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.

 

Go Back   Doom9's Forum > Hardware & Software > Software players

Reply
 
Thread Tools Search this Thread Display Modes
Old 7th July 2015, 00:40   #31561  |  Link
huhn
Registered User
 
Join Date: Oct 2012
Posts: 7,903
Quote:
Originally Posted by fairchild View Post
I use lav copyback and it forces high powerstates when the clip allows. But I just checked an HEVC clip which doesn't use copyback and it still had same results. 12ms with new path d3d9 and 23ms with old path d3d9.

I am using 14.12 official drivers if that makes a difference, also my settings are pretty pedestrian:
you should check the power states. dxva has rarely a clear effect on it. on the other side cuvid forces high power state.

just check it with GPU-z
huhn is offline   Reply With Quote
Old 7th July 2015, 17:57   #31562  |  Link
fairchild
Registered User
 
Join Date: Sep 2010
Posts: 321
Quote:
Originally Posted by aufkrawall View Post
I can hardly believe these figures. Almost twice as fast, just because of different API?
I'd really rather assume that is somehow a counting/calculation error.
You could check this with madVR settings which almost cost twice as much, e.g. more Needi3 neurons if interop performance is sufficient.
I'm just reporting my findings for my rig. Not trying to convince anybody or anything. Different hardware using different drivers using different settings will produce different results.

Quote:
Originally Posted by huhn View Post
you should check the power states. dxva has rarely a clear effect on it. on the other side cuvid forces high power state.

just check it with GPU-z
Well what I've noticed is that when I use D3D9 old-path it maintains 450mhz/1250mhz throughout the video, where as D3D9 new-path or D3D11 with present a frame for every vsync, it fluctuates from 450mhz/1250mz to 1050mhz/1250mhz. Not sure why this causes the reported rendering times by MadVR to go down so drastically.

I'm going to try not using FSE while using D3D11 windowed mode to see if that bug that always forces me to go back to FSE is squashed. (the composition rate changes randomly and this mismatch causes the videos to be choppy. eg when the composition rate is 23.976 and I am watching a 60p video clip)

You were right that the dxva copy-back doesn't seem to affect the clock rates much. I think it's best for the card to stay at a single clock rate, say the 450mhz/1250mhz mid state instead of constantly fluctuating. low power state for my card is 300mhz/150mhz
__________________
MPC-HC/MPC-BE, Lav Filters, MadVR
CPU: AMD Ryzen 5 1600, Video: AMD Radeon RX Vega 56 -> TCL S405 55", Audio: Audio-Technica M50S
fairchild is offline   Reply With Quote
Old 7th July 2015, 18:47   #31563  |  Link
aufkrawall
Registered User
 
Join Date: Dec 2011
Posts: 1,812
Quote:
Originally Posted by fairchild View Post
I'm just reporting my findings for my rig. Not trying to convince anybody or anything. Different hardware using different drivers using different settings will produce different results.
Yeah, but it's kind of esoterism to conclude D3D11 would be magically 100% faster than D3D9 just because of the rendering times (which are not a secure method to measure performance, even according to madshi) without really benchmarking this under clean conditions.
aufkrawall is offline   Reply With Quote
Old 7th July 2015, 21:12   #31564  |  Link
fairchild
Registered User
 
Join Date: Sep 2010
Posts: 321
Quote:
Originally Posted by aufkrawall View Post
Yeah, but it's kind of esoterism to conclude D3D11 would be magically 100% faster than D3D9 just because of the rendering times (which are not a secure method to measure performance, even according to madshi) without really benchmarking this under clean conditions.
I actually said in my original post that both D3D9 new-path and D3D11 were about equal. D3D9 old-path is the one that shows the huge increase in rendering times. Anyways... Just use whatever works best for you.
__________________
MPC-HC/MPC-BE, Lav Filters, MadVR
CPU: AMD Ryzen 5 1600, Video: AMD Radeon RX Vega 56 -> TCL S405 55", Audio: Audio-Technica M50S
fairchild is offline   Reply With Quote
Old 7th July 2015, 21:30   #31565  |  Link
aufkrawall
Registered User
 
