Question Handbrake 1.3.3 - Benchmark your System - New benchmark criteria

Page 4 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Hulk

Diamond Member
Oct 9, 1999
4,191
1,975
136
Back in 2018 ElFenix posted the original Handbrake Benchmark Thread. I thought it was cool so I started updating the thread with a results table. With his permission I'm posting an update to that thread. If you want to participate we're going to tighten up the testing methodology (see below) so the results provide data better for analysis.

I'll update the second post of this thread periodically with results.

The test file:
sdifox said:
et voila. we can all host the file and round robin it :p

LG New York HDR UHD 4K Demo.zip

drive.google.com
drive.google.com


1. Use the following version of Handbrake with the built-in h.265 mkv 2160p60 preset
HandBrake-1.3.3-x86_64-Win_GUI.exe
Don't forget to turn on logging in Handbrake so you can retrieve your time. Tools>Preferences>Advanced>Logging
Once this current version is replaced you'll be able to access this version from the following link.
HandBrake: Nightly Builds
Nightly builds of HandBrake
handbrake.fr

2. Report average effective clockspeed using HWinfo - https://www.hwinfo.com/
I'm sure there are better ways to do this using the logging feature but here's how I did it. I downloaded the portable version, which runs from the exe file.

Run the exe and select "sensors only." You only need this window open.
Minimize the sensors window so you can have it and Handbrake visible at the same time. Select the "average effective clockspeed" line so you can spot it easily. It will highlight.
Note the icon for "reset min/max/average values and timer" in the bottom right area of the window, it's the clock icon.

Start the handbrake but don't start/reset the timer until handbrake is actually encoding video. Record the average effective clockspeed right before handbrake finishes the encode. It should settle in pretty quickly after the encode starts and not change during the encode.

More info on "effective" vs "instant" clock: https://www.hwinfo.com/forum/threads/effective-clock-vs-instant-discrete-clock.5958/

3. Report "Total CPU Usage" (also from HWinfo)

4. Report your encoding time, CPU Model, and RAM specs (amount, speed, latency)

1665102318265.png
 
Last edited:
  • Like
Reactions: Kaluan and ElFenix

Arni90

Junior Member
Apr 7, 2021
6
10
51
Just want to confirm the score you posted was using AVX512 at 4.8 effective clock?
I have set an AVX512 offset of 4, and the all core boost was 52x, that should give 4.8 GHz during AVX512.
Two cores have a maximum multiplier of 54x, so they might run at 50x in AVX512?
I haven't really tested AVX512 offset extensively, I just set the offset to prevent an instant shutdown if I end up running programs with AVX512 (and it actually works, I can run Linpack just fine)

I don't think Handbrake's H.265 encoder makes use of AVX512, HWiNFO reported all cores to run at 5.2 GHz pretty much all the time during the run as you can see.

and using 346W at the socket!
Those 346W are just for the CPU die, there's a decent amount of power lost in the power plane, LGA1200 socket, and VRM at 200A current draw. Even in a high-end board like the Maximus XIII Hero
 
  • Like
Reactions: lightmanek

CakeMonster

Golden Member
Nov 22, 2012
1,384
482
136
5900X on Asus X570-E. No CPU OC whatsoever.
RAM: 2x16Gb 3600MHz @ CR1 16-16-16-36 (44-320)

Average Effective Clock during encoding: 3502MHz

encoded 1806 frames in 186.14s (9.70 fps), 11820.04 kb/s, Avg QP:29.09
 

Hulk

Diamond Member
Oct 9, 1999
4,191
1,975
136
I have set an AVX512 offset of 4, and the all core boost was 52x, that should give 4.8 GHz during AVX512.
Two cores have a maximum multiplier of 54x, so they might run at 50x in AVX512?
I haven't really tested AVX512 offset extensively, I just set the offset to prevent an instant shutdown if I end up running programs with AVX512 (and it actually works, I can run Linpack just fine)

I don't think Handbrake's H.265 encoder makes use of AVX512, HWiNFO reported all cores to run at 5.2 GHz pretty much all the time during the run as you can see.


Those 346W are just for the CPU die, there's a decent amount of power lost in the power plane, LGA1200 socket, and VRM at 200A current draw. Even in a high-end board like the Maximus XIII Hero

In order to enable AVX512 support in Handbrake you have to put "asm=avx512" in the advanced section of the video tab. It is disabled by default. Did you do that?

Your effective clock was reported at 4815. That is obtained by HWinfo as it constantly polls the clock of each core to find the actual average clock across all cores.
 

Arni90

Junior Member
Apr 7, 2021
6
10
51
In order to enable AVX512 support in Handbrake you have to put "asm=avx512" in the advanced section of the video tab. It is disabled by default. Did you do that?

Your effective clock was reported at 4815. That is obtained by HWinfo as it constantly polls the clock of each core to find the actual average clock across all cores.
I didn't in the previous run

I tried it now, and stability failed. However, all core 51x with AVX512 offset to -4 worked perfectly fine, that translated to
2 cores at 50x
4 cores at 49x
2 cores at 48x
So the AVX512 offset works on each core's maximum multiplier.

