Why is that an issue?
Please be more specific...you still buy snaphats for the controller RTC and you can still get the DALLAS L1 RTCs (see vvuk's thread). if both are gone...there is nothing to clone...if yours are dead...there is nothing to clone. You clone your data to a new chip to side-step NVRAM setup again...but that's minor unless you're running non-defaulted values. If you place SOMEONE else's cloned data in your NVRAM chip your system WILL simply either complain or overwrite it immediately because the serial wouldn't be right (serial mismatch). Plus you cannot clone a DEAD chip as it has no working battery to preserve contents...the only reason vvuk was even able to clone his is he mistakenly thought he had a dead L1 RTC..but the fact that is still has its data means...it's not dead...it was the snaphat (usual signs, wrong time in Irix and loss of NVRAM parameters are SNAPHAT not L1 RTC).
What do you expect to clone? You NEED a functional snaphat to run the controller RTC FIRST if you have both RTCs DEAD.
Once you have a working controller RTC (and have booted/power-on once), simply add a L1 RTC...if you want to keep a dead one in the O356 you can, you just need to be aware that the default parameters will be programmed on every boot, thus you may need a double boot (cold, then reboot) to start the system.
Modern SGIs can auto-program their serials (unless you're part of a rack system, where serials come from someplace like an L2). Serial is stored in the atmel chip with sysid and MAC address. The serial IS only copied to the RTCs, it's not it's native home. The Atmel chip beside the L1 RTC is its native home (non-volatile). However there is an order for the automation to work, so you CANNOT have BOTH RTCs with dead batteries, that confuses things.
This was proven on irixnet this year, cloning was done on several systems to prove there is only ONE failure point and it's on Tezro. If you place a USED L1 RTC from a Tezro into ANOTHER, different, Tezro it will NOT correctly wipe the chip and go weird without a working serial and halt, forcing you to fiddle with reprogramming the RTC or wiping it another SGI. Any other system will work correctly and even a Fuel RTC to a Tezro swap will wipe correctly. Only Tezro to Tezro will fail. O350 was tested as well, issue was not present...any old, used, L1 RTC can go into an O350 and will be wiped and provisioned correctly.
Unless it's part of a rack, then serial setup is 100% different and most people WANT dead L1 RTC due to this reason on racked O350.