[Long Haired Git]

Wow, thank's for that quote.
Right now for any pump I don't have test results from that french site for, I assume a heat dump into the coolant of between 50% (zero flow) to 70% (max flow). This was based on the normalised average of the pumps that were tested.
Compare this to the MCP350 which when tested at full flow dumped just 53% of its draw, and from other figures I know at zero flow it would drop just 42% assuming linear power-draw to heat-dump vs flow rate.

So, for a Storm G4 and HE120.2 (performs similar with fans @ 7v?) I get the following with 2m of tubing and ignoring any non-standard barbs on a 90W CPU (simulated die by Robotech actually).
All numbers are out of my approximator and thus can be, well, just plain wrong.

MCP350: 3.09 LPM, coolant temp at exit of radiator of 4.48 degrees Celsius above ambient (°Caa), CPU die temp at 17.63°Caa
MCP655 @ 5: 5.78 LPM, coolant temp at exit of radiator of 4.83 °Caa, CPU die temp at 17.11 °Caa
Iwaki 20RT: 6.79 LPM, coolant temp at exit of radiator of 4.65 °Caa, CPU die temp at 16.73°Caa
Iwaki 30RT: 7.97 LPM, coolant temp at exit of radiator of 5.3 °Caa, CPU die temp at 17.2°Caa

Let me know if you need other pumps (RD20 diffs to RT20) and where I can find PQ graph for them.
As per above, MCP pumps are measured heat dump, Iwaki's are assumed to range between 50% and 70% so if you have a reference to a better figure, please provide....