i7 CPU block.

[Petra]

These results are from a review on the i7 940 @ 4ghz w/a relatively average flow pump.

Not to put anyone down or anything here, but one glance at that and I already see a problem...

"3mounts each...no variance...ended up with same temp values..."

Sorry, but there's always mount variation and if a test system isn't precise enough to even detect it, then the system is not fit for waterblock performance evaluation. Far too uncontrolled/imprecise to be regarded as anything more than 'anecdotal user data'.

[Hondacity]

Not to put anyone down or anything here, but one glance at that and I already see a problem...

"3mounts each...no variance...ended up with same temp values..."

Sorry, but there's always mount variation and if a test system isn't precise enough to even detect it, then the system is not fit for waterblock performance evaluation. Far too uncontrolled/imprecise to be regarded as anything more than 'anecdotal user data'.

dunno how you mount your blocks..i just tightened them to a point im just hurting my fingers... remounts are easy... i get the same temperatures..over and over...try real temp on the i7.

after mounting then logging temps...remove the nut...clean the block..clean the cpu..put in some paste...hand tight..the repeat the process...

easy as 1-2-3...except it takes awhile

[Petra]

dunno how you mount your blocks..i just tightened them to a point im just hurting my fingers... remounts are easy... i get the same temperatures..over and over...try real temp on the i7.

Let me state this once more: you will always have some mount variation and your statement only implies an even less controlled test environment than I had previously thought. Whether you're properly using the included stock mounting hardware or a custom mounting configuration which applies a known and equal (or, as equal as possible) mounting load to all blocks being tested, you're still going to end up with some variation. You'd get this same information whether you asked me, Scott, Bill Adams, or Lee Garbutt. The best that any tester can do is take the mean of this data, whilst being sure to note the observed data range.

Moving right along, a CPU's internal temperature monitoring hardware is near worthless when it comes to controlled testing. You're working with a device of unknown calibration (that you can't calibrate properly), unknown and questionable accuracy, unknown and questionable precision, and unknown linearity. From a design consideration standpoint, the on-die [point] temperature sensors are little more than an overheat alarm--they just have to be good enough within a certain range so that the CPU doesn't fry. That's it.

[Hondacity]

Moving right along, a CPU's internal temperature monitoring hardware is near worthless when it comes to controlled testing. You're working with a device of unknown calibration (that you can't calibrate properly), unknown and questionable accuracy, unknown and questionable precision, and unknown linearity. From a design consideration standpoint, the on-die [point] temperature sensors are little more than an overheat alarm--they just have to be good enough within a certain range so that the CPU doesn't fry. That's it.

same mounting ..same cpu..same temp sensors...same fans speeds..same paste...same pump... oh well...i'm just a noob tester