NVEncC by rigaya - NVIDIA GPU encoding

[Selur]

Not that command line, but the help:

NVEncC (x86) 1.01 by rigaya, build Feb 14 2015 16:44:05
  avi reader: disabled
  avs reader: enabled
  vpy reader: enabled

使用方法: NVEncC.exe [オプション] -i <入力ファイル名> -o <出力ファイル名>

入力は avs, raw YUV, YUV4MPEG2(y4m)です。
Raw入力時には, fps, input-res の指定も必要です。

出力フォーマットは raw H.264/AVC または H.265/HEVC ESです。

実行例:
  NVEncC -i "<avsfilename>" -o "<outfilename>"
  avs2pipemod -y4mp "<avsfile>" | NVEncC --y4m -i - -o "<outfilename>"

オプション: 
-h,-? --help                      ヘルプの表示
-v,--version                      バージョン情報の表示
   --check-hw                     NVEncが使用可能かチェック
   --check-features               NVEncの使用可能な機能を表示
   --check-environment            NVEncの認識している環境情報を表示

-i,--input <filename>             入力ファイル名の指定
-o,--output <filename>            出力ファイル名の指定

 入力フォーマット (指定のない場合、拡張子から自動判定)
   --raw                          rawとしてファイルを読み込み
   --y4m                          y4mとしてファイルを読み込み
   --avs                          avsとしてファイルを読み込み
   --vpy                          vpyとしてファイルを読み込み
   --vpy-mt                       vpy(mt)としてファイルを読み込み

   --input-res <int>x<int>        入力解像度
   --crop <int>,<int>,<int>,<int> 左、上、右、下の切り落とし画素数
-f,--fps <int>/<int> or <float>   フレームレートの指定

-c,--codec <string>               出力コーデックの指定
                                    h264 (or avc), h265 (or hevc)
   --profile <string>             プロファイルの指定
                                    H.264: baseline, main, high(デフォルト)
   --level <string>               Levelの指定
   --sar <int>:<int>              SAR(PAR, 画素比)の指定
   --dar <int>:<int>              DAR(画面比)の指定

   --cqp <int> or                 固定量子化量でエンコード
         <int>:<int>:<int>          デフォルト: <I>:<P>:<B>=<20>:<23>:<25>
   --vbr <int>                    可変ビットレートのビットレート (kbps)
   --cbr <int>                    固定ビットレートのビットレート (kbps)
                                    デフォルト 7500 kbps

   --max-bitrate <int>            最大ビットレート(kbps) / デフォルト: 17500 kbps
   --gop-len <int>                GOPのフレーム数 / デフォルト: 0 frames (自動)-b,--bframes <int>                連続Bフレーム数 / デフォルト 3 フレーム
   --ref <int>                    参照距離 / デフォルト 3 フレーム
   --mv-precision <string>        動きベクトル精度 / デフォルト: Q-pel
                                    Q-pel    … 1/4画素精度 (高精度)
                                    half-pel … 1/2画素精度
                                    full-pel … 1  画素精度 (低精度)

H.264/AVC
   --interlaced                   インタレ保持エンコ
   --cabac                        CABACを使用する
   --cavlc                        CAVLCを使用する
   --deblock                      デブロックフィルタを有効にする
   --no-deblock                   デブロックフィルタを無効にする
   --fullrange                    fullrangeの指定
   --videoformat <string>         undef, ntsc, component, pal, secam, mac
                                   default: undef
   --colormatrix <string>         undef, auto, bt709, smpte170m, bt470bg
                                  smpte240m, YCgCo, fcc, GBR
                                   default: undef
   --colorprim <string>           undef, auto, bt709, smpte170m, bt470m
                                  bt470bg, smpte240m, film
                                   default: undef
   --transfer <string>            undef, auto, bt709, smpte170m, bt470m
                                  bt470bg, smpte240m, linear, log100, log316
                                   default: undef

H.265/HEVC
   --cu-max <int>                 CUの最大サイズを指定する
   --cu-min  <int>                CUの最小サイズを指定する
                                    8, 16, 32 を指定可能

does mention '--input' and '---output', which both do not work here.

[MaverickTse]

あらあら(oh my~)
I'll report to rigaya.
Seems to be a documentation miss.

[MaverickTse]

try
--input-file
and
--output-file

instead

[Selur]

Those two do work. Thanks for looking into it.

