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. |
9th November 2011, 17:55 | #10761 | Link |
Registered Developer
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,347
|
Thats exactly why the 7000 series won't be all that great. Their architecture is old and over-used, they need something fresh - just shrinking it won't help all that much. (Unless i missed the news on a new architecture)
On the other hand, they shouldn't take too much after Bulldozer - less performance then previous architecture is not good for selling stuff.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders |
9th November 2011, 18:25 | #10763 | Link | |
quack quack
Join Date: Apr 2009
Posts: 259
|
Quote:
BTW, HD7000 series will be 4way VLIW architecture across the board (6800 and below is 5 way VLIW - less efficient). Apart from the high end 79xx card wich will be a new architecture called GCN |
|
9th November 2011, 18:29 | #10764 | Link | |
Registered Developer
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,347
|
Quote:
PS: Half power consumption is rather unlikely.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders |
|
9th November 2011, 19:01 | #10765 | Link | |
Registered User
Join Date: Sep 2011
Posts: 48
|
Quote:
|
|
9th November 2011, 19:43 | #10766 | Link | |
quack quack
Join Date: Apr 2009
Posts: 259
|
Quote:
From one generation to the next, it has been normal to get twice the GPU for the money. 40nm to 28nm is a bigger step than normal, so it's reasonable to expect more than twice the GPU. If my calculations are correct, the 6870 would be only 44% of it's current die size at 28nm compared to 40nm. |
|
9th November 2011, 19:53 | #10767 | Link |
Registered User
Join Date: Apr 2009
Posts: 1,019
|
Kepler seems a lot more exciting than the 7000 series, and definitely the cards to be waiting on if you plan on gaming. (AMD drivers have been disastrous for a number of big releases recently, Rage in particular)
|
9th November 2011, 21:29 | #10768 | Link | ||||||||||||||||||
Registered Developer
Join Date: Sep 2006
Posts: 9,140
|
Quote:
(1) "LAV CUVID+EVR" vs. "DXVADecoder+EVR" (2) "LAV CUVID+madVR" vs. "SoftwareDecoder+madVR" I would expect you to get somewhat similar GPU numbers for "LAV CUVID+EVR" and "DXVADecoder+EVR". Not sure about (2). Depends on whether video decoding is counted as GPU usage. If it's not, I would expect you to get somewhat similar GPU numbers for "LAV CUVID+madVR" and "SoftwareDecoder+madVR", too. But of course the GPU numbers should be higher for (2) than for (1). Quote:
Quote:
Can you please grab an unscaled frame? We want to check out deinterlacing quality, so there's no need to scale at all. Quote:
It might be worth trying drivers 10.11. But I'd really be interested to hear your results when doing frame-by-frame stepping with your current drivers. Quote:
Quote:
But please check out my reply to mr.duck (see above). I'd like to get a log file from you, too. And I'd like you to try frame stepping to get to the bottom of the non-smoothness, too. Thx. Quote:
I've tried once again, this time in fullscreen exclusive mode (tested only in windowed mode before), still no problem. I'm testing in XP, though, where there's no DX11 presentation mode. What happens if you use DX9 instead of DX11 in the madVR settings? Does that change anything? The DX11 path is known to be not as stable (yet) as the DX9 path. Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
The left click menu is just a menu for the currently running media player, where you can do all kinds of things to influence the media playback. The right click menu contains all things you might ever want to do, including closing the tray icon process, changing tray icon settings etc. The reason why the right click menu has so many sub menus is that it contains more functionality. Having all menu items of the right click menu in one level would be a usability catastrophe. The left click menu is so nicely simple because it limits itself to only control the current media player, nothing else. |
||||||||||||||||||
9th November 2011, 21:30 | #10769 | Link | ||||||||||||||||||
Registered Developer
Join Date: Sep 2006
Posts: 9,140
|
Quote:
As to ATI vs. Nvidia, that's really a tough decision. NVidia seems to have big problems with presentation glitches in exclusive mode. ATI doesn't seem to have that problem, but then ATI has problems on its own. So it's really hard to decide... Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Ok, seriously. madVR has several queues where it stores frames that were already decoded, uploaded to GPU RAM, fully rendered and dithered to output resolution etc and passed on to Direct3D for presentation. Ideally, all queues should be full. Quote:
Quote:
If you're sure that the GPU clock is ok then please create a short log with v0.77 (no frame drops) and one with v0.78 (frame drops). Maybe 20 seconds of playback each. Thanks. Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
madVR v0.77 and older versions didn't support deinterlacing *at all*. You were required to deinterlace before sending the video to madVR. madVR v0.78 does not deinterlace itself, it lets DXVA do it. And DXVA usually does better than using Bob. |
||||||||||||||||||
9th November 2011, 21:53 | #10771 | Link |
Registered Developer
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,347
|
"Bob" is called that way because the image bobs up and down if its used. Its not a category of algorithms that all do the right thing and use both fields to create separate images.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders |
9th November 2011, 22:09 | #10773 | Link |
Registered User
Join Date: Aug 2010
Posts: 134
|
madshi, you were mentioning improving 24 fps clips on 60hz screens.
May I suggest an option to make 50% frame blends? (frame1)-(frame1)-(50%frame1+50%frame2)-(frame2)-(frame2)- (frame3)-(frame3)-(50%frame3+50%frame4) |
9th November 2011, 22:36 | #10774 | Link | |
Registered Developer
Join Date: Sep 2006
Posts: 9,140
|
Quote:
(Just to avoid misunderstandings: All of the above only applies to native video content. Native movie content is a whole different thing.) That is the solution I had in mind. Just haven't decided yet whether I'll blend in linear or gamma light. Linear light should be better, but it's also more expensive. |
|
9th November 2011, 22:42 | #10775 | Link |
Registered User
Join Date: Apr 2008
Posts: 418
|
I read up on DXVA deinterlacing and it uses the deinterlacing your GPU supports (instead of having a native deinterlacing method). DXVA can even support 3:2 pulldown if it is implemented in the renderer. Imo best feature EVAR!
Last edited by Gser; 9th November 2011 at 22:54. |
9th November 2011, 23:17 | #10777 | Link |
Registered User
Join Date: Dec 2002
Posts: 5,565
|
There's a file called "activate debug mode.bat" in the madVR folder. It will exchange "madvr.ax" with "madvr [debug].ax". MadVR will now create a log file on your desktop. Zip it and upload it to something like mediafire.
|
9th November 2011, 23:26 | #10778 | Link | |
Registered User
Join Date: Apr 2010
Posts: 163
|
Quote:
__________________
4gb DDR3/AMD Fx6300 Windows 8.1, GT750ti, Auzentech Meridian 7.1, LG 42LN5400 4:4:4 1080P LCD TV |
|
9th November 2011, 23:28 | #10779 | Link | |
Registered User
Join Date: Oct 2011
Posts: 8
|
Quote:
http://www.mediafire.com/?hqbsevc64eebesb Thanks. |
|
9th November 2011, 23:44 | #10780 | Link |
Registered User
Join Date: Aug 2011
Posts: 98
|
I meant how people hyped it up so much like that, saying how much it would beat the i5 2500k and it was one of the most disappointing things ever and underperformed compared to their Intel counterparts by heaps Well enough of that, this is hardly the place for hardware talk.
Last edited by golagoda; 9th November 2011 at 23:49. |
Tags |
direct compute, dithering, error diffusion, madvr, ngu, nnedi3, quality, renderer, scaling, uhd upscaling, upsampling |
|
|