Join Date: Dec 2011
Posts: 1,812
Oops, sorry. I misread that.
aufkrawall is offline   Reply With Quote
Old 7th July 2015, 21:57   #31566  |  Link
SithUK
Registered User
 
Join Date: Dec 2001
Posts: 16
I am using MPC-HC with madVR and a 3D LUT. I am using a laptop as the source. The laptop has an i5 quad core and ATI 4500 equivalent GPU. I am struggling with video playback FPS. Is there a guide that anyone can suggest that suggest how to increase performance madVR performance with a 3D LUT (or even without it)? The big win for me is running the 3D LUT to achieve an accurate Rec709 gamut.
SithUK is offline   Reply With Quote
Old 7th July 2015, 22:56   #31567  |  Link
Warner306
Registered User
 
Join Date: Dec 2014
Posts: 1,127
Quote:
Originally Posted by SithUK View Post
I am using MPC-HC with madVR and a 3D LUT. I am using a laptop as the source. The laptop has an i5 quad core and ATI 4500 equivalent GPU. I am struggling with video playback FPS. Is there a guide that anyone can suggest that suggest how to increase performance madVR performance with a 3D LUT (or even without it)? The big win for me is running the 3D LUT to achieve an accurate Rec709 gamut.
You need to lower your image scaling settings. I'd start by choosing a lower-quality chroma upscaler such as Bicubic because the difference can be neglible with most sources. Then I'd use image upscaling such as Lanczos3 or lower. Lastly, turn off debanding and any other extra processing. If still over the average rendering queue, start experimenting with the " Trade quality for performance" checkboxes.
Warner306 is offline   Reply With Quote
Old 8th July 2015, 02:38   #31568  |  Link
har3inger
Registered User
 
Join Date: Feb 2014
Posts: 139
I would actually start first with the trade quality for performance stuff. A better image upscaler or smooth motion on is more noticeable than madvr doing math in 10 bit instead of 16 bit. The setting to toggle 10 bit buffers gives quite a bit of performance for its negligible tradeoff.
har3inger is offline   Reply With Quote
Old 8th July 2015, 02:51   #31569  |  Link
seiyafan
Registered User
 
Join Date: Feb 2014
Posts: 162
I think ATI 4500 has a performance similar to HD 2000, if that's the case it would have a pretty hard time doing anything other than the bare minimum.
seiyafan is offline   Reply With Quote
Old 8th July 2015, 08:53   #31570  |  Link
SithUK
Registered User
 
Join Date: Dec 2001
Posts: 16
Quote:
Originally Posted by seiyafan View Post
I think ATI 4500 has a performance similar to HD 2000, if that's the case it would have a pretty hard time doing anything other than the bare minimum.
I've had a look and the Ati 540v that I have is basically a mobility hd 4550. Userbenchmark.com has it as comparable to an Intel hd2000.

Should I try selecting all the tick boxes in the trade performance for quality section? The default install has the first half of the options selected already.

The default chroma, up scaling, and down scaling options are already as suggested above. (Bicubic and lanczos)

I tried adjusting the 3d lut bit rate option on the trade performance for quality page to the lowest value of 6. I still had stuttering playback.

Is there anyway to divert the load from the gpu to the cpu?my processor is showing max 6%, and that is likely for only one of the 4 cores.

Is my hope of using a 3d lut with this laptop realistic? (Cpu is i5 m450 @ 2.4Ghz, 4Gb ram, Ati 540v gfx)

Last edited by SithUK; 8th July 2015 at 08:59.
SithUK is offline   Reply With Quote
Old 8th July 2015, 09:06   #31571  |  Link
Ver Greeneyes
Registered User
 
Join Date: May 2012
Posts: 447
Quote:
Originally Posted by SithUK View Post
Is there anyway to divert the load from the gpu to the cpu?my processor is showing max 6%, and that is likely for only one of the 4 cores.
Is the decoding being done on the GPU? If you're using the LAV filters (through MPC-HC or otherwise), check the Hardware Acceleration section in the video decoder settings.
Ver Greeneyes is offline   Reply With Quote
Old 8th July 2015, 09:59   #31572  |  Link
SithUK
Registered User
 
