Benchmark your computer @4K with Handbrake 1.1 and H265! NOW WITH ZEN3!

[William Gaatjes]

encoded 1806 frames in 577.26s (3.13 fps), 11820.04 kb/s, Avg QP:29.09

win10 home. build : 17134.165
Ryzen 2600 at stock clocks on ASROCK AB350M.
8GB ram@2933MHz. timings :

Voltage 1.3
tCL 18
tRCDRD 18
tRCDWR 18
tRP 18
tRAS 40
tRC 52
CR1 1
tRRD_S 7
tRRD_L 7
tCCDL 7
tFAW 35
tRFC 416
tRFC2 256
tRFC4 176

I am planning to start optimizing memory this night again, i want to reach 3200Mhz after all.


edit:
with stock clocks, i mean the all core boost of 3.6GHz(sometimes 3.625GHz), checked with ryzen master.

 

[kjboughton]

Overclocked dual E5-2696 v3 with watercooling (see sig)

Limited to 16 cores (out of 18) per CPU with no HT (total Cores: 32, total logical CPUs: 32) @
encoded 1806 frames in 192.66s (9.37 fps), 11820.04 kb/s, Avg QP:29.09

Full 18 cores per CPU with HTT (total Cores: 36, total logical CPUs: 72) @
encoded 1806 frames in 174.04s (10.38 fps), 11820.04 kb/s, Avg QP:29.09

Dual Xeon for the win?

 

[PeterScott]

encoded 1806 frames in 2631.13s (0.69 fps), 11820.04 kb/s, Avg QP:29.09
Gemini Lake J5005 NUC.

Dang, I was just a smidgen behind an Atom based NUC:

HandBrake 1.1.1 (2018061800) - 64bit
OS: Microsoft Windows NT 6.1.7601 Service Pack 1 - 64bit
CPU: Intel(R) Core(TM)2 Quad CPU Q9400 @ 2.66GHz (Actually 3.2GHz OC)

encoded 1806 frames in 2655.10s (0.68 fps), 11820.04 kb/s, Avg QP:29.09

 

[kjboughton]

No one can beat my 3+ year-old dual Haswell-E Xeon system?

Where are all the AMD ThreadRipper fans?

 

[Markfw]

No one can beat my 3+ year-old dual Haswell-E Xeon system?

Where are all the AMD ThreadRipper fans?

And we would expect that your 36 cores could easily beat 16 cores, but there are some threadripper results in here anyway,

 

[eek2121]

And we would expect that your 36 cores could easily beat 16 cores, but there are some threadripper results in here anyway,

Yep, and you'll notice that nobody except those with more cores have beaten the results at any rate.

That's the whole reason I got the 1950x, vdeo encoding.

No one can beat my 3+ year-old dual Haswell-E Xeon system?

Where are all the AMD ThreadRipper fans?

getting basically the same score as you...with LESS THAN HALF as many cores (edit: and you'll note that our CPUs were only 200 MHz apart according to your sig).

 

[Batboy88]

I don't need to bench anything anymore already ran the 7700k through everything...Yeah it does well in Handbrake....but will get destroyed in Multithreaded stuff....again it's a Quad core....

 

[DooKey]

encoded 1806 frames in 320.75s (5.63 fps), 11820.04 kb/s, Avg QP:29.09

7820X@4.5ghz

Windows 10, 32GB 3200 CL15

 

[Batboy88]

encoded 1806 frames in 320.75s (5.63 fps), 11820.04 kb/s, Avg QP:29.09

7820X@4.5ghz

Windows 10, 32GB 3200 CL15

Exactly Why i said the 7820x is Excellent and can take it up to a good Clock. Just caught Everyone and mobo manufactures at the wrong time with it...

 

[DooKey]

Exactly Why i said the 7820x is Excellent and can take it up to a good Clock. Just caught Everyone and mobo manufactures at the wrong time with it...

