(I'll just comment on quotes from everyone here in one reply)
You're forgetting that controllers are involved in all of this. Said controller won't speak to IC's just like that unless they're programmed to.
So... Is it possible to reprogram it to "accept" the "new" RAM?!
The other aspect would be practicality/feasibility.
(This is only out of curiosity...)
DDR3 and DDR5 chips have completely different pinouts. They use different voltages.
I understand about different pinouts, so if you were to (re-)read my post more carefully, you'd see that what I'm saying is that if you were to unsolder everything from DDR3 and DDR5 and swap components from one stick to another, would that in some way make it work on the respective CPU? That is, say, DDR3 on i7 14700K and DDR5 on i7 4770K
About voltage - there are quite a few 1.35, 1.5, and other voltages on both DDR3 and DDR5. Since there may be a lot of technicalities over "how" RAM gets it, that may be another factor. But either way, is it possible to make voltage regulator work like the same way as the one on the other RAM?
OK... still, would the computer start with the swapped components on the RAM sticks?
IIRC Core2 could support DDR2 or DDR3 and X99 boards could come with DDR3 or DDR4 with different types to boot but the board itself would support one or the other.
I think there was the same situation with 10-series CPUs or something like this, where a CPU had IMC for both DDR4 and DDR5, allowing users to choose either DDR4 or DDR5 motherboards, but there was no way it could support a mix of DDR4 and DDR5 on one mobo. It's just unpractical. What I'm asking is not about practicality or reasoning or whatever else that may be smart, it's... to some extent, the opposite, just out of curiosity.
I understand about pin layouts and everything, and that you can't put a DDR3 into a DDR5 slot, and so this is not like "let's do it and test it" type of situation. It's cumbersome to unsolder all these small capacitors and stuff, and there's no real need to, unless you're doing it for an experiment (just like what I'm asking about; asking questions, not to do it). This wouldn't work with a CPU, that is, if you were to remove a die from the same 4770K and 14700K and swap those... that would never work because of different LGA pins and pin layout, like, if there's one pin (or land, or whatever) is ground on the die and the same pin is power... so that's a bit more complicated (and by 'bit' I mean 'everything') so... but with RAM... I understand that they are different types of RAM and everything, but chips are chips, capacitors are capacitors... Or not? Say, if the capacitors on DDR5 can store more and upon soldering them onto DDR3 they blow up because of storing excessive amount of whatever... that's one thing. If it can work (in theory) then... What happens? Does it just take all the energy it can (or whatever it capacitates) and work the same way as on DDR5? Same with voltage regulator and other components.
Finally, here's what I found on
Wikipedia: When the memory controller is not on-die, the same CPU may be installed on a new motherboard, with an updated northbridge to use newer memory. (I understand now with IMC's it's irrelevant, but still...) So in theory, at least, say with DDR and DDR3... Would this work?
Again, I understand that with each new generation RAM, as well as any other component, becomes more and more complicated, and while DDR3 may utilize, say, 16 chips for 32 GB, DDR5 may utilize, for example, 4, so that does make it a bit more complicated, but... can it work?! If, say, there are both DDR3 and DDR5 with the same amount of chips and capacity and voltage
Either way, to finally finalize this reply,
@Lutfij 's words make the most sense so far... Yet they didn't say much either...