The 7k/m per rad collector benchmark was with your PR that bumped the power to research point conversion rate from 0.00001 to 0.01. And that value was from an actual test with TGcode. For those completely unaware,
here is OP's original PR that insanely overbuffed rad collectors, and
here is my PR that reverted it and went through how completely ridiculous it was.
That being said though, I've noticed that the supermatter's power output is actually lower than what's supposed to be intended, I'm not sure if the supermatter or rad collectors that were nerfed some time between last week and last year, but the setup I had that was able to produce a peak of 5k research points per minute per rad collector while also heavily irradiating literally 60% of the station now only produces a measly 500 a minute last time I toyed around with TG code. I'd actually be completely unopposed to bumping the research point conversion rate from 0.00001 to 0.0001, since it seems like rad collector power production was reduced by *0.1 across the board on TG. Or, hell, bump up the base conversion rate to 0.0005, and make the conversion rate decline linearly with power production in such a way that the expected maximum still remains 5k per minute per rad collector, and if someone calls you insane for making the baseline conversion rate be 0.0005, just take a screenshot of this post and flood my DMs with links to the PR. The conversion rate is balanced with engineering autism in mind, and as such, it should take no less than 30 minutes to complete RND with rad collectors alone if you really know what you're doing, and that's a good idea to always keep that in mind when adjusting the values. If you want some basic pointers on how to produce obscene amounts of research points with rad collectors, take a look at how fusion and tritium both work, and then look at ways to make the supermatter handle an intense hellburn. Just be sure to have your doctor prescribe you copious amounts of ritalin before you fully sperg out.
Another potential option for rad collector research point conversion would be to make research point production scale exponentially rather than linearly, which would be a huge, *huge* buff to engineering autism but would make it much easier for the unrobust to see the insane potential it has just from basic experimentation, as increases in radiation would have a much more intense impact on the research point production of rad collectors
EDIT: Also, you do not need to use the incinerator to make tritium. Radiation collectors turn plasma into tritium for you, you do not need to exit the engine room at all if you want tritium. Literally all you need to do is premix o2 into the roundstart plasma tanks to have enough tritium for a good 10 minutes of research point production after about 10 minutes of power production. Unless you're able to enslave the science department to mass produce tritium for you, making tritium in the incinerator simply is not worth it when you're speedrunning RND with the supermatter.
