The short answer is: it depends.
If you setup the blocks in parallel, you'll have less restriction in the total loop, which will allow the pump to give more flow, but it will be split between the GPU and CPU block, and it usually turns out in favor of the less restrictive GPU block, so you have to add a restriction (valve) to balance it.
If you setup the blocks in series, the pump will have to fight a higher restriction, so you won't have as much flow, but all of it will go into the blocks.
Given that most pumps are already running pretty high up on their PQ curve, reducing the pressure may not increase the flow enough, so that a 50/50 split is actually better, for each individual block.
And this is where the answer is "it depends".
You'd have to measure the effective flow rate, in each block, for both configurations. Just lay it out on a workbench (i.e. kitchen table
), run a loop like this:
#1: bucket->pump->rad->y splitter>blocks (and valve)->bucket
#1: bucket->pump->rad->block1>block2->bucket
Of course if you tell us which blocks and pump you have in mind, we might be able to tell you outright.