Depends on the situation.
If you are to test a rad, in a given setup, comparing it to another rad is easy.
If you are to compare waterblocks it's a more delicate matter...
What i'd like to see, is a max OC freq, defined by the first BSOD in a set of CPU loading programs.
Benchmarks are useless.
Take that same 1600+ (for example) and see how far it goes with one block, then another (well repeat etc., you know the drill).
The ideal would be a set of CPUs, one for each core variant. One TBredA, one TBredB, one P4 Willamette, etc...
And see how far in MHz they go.
Obviously, if a block allows consistent higher overclocks, why bother about temps ?
(edit) ah also i want price, avaibility, ease of maintenance, and mounting options.. oh and pressure drop in mH2O
|