery graphs I get with evr-cp and the fairly frequent frame drops with madVR (maybe 1 every 20 seconds). Do you think this is something you can workaround with the renderer? I can post a link to the file in question if you're interested... Razoola, I have just clean installed nvidia 267.24 drivers, and of course got the grey / washed out desktop. After changing the option in the control panel you speak of to "Full-screen videos" or "Desktop programs" I get no change at all. I assume this is something being sent to the display.. which my TV at least is ignoring. I have my tv set to display full range, so maybe some tv's have an auto setting also which would "listen" to nvidias output. Not mine anyways. I have to revert to the inf mod Last edited by mark0077; 14th March 2011 at 20:37. |
Tags |
direct compute, dithering, error diffusion, madvr, ngu, nnedi3, quality, renderer, scaling, uhd upscaling, upsampling |
Thread Tools | Search this Thread |
Display Modes | |
|
|