Last edit: 21. Aug. 2017, 00:57h

AMD Ryzen – My new PC for content creation (with Hash Animation:Master) and Games (Part 3/3)

blog bild am bench 2017 ryzen at PatchWork3d

Have you been eagerly waiting? Here after a "long" wait I present the fresh benchmarks of my AMD Ryzen 1700 and A:M.
And because of this, I have even created a new benchmark test for Hash Animation:Master, which should test a lot of newer features too.

It is called "A:M Bench 2017" and it will utize among other things:

  • Ambient Occlusion (CPU based)
  • SubSurfaceScattering (SSS)
  • Particle (Hair)
  • Displacement Maps
  • Bump Maps
  • Raytraced Shadows
  • Complex, procentual Materialies
  • Volumetric Light
  • Reflections
  • ...and much more.

Apart from that, I tested the well-known "ThreeTeaPots"-Benchmark from 2009.

And since the AMD Ryzen 7 1700 is a 8-core/16-threads processor and A:M does only use 1 core when rendering internally, I have additional to the A:M internal rendering tested it with A:M Netrender with 26 frames to test the multi-core abillities too. (in A:M netrender we can use all of the cores/threads at once)

The tested computer system is described here: AMD Ryzen – Mein Neuer PC für Content Creation (u.a. mit Hash Animation:Master) und Spiele (Part 2/3)

A:M Bench 2017 Results

360p (Single Thread)

A:M Bench 2017 (Single-360p), 1 Slave / Core active:
1h 17min 23s

A:M Bench 2017 (Single-360p), 1 Slave / Core active:
6-7% workload

180p (Single Thread)

A:M Bench 2017 (Single-180p), 1 Slave / Core active:
0h 19min 12s

A:M Bench 2017 (Single-180p), 1 Slave / Core active:
6-7% workload

180p (16 Threads, 26 frames)

A:M Bench 2017 (Multi-180p), 15 Slaves active:
0h 38min 34s

A:M Bench 2017 (Multi-180p), 15 Slaves active:
94-95% workload

180p (8 Threads, 26 frames)

A:M Bench 2017 (Multi-180p), 8 Slaves active:
0h 59min 31s

A:M Bench 2017 (Multi-180p), 8 Slaves active:
50% workload

ThreeTeaPots 2009 Results

240p (Single Thread)

ThreeTeaPots 2009 (Single-240p), 1 Slave active:
0h 01min 54s

ThreeTeaPots 2009 (Single-240p), 1 Slave active:
6-7% workload

And there we go... the AMD R7 1700 at 3.9 GHz overclocking in differen benchmarks with Hash Animation:Master.
I am very happy with the machine. It outperformance my old AMD PII 1090T in single-thread-performance by about 100% and in Multicore it is even better :).

Why I used 15 threads, not 16? Netrender as a software does need some performance too. Because of that it is good practice to leave a little bit of processing power for it. If you use for instance another computer in your LAN to run Netrender on it, you can of course use 16.

And why 15 and 8 threads? AMD calls it "Simultanous Multi Threading", Intel calles it "Hyper-Threading". Both do more or less the same: They will split a real core into two simulated once to push the performance limits even further. But such cores are each only at about 60% performance of a real core in general. To compare those, I have put up both benchmarks.

And if you want to test the benchmarks on your own computer too, you can do that with this files in this download:
Hash Animation:Master Benchmark Suite 2009 & 2017(ZIP, 1.64 MB)

Have fun! :)

If you have any question, please leave a comment or use the contact form of the website :).

 
X

If my website sucks, you do not need to visit it.
Have a nice day and think about how much you have payed to visit it? Nothing...
How many ads do you see at it? None.

How much money do I earn from it? Nothing.
Does it cost me something to host it? Hell yeah.

I do not see why I should provide it to you to be insulted?