This may have been discussed but is there enough room to shift the gpu mount up 20 or so mil and get two rads into the 580? I was eyeballing the renders and with 3 slots worth of space on the GPU side a pump and small res like an old school Swiftec Micro res and pump could fit with a single slot WC GPU,
Micro Res
Pump
Or even the Alphacool Solo used purely as a pump/res for a home brew GPU AIO with rad and waterblock for the gpu. Not sure if the pump in this could push a separate cpu and gpu block and say 2 240 rads? Brainstorming on how to do a full loop in one of these cases. OR just a GPU WC loop in the 570 and air cool the CPU.
If something like this were possible I could shrink my CerbX custom loop build down to 16 or so L with a mobo swap. Pretty sweet
Based on previous discussion with Ksliger, my current plan for 2x280mm rad cooling is Top with rad+fans and bottom with internal rad and external slim fans and fan cover. While not the *most* ideal setup, it should be more than adequate for most builds. Ideally I'll be running a temp setup with my current build (8700k+2080) using cheaper water blocks (instead of full face motherboard, etc) and soft tubing, before I upgrade when PCI-e 5.0 and DDR5 hit consumer products, likely 2020/21 (PCI-e 4.0 adoption was stunted aside from in data center applications due to the quick turnaround to PCI-e 5.0).
Once that happens, my ideal goal is:
High Core/Clock AMD CPU (likely AM5)
NVLink/XFire/QPI(?) GPUs (depends on workload at the time)
Full face monoblock
EKWB(?) GPU blocks
Pump/Res as displayed in some renders.
Unfortunately I'm likely going to have to order different PCBs with clock buffers or design my own as the one offered with the case are side-by-side and don't offer enough space for the NVLink bridge (Alternatively, it's possible that the NVlink connector is passive, in which case making my own side-by-side would be simple enough, though additional research suggests a clock buffer chip also exists, so I'll have to do some PCB shot comparisons to see the difference between PCBs [and reading the datasheet] to try to interpolate any required changes, should any be necessary.) should I go the Nvidia path, which is increasingly unlikely as I really don't care to do driver/registry/bios trickery to get SLI to be recognized. I'm very interested to see Intel's GPU offering, and hope they go chiplet or 3d/stacked rather than monolithic.