[MaverickTse]

For rigaya's stuff, you can take a look at the bundled source code next time when you bump into problem. Thought some message text are Japanese, most are just C++ in English

[stax76]

Has anybody else problems with German 'Umlaute' (ä ö ü)?

[Lupissimo]

Double posting. sorry!

[Lupissimo]

I have been using Staxrip for a long time (not being familiar with the encoding commands and settings!). Since the advent of Staxrip 1.2.0... I wanted to use the NVEnc feature for both x264 and x265. I installed a passive Zotac GTX750 (Marvell) and have done numerous 1080p, HDTV 720p 50Hz, and SD encodes. Compared to the "normal" x264 NVEnc runs up to 7..8x faster. Files size is bigger e.g. SD File 2.238GB vs 1.645GB. After 2 reencodes the size goes down to 1.754 GB. For that I have used the Staxrip default command line:

"%app:avs2pipemod%" -y4mp "%avs_file%" | "%app:NVEncC%" --y4m -c h264 -i - -o "%encoder_out_file%"

With the "normal" x264 I have been using the Staxrip settings: SD: Quality 21 high, preset medium, Tune Film and device disabled (I use a hardware SAP mede8er 600x3d for playback) ; for HD: I change the quality to 22 higher.

Is it possible to change the NVEnc command line for these 2 settings to come close to the 2 settings for the x264?? If so please give me a hint what needs to be changed!

[MaverickTse]

Not that command line, but the help:

NVEncC (x86) 1.01 by rigaya, build Feb 14 2015 16:44:05
  avi reader: disabled
  avs reader: enabled
  vpy reader: enabled

使用方法: NVEncC.exe [オプション] -i <入力ファイル名> -o <出力ファイル名>

入力は avs, raw YUV, YUV4MPEG2(y4m)です。
Raw入力時には, fps, input-res の指定も必要です。

出力フォーマットは raw H.264/AVC または H.265/HEVC ESです。

実行例:
  NVEncC -i "<avsfilename>" -o "<outfilename>"
  avs2pipemod -y4mp "<avsfile>" | NVEncC --y4m -i - -o "<outfilename>"

オプション: 
-h,-? --help                      ヘルプの表示
-v,--version                      バージョン情報の表示
   --check-hw                     NVEncが使用可能かチェック
   --check-features               NVEncの使用可能な機能を表示
   --check-environment            NVEncの認識している環境情報を表示

-i,--input <filename>             入力ファイル名の指定
-o,--output <filename>            出力ファイル名の指定

 入力フォーマット (指定のない場合、拡張子から自動判定)
   --raw                          rawとしてファイルを読み込み
   --y4m                          y4mとしてファイルを読み込み
   --avs                          avsとしてファイルを読み込み
   --vpy                          vpyとしてファイルを読み込み
   --vpy-mt                       vpy(mt)としてファイルを読み込み

   --input-res <int>x<int>        入力解像度
   --crop <int>,<int>,<int>,<int> 左、上、右、下の切り落とし画素数
-f,--fps <int>/<int> or <float>   フレームレートの指定

-c,--codec <string>               出力コーデックの指定
                                    h264 (or avc), h265 (or hevc)
   --profile <string>             プロファイルの指定
                                    H.264: baseline, main, high(デフォルト)
   --level <string>               Levelの指定
   --sar <int>:<int>              SAR(PAR, 画素比)の指定
   --dar <int>:<int>              DAR(画面比)の指定

   --cqp <int> or                 固定量子化量でエンコード
         <int>:<int>:<int>          デフォルト: <I>:<P>:<B>=<20>:<23>:<25>
   --vbr <int>                    可変ビットレートのビットレート (kbps)
   --cbr <int>                    固定ビットレートのビットレート (kbps)
                                    デフォルト 7500 kbps

   --max-bitrate <int>            最大ビットレート(kbps) / デフォルト: 17500 kbps
   --gop-len <int>                GOPのフレーム数 / デフォルト: 0 frames (自動)-b,--bframes <int>                連続Bフレーム数 / デフォルト 3 フレーム
   --ref <int>                    参照距離 / デフォルト 3 フレーム
   --mv-precision <string>        動きベクトル精度 / デフォルト: Q-pel
                                    Q-pel    … 1/4画素精度 (高精度)
                                    half-pel … 1/2画素精度
                                    full-pel … 1  画素精度 (低精度)