Anyway, log output now:
x265 [info]: using cpu capabilities: MMX2 SSE2Fast LZCNT SSSE3 SSE4.2 AVX FMA3 BMI2 AVX2 AVX512
...
encoded 1806 frames in 215.48s (8.38 fps), 11820.04 kb/s, Avg QP:29.09


In other words: AVX512 improved performance very slightly
 

Attachments

  • Skjermbilde (71).png
    Skjermbilde (71).png
    1.3 MB · Views: 15
  • Skjermbilde (72).png
    Skjermbilde (72).png
    1.3 MB · Views: 13

Arni90

Junior Member
Apr 7, 2021
6
10
51
And just for comparison, a run at 5.1 GHz with plain AVX2:

encoded 1806 frames in 224.04s (8.06 fps), 11820.04 kb/s, Avg QP:29.09

As you can see from power usage and temperatures, this is quite a bit more efficient.
 

Attachments

  • Skjermbilde (73).png
    Skjermbilde (73).png
    1.5 MB · Views: 10
  • Skjermbilde (74).png
    Skjermbilde (74).png
    1.6 MB · Views: 8
  • Like
Reactions: Hulk and moinmoin

Hulk

Diamond Member
Oct 9, 1999
4,191
1,975
136
And just for comparison, a run at 5.1 GHz with plain AVX2:

encoded 1806 frames in 224.04s (8.06 fps), 11820.04 kb/s, Avg QP:29.09

As you can see from power usage and temperatures, this is quite a bit more efficient.

For non-AVX512 I left your original scores since it was faster and made a separate chart for AVX512 scores. Compute efficiency increased about 7.9% with AVX512 enabled on your rig.
 
  • Like
Reactions: Makaveli

brinstar117

Senior member
Mar 28, 2001
954
4
91
I have another result, this time with a stock 8700k with default DDR4-2133 speed and timings and again with XMP DDR4-3600 speed and timings. Got a performance boost of about 9%. I only changed it by applying XMP settings on my Asus ROG Strix Z370-I.

I'm not sure if it changed any CPU settings as the 2nd run with the XMP settings seems to have about a 125MHz higher average CPU clockspeed. Though that alone would not have accounted for the performance boost.

I kind of want to try a Core 2 era system but I don't want to babysit it to grab a screenshot right at the end. I think it might take an hour or so to complete.

First result:

Intel Core i7-8700K Coffee Lake
16GB DDR4-2133 CL15-15-15-36
encoded 1806 frames in 429.08s (4.21 fps), 11820.04 kb/s, Avg QP:29.09
Average Effective Clock 4049 MHz
Average CPU Usage 96.9%

8700k handbrake DDR4-2133-15.png

8700k handbrake DDR4-2133-15 log.PNG

Second Result:

Intel Core i7-8700K Coffee Lake
16GB DDR4-3600 CL18-22-22-42
encoded 1806 frames in 393.74s (4.59 fps), 11820.04 kb/s, Avg QP:29.09
Average Effective Clock 4174.8 MHz
Average CPU Usage 96.8%

8700k handbrake DDR4-3600-18.png

8700k handbrake DDR4-3600-18 log.PNG
 

brinstar117

Senior member
Mar 28, 2001
954
4
91
I tried my LGA 771 to 775 modded Intel Xeon X5460 (equivalent to a Core 2 Quad QX9750 Yorkfield). The results were a little better than I was expecting but not by much. I threw in a few CPU-Z benchmarks at the end. Modern CPUs are about twice as fast in single threaded performance.

Intel Xeon X5460 3.17 GHz quad-core Harpertown
16GB DDR2-800 CL5-5-5-15 (4x4GB G.Skill F2-6400CL5-4GBPQ)
encoded 1806 frames in 2358.71s (0.77 fps), 11820.04 kb/s, Avg QP:29.09
Average Effective Clock 3133.2 MHz
Average CPU Usage 99.4%

Xeon X5460.png

Handbrake Log.PNG

Xeon X5460 single threaded bench.PNG
 

ch33zw1z

Lifer
Nov 4, 2004
37,733
18,003
146
Gave this a shot on my 3700X (no OC)


From the Logged output:

encoded 1806 frames in 313.10s (5.77 fps), 11820.04 kb/s, Avg QP:29.09

1618582027861.png
 

ElFenix

Elite Member
Super Moderator
Mar 20, 2000
102,414
8,356
126
5900HS
encoded 1806 frames in 318.68s (5.67 fps), 11820.04 kb/s, Avg QP:29.09
3482.4 mhz avg with 90.4% usage
16 GB DDR4 3200 22/22/22 (ugh lets see if we can get that a bit faster) (edit: no laptops lol)
 
Last edited:

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
Not trying to necro but I think it's time to bring this thread back to life and get more data on 1.4 if possible.

Not only should we be getting reports from the Alder Lake buyers soon, but Handbrake also supports M1 now. Just have to make sure we're getting pure SW results from the M1/Pro/Max crowd instead of hw accelerated results.
 