Join Date: Dec 2001
Posts: 16
Quote:
Originally Posted by Ver Greeneyes View Post
Is the decoding being done on the GPU? If you're using the LAV filters (through MPC-HC or otherwise), check the Hardware Acceleration section in the video decoder settings.
Mpc-hc
Lav filter video settings
Hardware acceleration is using Dxva2 (native)

Even if I set the option to none, would it transfer the madvr 3d lut related processing off my gpu?
SithUK is offline   Reply With Quote
Old 8th July 2015, 10:15   #31573  |  Link
nevcairiel
Registered Developer
 
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,344
CPUs are incredibly bad at video processing (or maybe more accurate, GPUs are extremely good at it), so any video processing on the CPU would be extremely inefficient.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders
nevcairiel is offline   Reply With Quote
Old 8th July 2015, 11:56   #31574  |  Link
chros
Registered User
 
chros's Avatar
 
Join Date: Mar 2002
Posts: 2,323
Quote:
Originally Posted by fairchild View Post
Well what I've noticed is that when I use D3D9 old-path it maintains 450mhz/1250mhz throughout the video, where as D3D9 new-path or D3D11 with present a frame for every vsync, it fluctuates from 450mhz/1250mz to 1050mhz/1250mhz. Not sure why this causes the reported rendering times by MadVR to go down so drastically.
Quote:
Originally Posted by fairchild View Post
I actually said in my original post that both D3D9 new-path and D3D11 were about equal. D3D9 old-path is the one that shows the huge increase in rendering times. Anyways... Just use whatever works best for you.
The 1st sentence in your 1st quote is the answer to your question: if you use Fullscreen Exclusive Oldpath your gpu can do the same processing with slightly worse rendering times on much lower GPU frequency!
That means Oldpath is way faster than the rest.
For me it's important with my laptop, that can be easily as loud as a hover, lower power states/frequencies means less heat which means less noise.

Quote:
Originally Posted by seiyafan View Post
I think ATI 4500 has a performance similar to HD 2000, if that's the case it would have a pretty hard time doing anything other than the bare minimum.
As we talk about this, try out the Fullscreen D3D9 (Oldpath) presentation, it's faster then the rest. (Untick 'present frames in advance' in Fullscreen Exclusive setting. You can check what madvr is using with CTRL+J)

And if you use subtitles (I do) and MPC/MPC-HC , than there's 1-2 subtitle related settings (like animation related) which can affect speed as well. Try them out and check your gpu usage in the meantime with GPU-Z.

Quote:
Originally Posted by nevcairiel View Post
CPUs are incredibly bad at video processing (or maybe more accurate, GPUs are extremely good at it), so any video processing on the CPU would be extremely inefficient.
More than that, they use separate part of the gpu (vpu) which nothing to do with other madvr settings, at least with nvidia GPUs (I don't know anything about ATI cards).
__________________
Ryzen 5 2600,Asus Prime b450-Plus,16GB,MSI GTX 1060 Gaming X 6GB(v398.18),Win10 LTSC 1809,MPC-BEx64+LAV+MadVR,Yamaha RX-A870,LG OLED77G2(2160p@23/24/25/29/30/50/59/60Hz) | madvr config
chros is offline   Reply With Quote
Old 8th July 2015, 12:06   #31575  |  Link
nevcairiel
Registered Developer
 
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,344
Quote:
Originally Posted by chros View Post
More than that, they use separate part of the gpu (vpu) which nothing to do with other madvr settings, at least with nvidia GPUs (I don't know anything about ATI cards).
Wat, processing the video just runs on the normal shaders.
Processing is not decoding.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders
nevcairiel is offline   Reply With Quote
Old 8th July 2015, 12:33   #31576  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by THX-UltraII View Post
It s been a while since I asked this question but are there ANY plans at all in the future to get madVR work with PowerDVD? I would LOVE to use a 3Dlut file with powerDVD!!!
You'd have to ask the PowerDVD devs to add support for madVR.

