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 10th May 2013, 22:33   #18781  |  Link
Dodgexander
Registered User
 
Join Date: Jul 2008
Posts: 157
Quote:
Originally Posted by maco07 View Post
Hi madshi. Can you share with us in wich new features are you working for next madVR version?

Great work!! I'm using madVR sin 0.3x!
Too many to post I think

Sent from my Blade S using Tapatalk 2
Dodgexander is offline   Reply With Quote
Old 10th May 2013, 23:02   #18782  |  Link
leeperry
Kid for Today
 
Join Date: Aug 2004
Posts: 3,477
Quote:
Originally Posted by petran79 View Post
I did upgrade the GPU, from an Nvidia GTS450 to a GTX660

Now in Windows 7 there are no issues with MadVR.

But on Windows XP, after the upgrade any video I play with MadVR shows dropped frames for like 10-15 seconds, then plays normally and after a while dropped frames appear again for the same length. If I choose other renderers videos play normally. If I choose lower settings in Madvr, delay may appear much later or not at all if the video is lower in quality.
"lower in quality"

Is the refresh rate properly detected? Is the problem still there with the old rendering path?

Anyway, thanks for the feedback as you are the second person mentioning that 660's are unusable with mVR on XP and I was just about to order one.....I guess I'll have to find the motivation to install and thorougly testbed W7/W8 before ordering a new GPU, bummer....but apparently W8 comes with its own bag of new problems, m$ doesn't want to release W7 SP2(critical hotfixes were released for audio & USB since SP1) and madshi might very well not care all that much about XP anymore. Oh well, don't fix it if it's ain't broken and I'll be saving money in the process too I guess ^^

Quote:
Originally Posted by XRyche View Post
Last time I checked everyone was saying you still couldn't us madVR with TV.
I'm wet dreaming about HD DVB-T with 1080@50i CUVID deinterlacing, J3AR scaling, inverse-palspeedup with Reclock to 24/48p and smooth-motion with mVR in 140Hz on top of it .....but if the 660 is no workee on XP with mVR then I'll have to bite the bullet and upgrade, hah.

I see that you can specify a DVB-T input device in PotP, and as a worst case scenario DragonQ told me that JRiver can do DVB-T with mVR.

Last edited by leeperry; 11th May 2013 at 16:59.
leeperry is offline   Reply With Quote
Old 11th May 2013, 03:32   #18783  |  Link
pie1394
Registered User
 
Join Date: May 2009
Posts: 212
Quote:
Originally Posted by leeperry View Post
I'm wet dreaming about HD DVB-T with 1080@50i CUVID deinterlacing, J3AR scaling, inverse-palspeedup with Reclock to 24p and smooth-motion with mVR in 140Hz on top of it .....but if the 660 is no workee on XP with mVR then I'll have to bite the bullet and upgrade, hah.
My 4 years-old HTPC's MB just suddenly becomes unstable. But it is not a good timing to buy new PC components since DDR3 becomes twice expensive and Intel next-generation CPU will come out soon.

Anyway I have no choice but to upgrade the CPU+MB+RAM from

C2D E8400 + Gigabyte EP45-UD3P + Kingbox DDR2-1066 2GB *4

to

i5-3570K + MSI Z77A-G41 + Transcend Axe DDR3-2400 XMP 4GB * 2.
(about US$420 in my country, 5% tax included)

It is not cheap for this upgrade, but guaranteed to handle super-high bit-rate FHD H.264 Hi10, H.264 / HEVC 4K contents -- and new games with my previously upgraded Lantic HD7970.

Hyperthreading and extra 2 MB L3 cache on E3-1230v2 will not show the obvious advantage over i5-3570K for multimedia / gaming purposes. Instead the higher boosted clock rate on i5-3570K gains more actual speed. So I still decided to choice the later one.

Actually Haswell will not have big gain on CPU part's calculation performance unless the program contains a lot of AVX SQRT calculation. The current Sandy/Ivy-bridge's are not fully 256-bit implementation for all AVX instructions. Multi-threads > 4 on 4-core/8-HT CPU will NOT boost the SSE4/AVX performance either since all-HT in the same core still shares the same set of SSE4/AVX arithmetic units. Only those cheap-cost scalar instructions get extra arithmetic units.

I feel the main improvement on Haswell x86 part is for multi-threaded and multi-processed program. The common mutex / semaphore / critical section operations are implemented as x86 instructions now. It is no more needed to perform EXTERNAL memory bus locking / reading access in order to check if a mutex is locked by any other CPU core. So other cores running totally different memory access will not be held/blocked for unnecessary CPU cycles.