CakeMonster

Golden Member
Nov 22, 2012
1,384
482
136
Quick test, very similar result to 1.3.3 on 1.4.2 with my 5900X.

Also, Handbrake complained about .NET Framework, if you go and download it, make sure to get the 5.0 Desktop Runtime version, else it won't help.
 

Makaveli

Diamond Member
Feb 8, 2002
4,715
1,049
136
Ryzen 5800X 32GB DDR4 @ 3200 14-14-13-28-1T
Windows 21H1 HB 1.4.2
encoded 1806 frames in 241.84s (7.47 fps), 11821.55 kb/s, Avg QP:29.08

Previous score was
Windows 20H2 HB 1.3.3
encoded 1806 frames in 239.83s (7.53 fps), 11820.04 kb/s, Avg QP:29.09


1635699222104.png


 

Makaveli

Diamond Member
Feb 8, 2002
4,715
1,049
136
@Makaveli

Looks like you gained some performance. I should redo mine later . . .

There are differences in my runs though. Windows build is newer and first run was done on 16GB of ram 2nd run on 32GB's of ram. But give it a go and see if you notice anything on your rig.

My C0 tweaks are the same between both runs. Going to make a small change here and test that again.

My current C0 best core -20 2nd best core -15 then the rest -10

And I just did -20 best core and -15 on the rest the of cores give me this now

This is my best result so far in this test, this recent C0 changes seems to have boost multicore performance abit. If we look at chart on page 1 my original score had me just slighly slower then the 10 core comet lake chip with this C0 change now pretty much even with it.

encoded 1806 frames in 234.89s (7.69 fps), 11821.55 kb/s, Avg QP:29.08

1635702618300.png
 
Last edited:

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
@Makaveli

My scores are not much different in Handbrake 1.42. My old score:


My new score is

encoded 1806 frames in 191.23s (9.44 fps), 11821.55 kb/s, Avg QP:29.08

I'm Running Win11 22489.1000 now. My old score was from a Win10 build (don't remember) with Handbrake 1.1

Same clockspeed: 4.35 GHz static on a 3900x.
 
  • Like
Reactions: Makaveli

tamz_msc

Diamond Member
Jan 5, 2017
3,708
3,554
136
With my current setup in my flair -

encoded 1806 frames in 749.42s (2.41 fps), 11821.55 kb/s, Avg QP:29.08

Avg. effective clock speed - 2857 MHz
Avg. total CPU usage - 95%

EDIT: with Handbrake 1.4.2

Now with AVX-512 enabled -

encoded 1806 frames in 710.76s (2.54 fps), 11821.55 kb/s, Avg QP:29.08

Avg. effective clock speed - 2780 MHz
Avg. total CPU usage - 97%

Huh, so I got better scores with AVX-512 after all. Looks like this Dell Inspiron has better cooling than the HP machine.
 
Last edited:
  • Like
Reactions: Makaveli

Makaveli

Diamond Member
Feb 8, 2002
4,715
1,049
136
@Makaveli

My scores are not much different in Handbrake 1.42. My old score:


My new score is

encoded 1806 frames in 191.23s (9.44 fps), 11821.55 kb/s, Avg QP:29.08

I'm Running Win11 22489.1000 now. My old score was from a Win10 build (don't remember) with Handbrake 1.1

Same clockspeed: 4.35 GHz static on a 3900x.

My scores are pretty close between the two builds. Its only my adjusted C0 score that shows improvement.
 

Roumaldo

Junior Member
Nov 9, 2021
3
0
11
Someone with i5-12600K can do this benchmark? I see benchsmarks in some sites that says that i5 is more powerful than 5900X and 3950X but I don´t know if this benchsmarks are fake or not. If they are true, the i5 is a real beast in this bench. Thanks a lot! :)
 

Roumaldo

Junior Member
Nov 9, 2021
3
0
11
Someone with i5-12600K can do this benchmark? I see benchsmarks in some sites that says that i5 is more powerful than 5900X and 3950X but I don´t know if this benchsmarks are fake or not. If they are true, the i5 is a real beast in this bench. Thanks a lot! :)
 

Makaveli

Diamond Member
Feb 8, 2002
4,715
1,049
136
Someone with i5-12600K can do this benchmark? I see benchsmarks in some sites that says that i5 is more powerful than 5900X and 3950X but I don´t know if this benchsmarks are fake or not. If they are true, the i5 is a real beast in this bench. Thanks a lot! :)

This info is already available.

 

ElFenix

Elite Member
Super Moderator
Mar 20, 2000
102,414
8,356
126

S3am

Junior Member
Mar 25, 2017
3
0
66
CPU: Ryzen 5 5600G on A520M-ITX/ac. No CPU OC (default PBO).
RAM: 2x8Gb 3600@16-18-18-38 (XMP)

Average Effective Clock during encoding: 4096MHz
Avg. total CPU usage - 96%

encoded 1806 frames in 328.81s (5.49 fps), 11820.04 kb/s, Avg QP:29.09