H.264/AVC
   --interlaced                   インタレ保持エンコ
   --cabac                        CABACを使用する
   --cavlc                        CAVLCを使用する
   --deblock                      デブロックフィルタを有効にする
   --no-deblock                   デブロックフィルタを無効にする
   --fullrange                    fullrangeの指定
   --videoformat <string>         undef, ntsc, component, pal, secam, mac
                                   default: undef
   --colormatrix <string>         undef, auto, bt709, smpte170m, bt470bg
                                  smpte240m, YCgCo, fcc, GBR
                                   default: undef
   --colorprim <string>           undef, auto, bt709, smpte170m, bt470m
                                  bt470bg, smpte240m, film
                                   default: undef
   --transfer <string>            undef, auto, bt709, smpte170m, bt470m
                                  bt470bg, smpte240m, linear, log100, log316
                                   default: undef

H.265/HEVC
   --cu-max <int>                 CUの最大サイズを指定する
   --cu-min  <int>                CUの最小サイズを指定する
                                    8, 16, 32 を指定可能

does mention '--input' and '---output', which both do not work here.

Fixed :
http://rigaya34589.blog135.fc2.com/blog-entry-590.html

but try to see if --input-file --output-file work

[Selur]

but try to see if --input-file --output-file work

already did see #34 and they did

[Selur]

Argh, seems like 'NVEncC --check-features' freezes on systems without an NVIDIA card.

Cu Selur

[Selur]

btw. NVEncC 0.8 is available and a small warning I noticed that when I call '-check-hw' instead of '--check-hw', NVEncC crashes. Seems like the parameter parser isn't that robust.

2015.04.29 (1.08)
[共通]
・環境によって例外:0xc0000096で落ちることがあるのを回避。
[NVEnc]
・音声エンコ前後にバッチ処理を行う機能を追加。
[NVEncC]
・64bit版にavsリーダーを追加。

2015.04.16 (1.07)
[NVEnc]
※要NVEnc.ini更新
・neroを使用すると途中でフリーズする問題を修正。
・いくつかの音声エンコーダの設定を追加。

2015.04.12 (1.06)
[共通]
・VBVバッファサイズを指定するオプションを追加。
・Bluray用出力を行うオプションを追加。
  Bluray用に出力する際には、必ず指定してください。

translated:

2015.04.29 (1.08)
[Common]
- By the environment Exception: avoid it may fall in the 0xc0000096.
[NVEnc]
Add a function to perform batch processing and audio encoder around.
[NVEncC]
Add avs leader in · 64bit version.

2015.04.16 (1.07)
[NVEnc]
※ required NVEnc.ini update
Fixed a problem to freeze the nero · on the way when you use.
- Add some of the settings of the speech encoder.

2015.04.12 (1.06)
[Common]
- Added an option to specify the VBV buffer size.
- Added option to Bluray for output.
   When output for Bluray, please be sure to specify.

[sajtipajti]

Hi!
I have a GeForce 840M card, and when I run the NvencC program (v1.08), I get an unsupported device error.
If I run "NVEncC.exe --check-features > x", I get:

Environment Info
OS : Windows 7 (x64)
CPU: Intel Core i3-4030U @ 1.90GHz (2C/4T)
GPU: GeForce 840M (3 EU) @ 1124 MHz (347.52)
RAM: Total 4023 MB / Used 1395 MB

No NVEnc support.

How can I solve this problem?
TY for any help.

[Selur]

Did you install the latest drivers from NVIDIA?

[sajtipajti]

With latest driver:

Environment Info
OS : Windows 7 (x64)
CPU: Intel Core i3-4030U @ 1.90GHz (2C/4T)
GPU: GeForce 840M (3 EU) @ 1124 MHz (353.06)
RAM: Total 4023 MB / Used 1148 MB

No NVEnc support.

Unfortunately nothing changed

[MaverickTse]

Hi!
I have a GeForce 840M card, and when I run the NvencC program (v1.08), I get an unsupported device error.
If I run "NVEncC.exe --check-features > x", I get:

Environment Info
OS : Windows 7 (x64)
CPU: Intel Core i3-4030U @ 1.90GHz (2C/4T)
GPU: GeForce 840M (3 EU) @ 1124 MHz (347.52)
RAM: Total 4023 MB / Used 1395 MB

No NVEnc support.

