it is for sure MXT2001, ADC08D502 looks complettly different (i have them here and they looks exact the same as ADC08D500 on the picture), where MXT2002 in Owon looks like the unknown ADC used in Rigol. Just check the markers on picture.
I agree with Tinhead here, it really looks like it could be the MXT2002 (or a sister part with same pinout) used as ADC in both DS2k and DS4k.
I agree, but NOT based on the markings from the plastic mould, but because of matching when comparing the pinning of the MXT2002 datasheet with the routing of tracks in both the DS2000 and the DS4000 teardown videos.
I can not find one place where tracks go where they should not go, every place there is a routed diff pair or vias looking like it is a start of a pair it goes to to pins that are also pairs on the MXT2002.
The few exceptions to the paired track routing are for signals that are either GND, VDR or VA (routed wider than signals also fits with it being power) or are logic control signals from series termination resistors (SCLK, SDATA, SC) or other single signals.
To get 4Gsps in the DS4k one would combine 2 of these ADCs clocked at 1000MHz on one channel (4 separate ADCs interleaved).
MXT2002 has two input ports and seems to have internal SW controlled muxes so cross routing two channels to get double sample rate from two of these ADCs should be simple.
Operating an MXT2002 at 1000MHz is already over clocking vs the 500MHz to 800MHz mentioned in the datasheet.
I start to doubt if MXT2002 can be used also in the DS6k (@1250MHz) but it may still be a possibility.
Edit: Just noticed there is a 1-1.3GHz version of the same (MTX2001) so no overclock needed for that, and it also then fits the DS6k sample rate.
Edit2 : Did not see until now how closely the pinning matches other types (should investigate better then write), still tinheads comment about pin 2 not being connected to GND looks sound so MTX2001 might still be correct.