[mythtv-users] low-power FE: AMD now better than Nvidia?

[Warpme]

On 23/05/14 23:29, merc1984 at f-m.fm wrote:
> On Fri, May 23, 2014, at 12:47, Warpme wrote:
>> Recently I do some tests on Brazos (E350) and I must say for me FOSS AMD
>> support for video reaches nearly Nvidia level (which is reference for
>> me).
>> HW video decode works for all content I tested. But what surprise me is
>> DI quality.
>> Visually I can't see difference between Nvidia 2xTemporal and AMD 2xHW.
>> What surprise!
>> All this was on FOSS without any monolithic BLOBs like Nvidia
>> proprietary drivers.
>> Moreover, AMD hasn't some visual glitches with notification GL
>> animations which Nvidia has so I can say AMD works for me even better
>> than Nvidia - go figure?!??!
>> Taking into account that Brazos can do full DTS-HD MA and Dolby TrueHD
>> bit streaming - it can become to be platform of choice for low-power FE.
>> I'm really surprised....
> I've been running Intel Haswell (E3-1265L v3) with
> xserver-xorg-video-intel almost since Linus gave nVidia the finger.
> Full opengl menus, and hardware decode.
>
> Flawless, except when Debian decides to install fglrx drivers during
> dist-upgrade, which step on it...
>
> My whole (large) server draws 110w when I'm watching recordings... and
> that's with 3 WD Green 2TB drives, a WD Black 512GB, and Teac BD
> slot-loader.  Not to mention, silent.  Understand, this is a Silverstone
> GD02-MT
> (http://www.overclockers.com/silverstone-grandia-gd02mt-case-review/)
> with Asus ATX mobo.
>
>
>
>
Oh - give me break with Intel as video playback solution.
I recently tested IntelNUC 53427 together with ION1 & Brazos1.
While ION/Brazos are going head-to-head (they work exactly as I'm 
expecting:
(1) all my content is hw decoded
(2) display refresh rates are correctly changing
(3) have x2 DI

In this context Intel lags far. It looks following:
(1) on Intel seems to be working now. IIRC only for:

libva info: VA-API version 0.35.1
libva info: va_getDriverName() returns 0
libva info: Trying to open /usr/lib/va/i965_drv_video.so
libva info: Found init function __vaDriverInit_0_35
libva info: va_openDriver() returns 0
vainfo: VA-API version: 0.35 (libva 1.3.1)
vainfo: Driver version: Intel i965 driver - 1.3.1
vainfo: Supported profile and entrypoints
       VAProfileMPEG2Simple            : VAEntrypointVLD
       VAProfileMPEG2Simple            : VAEntrypointEncSlice
       VAProfileMPEG2Main              : VAEntrypointVLD
       VAProfileMPEG2Main              : VAEntrypointEncSlice
       VAProfileH264ConstrainedBaseline: VAEntrypointVLD
       VAProfileH264ConstrainedBaseline: VAEntrypointEncSlice
       VAProfileH264Main               : VAEntrypointVLD
       VAProfileH264Main               : VAEntrypointEncSlice
       VAProfileH264High               : VAEntrypointVLD
       VAProfileH264High               : VAEntrypointEncSlice
       VAProfileVC1Simple              : VAEntrypointVLD
       VAProfileVC1Main                : VAEntrypointVLD
       VAProfileVC1Advanced            : VAEntrypointVLD
       VAProfileNone                   : VAEntrypointVideoProc
       VAProfileJPEGBaseline           : VAEntrypointVLD

Brazos gives me:

Decoder capabilities:

name               level macbs width height
-------------------------------------------
MPEG1                 0  9216  2048  1152
MPEG2_SIMPLE          3  9216  2048  1152
MPEG2_MAIN            3  9216  2048  1152
H264_BASELINE        41  9216  2048  1152
H264_MAIN            41  9216  2048  1152
H264_HIGH            41  9216  2048  1152
VC1_ADVANCED          4  9216  2048  1152
MPEG4_PART2_SP        3  9216  2048  1152
MPEG4_PART2_ASP       5  9216  2048  1152

(2) playback in MythTV is broken (video area is shifted down by 1/4 of 
screen). Beside that IIRC fractional refresh rates are still unclear for 
me on Intel.

(3) is total mess (no x2 DI, what works is still worse than x1 DI on 
Nvidia/AMD)

While (1) is OK, (2) is MythTV bug - (3) is total unacceptable for me.
It is even worst - Intel's VPP seem to be broken. Haswell architecture 
makes it even difficult to workaround.
For me simply: Intel as video playback solution is mess.

Intel as server platform however shines.

Here are some numbers from my server:
It has 8G RAM, 12T storage. I'm using efficient Intel main board (Intel 
DH67BL) and low power i3 Intel CPU (3220T).
I was able to achieve following server power consumption:

Idle: 270-280mA at 220V AC = 59-62W
recording playback on one frontend: 290mA = 64W
1 LiveTV session on 1 frontend: 300mA = 66W
1 playback and 1 LiveTV :320mA = 70W
1 playback, 1 recording and 1 LiveTV: 335mA = 74W
1 playback, 2 recordings and 1 LiveTV: 350mA = 77W
1 playback, 3 recordings and 1 LiveTV: 360mA = 79W
1 playback, 4 recordings and 1 LiveTV: 360mA = 79W
16 concurrent HD recordings: 370mA = 81W
CPU load for 16 concurrent HD recordimgs is 35-38% measured on active 
core (it is normalized to full load on both cores = 200%)

Well - here in Poland  business case for switching from AMD 235e to 
3220T is 3.5-4.5y

br