The R&S can go up to 160Mpts memory depth, which is more than the Siglent has.
I think that's across all channels, i.e. 10M/channel max
Without doubt R&S is higher class scope. I do not want make it questionable. Also personally I have been tens of years R&S fan when talk T&M (older) "Rolls Royces"
I do not know what Siglent here is now "compared". ( I do not want compare apples and oranges)
But also I do not like "alternative truths aka trumpling".
R&S this machine here have max 10M for all channels simultaneously and 20M for channel (not interleaved)
Siglent
SDS2000 have max 35M for all channels simultaneously and 70M for channel not interleaved. 4ch models have 2x70M
SDS2000X max 70M for all channels simultaneously and 140M for channel not interleaved. 4ch models have 2x140M
For segmented memory acq there is available up to 180M (least in SDS2000, just checked)
Max speed 500 ksegment/s.
Example (checked w SDS2304: 4ch on. 1us/div 1GSa/s. Segment length 14k for every channel. 3323 segment 4x14000x3323= ~186 Msample.
(Maximum memory for history/segments is not available with all t/div and segment sizes.)
For waveform history same amount as segmented. (R&S and Siglent have this very nice feature)
waveform history buffer works just same speed what is current wfm/s speed in use (up to 140kwfm/s)
just for right facts, not like this is better etc...