Benchmark your computer with Handbrake

[Mr. Fox]

Testing a new EUROCOM Sky X9 Desktop Laptop with 6700K and GTX 980 and fully unlocked BIOS from PremaMod.com

HandBrake 0.9.9.5530 - 64bit Version
OS: Microsoft Windows NT 6.1.7601 Service Pack 1
CPU: Intel(R) Core(TM) i7-6700K CPU @ 4.00GHz
Ram: 16334 MB, Screen: 1920x1080
-------------------------------------------
x264 [info]: using SAR=43127/43200
x264 [info]: using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.2 AVX AVX2 FMA3 LZCNT BMI1 BMI2
x264 [info]: profile Main, level 3.0
[09:45:32] reader: done. 1 scr changes
[09:45:33] work: average encoding speed for job is [B][COLOR="Red"][SIZE="4"]330.553314[/SIZE][/COLOR][/B] fps
Encode done!
HandBrake has exited.

Not bad for an air-cooled laptop.

 

[lufeig]

@ 4.3GHz

HandBrake 0.9.9.5530 - 64bit Version
OS: Microsoft Windows NT 6.2.9200.0
CPU: Intel(R) Core(TM) i7-3770K CPU @ 3.50GHz
Ram: 16258 MB, Screen: 1920x1080

x264 [info]: using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.2 AVX
x264 [info]: profile Main, level 3.0
[21:21:44] reader: done. 1 scr changes
[21:21:45] work: average encoding speed for job is [B][U]245.099396[/U][/B] fps

 

[Conroe]

HandBrake 0.9.9.5530 - 64bit Version
OS: Microsoft Windows NT 6.2.9200.0
CPU: Intel(R) Core(TM) i5-6500 CPU @ 3.20GHz

x264 [info]: using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.2 AVX AVX2 FMA3 LZCNT BMI1 BMI2

[19:01:32] work: average encoding speed for job is 188.286392 fps

My old rig with Core2 based Xeon was faster. Maybe I will be able to overclock this soon. My mobo is a Asrock H170M-ITX/ac.

 

[bystander36]

HandBrake 0.10.2.7286 - 64bit Version
OS: Microsoft Windows NT 6.2.9200.0 - 64bit
CPU: Intel(R) Core(TM) i7-5820K CPU @ 4.375GHz
Ram: 16280 MB, 
[19:16:42] work: average encoding speed for job is [B]477.024353 fps[/B]

6 cores/12 threads has advantages.

 

[Deders]

i7 6700K @4.6

DDR4 3200 16-16-16-2T

HandBrake 0.9.9.5530 - 64bit Version
OS: Microsoft Windows NT 6.2.9200.0
CPU: Intel(R) Core(TM) i7-6700K CPU @ 4.00GHz
Ram: 16231 MB, Screen: 2560x1080

[03:41:21] reader: done. 1 scr changes
[03:41:21] work: average encoding speed for job is 327.133850 fps
[03:41:22] sync: got 14315 frames, 14315 expected
[03:41:22] render: 14315 frames output, 0 dropped and 0 duped for CFR/PFR
[03:41:22] render: lost time: 0 (0 frames)
[03:41:22] render: gained time: 0 (0 frames) (0 not accounted for)
[03:41:22] h264-decoder done: 14315 frames, 0 decoder errors, 0 drops

Same job with latest version, default android settings...

HandBrake 0.10.2.7286 - 64bit Version
OS: Microsoft Windows NT 6.2.9200.0 - 64bit
CPU: Intel(R) Core(TM) i7-6700K CPU @ 4.00GHz
Ram: 16231 MB,
GPU Information:
Intel(R) HD Graphics 530 - 10.18.15.4274
NVIDIA GeForce GTX 780 - 10.18.13.5906
Screen: 2560x1080

x264 [info]: using SAR=21845/21774
x264 [info]: using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.2 AVX AVX2 FMA3 LZCNT BMI2
x264 [info]: profile Main, level 3.0
[04:01:09] reader: done. 1 scr changes
[04:01:09] work: average encoding speed for job is 385.434723 fps
[04:01:09] sync: got 14315 frames, 14315 expected
[04:01:09] render: 14315 frames output, 0 dropped and 0 duped for CFR/PFR
[04:01:09] render: lost time: 0 (0 frames)
[04:01:09] render: gained time: 0 (0 frames) (0 not accounted for)
[04:01:09] h264-decoder done: 14315 frames, 0 decoder errors, 0 drops

 

[Carfax83]

Over two years ago I benchmarked the exact same file in the OP with my 3930K using the nightly build at the time. This was my score (384.958954 fps)