Of course it still needs these OS API call or program itself implemented in the new instruction set to gain the advantage.
pie1394 is offline   Reply With Quote
Old 11th May 2013, 19:18   #18784  |  Link
JarrettH
Registered User
 
Join Date: Aug 2004
Location: Canada
Posts: 860
I don't know who else is using Intel HD Graphics here, but I just lowered my rendering times by 10 ms on this PC:

Intel i3 2100
4GB DDR3
Intel HD Graphics 2000 on latest drivers
Win 7 64-bit
latest official mpc, madvr, lav (quicksync)

madvr settings

image scaling: DXVA2
chroma scaling: bicubic75 with AR

general settings: cpu queue 8, gpu queue 4 <<< that made a tremendous difference from the default of 12/8

backbuffers: 4
present queue: 6

smooth motion: on <<< allowing me to now turn on smooth motion

Last edited by JarrettH; 12th May 2013 at 00:42.
JarrettH is offline   Reply With Quote
Old 11th May 2013, 21:03   #18785  |  Link
digitech
Registered User
 
Join Date: Jun 2012
Posts: 54
Quote:
Originally Posted by JarrettH View Post
I don't know who else is using Intel HD Graphics here, but I just lowered my rendering times by 10ms on this PC:

Intel i3 2100
4GB DDR3
Intel HD Graphics 2000
Win 7 64-bit
latest official mpc, madvr, lav (quicksync)

madvr settings

image scaling: DXVA2
chroma scaling: bicubic75 with AR

general settings: cpu queue 8, gpu queue 4 <<< that made a tremendous difference from the default of 12/8

backbuffers: 4
present queue: 6

smooth motion: on <<< allowing me to now turn on smooth motion
Nice tip, i'd like to know what kind of files you usually play, sd, 720p to 1080p? thanks
digitech is offline   Reply With Quote
Old 11th May 2013, 21:42   #18786  |  Link
JarrettH
Registered User
 
Join Date: Aug 2004
Location: Canada
Posts: 860
Mostly 720p. Standard definition stuff (720 x 400...ish) my rendering times were 12 ms lower.

I should have mentioned they are only being scaled up to 1680x1050. Does going to 1080p have a bigger impact? This isn't my home machine.

Last edited by JarrettH; 11th May 2013 at 22:08.
JarrettH is offline   Reply With Quote
Old 12th May 2013, 00:12   #18787  |  Link
iSunrise
Registered User
 
Join Date: Dec 2008
Posts: 496
Quote:
Originally Posted by JarrettH View Post
Mostly 720p. Standard definition stuff (720 x 400...ish) my rendering times were 12 ms lower.

I should have mentioned they are only being scaled up to 1680x1050. Does going to 1080p have a bigger impact? This isn't my home machine.
FYI, itīs not important of your rendering times are 12ms lower, when you donīt actually have dropped frames or other performance problems with the default settings. Decreasing your queues is only useful for tweaking on high-performance CPUs or graphics hardware, where you have a lot of headroom left. I would stay away from altering the settings if all you accomplish is lowering your rendering times. The only positive side-effect is that decreasing your queues (on capable hardware) leads to faster switching and skipping.

The higher your source:target resolution difference, the more taxing it is for your hardware, as madVR needs to upscale a lot more. Depeding on your scaling (quality) settings, madshi implemented optimized algorithms that are especially fast if you only need to scale certain factors like 2x. If you need to scale to some uncommon factors, you could potentialy lose a lot of speed.

Last edited by iSunrise; 12th May 2013 at 00:19.
iSunrise is offline   Reply With Quote
Old 12th May 2013, 00:23   #18788  |  Link
digitech
Registered User
 
Join Date: Jun 2012
Posts: 54
Quote:
Originally Posted by JarrettH View Post
Mostly 720p. Standard definition stuff (720 x 400...ish) my rendering times were 12 ms lower.

I should have mentioned they are only being scaled up to 1680x1050. Does going to 1080p have a bigger impact? This isn't my home machine.
You should try the 1680x945 custom ratio to avoid any distortion if your primary monitor belongs to the 16:9 standard, actually i have a 1440x810 custom resolution with lanczos3 and anti ringing option activated in chroma, dxva with image, only in that resolution i can get away with the smooth motion functionality, if i try 1680x945 i have lots of ghosting and too much dropped and delayed frames. I have an nvidia ion gpu and a intel atom htpc, a little too low to be able to handle more demanding upscaling methods and resolutions, but the compromise looks awesome to my eyes, I think if you are going for the 1080p route you can have dropped frames but try it maybe tour gpu/cpu can handle it, i recommend to try setting up your resolution in a 100 pixel increase/decrease basis so you can find a sweet spot where you can have a smooth playback.

