Notebookcheck is testing the Dell Precision 7875 Tower Workstation with AMD Ryzen Threadripper Pro 7995WX, 512GB RAM, 4TB SSD, and Nvidia RTX 6000 Ada Generation.
Well matjeh already answered, but IMO it represents a real workload IF you run it on the GPU.
There’s a table out that shows the 14900 and 7950X scoring around 2.2k pts, which is like 1/5th of a 4060’s 10k pts. The 4090 scores just under 35k pts.
Running CB24 on this beast of a system will show up where the best CPU currently is vs the best GPU score considering the article said the A6000 Ada can score over 90% above a 4090 in certain workloads.
Cinebench uses the same rendering engine as the commercial Cinema 4D software which is used to make movies and games.
It gives a good idea of how that software will run, but also how 3D rendering software (rasterization, ray tracking, path tracing, radiosity, volumetics) in general will perform, and the workload of heavy floating-point vector instructions coupled with moderate memory load (mostly stressing the cache levels, with the large L3 per-CCX cache of Threadripper helping a lot).
Would have been nice to see CB24 running on both the CPU and GPU.
Why? It’s nothing but a 3d mark… test. Nothing real world… just… hey I got this score… bragging right software
Well matjeh already answered, but IMO it represents a real workload IF you run it on the GPU.
There’s a table out that shows the 14900 and 7950X scoring around 2.2k pts, which is like 1/5th of a 4060’s 10k pts. The 4090 scores just under 35k pts.
Running CB24 on this beast of a system will show up where the best CPU currently is vs the best GPU score considering the article said the A6000 Ada can score over 90% above a 4090 in certain workloads.
Source: https://www.cgdirector.com/cinebench-2024-scores/
Sometimes it’s just nice to see big numbers go brrr!
Number bigger better!
Cinebench uses the same rendering engine as the commercial Cinema 4D software which is used to make movies and games.
It gives a good idea of how that software will run, but also how 3D rendering software (rasterization, ray tracking, path tracing, radiosity, volumetics) in general will perform, and the workload of heavy floating-point vector instructions coupled with moderate memory load (mostly stressing the cache levels, with the large L3 per-CCX cache of Threadripper helping a lot).
The previous iterations of cinebench showed little to no scaling with memory performance of 3D v cache. Did something change with 2024?