But aside from using RAMDIsk for that task, here is my recommendation for cache-configuration:
Cache Task for C:\ (SSD): R/W
L1-Size: 12 GiB (or higher maybe 18GiB if you do not need another seperate cache-task for HDDs)
Block-Size: 16 Kb block-size (or lower depending how much additioonal CPU you would like to waste for little IO improvement-impact).
No L2 because this one is already a SSD.
If you have additional HDD for media then another Cache-Task comes into play:
Cache Task for D:\, E:\, etc. (HDD): R/W
L1-Size: 6 GiB
Block-Size: 64kb or 128kb (varies on chipset, might try different also)
L2-Size: 64 GiB (perfect use-case for the spare SSD)
Oh wait we only consume 18GB...so 2 left.... hmmmmm, just use more RAMdisk ... these work perfectly seperated side-a-side.
Whatever you put inside a RAM-disk and symlink the folder in NTFS to the RAM-Drive...all of it excluded from CACHE, not interfering with CACHE and is permanently accelerated.
Also If a app uses 2 gib ram from content...and you put these ressources to ram-disk, then it uses no more ram for content itself, only the EXE size

. ... if the exe is also on the ram-disk... system ram usage would show 0 bytes for the process......

try that with cache on block-level , impossible
Better think of "Cache" beeing a general-on-demand "RAM-Buffer" and "RAMDisk" beeing a specific permanent "RAM-Storage".
If you realise that, everthing gets clear.
I personally promote renaming those
Most people are not aware of the power of "mklink /J" for ram-disks
Hope my words helped you!