Quote:
Originally Posted by aufkrawall View Post
Have the values for SuperRes for chroma changed? I can't remember it worsening ringing with default values, but "now" it does.
Changed compared to what?

Quote:
Originally Posted by viewer View Post
i prefer super-xbr strict AR. super-xbr high-quality AR is something ringing.
Yes, it has more ringing in some situations, but it's also better in some situations. It's not a good idea to judge an algorithm by only using one single image.

Quote:
Originally Posted by xabregas View Post
I must say super-xbr is amazing, but why not put it in luma upscaling below jinc?
Because it can only double the resolution. It cannot do anything else.

Quote:
Originally Posted by ryrynz View Post
Woah, that's messed up! Look at all those green lines.. Madshi, any room for improvement there?
Maybe. But there's a limit at some point. Either I supress all ringing, or I allow some. If I do allow some, there will always be some situations where the allowed ringing is harmful. But if I surpress all ringing, that can sometimes be harmful, too.

Quote:
Originally Posted by michkrol View Post
Thanks for the new version of madVR, especially this.

After some testing, Bilateral is now my chroma upscaling algorithm for anime.
It's pure magic, especially with high quality sources
Quote:
Originally Posted by Vyral View Post
Really !?

I didn't even try this algorithm because many users here said it was horrible.
Quote:
Originally Posted by aufkrawall View Post
With lowres sources, heavy color bleeding can be observed.
Bilateral can be horrible, or it can leave all other algos in the dust. It all depends on the source.

Quote:
Originally Posted by chros View Post
And 1 more interesting thing: D3D9 exclusive Old path is way-way faster (~40%) then the New path with the same settings (when you switch on several frames in advance)! Is there any advantage (quality wise) to use the New path if everything else seems to work fine with the old one?
No quality advantage, but should be produce more reliable smoothness. If the old mode works fine you, by all means use it. The new path shouldn't really be much slower, though - except maybe if you have a rather slow GPU and if your refresh rate is much higher than the movie frame rate.
madshi is offline   Reply With Quote
Old 8th July 2015, 12:39   #31577  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
madVR v0.88.16 released

http://madshi.net/madVR.zip

Code:
* modified SuperRes algorithm slightly (faster, less artifacts, less sharp)
* added SuperRes "use alternative color space" option
* replaced SuperRes "HQ downscaling" option with new "algo" multi option
* removed SuperRes "softness" option (now always 0.00)
* fixed: decoder sending BT.2020 decoding matrix wasn't properly detected
FEEDBACK

All SuperRes users, please play with this new build. There are 2 new options: "algo" and "use alternative color space". I've intentionally named them in such a way that you can't know which does which exactly. I would like to hear your feedback about which settings you like.

This time the upscaling factor is less important (higher factor might make differences more obvious, though). Also this time don't try to find optimal passes/sharpness values. Instead it might make sense to dial passes/sharpness up as much as your GPU can handle, so you can better judge the differences between the various new options. Higher passes/strength values will make the differences more obvious.

Please make sure you have the other sharpening algos (FineSharp, LumaSharpen, AdaptiveSharpen) disabled. And please make sure you have "refine the image only once after upscaling is complete" activated.
madshi is offline   Reply With Quote
Old 8th July 2015, 13:31   #31578  |  Link
leandronb
Guest
 
Posts: n/a
dropped frames specific file and nnedi3

Hi guys. today i started watching a movie and noticed that was not smooth, i saw the stats on madvr and was showing a lot of dropped frames. tried another file and playback was smooth, so i thought it was file specifc.
i noticed that when in windowed mode playback is smooth even in 60hz mode, but when i maximize window it have the same problem than in fullscreen exclusive.
there is some pictures for explanation.
i noticed that rendering time increased drastically when in fullscreen or maximized window but only for this file. i fixed the problem by disabling double luma resolution, i was using always and nnedi3 64 neurons, but i always used this option and no problem happened before. also all my used settings i think is in the screenshots, i am using madvr v0.88.8, nvidia gtx660ti, i5 3570k, 8gb ddr3.
even i lower resolution movies that i think is the heaviest for doubling resolution played fine with no dropped frames and rendering time under 14ms. so what coulb be the problem?