How can I solve this problem?
TY for any help.

Ref: https://www.reddit.com/r/nvidia/comments/36hryu/geforce_840m_4gb_nvenc_shadowplay/

Hardware constraint...

[Selur]

-c,--codec <string>               set ouput codec
                                    h264 (or avc), h265 (or hevc)

atm. only hevc does work, when h265 is used one ends up with:

Failed to Initialize the encoder
.: 8 (NVENC indicates that one or more of the parameter passed to the API call is invalid.)

Cu Selur

[ValenNC]

-c,--codec <string>               set ouput codec
                                    h264 (or avc), h265 (or hevc)

atm. only hevc does work, when h265 is used one ends up with:

Failed to Initialize the encoder
.: 8 (NVENC indicates that one or more of the parameter passed to the API call is invalid.)

Cu Selur

Just an update from Selur's and my testing, it does work with both HEVC/H265, it was a level 1 argument that caused the parameter error, changing to level 5 worked.

I've been using both StaxRIP and Hybrid with a GTX 970 and h.265 encoding, NVENC enables 1080p h.265 encoding at 180 fps vs x.265 where I get 5 Fps on the same 1080p vid and have seen 450fps with 720p. It's strange that encoding doesn't seem to max out either CPU or GPU with GPU sitting anywhere from 8-23% and CPU 25-60% max. I've yet to figure out where the bottleneck is, it's not HDD, I'm guessing PCI-E Bus. I'm using a i7 950 3Ghz x58 system with 4 Raid0 volumes and win 10x64.

Filesize on the h.264 1080p was 1,755,846KB -> h.265 1080p it dropped to 1,072,632KB with muxed audio. My eyes can't see a difference. There is a difference in seeking, there's a bit of a pause before video begins to play after a seek, I'm guess that's just the GPU having to work harder.

[ValenNC]

I have been using Staxrip for a long time (not being familiar with the encoding commands and settings!). Since the advent of Staxrip 1.2.0... I wanted to use the NVEnc feature for both x264 and x265. I installed a passive Zotac GTX750 (Marvell) and have done numerous 1080p, HDTV 720p 50Hz, and SD encodes. Compared to the "normal" x264 NVEnc runs up to 7..8x faster. Files size is bigger e.g. SD File 2.238GB vs 1.645GB. After 2 reencodes the size goes down to 1.754 GB. For that I have used the Staxrip default command line:

"%app:avs2pipemod%" -y4mp "%avs_file%" | "%app:NVEncC%" --y4m -c h264 -i - -o "%encoder_out_file%"

With the "normal" x264 I have been using the Staxrip settings: SD: Quality 21 high, preset medium, Tune Film and device disabled (I use a hardware SAP mede8er 600x3d for playback) ; for HD: I change the quality to 22 higher.

Is it possible to change the NVEnc command line for these 2 settings to come close to the 2 settings for the x264?? If so please give me a hint what needs to be changed!

Install the beta version of StaxRIP, it's much easier to adjust NVENC encoding parameters and you can also choose command line NVENC and enter them yourself. Also NVENC 1.10 is out now.

[MaverickTse]

Note that the current NVENC(The Nvidia's API) uses NO B-frame. That's why it is so fast, but compression rate is not always satisfactory.

[ValenNC]

Note that the current NVENC(The Nvidia's API) uses NO B-frame. That's why it is so fast, but compression rate is not always satisfactory.

I was assuming the 0 B frames was a hardware limitation, that would be great it it's just software, if so they need to throw more chefs in the software kitchen and get that resolved and ad HW de-interlacing and resizing while they're there. I've yet to see it not at least do 40% compression with h264 media. I've been using it to take 720p h264 5.1 1Gb media down to 720x400 - 5.1 300K files that look just as good.

[MaverickTse]

I'm not certain if it is HW or SW limitation, just saw a post on Nvidia forum where someone ask about this.
The answer is simply "B frame not currently supported"

but I assume Nvidia can utilize the CUDA cores even if there is no dedicated HW to handle B-frame encode.

[ValenNC]

I'll bet it is a hardware limitation with the first iteration of HW based h.265, otherwise it would be pretty sad to have your software lagging on a major feature, then again one never knows in this day and age where get it out to market and get that $$ fast seems to be the rule vs exception.