Mine will run at 4.7ghz with no problems, but I run it at low voltage (1.14v) 4.5ghz to keep my fans quiet.

 

[Batboy88]

Mine will run at 4.7ghz with no problems, but I run it at low voltage (1.14v) 4.5ghz to keep my fans quiet.

I know...I want to throw it all together.....good Price now for THAT performance....Get some Block for it and run it on the Loop. And probaly like yeah the Strix one agian, Even though yeah a little wierder Bios/ UEFI like them...

I had the F board Strix...I actually really liked it but was some redacted wrong with it...and I guess couldn't be Fixed....got the Gigabyte Still it's ok...and yeah really like the EVGA and Supermicro.

Then I don't know..a lot redacted happened...got robbed at the House and threw a bunch junk and Old stuff Out Lol.

Profanity is not allowed
in the tech areas.

AT Mod Usandthem

 

[Markfw]

Exactly Why i said the 7820x is Excellent and can take it up to a good Clock. Just caught Everyone and mobo manufactures at the wrong time with it...

encoded 1806 frames in 404.44s (4.47 fps), 11820.04 kb/s, Avg QP:29.09
Yes a little slower, but at 2/3rds of the cost.
2700x@4.1 ghz win 10

 

[kjboughton]

getting basically the same score as you...with LESS THAN HALF as many cores (edit: and you'll note that our CPUs were only 200 MHz apart according to your sig).

I like your spunk; however, at greater than 30% slower than me you are not "getting basically the same score" as me.

Edit: Also, I should add -- frequency shown in sig is max turbo boost. All cores loaded is closer to 3.3~3.4Ghz with 36 cores active.

FCG

 

[eek2121]

I like your spunk; however, at greater than 30% slower than me you are not "getting basically the same score" as me.

Edit: Also, I should add -- frequency shown in sig is max turbo boost. All cores loaded is closer to 3.3~3.4Ghz with 36 cores active.

FCG

I haven't posted my 4.25 GHz results due to a failing pump in my AiO. Also note that I am using Hynix MFR memory, I could easy surpass your rig if mine was properly cooled and used Samsung b die...and while using less cores and power to boot.

 

[PeterScott]

I like your spunk; however, at greater than 30% slower than me you are not "getting basically the same score" as me.

Edit: Also, I should add -- frequency shown in sig is max turbo boost. All cores loaded is closer to 3.3~3.4Ghz with 36 cores active.

FCG

Did you note your CPU usage percentage. I remember on the previous Handbrake benchmark run here, that it started to plateau on very high core counts, though I think higher resolution helps get more work out of more cores and the last test was at 1080p IIRC.

 

[kjboughton]

I haven't posted my 4.25 GHz results due to a failing pump in my AiO. Also note that I am using Hynix MFR memory, I could easy surpass your rig if mine was properly cooled and used Samsung b die...and while using less cores and power to boot.

Talk is cheap...

 

[kjboughton]

Did you note your CPU usage percentage. I remember on the previous Handbrake benchmark run here, that it started to plateau on very high core counts, though I think higher resolution helps get more work out of more cores and the last test was at 1080p IIRC.

100% more or less

 

[William Gaatjes]

I now tested it with memory frequency at 3200MHz.
encoded 1806 frames in 519.91s (3.47 fps), 10567.33 kb/s, Avg QP:28.58

win10 home. build : 17134.165
Ryzen 2600 at stock clocks(all core boost) on ASROCK AB350M.
8GB ram@3200MHz. timings :

Voltage 1.3
tCL 18
tRCDRD 18
tRCDWR 18
tRP 18
tRAS 40
tRC 52
CR1 1
tRRD_S 7
tRRD_L 7
tCCDL 7
tFAW 35
tRFC 416
tRFC2 256
tRFC4 176

 

[William Gaatjes]

Overclocked dual E5-2696 v3 with watercooling (see sig)

Limited to 16 cores (out of 18) per CPU with no HT (total Cores: 32, total logical CPUs: 32) @
encoded 1806 frames in 192.66s (9.37 fps), 11820.04 kb/s, Avg QP:29.09

Full 18 cores per CPU with HTT (total Cores: 36, total logical CPUs: 72) @
encoded 1806 frames in 174.04s (10.38 fps), 11820.04 kb/s, Avg QP:29.09

Dual Xeon for the win?

