The standard board planned and validated for Solidworks is more than enough to use Visualize.
Indeed, Visualize is actually the Bunkspeed software which was already particularly fast because it uses a particular technique to make very fast displays. (it doesn't use ray tracing)
The size of the ASMs plays very little role, since only the outer faces are rendered, the complexity of the shape remains managed by SW. Note that only the PRO version allows you to make video or animated images, with the first level version, you can only do one ASM in static.
Thank you for your answer, in fact, I use the Nvidia graphics card (recommended for solidworks) but yet the renderings take a long time (it reaches almost 9 days of the estimated time), and so I decided to buy a new computer for this I will need your advice for a fairly sufficient configuration.
The most important will be the graphics card, then the machine's RAM and only the processor (to a lesser extent). There is no direct relationship between the number of components and the memory that will be required. It is the number of textured triangles that consumes memory. It then also depends on the textures, rendering options, etc... A very generic rule of three can be used to estimate the memory needed to load an assembly and work with: 5 Million polygons per GB of graphics RAM. You also need about 2Gb re Graphics RAM for the denoiser if it is used... You need at least as much free RAM as VRAM on the graphics card to work properly.
The import method is also very important for large blends, among other things:
Limit the complexity of the mesh as low as possible (options/Document properties/Image quality/Shaded image resolution). If there are no close-ups of the objects, the quality can really be lowered without it being noticed. Otherwise, we have to find a happy medium
Favor the Simple export from SOLIDWORKS which will optimize the geometries by grouping the objects by material. This will limit the number of objects to be put into memory and therefore limit the resources needed to display the assembly.
As far as the rendering time is concerned, it will mainly depend on the speed of the graphics card. The average processor of a machine will only participate in about 10% of the calculations of a Visualize rendering on a machine with a mid-range graphics card. It is therefore preferable to favor the graphics card. It is indeed very difficult to double the speed of a processor, while it is possible to multiply the speed of a graphics card by 5 (or more). To help choose the model, here is a page referencing the main certified cards and their performance on the same stage (downloadable to locate a specific machine): https://www.solidworks.com/sw/support/visualize-hardware-benchmarks.htm
If that's not enough, it's finally possible (with Visualize Pro) to create a "compute farm" of several machines thanks to additional Visualize Boost licenses.