Rendering with Cinebench, Blender and LuxRender
Even though I don’t really like it because the Cinebench R23 delivers rather inconsistent results, you can of course still make a correct statement in total. The performance of the Ryzen 7 7800X3D is lower than that of the Ryzen 7 7700X due to the clock. Although the TDP was specified at 120 watts, the motherboard caps out at almost exactly 90 watts. What is rather unimportant in games certainly costs a few score points here. But it’s also more at the sweet spot, so you should be happy to use that with 90 watts as well. Yes, more is also possible, but then I would have to manually raise the board defaults.
The single-thread performance logically shows the expected picture.
It’s the same as always, of course: a good renderer needs invigorating core feed, it always has. My beloved igoBOT is a grateful task there, even if rendering on the CPU is slowly going out of fashion. But before I take things like Cinebench as the sole benchmark, I’d rather run something like that, which also sometimes causes a few minutes of work and delivers very consistent results. And it can also heat.
Once again, we see the Ryzen 9 7950X ramming and rendering virtually everything into the ground that didn’t make it to Three on the Trees. The new CPU also proves to be quite a match for the 30-watt thirstier Ryzen 7 7700X here, at least, and benefits from the cache (denoiser, textures). It could hardly be more efficient, but we’ll see about that in a moment….
The Luxmark as a decoupling of the LuxRender suite again shows the same behavior with the score as with the CB23.
LTspiceXVII
New in my benchmark suite is LTspiceXVII, a circuit simulation program. The simulator is designed to run industry standard semiconductor and behavioral models. New circuits can be designed with the integrated schematic capture. Simulation commands and parameters are placed as text on the schematic using common SPICE syntax. Waveforms of circuit nodes and device currents can be recorded by mouse click on the nodes in the schematic during or after simulation.
My thanks here go to our forum member Deridex, who contributed the workload as well as the idea. A total of 16 threads are used in the benchmark, which naturally makes the CPUs with 8 cores and more slip closer together at the top and shifts the focus to the clock. All the X3D models can only lose here.
Encoding, financial service and programming
The first two benchmarks also benefit many cores again, with FSI being pure compute. The Ryzen 7 7800X3D is clearly behind the model without cache. This fits the character of the tasks, whereby the power consumption of the X3D is of course considerably lower.
In Python and even more so Octave, Intel used to be the measure of all things, now they are no longer. Python, like Math Lab, relies on Intel’s Math Kernel Library (MKL) in many areas. NumPy in particular suffered a bit here in the past. The Ryzen 7 7800X3D is at the back. Unfortunately, the cache doesn’t help much if the clock is missing.
The next workload uses Octave, a programming language for scientific computing, to solve a variety of mathematical operations. The Ryzen 9 7950X3D really collapses, for whatever reason. Wrong CCD and thus too little clock? It is reproducible, unfortunately.
- 1 - Introduction, preliminary remark and CPU data
- 2 - Chipset, motherboard and the more complex test setup
- 3 - Gaming Performance HD Ready (1280 x 720 Pixels)
- 4 - Gaming Performance Full HD (1920 x 1080 Pixels)
- 5 - Gaming Performance WQHD (2560 x 1440 Pixels)
- 6 - Gaming Performance Ultra-HD (3840 x 2160 Pixels)
- 7 - Autodesk AutoCAD 2021
- 8 - Autodesk Inventor 2021 Pro
- 9 - Rendering, Simulation, Financial, Programming
- 10 - Science and mathematics
- 11 - Power consumption and efficiency
- 12 - Summary and conclusion
330 Antworten
Kommentar
Lade neue Kommentare
Mitglied
Urgestein
Mitglied
Urgestein
1
Mitglied
1
Neuling
Moderator
Veteran
Urgestein
1
1
Urgestein
Mitglied
Urgestein
Urgestein
Mitglied
Urgestein
Alle Kommentare lesen unter igor´sLAB Community →