Last edited by digitech; 12th May 2013 at 00:27.
digitech is offline   Reply With Quote
Old 12th May 2013, 00:47   #18789  |  Link
JarrettH
Registered User
 
Join Date: Aug 2004
Location: Canada
Posts: 860
Quote:
Originally Posted by iSunrise View Post
FYI, itīs not important of your rendering times are 12ms lower, when you donīt actually have dropped frames or other performance problems with the default settings. Decreasing your queues is only useful for tweaking on high-performance CPUs or graphics hardware, where you have a lot of headroom left. I would stay away from altering the settings if all you accomplish is lowering your rendering times. The only positive side-effect is that decreasing your queues (on capable hardware) leads to faster switching and skipping.

The higher your source:target resolution difference, the more taxing it is for your hardware, as madVR needs to upscale a lot more. Depeding on your scaling (quality) settings, madshi implemented optimized algorithms that are especially fast if you only need to scale certain factors like 2x. If you need to scale to some uncommon factors, you could potentialy lose a lot of speed.
I was only tweaking because I wanted to use smooth motion on this PC. Before I couldn't because with smooth motion on, it was over the movie frame interval time which I think you're supposed to be under for proper playback.
JarrettH is offline   Reply With Quote
Old 12th May 2013, 09:32   #18790  |  Link
fallengt
Registered User
 
Join Date: Oct 2012
Posts: 22
Disable Gamma Ramps in madvr doesn't for me so I use http://www.xrite.com/product_overview.aspx?ID=789&Action=support&SoftwareID=546 to reset GPU gamma curve. Does it work the same way as "dispwin -c" ?
fallengt is offline   Reply With Quote
Old 12th May 2013, 09:49   #18791  |  Link
DragonQ
Registered User
 
Join Date: Mar 2007
Posts: 934
Quote:
Originally Posted by iSunrise View Post
The higher your source:target resolution difference, the more taxing it is for your hardware, as madVR needs to upscale a lot more. Depeding on your scaling (quality) settings, madshi implemented optimized algorithms that are especially fast if you only need to scale certain factors like 2x. If you need to scale to some uncommon factors, you could potentialy lose a lot of speed.
Is it? I thought 1440x1080i was generally the most taxing when outputting at 1080p.
__________________
TV Setup: LG OLED55B7V; Onkyo TX-NR515; ODroid N2+; CoreElec 9.2.7
DragonQ is offline   Reply With Quote
Old 12th May 2013, 11:05   #18792  |  Link
dansrfe
Registered User
 
Join Date: Jan 2009
Posts: 1,210
Sometimes the render queue inexplicably drops to 0/8 when I go from exclusive to windowed mode. In order to resolve this I either have to pause the video for 5 - 10 seconds and play or I have to restart the player. I think this may be connected to some bug in the smooth motion but I'm not really sure.
dansrfe is offline   Reply With Quote
Old 12th May 2013, 14:29   #18793  |  Link
callannn
Registered User
 
Join Date: Apr 2013
Posts: 6
Hi, first time poster here so please bear with me. I've been using MPC-HC and madVR for a while now but know next to little about it, and have used Niyawa's Guide exclusively for anime playback. I'll give you the specs of my laptop

Samsung NPC700G7C
Intel Core i7 3610QM @2.30GHz
16.0GB RAM Dual Channel DDR3
NVIDIA GeForce GTX 675M

I have followed Niyawa's guide to the word apart from a few things (I don't use ffdshow and have Smooth Motion turned off) and I am using the Highest settings under scaling algorithms. I recently had a few problems with dropped/delayed frames getting into the hundreds but realised that was due to me not disabling f.lux. Now I have 0 dropped/delayed frames during playback, but I have noticed another problem. When I play 480p and lower content and lower, my average rendering time seems to skyrocket, but when I watch 720p/1080p content, it stays at around 4-6ms depending on if i'm watching 8bit or 10bit content. Surely it should be a lot than this?
I also seem to be having problems with exclusive mode. Now, even though I don't particularly use it, I was just wondering why when I switch to it the playback becomes sluggish and redering time skyrockets once again? Would this be because I haven't changed anything under 'exclusive mode' in madVR settings?