Now I just benchmarked it again at the same clock speed (4.4ghz) using the latest nightly, but on my Haswell-E machine:

HandBrake Nightly 20151227221722-1e56395-master (2016010201) - 64bit Version
OS: Microsoft Windows NT 6.2.9200.0 - 64bit
CPU: Intel(R) Core(TM) i7-5930K CPU @ 3.50GHz
Ram: 32668 MB
x264 [info]: using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.2 AVX FMA3 AVX2 LZCNT BMI2
x264 [info]: profile Main, level 3.0
[01:11:17] reader: first SCR 0 id 0x2 DTS 0
[01:11:46] reader: done. 1 scr changes
[01:11:47] work: average encoding speed for job is 483.826721 fps

Not bad Picked up almost 100 FPS, courtesy of AVX2 and FMA3.. I must say Haswell is starting to impress.. My 5930K uses only 1.21v at 4.4ghz, whereas my 3930K at that time used 1.285v at the same clock speed, but the Haswell-E is 26% faster than Sandy Bridge-E..

On higher quality presets, the percentage would be even higher.. So it looks like AVX2 and FMA3 are finally coming home to roost Now if only games would start using them....

 

[dclive]

Using the following:
.\HandBrakeCLI.exe -i .\big_buck_bunny_480p_h264.mov -o bbb -Z "Android"

I get:
x264 [info]: using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.2
x264 [info]: profile Main, level 3.0
Encoding: task 1 of 1, 99.31 % (271.41 fps, avg 291.14 fps, ETA 00h00m00s)[11:10:08] reader: done. 1 scr changes
Encoding: task 1 of 1, 99.92 % (271.41 fps, avg 291.14 fps, ETA 00h00m00s)[11:10:08] work: average encoding speed for jo
b is 291.135559 fps
Encoding: task 1 of 1, 99.99 % (271.41 fps, avg 291.14 fps, ETA 00h00m00s)[11:10:09] sync: got 14315 frames, 14315 expec
ted
[11:10:09] render: 14315 frames output, 0 dropped and 0 duped for CFR/PFR
[11:10:09] render: lost time: 0 (0 frames)
[11:10:09] render: gained time: 0 (0 frames) (0 not accounted for)
[11:10:09] h264-decoder done: 14315 frames, 0 decoder errors, 0 drops
Encoding: task 1 of 1, 99.99 % (271.41 fps, avg 291.14 fps, ETA 00h00m00s)x264 [info]: frame I:149 Avg QP:17.16 size:
50373 PSNR Mean Y:46.90 U:50.81 V:51.26 Avg:47.70 Global:46.37
x264 [info]: frame P:5300 Avg QP:21.50 size: 6792 PSNR Mean Y:42.70 U:47.27 V:47.69 Avg:43.60 Global:42.57
x264 [info]: frame B:8866 Avg QP:25.80 size: 1000 PSNR Mean Y:42.91 U:47.05 V:47.58 Avg:43.79 Global:42.63
x264 [info]: consecutive B-frames: 9.5% 18.7% 14.8% 56.9%
2E6n4c o[diinnfgo:] :t amsbk I1 oIf1 61.,. 49:9 .2909. 5%% ( 207.10.%4 17 9f.p5s%,
xa2v6g4 2[9i1n.f1o4] :f pmsb, PE T AI 1060.h.040:m 0 02s.)9% 0.0% 4.4% P16..4: 31.2% 13.6% 9.2% 0.0% 0.0% sk
ip:38.7%
x264 [info]: mb B I16..4: 0.1% 0.0% 0.6% B16..8: 22.3% 2.5% 0.7% direct: 1.1% skip:72.7% L0:36.7% L1:56.2% BI:
7.1%
x264 [info]: coded y,uvDC,uvAC intra: 59.0% 70.2% 46.6% inter: 9.8% 12.4% 2.6%
x264 [info]: i16 v,h,dc,p: 42% 24% 8% 26%
x264 [info]: i4 v,h,dc,ddl,ddr,vr,hd,vl,hu: 29% 16% 13% 6% 8% 9% 7% 7% 6%
x264 [info]: i8c dc,h,v,p: 49% 22% 19% 10%
x264 [info]: Weighted P-Frames: Y:4.7% UV:2.5%
x264 [info]: ref P L0: 63.4% 16.2% 13.9% 6.3% 0.2%
x264 [info]: ref B L0: 82.9% 15.3% 1.8%
x264 [info]: ref B L1: 92.9% 7.1%
x264 [info]: SSIM Mean Y:0.9849442 (18.223db)
x264 [info]: PSNR Mean Y:42.872 U:47.171 V:47.660 Avg:43.756 Global:42.635 kb/s:702.42
[11:10:09] aac-decoder done: 0 frames, 0 decoder errors, 0 drops
[11:10:09] mux: track 0, 14315 frames, 52364232 bytes, 702.29 kbps, fifo 4096
[11:10:09] mux: track 1, 27961 frames, 9542025 bytes, 127.97 kbps, fifo 8192
Muxing: this may take awhile...[11:10:09] libhb: work result = 0

....looks like 271 fps.... with a Xeon 5675 @ 4.2 ghz, 1600 mhz RAM.