I did find out why GPU /CPU performance isn't maxing out, I wasn't monitoring a whole other aspect of GPU utilization... that being the Video Engine where much of the work is done with video acceleration/codecs. I guess you could think of the Video Engine as the 2D mode vs the 3D rending mode usually monitored. GPU-Z will report utilization for both.

[MaverickTse]

FYI, HW(or hybrid) resizing is there on many existing video card, but limit to bilinear resize.
deinterlacing, I'm not sure if it is HW, but there are decoding-related API function in CUDA SDK with deinterlacing option.

Intel Media SDK also has these, but also up to Intel to decide which work goes to HW/iGPU, which goes to CPU

[Selur]

deinterlacing on NVIDIA cards is done by the vpu (like with the encoding there's normally no real usage of cuda or the gpu for that matter involved); no clue for Intel, but I also guess that they use a chip for this.

[flying2ndsky]

Hello, guys..

ii have a question about crop usage in nvencc.
following is my commands in cmd line, but the returned was error message.

Y:>NVEncC64 --codec h264 --bframes 5 --ref 9 --max-bitrate 80000 --cqp 20:25:30 -i temp.mkv -o temp.264 --crop 0,92,12
80,628
Invalid crop value.: 0x0, Crop [0,92,1280,628]

so, i changed the crop option slightly with [ ] , as below..

Y:>NVEncC64 --codec h264 --bframes 5 --ref 9 --max-bitrate 80000 --cqp 20:25:30 -i temp.mkv -o temp.264 --crop [0,92,1
280,628]
Invalid value. crop : [0,92,1280,628]

but, the error message was returned too.

what should i do to fix this crop option in commands ?
Is there anyone to give me a tip to fix it out ?

thanks in advance.

[poisondeathray]

You cannot use mkv input directly. Only raw,y4m,vpy,avs

Crop works for me on a 1920x1080 avs source - so the crop isn't the problem

or you can pipe using ffmpeg to decode, yuv4mpegpipe to NVEncC64

[flying2ndsky]

You cannot use mkv input directly. Only raw,y4m,vpy,avs

Crop works for me on a 1920x1080 avs source - so the crop isn't the problem

or you can pipe using ffmpeg to decode, yuv4mpegpipe to NVEncC64

thank you for reply

so, is the crop option term, --crop 0,92,1280,628, in right ?


i've just reruned commands which were consisting of qsvencc decoding +nvencc encoding with piping method.

below is the result of that.

Z:\>G:\test\TOOLS\qsvencc64 --avqsv -i temp.mkv -o - --crop 0,92,1280,628 -c raw | NVEncC64 --codec h264 --bframes 5 --
ref 9 --max-bitrate 80000 --cqp 20:25:30 --y4m -i - -o temp.264
###   encoding process with crop option in qsvencc's decoding part   ###

Failed to get required buffer size for vpp. : invalid video parameters.

QSVEncC.exe finished with error!
Failed to open input file.

Z:\>G:\test\TOOLS\qsvencc64 --avqsv -i temp.mkv -o - -c raw | NVEncC64 --codec h264 --bframes 5 --ref 9 --max-bitrate 8
0000 --cqp 20:25:30 --crop 0,92,1280,628 --y4m -i - -o temp.264
###   encoding process with crop option in nvencc's encoding part   ###

QSVEnc 2.18 (x64), based on Intel(R) Media SDK Encoding Sample 6,0,0,98947
OS             Windows 7 (x64)
CPU Info       Intel Core i7-4700HQ @ 2.40GHz [TB: 3.20GHz] (4C/8T) <Haswell>
GPU Info       Intel HD Graphics 4600 (20EU) 200-1200MHz [47W] (10.18.14.4264)
Media SDK      QuickSyncVideo (hardware encoder), 1st GPU, API v1.16
Async Depth    4 frames
Buffer Memory  system, 1 input buffer, 9 work buffer
Input Info     avcodec video: H.264/AVC, 1280x720, 30000/1001 fps
Output         1280x720p 1:1 29.970fps (30000/1001fps)
Invalid crop value.: 1280x720, Crop [0,92,1280,628]
unknown error.
undefined behavior.
undefined behavior.
Error in encoding pipeline. : undefined behavior.

error at encode thread.

encoded 0 frames, 0.00 fps, -nan(ind) kbps, 0.00 MB
encode time 0:00:10 / CPU Usage: 0.02

QSVEncC.exe finished with error!