Thank you
callannn is offline   Reply With Quote
Old 12th May 2013, 15:06   #18794  |  Link
iSunrise
Registered User
 
Join Date: Dec 2008
Posts: 496
Quote:
Originally Posted by DragonQ View Post
Is it? I thought 1440x1080i was generally the most taxing when outputting at 1080p.
I didnīt say what is the most taxing, I said that the higher your source:target resolution difference is, the more taxing it is for your hardware. That is only true for the cases JarettH has mentioned, though, not in general. The higher your source resolution, the more pixels your hardware has to process and if you add interlaced resolutions that need to be converted to progressive, that is going to tax your hardware even more, which should be obvious. You donīt need your PC for that though, deinterlacing can also be accomplished with your display, so you have more headroom for scaling in madVR.

Last edited by iSunrise; 12th May 2013 at 15:30.
iSunrise is offline   Reply With Quote
Old 12th May 2013, 15:09   #18795  |  Link
nevcairiel
Registered Developer
 
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,342
Quote:
Originally Posted by iSunrise View Post
I didnīt say what is the most taxing, I said that the higher your source:target resolution difference is, the more taxing it is for your hardware
But thats not true.
The performance requirements for scaling 1919x1079 to 1920x1080 is a lot higher then scaling 192x108 to 1920x1080

The more input pixels it has to read for scaling, the more taxing it will be. A higher scaling factor doesn't matter much.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders

Last edited by nevcairiel; 12th May 2013 at 15:11.
nevcairiel is online now   Reply With Quote
Old 12th May 2013, 15:15   #18796  |  Link
iSunrise
Registered User
 
Join Date: Dec 2008
Posts: 496
Quote:
Originally Posted by nevcairiel View Post
But thats not true.
The performance requirements for scaling 1919x1079 to 1920x1080 is a lot higher then scaling 192x108 to 1920x1080

The more input pixels it has to read for scaling, the more taxing it will be.
So it is not true that scaling SD or 720p to 1080p instead of 1680x1050 is more taxing? This is not about a general rule, my posts were specifically directed to JarettH, who asked about performance requirements if you would scale to 1080p instead of 1680x1050. Please stop selectively taking apart my post by quoting only one sentence of it and as a result, quoting me completely out of context.

Quote:
Originally Posted by JarrettH View Post
Mostly 720p. Standard definition stuff (720 x 400...ish) my rendering times were 12 ms lower.

I should have mentioned they are only being scaled up to 1680x1050. Does going to 1080p have a bigger impact? This isn't my home machine.

Last edited by iSunrise; 12th May 2013 at 15:39.
iSunrise is offline   Reply With Quote
Old 12th May 2013, 15:38   #18797  |  Link
nevcairiel
Registered Developer
 
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,342
A higher target resolution increases the cost as well of course. The performance depends pretty much always on the number of pixels, both input and output. More = slower.
The difference between the two isn't that big, though.

Your post didn't make it clear, and it sounded quite different to your revised version now.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders

Last edited by nevcairiel; 12th May 2013 at 15:41.
nevcairiel is online now   Reply With Quote
Old 12th May 2013, 15:43   #18798  |  Link
iSunrise
Registered User
 
Join Date: Dec 2008
Posts: 496
Quote:
Originally Posted by nevcairiel View Post
A higher target resolution increases the cost as well of course. The performance depends pretty much always on the number of pixels, both input and output. More = slower.
The difference between the two isn't that big, though.
I guess I should have made the bolded part more clear in my answers. You had me there for a moment.

Itīs quite the bad habit that I always see mistakes only after I clicked on submit. And when I want to perfect my posts, thereīs always the risk of someone that jumps at me and points to mistakes, even before I have time to correct my mistakes. Itīs not intentional at all.

Last edited by iSunrise; 12th May 2013 at 15:54.
iSunrise is offline   Reply With Quote
Old 12th May 2013, 18:40   #18799  |  Link
MSL_DK
Registered User
 
Join Date: Nov 2011
Location: Denmark
Posts: 137
Quote:
Originally Posted by fallengt View Post
Disable Gamma Ramps in madvr doesn't for me so I use http://www.xrite.com/product_overview.aspx?ID=789&Action=support&SoftwareID=546 to reset GPU gamma curve. Does it work the same way as "dispwin -c" ?
No ...
MSL_DK is offline   Reply With Quote
Old 12th May 2013, 19:21   #18800  |  Link
e-t172
Registered User
 
Join Date: Jan 2008
Posts: 589
Quote:
Originally Posted by MSL_DK View Post
No ...
I believe it does. It probably just calls SetDeviceGammaRamp() just like "dispwin -c". What makes you think it doesn't?
e-t172 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 11:06.


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