I poked around looking for prior 8bit vs 12bit discussions but couldn't find as many on topic as I expected... My gut says obviously 12bit is better than 8bit, but I seem to recall many folks saying that given the size of the display you'd be hard pressed to differentiate much improvement on the screen.
But the biggest thing is the single ADC - the Siglent SDS1104X-E has dual 1G sampling, so 4 channels at 500M. If I understand correctly, the new Rigol's are 1.25G X 1 and then get divided down the more channels you use. It seems like a higher sampling rate might offer more benefits compared to larger bit depth for hobby use?
Still - great to see this level of performance coming at such competitive pricing. And the possibility of portable power is always welcomed.
I will attempt a comparison with a digital photocamera
The resolution of an oscilloscope is made out of ADC resolution on vertical so 8bit (256pixels) or 12bit (4096pixels) and sample rate on horizontal 1Gsps (1 billion) or 1.25Gsps (1.25billion)
So same as you can not see the full resolution of the camera sensor on the small build in preview LCD you can not see the full 12bit resolution unless you have an 8K screen (8000x4000pixels) but the information is available in the camera and oscilloscope memory and you can take that out and watch it on a large screen or zoom around on a smaller screen to see the details in some parts of the photo/waveform.
So the LCD resolution on an oscilloscope is about as irrelevant as the LCD screen on a photo-camera.
Now for for the 4 channel that will be equivalent to 4 lenses on a single photo camera that can be oriented in any direction and the ADC will be the sensor.
If you have a single sensor 4096 x 1 250 000 000 pixels then when you want to use a single lens you have access to entire resolution with two lenses half on the horizontal for each and 3 or all 4 just a quarter.
If you have two ADC/sensors each with 256 x 1 000 000 000 pixels then when you use a lens you have a lower resolution as it can project on only one of those sensors and resolution will be lower than on the Rigol.
When all 4 channel's are used then each can see 256 x 500 000 000 pixels on that Siglent and that is better on horizontal but not on vertical 4096 x 312 500 000
At 70 or 100Mhz it will not make much of a difference if you have 500Msps or 312.5Msps but if you are interested in the vertical resolution then 4096 will be quite a bit more than 256.
You can always turn off the other 3 channels and then you still have slightly better horizontal resolution with 1.25Gsps vs 1Gsps
In therms of capturing a waveform you will be limited by the available memory for the horizontal.