that xeon is cooking man hahaha, I hope you can get what you want out of that platform but I guess the TDP of those chips is pretty high compared to newer platforms. I think Ivy bridge might be the sweet spot of overclockability and low heat load that is perfect for TEC cooling, ie more likely to reach 5ghz and also pretty low heat load, sandy bridge is better for 5ghz but higher heat load, haswell is similar heat load but less likely to reach 5ghz. Skylake is just too expensive but it seems like a good option for tec cooling too.
thats the funny thing, I can boot 5ghz at 1.475v, maybe even 1.45v, I can't remember now. So it's possible I just need to play with things a bit more to find some stability. play with the input voltage and cache voltage and maybe system agent, might need to play with the FIVR frequency etc. It's also possible that it's my mismatched ram that is causing bsod's.
I bought the Hardware Labs Alpha Extreme III 360mm rad. It should help me hold my temps far better. Also I've worked out a way to fit all 3 rads in plus my chillers when it all gets here, it will be cramped, but it'll fit, just. So I should be able to get away with a bigger TEC. I have my eye on a 4670k for a reasonable price. 4770k would be nice though as I use a few programs that could do with 4 cores and 8 threads.
My water block isn't ideal. I can't even mod it as I think it would make it worse, I wanted to tap a G 1/4 port in the top left of block for new inlet, outlet is at bottom right, but the inside of the block has a channel that goes right around the outside next to the o-ring, so the water would actually divert round the outside rather than going through the very high restriction micro pin structure. As it is now the water is blasted into the center of the pin structure and is then forced out in all directions and gathers in that perimeter channel then goes out the outlet, I know it won't work perfectly but it is better than modding the top. However I have thought about making a new top myself from some acrylic that has two ports on opposite sides that has a channel spreading the full width across the block but not connected to the other outlet, this way the water will flow in, spread across the full width in the channel in the top, then be forced through the pins in the base to the outlet of the block on the other side. I think I can do that with a drill press if I take my time.
this is how my block works now http://www.legitreviews.com/images/reviews/1160/swiftech_apogee_xt_014.jpg
I have revision 1 version of the top, https://martinsliquidlab.files.wordpress.com/2011/05/apogeextrev2top6.jpg?w=614&h=364
so you can see if i tap a g 1/4" port at the top left corner it will just flow around that outside channel to the outlet port and not through the copper pins, so I'd lose cooling in the center and most likely it would be very bad. revision 2 probably even worse for TEC.
The base of my apogee xt is a very fine pin matrix over about 45x45mm http://minidriven.com/BlacknBlueforIB/PoorlyMAdeApogeeXT.jpg
so if i make my own top for it similar to the newer apogee xl top it should work better for a TEC http://tmz-media.com/images/reviews/swiftech/apogeexl/the_mod_zoo_review_swiftech_apogee_xl_11.jpg
I want to cut channels in the acrylic like the apogee xl top but keep the pins for the full 45x45mm like my apogee xt base. If' I could find the apogee XL to buy anywhere in australia I would but it's only available as part of an AIO here, and also the pins only cover like a 30x40mm section of the base so it's not ideal anyway.
they way the block is now may actually be working okay as the water is forced out from center in all directions so it should be getting flow to most of the pins above the TEC but I'm pretty sure the majority of flow would hit the center of the base and then just flow right towards the outlet so one side of the TEC may not be getting cooled efficiently.
I ran some Cinebench R15 today, its funny, in the CPU multi thread test my cold plate temp rises a few degrees, but in the single thread test it stays the same temp the whole way through. So it's a combination of not enough Qc from TEC and not enough rad space for now. Sorted the rad space, now time to find a bigger TEC. I'm trying to work out how to order a TEC from thermonamic lol.