Z:\>G:\test\TOOLS\qsvencc64 --avqsv -i temp.mkv -o - -c raw | NVEncC64 --codec h264 --bframes 5 --ref 9 --max-bitrate 8
0000 --cqp 20:25:30 --y4m -i - -o temp.264 --crop 0,92,1280,628
###   encoding process with crop option in nvencc's encoding part, 2nd   ###

QSVEnc 2.18 (x64), based on Intel(R) Media SDK Encoding Sample 6,0,0,98947
OS             Windows 7 (x64)
CPU Info       Intel Core i7-4700HQ @ 2.40GHz [TB: 3.20GHz] (4C/8T) <Haswell>
GPU Info       Intel HD Graphics 4600 (20EU) 200-1200MHz [47W] (10.18.14.4264)
Media SDK      QuickSyncVideo (hardware encoder), 1st GPU, API v1.16
Async Depth    4 frames
Buffer Memory  system, 1 input buffer, 9 work buffer
Input Info     avcodec video: H.264/AVC, 1280x720, 30000/1001 fps
Output         1280x720p 1:1 29.970fps (30000/1001fps)
Invalid crop value.: 1280x720, Crop [0,92,1280,628]
unknown error.
undefined behavior.
undefined behavior.
Error in encoding pipeline. : undefined behavior.

error at encode thread.

encoded 0 frames, 0.00 fps, -nan(ind) kbps, 0.00 MB
encode time 0:00:10 / CPU Usage: 0.00

QSVEncC.exe finished with error!

Z:\>G:\test\TOOLS\qsvencc64 --avqsv -i temp.mkv -o - -c raw | NVEncC64 --codec h264 --bframes 5 --ref 9 --max-bitrate 8
0000 --cqp 20:25:30 --y4m -i - -o temp.264
###   encoding process without crop option  ###


QSVEnc 2.18 (x64), based on Intel(R) Media SDK Encoding Sample 6,0,0,98947
OS             Windows 7 (x64)
CPU Info       Intel Core i7-4700HQ @ 2.40GHz [TB: 3.20GHz] (4C/8T) <Haswell>
GPU Info       Intel HD Graphics 4600 (20EU) 200-1200MHz [47W] (10.18.14.4264)
Media SDK      QuickSyncVideo (hardware encoder), 1st GPU, API v1.16
Async Depth    4 frames
Buffer Memory  system, 1 input buffer, 9 work buffer
Input Info     avcodec video: H.264/AVC, 1280x720, 30000/1001 fps
Output         1280x720p 1:1 29.970fps (30000/1001fps)
Max B frames are 4 frames.
NVEnc 1.10 (x64), using NVENC API v5.0
OS Version              Windows 7 (x64)
CPU                     Intel Core i7-4700HQ @ 2.40GHz [TB: 3.20GHz] (4C/8T)
GPU                     GeForce GTX 860M (5 EU) @ 1019 MHz (355.60)
Input Buffers           CUDA, 16 frames
Input Info              y4m (yv12) -> nv12 [AVX2], 1280x720, 30000/1001 fps
Output Info             H.264/AVC high
                        1280x720p 1:1 29.970fps (30000/1001fps)
Rate Control            CQP
CQP                     I:20  P:25  B:30
GOP length              300 frames
B frames                4 frames
Ref frames              9 frames
MV Quality              Q-pel
CABAC/deblock           cabac / on
18698 frames: 522.68 fps, 805.18 kbps  .56 kb/s, remain 0:06:46  --->  in processing of encoding with 522.68fps(qsvencc' decoding, perhaps), 805.15fps(nvencc' encoding, perhaps) speed.(without crop option)

where is the fault in my commands for appling the crop option in qsvencc or nvencc ?

how can i fix it ?

thanks in advance.