 

[dclive]

And latest version (latest *full release*, *not* latest nightly) get this:

HandBrake 0.10.2 (2015060900) - MinGW x86_64 - https://handbrake.fr
12 CPUs detected
[11:15:42] work: average encoding speed for job is 330.657837 fps

So latest full release = 55 fps per Xeon 5675@4.2 core.

 

[dclive]

...and with DXVA enabled.... (with nVidia 970 GPU)

[11:18:45] dxva2:Available decoder output format 61 (AV_PIX_FMT_DXVA2_VLD)
[11:19:25] reader: done. 1 scr changes
[11:19:26] work: average encoding speed for job is 348.689148 fps

 

[dclive]

From MacOS 10.11.2 Handbrake 9.9 (version in the OP's MS link):

x264 [info]: using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.2 AVX
x264 [info]: profile Main, level 3.0
[19:40:31] reader: done. 1 scr changes
[19:40:31] work: average encoding speed for job is 193.978653 fps

So 193.97 fps for a mobile i7 @ 2.3 ghz (Macbook Pro 15" 2012, 16GB/500GB) with the 2013 release.

 

[dclive]

From MacOS 10.11.2 Handbrake (Nightly from 1/2/2016):

Encoding: task 1 of 1, 99.99 % (226.25 fps, avg 206.19 fps, ETA 00h00m00s)[19:46:51] work: average encoding speed for job is 206.185715 fps

So 206.18 fps for the latest nightly.

This for the same i7 @ 2.3 (I7 - 3615QM).

So latest nightly = 51.54 fps per i7@2.3 core.

An Intel i7@2.3 (3615QM) is 94% of the speed of an Xeon 5675@4.2 per core (average), even though the newer i7 has only 55% of the clockspeed!

Wow. I'm surprised how poorly the older Xeon does against what's, honestly, "just" a laptop CPU. Yes, we're comparing latest nightly vs. latest release, but still - that's a big difference in the newer silicon, and it's really just 2 generations newer (Xeon 5675 series -> Sandy Bridge 2xxx series -> Ivy Bridge 3xxx series).

Does anyone have an even newer i7 they'd care to post specs on? Particularly a laptop i7?

 

[dclive]

(repost)

 

[dclive]

Macbook Air 2013 i5/1.3 :
18:02:43] CPU: Intel(R) Core(TM) i5-4250U CPU @ 1.30GHz
[18:02:43] - Intel microarchitecture Haswell
[18:02:43] - logical processor count: 4
[18:02:43] OpenCL device #1: Intel HD Graphics 5000
[18:02:43] - OpenCL version: 1.2
[18:02:43] - driver version: 1.2(Nov 18 2015 20:57:43)
[18:02:43] - device type: GPU
[18:02:43] - supported: YES


.99 Version:
Encoding: task 1 of 1, 99.99 % (118.34 fps, avg 79.84 fps, ETA 00h00m00s)[17:50:19] work: average encoding speed for job is 79.838272 fps

Latest Nightly:
Encoding: task 1 of 1, 99.99 % (129.52 fps, avg 93.68 fps, ETA 00h00m00s)[18:05:16] work: average encoding speed for job is 93.679573 fps

Interesting - 46.8 fps per 1.3 ghz core - vs the Xeon X5675 getting 55.10 fps per 4.2 ghz core. 31% of the CPU speed and getting 85% of the Handbrake rendering speed per core - wow.

Vs the 3615QM, 46.8 fps per 1.3 ghz core vs the 3615QM getting 51.54 fps - 56% of the core speed getting 91% of the per-core Handbrake rendering speed per core.

 

[dclive]

2014 Macbook Pro 15" i7/2.5 :

[18:20:33] CPU: Intel(R) Core(TM) i7-4870HQ CPU @ 2.50GHz
[18:20:33] - Intel microarchitecture Haswell
[18:20:33] - logical processor count: 8
[18:20:33] OpenCL device #1: Intel Iris Pro
[18:20:33] - OpenCL version: 1.2
[18:20:33] - driver version: 1.2(Nov 18 2015 20:57:43)
[18:20:33] - device type: GPU
[18:20:33] - supported: YES
[18:20:33] OpenCL device #2: NVIDIA GeForce GT 750M
[18:20:33] - OpenCL version: 1.2
[18:20:33] - driver version: 10.8.14 310.42.15f01
[18:20:33] - device type: GPU
[18:20:33] - supported: no

.99:
Encoding: task 1 of 1, 99.88 % (210.11 fps, avg 209.95 fps, ETA 00h00m00s)[18:19:31] work: average encoding speed for job is 209.951889 fps - 52.48 fps/core.

