I’m not aware of any measurements on this layout. Or on rotated air coolers with recent gens, either with or without dGPU obstruction. So if you want a cookbook it looks likely you’d need to write it. The absence of data makes guesstimating dB(A) and °C disadvantages difficult, so it’s hard to quantify the size-cooling tradeoff.
A caution I’d make is case perforation and pattern have strong influences on flow resistance and load noise. From the various cases I’ve built and measured and Jonsbo’s product photography I think it’s possible the D41 may have greater build issues there than it does layout limitations.
I’ve been watching for a good mATX-ish airflow case implementation for a while and there kind of isn’t one. Prodigy M and Meshify 2 Mini aren’t bad. AP201 might be a little better than the D41, Pop Air Mini’s limited for dGPU or 3.5s. IIRC Lian Li showed a 205M refresh at Computex but that doesn’t matter until it’s shipping.
For shrink without getting into ITX the Jonsplus Z20 strikes me as something of a logical endpoint. Might be worth a look, though since you didn’t pick the D31 I’m guessing this an ATX mobo build rather than mATX. There might also be small tower options of interest but, among the ones I’m aware of, nothing stands out to me.
Either way a dGPU’s a substantial mid-case obstruction and the CPU cooling’s running in its exhaust. It’ll work but IMO part of getting a build right is being mindful of size-noise-heat tradeoffs. Hence the questions around how much of a dGPU. Like a 4060 dual that’s occasionally hitting 50 W’s probably not too much of an issue. For a 7900 XTX at 350 W sustained, Lancool (or similar). The design question’s where’s the breakpoint for going with a larger case but combination of data and build priorities is needed to answer that.
For a dGPU- and 3.5-less build with the D41 I’d probably use front to back dual tower air as a baseline. While PSU’s in the way and Jonsbo’s communication leaves something to be desired, top mid/front, front bottom, plus bottom front intake with reasonable curves and fans that are ok in pull will get air to the cooler (Jonsbo seems to have the PSU exhausting upwards). Rotating the cooler and measuring bottom to top airflow should then give decent initial data to work from. If you already have a dGPU around you could also assess its effects at idle and active.
Alternatively, going with AMD’s liquid cooling recommendation for the 9900X (I won’t be in my build) pretty well indicates for a 360 in top exhaust with 3x120 bottom intake. Since air coold GPU exhaust would mostly go through the radiator in this config, a proxy approach I’d use would be to add the dGPU’s heat load to the CPU’s. At common ambients with IMO a noise tolerance towards the higher side a good 360 AIO can handle around 250 W. Subtract out whatever you’re planning for the 9900X and that sets a dGPU power budget. With the 9900X at stock (162 W PPT) and a bit of margin it’s maybe 75 W GPU, though probably DDR should be included. Crossflow from front bottom intake and rear exhaust should get some heat out of the back of the case but I don’t think I’d want to try for more than a 4060 here.
For ambients over 25 °C or so, derate or accept additional noise accordingly. If it’s a warmer day, looks to me like the dGPU power budget could potentially close down to idle under a constant noise constraint with the 9900X at stock PPT. If that’s a likely operating point then a larger case looks to me like it’d probably be useful.
Pretty much anything vaguely current will, though it seems not uncommon manufacturers neglect to specify gen 1 versus gen 2. There’s starting to be some mobo 2x2 availability but offhand I can’t think of any cases indicating 2x2 support.
This is supposed to be my small desktop.