another question i have is an older one in fact. when i started tweaking a little more with madvr settings read that nnedi3 improves even more the image quality. i noticed that when using this option the image gets a little glitchy with some random dots appearing, i noticed that this was discussed here. what i did to fix was using the mpc x64 with madvr, i don't know if i did it right but the glitches are gone. but my question is, in current versions there is still this problem? do you guys think that nnedi3 worth the performance hit from 4ms to 16ms rendering time?



thanks in advance.
Attached Images
    

Last edited by leandronb; 8th July 2015 at 18:50. Reason: add images
  Reply With Quote
Old 8th July 2015, 13:57   #31579  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by SecurityBunny View Post
Here you go.

Went from windowed (1s) -> fullscreen (1s) -> windowed (1s) -> fullscreen (~10s) -> alt + f4 to close player in fullscreen. Queues not filling occurred, as well as playback appearing to stutter but not showing dropped frames.

https://www.dropbox.com/s/hi8yiom7xc...20log.txt?dl=0

Not sure if it would make much of a difference, but here is another debug where I waited for queues to completely fill before switching modes. The bug occurred in both attempts and is easily repeatable by just toggling between windowed and fullscreen exclusive mode twice.

Windowed (3s) -> fullscreen (3s) -> windowed (3s) -> fullscreen (~10s) -> alt + f4.

https://www.dropbox.com/s/cqo5yif5vy...0log2.txt?dl=0

Again as a reminder, this only seems to occur with D3D11 10bit. D3D9 10bit and D3D11 8bit both seem to be fine.
Your log says that copying the frames to the backbuffer takes ages (about 15ms per frame). Which seems to be a GPU driver or D3D11 issue. Probably nothing I can do about it, unfortunately...

Quote:
Originally Posted by leandronb View Post
Hi guys. today i started watching a movie and noticed that was not smooth, i saw the stats on madvr and was showing a lot of dropped frames. tried another file and playback was smooth, so i thought it was file specifc.
i noticed that when in windowed mode playback is smooth even in 60hz mode, but when i maximize window it have the same problem than in fullscreen exclusive.
there is some pictures for explanation.
i noticed that rendering time increased drastically when in fullscreen or maximized window but only for this file. i fixed the problem by disabling double luma resolution, i was using always and nnedi3 64 neurons, but i always used this option and no problem happened before. also all my used settings i think is in the screenshots, i am using madvr v0.88.8, nvidia gtx660ti, i5 3570k, 8gb ddr3.
even i lower resolution movies that i think is the heaviest for doubling resolution played fine with no dropped frames and rendering time under 14ms. so what coulb be the problem?
Attaching stuff to this forum is not a good idea. It usually takes ages for the attachments to get approved.

In any case, what does the debug OSD (Ctrl+J) say about the movie framerate? And is deinterlacing on or off for the problematic file? My best guess is that it's a 60i file, which madVR has to deinterlace to 60p. Which consumes twice as much processing power as a 24p file does.

Quote:
Originally Posted by leandronb View Post
another question i have is an older one in fact. when i started tweaking a little more with madvr settings read that nnedi3 improves even more the image quality. i noticed that when using this option the image gets a little glitchy with some random dots appearing, i noticed that this was discussed here. what i did to fix was using the mpc x64 with madvr, i don't know if i did it right but the glitches are gone. but my question is, in current versions there is still this problem?
This is a GPU driver issue. I don't know in which driver versions this occurs and in which not.
madshi is offline   Reply With Quote
Old 8th July 2015, 14:04   #31580  |  Link
MS-DOS
Registered User
 
Join Date: Sep 2012
Posts: 77
Tried new SR with SXBR, 10 passes 1.00 str. Alternative color space makes the edges darker, a bit blurrier and maybe a little more artificial on pretty much everything I test. Subjectively it looks worse. Algo options don't seem to do anything at all, as I can't notice even a slightest difference.
MS-DOS is offline   Reply With Quote
Reply

Tags
direct compute, dithering, error diffusion, madvr, ngu, nnedi3, quality, renderer, scaling, uhd upscaling, upsampling

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 21:09.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.