[flying2ndsky]

Here, i'm adding the result of the re-running with yuv4mpegpipe

>ffmpeg -i temp.mkv -threads 12 -pix_fmt yuv420p -f yuv4mpegpipe - | NVEncC64 --codec h264 --bframes 5 --ref 9 --max
-bitrate 80000 --cqp 20:23:28 --crop 0,92,1280,628 --y4m -i - -o temp.264
##############      encoding with crop option        #############################

ffmpeg version N-72805-g913685f Copyright (c) 2000-2015 the FFmpeg developers
  built with gcc 4.9.2 (GCC)
  configuration: --enable-gpl --enable-version3 --disable-w32threads --enable-avisynth --enable-bzlib --enable-fontconfig --enable
-frei0r --enable-gnutls --enable-iconv --enable-libass --enable-libbluray --enable-libbs2b --enable-libcaca --enable-libdcadec --e
nable-libfreetype --enable-libgme --enable-libgsm --enable-libilbc --enable-libmodplug --enable-libmfx --enable-libmp3lame --enabl
e-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-libopus --enable-librtmp --enable-libschroedinger --e
nable-libsoxr --enable-libspeex --enable-libtheora --enable-libtwolame --enable-libvidstab --enable-libvo-aacenc --enable-libvo-am
rwbenc --enable-libvorbis --enable-libvpx --enable-libwavpack --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxavs
--enable-libxvid --enable-lzma --enable-decklink --enable-zlib
  libavutil      54. 27.100 / 54. 27.100
  libavcodec     56. 41.100 / 56. 41.100
  libavformat    56. 36.100 / 56. 36.100
  libavdevice    56.  4.100 / 56.  4.100
  libavfilter     5. 16.101 /  5. 16.101
  libswscale      3.  1.101 /  3.  1.101
  libswresample   1.  2.100 /  1.  2.100
  libpostproc    53.  3.100 / 53.  3.100
Input #0, matroska,webm, from 'temp.mkv':
  Metadata:
    COMPATIBLE_BRANDS: isomiso2avc1mp41
    MAJOR_BRAND     : isom
    MINOR_VERSION   : 512
    ENCODER         : Lavf56.36.100
  Duration: 02:06:52.48, start: 0.000000, bitrate: 3185 kb/s
    Stream #0:0(und): Video: h264 (Constrained Baseline), yuv420p, 1280x720 [SAR 1:1 DAR 16:9], 29.97 fps, 29.97 tbr, 1k tbn, 59.9
4 tbc (default)
    Metadata:
      CREATION_TIME   : 1970-01-01 00:00:00
      LANGUAGE        : und
      HANDLER_NAME    : VideoHandler
    Stream #0:1(und): Audio: aac (LC), 44100 Hz, stereo, fltp (default)
    Metadata:
      CREATION_TIME   : 1970-01-01 00:00:00
      LANGUAGE        : und
      HANDLER_NAME    : SoundHandler
Output #0, yuv4mpegpipe, to 'pipe:':
  Metadata:
    COMPATIBLE_BRANDS: isomiso2avc1mp41
    MAJOR_BRAND     : isom
    MINOR_VERSION   : 512
    encoder         : Lavf56.36.100
    Stream #0:0(und): Video: rawvideo (I420 / 0x30323449), yuv420p, 1280x720 [SAR 1:1 DAR 16:9], q=2-31, 200 kb/s, 29.97 fps, 29.9
7 tbn, 29.97 tbc (default)
    Metadata:
      CREATION_TIME   : 1970-01-01 00:00:00
      LANGUAGE        : und
      HANDLER_NAME    : VideoHandler
      encoder         : Lavc56.41.100 rawvideo
Stream mapping:
  Stream #0:0 -> #0:0 (h264 (native) -> rawvideo (native))
Press [q] to stop, [?] for help
[yuv4mpegpipe @ 0000000005c829a0] Encoder did not produce proper pts, making some up.
Invalid crop value.: 1280x720, Crop [0,92,1280,628]
av_interleaved_write_frame(): Broken pipe
frame=    1 fps=0.0 q=0.0 Lsize=    1350kB time=00:00:00.03 bitrate=331458.3kbits/s
video:0kB audio:0kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 1439974.000000%
Conversion failed!

>ffmpeg -i temp.mkv -threads 12 -pix_fmt yuv420p -f yuv4mpegpipe - | NVEncC64 --codec h264 --bframes 5 --ref 9 --max
-bitrate 80000 --cqp 20:23:28 --y4m -i - -o temp.264
##############      encoding without crop option        #############################

