Speed of MSA Server

Hey,

I deployed only the MSA server on a A100 host. I used the GPU mode, it takes some minutes to compute the alignments, is it normal ?
If GPU mode, should i also use the vmtouch command to put the db in cache ?
Is it also normal that during the request the process are present in the GPU but doesnt consume utilization but only memory ?

[opc@vmpsiloaf2dispsildv01 ~]$ nvidia-smi
Fri Oct 10 15:17:30 2025       
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 575.57.08              Driver Version: 575.57.08      CUDA Version: 12.9     |
|-----------------------------------------+------------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
|                                         |                        |               MIG M. |
|=========================================+========================+======================|
|   0  NVIDIA A10                     Off |   00000000:00:04.0 Off |                    0 |
|  0%   43C    P0             57W /  150W |   22265MiB /  23028MiB |      0%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+
                                                                                         
+-----------------------------------------------------------------------------------------+
| Processes:                                                                              |
|  GPU   GI   CI              PID   Type   Process name                        GPU Memory |
|        ID   ID                                                               Usage      |
|=========================================================================================|
|    0   N/A  N/A           66125      C   mmseqs                                12506MiB |
|    0   N/A  N/A           66127      C   mmseqs                                  348MiB |
|    0   N/A  N/A           66128      C   mmseqs                                 9392MiB |
+-----------------------------------------------------------------------------------------+

No repository file, test, or entry point is named. Start by locating the MSA server's GPU-mode request path and the documentation for vmtouch and database caching, then reproduce the reported behavior using the supplied nvidia-smi output. Done means determining whether the runtime, GPU utilization, and memory usage are expected or identifying a specific fix.