Latest Nightly:
Encoding: task 1 of 1, 99.90 % (238.68 fps, avg 249.09 fps, ETA 00h00m00s)[18:21:32] work: average encoding speed for job is 249.089340 fps - 62.27 fps/core

A 4870HQ at 2.5 ghz does 62.27 fps/core; the Xeon 5675 at 4.2 ghz does 55.10 fps per core. The 4870HQ is 59% of the CPU clock but 13% *faster* at Handbrake rendering per core.

Lesson learned: for rendering, don't even consider any other tool other than the very latest version of the software product you're using. I suspect the main change is simply newer versions of ffmpeg in each of the later builds.

 

[dclive]

...and with DXVA enabled.... (with nVidia 970 GPU)

[11:18:45] dxva2:Available decoder output format 61 (AV_PIX_FMT_DXVA2_VLD)
[11:19:25] reader: done. 1 scr changes
[11:19:26] work: average encoding speed for job is 348.689148 fps

Wow! But if I go from that older 101 version from mid-2015 to the latest nightly, I jump from 348 to 412:

[23:11:41] work: average encoding speed for job is 412.364990 fps

* That's 412.36/6 = 68.73 fps per core (w/970)

Machine:
[23:11:06] CPU: Intel(R) Xeon(R) CPU X5675 @ 3.07GHz (@4.2)
[23:11:06] - logical processor count: 12
[23:11:06] OpenCL device #1: NVIDIA Corporation GeForce GTX 970
[23:11:06] - OpenCL version: 1.2 CUDA
[23:11:06] - driver version: 361.43
[23:11:06] - device type: GPU
[23:11:06] - supported: no
[23:11:06] Intel Quick Sync Video support: no


...but if I forgo DXVA (nVidia 970) and go to straight hex-core CPU:

[23:14:37] work: average encoding speed for job is 398.551971 fps

* That's 398.55 / 6 = 66.42 fps / core

...so that nV970 with the latest nightly gives me about 14 more FPS, or perhaps 5% more speed.

 

[YBS1]

5960X@4.5
[22:19:00] work: average encoding speed for job is 623.418579 fps

Latest official release, might be able to just squeak by 700fps if I ran it at benchmarking settings.

 

[Carfax83]

Looks like the official release is slightly faster than the nightly build that I used before:

HandBrake 0.10.2.7286 - 64bit Version
OS: Microsoft Windows NT 6.2.9200.0 - 64bit
CPU: Intel(R) Core(TM) i7-5930K CPU @ 3.50GHz
Ram: 32668 MB
x264 [info]: using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.2 AVX AVX2 FMA3 LZCNT BMI2
x264 [info]: profile Main, level 3.0
[21:44:17] reader: done. 1 scr changes
[21:44:18] work: average encoding speed for job is 485.064758 fps

 

[Burpo]

5960X@4.5
[22:19:00] work: average encoding speed for job is 623.418579 fps

Latest official release, might be able to just squeak by 700fps if I ran it at benchmarking settings.

Damn it man.. That's ridiculous.
I need your setup

 

[alexanderdiehl]

You can re-code this (click Download) to mp4 using Android Preset and note the times, to make it comparable to other systems.

I'd love to benchmark my new hardware setup, but I can't access the file in the link above. I get a OneDrive error message that says the file has beend moved or deleted. Can anyone help me to get a copy? THANKS!

 

[Makaveli]

Nice necro dude.

The last response in this thread is 8 months old.

 

[alexanderdiehl]

Nice necro dude.

The last response in this thread is 8 months old.

The thread ran over 3+ years, so I thought I give it a try.
I'd find it helpful to understand if I have configured my new system in a way to get the best performance ....

 

[YBS1]

Usually Makaveli isn't quite so snappy, maybe he was having a bad day. I'm thinking there was an updated link somewhere in this thread, if not I'm sure several of us have the file. I'd have to find a host for it later if you have no luck.

Thread needs to be restarted anyway with newest version.

 

[Burpo]

Here's the originals..

http://download.blender.org/peach/bigbuckbunny_movies/

 

[FlanK3r]

which one of files in your link we can to use for encoding?

 

[Burpo]

big_buck_bunny_480p_h264.mov