AMD 9900X3D Performance, Metro 5.8.4

Hello everyone,

I recently switched from the 5800X3D to the 9900X3D, and in my opinion, the core utilization was better with the 5800X3D.

In the attached screenshot, I calculate the lattice body in 0.25 mm and the process only uses 2 cores instead of 6 (without 3D cache).

Are there other AMD users who are not using all cores?

I have a 4070 GPU, which should be sufficient.

Thanks in advance.
Translated from German*

Hi @Mrvn

that’s really slow , plus you open another window on top what even more slowed up the processor , Revo Metro need to be maximize for full processing power .

My CPU i9 14Gen @ 5GGz runs around 75% while processing and the meshing is the slowest one.

RM need better optimization for AMD based CPUs

1 Like

Hi @Mrvn. I have a Ryzen 9 5900x 12 cores. With latest updates cpu utilization went up from using 2 to 4 logical cores (out of 24) and from about 12% to 20%. Still lot of room from improvement but already better. I also maximize utilization for RevoMetro in taskmanger to “realtime”.

1 Like

Looks like they are not yet there to full take the potential of the processors . It getting better but it could already be better after so long period of time. Or at least let the GPU handle it too. :wink:

2 Likes

On a desktop PC with an Intel i9-10900 processor, all cores were utilized during the Fusion process. I haven’t tried it on a laptop with a Ryzen 9-8940HX yet.

3 Likes

The CPU on my PC and laptop works as expected i9 10th Gen and i9 14th Gen.

But I am suspecting the software is not yet optimized for the AMD to utilize all cores , it is already better than before but not yet optimal.

1 Like

My Ryzen 9 7950X also only uses only 2 of the 16 cores.

1 Like

Hi, Cath!

Do you mind ask the tech team:

  1. if and when the utilization for AMD CPUs will be further improved

and

  1. if the same core utilization for AMD like for Intel counterparts is even achievable resp. to what extent can it - at least theoretically - be improved at all

To be honest, I actually am very interested in reason(s) why Intel CPUs seem to perform so much better than AMD for fusion and other processing steps, too. I do get it for AMD vs NVIDIA GPUs (CUDA) but less for CPUs and I am sure there must be a reason for it.

At the moment it is bit a bummer having upgraded the CPUs just for MetroX and the processing seems to take much less time on Intel systems :face_without_mouth:

But hey, it works ,my projects aren’t as big and I can live with the processing times - for now😁

Thanks in advance!:folded_hands:t2:

1 Like

Hi Ivan ,

Ok no problem I will ask .

1 Like

I second these questions. I use laptops for my scanning and in that context AMD processors are better then Intel for multi core operations. The AMD processors run all cores at full speed while Intel processors now have efficiency cores that prioritize power saving over performance. My Ryzen 7 7735HS (8C/16T) out-performs a Intel Core Ultra 9 285H (6 performance cores, 8 efficient cores, 2 low power cores) on scanning tasks without GPU. I purchased the Intel because it does have an Nvidia GPU, so it’s a net win for the Intel laptop. But if I was foolish enough to do fusion, merges, etc. on the laptop, the AMD would/should be better with all cores being performance.

1 Like

Hi @Dehidding

The slow lack of support evolved from past struggles to modern strength, but challenges remain due to Intel’s long dominance, historical power issues, Windows 11 scheduler favoring Intel, and some OEM/software gaps, though AMD’s Ryzen era has drastically improved things, making them very competitive the rest is still much behind and take a much longer time in development to reach the same level.

I left a note already , hopefully I can tell you all more about tomorrow.

1 Like