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. |
4th July 2019, 12:20 | #56761 | Link | |||
Registered User
Join Date: Mar 2002
Posts: 2,323
|
Quote:
Quote:
Quote:
So, in summary, if you are still interested testing this on your system, I can provide you the test file and all the madvr setting to use (!) during the weekend. (Otherwise I don't see the point to talk about this anymore , no offense.)
__________________
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 |
|||
4th July 2019, 12:31 | #56762 | Link |
Registered User
Join Date: Oct 2012
Posts: 7,903
|
so you can't read the memory load?
and what bigger data the decoded frame size from a youtube video with the same size as a BD have 100 % the same size. system ram what are you talking about... yeah everyone has to use my test settings and file or there test is wrong... |
4th July 2019, 12:47 | #56763 | Link |
Registered User
Join Date: Mar 2002
Posts: 2,323
|
dxva2 copyback uses sytem ram, doesn't it?
__________________
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 |
4th July 2019, 14:07 | #56765 | Link |
Registered User
Join Date: Oct 2012
Posts: 7,903
|
the system memory shouldn't have an impact on the GPU performance as long as it can do it in realtime. maybe if it is still not clear i tested an UHD file not an SD file here.
the down and upload operation are effecting the GPU. |
4th July 2019, 14:08 | #56766 | Link | |
*****
Join Date: Feb 2005
Posts: 5,643
|
Quote:
That would be compatible with AMD Navi, NVIDIA RTX, and Intel Gen11 (IceLake). |
|
4th July 2019, 14:35 | #56767 | Link | ||
Registered User
Join Date: Mar 2002
Posts: 2,323
|
Quote:
No, it wasn't for me, maybe I missed something on the screenshot. Quote:
Even the cropped picture (with black bar detection) using dxva2 copyback is slower (!) than processing the whole image with d3d11 native (using the same settings in madvr) Anyway, I stop this conversion for now, this is how it works on our systems, I/we don't want to convince anybody, everyone can try it for themselves and do/think/believe whatever they like
__________________
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 |
||
4th July 2019, 16:05 | #56768 | Link |
Registered User
Join Date: Oct 2016
Posts: 896
|
Maybe there's some kind of 'stall' somewhere that blocks the GPU doing other tasks while it's doing texture transfer for copyback, and you don't notice the difference unless you have heavy rendering settings?
On my old Core 2, copyback with 2160p24 HDR maxes out the GPU and rendering times shoot up to 50-65 ms (can't test the Haswell unfortunately as my Radeon doesn't do HEVC). System RAM definitely has a big impact on copyback performance, so I think it doesn't depend on if you have a 960 or a 1080 Ti but on how fast your memory subsystem is. Maybe also the CPU being busy with other stuff has an impact? For example, did you test if disabling black bar detection changes anything?
__________________
HTPC: Windows 10 22H2, MediaPortal 1, LAV Filters/ReClock/madVR. DVB-C TV, Panasonic GT60, Denon 2310, Core 2 Duo E7400 oc'd, GeForce 1050 Ti 536.40 |
4th July 2019, 16:26 | #56769 | Link |
Registered User
Join Date: Oct 2012
Posts: 7,903
|
there are multiply issues with that too.
you don't have PCIe 3 and so your ram should be faster then your PCIe 2.0. the CPU speed could cripple it too but i doubt blackbar detection could effect it much because it is not the same program and you have more then 1 core there still worth investing. the first consumer grade CPU with PCIe 3 was ivy bridge AFAIK. and just have a look at the bus load. you don't have to test HEVC and and HDR doesn't help here anyway h264 should be good enough. |
4th July 2019, 22:03 | #56770 | Link | |
Registered User
Join Date: May 2013
Posts: 706
|
Quote:
__________________
Ghetto | 2500k 5Ghz |
|
5th July 2019, 03:52 | #56772 | Link |
Registered User
Join Date: Feb 2002
Location: San Jose, California
Posts: 4,406
|
For which settings?
My 2080 Ti can do 4Kp60 to 8Kp60 with NGU medium for both chroma and luma (~12ms), NGU high takes ~16ms, while very high takes ~44ms. Using Bicubic for chroma upscaling instead drops rendering times by about 2.5ms. This is without HDR, artifact removal, post processing, or trade quality for performance options.
__________________
madVR options explained Last edited by Asmodian; 5th July 2019 at 03:54. |
5th July 2019, 09:37 | #56775 | Link |
Registered Developer
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,344
|
Noone here has touched either of those, so we don't know. That said CPU is mostly irrelevant for madVR. GPU, we'll have to see. Polaris was somehow rather bad for madVR, if that continues with NAVI we won't know until someone tests.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders |
5th July 2019, 10:05 | #56776 | Link |
Registered User
Join Date: Mar 2002
Posts: 2,323
|
I don't remember But for me the only advantage of using copyback would be to utilise black bar detection+cropping to save performance, and that's not the case. Otherwise I don't mind the full image processing and it will make to write profile rules easier.
__________________
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 |
5th July 2019, 13:08 | #56777 | Link | |
Registered User
Join Date: May 2013
Posts: 706
|
Quote:
__________________
Ghetto | 2500k 5Ghz Last edited by tp4tissue; 5th July 2019 at 13:11. |
|
5th July 2019, 16:48 | #56780 | Link |
Registered User
Join Date: Apr 2018
Location: Paris, France
Posts: 89
|
Trust your own eyes. Can you see the difference ? If you can't, then it's good enough
__________________
Charky "Rule #1 : If it works, don't change anything." Last edited by Charky; 5th July 2019 at 21:23. |
Tags |
direct compute, dithering, error diffusion, madvr, ngu, nnedi3, quality, renderer, scaling, uhd upscaling, upsampling |
Thread Tools | Search this Thread |
Display Modes | |
|
|