I think cpu progress has advanced considerably, it is interesting to see.
you have 36 cores(with HT disabled) (going for the 18 core per cpu number) without hyperthreading and i have 6 cores with hyper threading enabled for 12 logical cores.
That is 1/3 and even with SMT/HT, I do get 33% of your score with an allcore clock of 3600MHz.
It seems handbrake does scale linear with cores.

The I7 8700K systems that IEC and epsilon84 and mjduduis have are insane.
overclocked to 4.7GHz it gets 48% of your 36 core score but with only 1/3 of the logical cores.
overclocked to 4.8GHz it gets 50% of your 36 core score but with only 1/3 of the logical cores.
overclocked to 4.9GHz it gets 51.1% of your 36 core score but with only 1/3 of the logical cores.

And from my own testing, it seems handbrake loves memory speed as well. I got a 10% increase just from moving from 2933MHz to 3200MHz.
The I7 8700K systems described above have very fast memory.

From looking at the threadripper results, i get the impression that handbrake does not recognize numa correctly in ryzen chips. Since you have a 2socket system.

edit:
added cursive text.

edit much later :
aarrg. I did not see that you had HTT enabled with the 36 cores. Well, if i take the 32 cores with HTT disabled score, it still applies what i wrote.

 

[William Gaatjes]

Exactly Why i said the 7820x is Excellent and can take it up to a good Clock. Just caught Everyone and mobo manufactures at the wrong time with it...

I would rather buy an I7- 8700K vs a 7820X if i had to choose between the two. Looks a lot better for my usage pattern.

 

[CakeMonster]

8700K @ 4.9GHz (4.6GHz AVX offset)
DDR 3600 @ cas16

encoded 1806 frames in 379.09s (4.76 fps), 11820.04 kb/s, Avg QP:29.09

 

[Franz316]

encoded 1806 frames in 429.95s (4.20 fps), 11820.04 kb/s, Avg QP:29.09

Stock 2700x
32GB Trident Z @ 3200 & cas16

The Prism cooler was really struggling at around 80c and causing speeds to hover from 3750 to 3900. I may upgrade that before the end of the year.

 

[William Gaatjes]

encoded 1806 frames in 429.95s (4.20 fps), 11820.04 kb/s, Avg QP:29.09

Stock 2700x
32GB Trident Z @ 3200 & cas16

The Prism cooler was really struggling at around 80c and causing speeds to hover from 3750 to 3900. I may upgrade that before the end of the year.

The Shadowrock TF 2 from be quite is a good air cooler. It is rated for 160W TDP and very silent.
But watercooling seems to be the best for ryzen chips.

 

[imported_stryfe]

Sandy Bridge and Lynnfield

i7 2600k @ 4.5ghz, 16gb DDR3-1866mhz CL10, Win 10 Pro
encoded 1806 frames in 951.72s (1.90 fps), 11820.04 kb/s, Avg QP:29.09

i5 750 @ 2.8ghz, 8gb DDR3-1333mhz CL8, Win 10 Pro
encoded 1806 frames in 2012.80s (0.90 fps), 11820.04 kb/s, Avg QP:29.09

 

[CrackRabbit]

encoded 1806 frames in 440.29s (4.10 fps), 11820.04 kb/s, Avg QP:29.09

Stock 2700x, 16gb Flare X @ XMP 14-14-14-34 3200
Had the same problem as Franz316, the Prism just couldn't keep up with the temps.
Still a lot faster than the Haswell based system it replaced.