ffmpeg version N-72805-g913685f Copyright (c) 2000-2015 the FFmpeg developers
  built with gcc 4.9.2 (GCC)
  configuration: --enable-gpl --enable-version3 --disable-w32threads --enable-avisynth --enable-bzlib --enable-fontconfig --enable
-frei0r --enable-gnutls --enable-iconv --enable-libass --enable-libbluray --enable-libbs2b --enable-libcaca --enable-libdcadec --e
nable-libfreetype --enable-libgme --enable-libgsm --enable-libilbc --enable-libmodplug --enable-libmfx --enable-libmp3lame --enabl
e-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-libopus --enable-librtmp --enable-libschroedinger --e
nable-libsoxr --enable-libspeex --enable-libtheora --enable-libtwolame --enable-libvidstab --enable-libvo-aacenc --enable-libvo-am
rwbenc --enable-libvorbis --enable-libvpx --enable-libwavpack --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxavs
--enable-libxvid --enable-lzma --enable-decklink --enable-zlib
  libavutil      54. 27.100 / 54. 27.100
  libavcodec     56. 41.100 / 56. 41.100
  libavformat    56. 36.100 / 56. 36.100
  libavdevice    56.  4.100 / 56.  4.100
  libavfilter     5. 16.101 /  5. 16.101
  libswscale      3.  1.101 /  3.  1.101
  libswresample   1.  2.100 /  1.  2.100
  libpostproc    53.  3.100 / 53.  3.100
Input #0, matroska,webm, from 'temp.mkv':
  Metadata:
    COMPATIBLE_BRANDS: isomiso2avc1mp41
    MAJOR_BRAND     : isom
    MINOR_VERSION   : 512
    ENCODER         : Lavf56.36.100
  Duration: 02:06:52.48, start: 0.000000, bitrate: 3185 kb/s
    Stream #0:0(und): Video: h264 (Constrained Baseline), yuv420p, 1280x720 [SAR 1:1 DAR 16:9], 29.97 fps, 29.97 tbr, 1k tbn, 59.9
4 tbc (default)
    Metadata:
      CREATION_TIME   : 1970-01-01 00:00:00
      LANGUAGE        : und
      HANDLER_NAME    : VideoHandler
    Stream #0:1(und): Audio: aac (LC), 44100 Hz, stereo, fltp (default)
    Metadata:
      CREATION_TIME   : 1970-01-01 00:00:00
      LANGUAGE        : und
      HANDLER_NAME    : SoundHandler
Output #0, yuv4mpegpipe, to 'pipe:':
  Metadata:
    COMPATIBLE_BRANDS: isomiso2avc1mp41
    MAJOR_BRAND     : isom
    MINOR_VERSION   : 512
    encoder         : Lavf56.36.100
    Stream #0:0(und): Video: rawvideo (I420 / 0x30323449), yuv420p, 1280x720 [SAR 1:1 DAR 16:9], q=2-31, 200 kb/s, 29.97 fps, 29.9
7 tbn, 29.97 tbc (default)
    Metadata:
      CREATION_TIME   : 1970-01-01 00:00:00
      LANGUAGE        : und
      HANDLER_NAME    : VideoHandler
      encoder         : Lavc56.41.100 rawvideo
Stream mapping:
  Stream #0:0 -> #0:0 (h264 (native) -> rawvideo (native))
Press [q] to stop, [?] for help
[yuv4mpegpipe @ 0000000005b329a0] Encoder did not produce proper pts, making some up.
Max B frames are 4 frames.
NVEnc 1.10 (x64), using NVENC API v5.0
OS Version              Windows 7 (x64)
CPU                     Intel Core i7-4700HQ @ 2.40GHz [TB: 3.40GHz] (4C/8T)
GPU                     GeForce GTX 860M (5 EU) @ 1019 MHz (355.60)
Input Buffers           CUDA, 16 frames
Input Info              y4m (yv12) -> nv12 [AVX2], 1280x720, 2997/100 fps
Output Info             H.264/AVC high
                        1280x720p 1:1 29.970fps (2997/100fps)
Rate Control            CQP
CQP                     I:20  P:23  B:28
GOP length              300 frames
B frames                4 frames
Ref frames              9 frames
MV Quality              Q-pel
CABAC/deblock           cabac / on
frame=10603 fps=513 q=0.0 size=14314112kB time=00:05:53.78 bitrate=331445.7kbits/s    6997 frames: 525.73 fps, 913.10 kbps   --->>  in processing of encoding in 513fps speed  without crop option

[poisondeathray]

Your crop values are incorrect - if your source video is 1280x720 , then you cannot crop 1280 pixels from the right (you're left with "zero" ). It works on a 1920x1080 mkv source with those values . The syntax is --crop left,top,right,bottom