I wanted to post my overclocking and undervolting results with this specific card. Hoping this might give other folks a head start in their efforts to do the same or provide others considering buying this card reasonable expectations with respect to performance.
 
First, here are my system specs. The case, an NCASE M1 is a 12.7L case and is very space limited. This GPU is known to be a hot one due to its compact size, so I wasn't expecting temperatures to be any where near as cool as the FE version or other AIB's. My goal was to pack as much performance in as small a case as possible.
 
System Specs:

• Case: NCASE M1 v. 6.1
• Motherboard: MSI MEG Unify Z490 LGA 1200 Intel Mini-ITX
• CPU: Intel Core i9-10900 10-Core 3.7 GHz LGA 1200 (game boosted)
• Memory: Crucial Ballistix 3200 MHz DDR4 DRAM 32GB (16GB x 2) CL16
• SSD: Western Digital WD BLACK SN750 NVMe M.2 2280 2TB PCI-Express 3.0
• PSU: CORSAIR SF750 CP-9020186-NA 750W SFX 80 PLUS PLATINUM
• CPU Cooler: NZXT Kraken X53 240mm - RL-KRX53-01 - AIO 
• Case Fans (Bottom Intake): Noctua NF-A12x15 PWM chromax.black.swap (120x15mm, Black)

I have the fans on the CPU radiator and the Noctua fans mounted on the bottom of the case set to intake using custom fan profiles trying to minimize noise from the system as much as possible.
 
For overclock testing I used MSI Afterburner, HWInfo, and 3DMark's Port Royal GPU benchmark software. First, here are the testing results using GPU stock settings: 
 
GPU Stock Settings Performance

• 3DMark Port Royal Score: 12,468
• Power Limit: 100%
• Temp Limit: 83C
• GPU Temp Max: 74C
• GPU Power Max: 342.5W

For overclocking I first maxed the power and temp limits for the card in MSI Afterburner (104% power limit and 91C temp limit). I then stepped up the core overclock by 50 MHz, ran 3DMark Port Royal 3 times and if the tests completed and no crashes I stepped up another 50MHz until i had a fail to complete event or computer crash (BSOD). Using this method I was able to achieve three complete benchmarks in Port Royal up to a +250 MHz core overclock. Next I started up Call of Duty: Cold War to see if I could get these higher overclocks to remain stable for at least 30 minutes of play time. This test brought me down to a preliminary "stable" overclock of +155 MHz. 
 
Best Stable Core Overclock Performance

• 3DMark Port Royal Score: 13,358 (+7% over stock)
• Core Overclock: +155MHz
• Power Limit: 104%
• Temp Limit: 91C
• GPU Temp Max: 75C
• GPU Power Max: 365.3W

Having established my best core overclock, I then proceeded to see what memory overclock would be stable as well using the same testing methodology. I began with a +600 MHz memory overclock which passed fine and then proceeded to test at +800MHz. +800MHz was not stable in 3DMark so backed down to +700MHz, which passed fine, and stepped up to +750MHz, which was also fine. From here I stepped up the memory overclock at +5MHz intervals until I reached an unstable setting. Best setting ended up being +770MHz.
 
Best Stable Memory Overclock Performance:

• 3DMark Port Royal Score: 13,373 (+0.1% over core overclock results)
• Core Overclock: +155MHz
• Power Limit: 104%
• Temp Limit: 91C
• GPU Temp Max: 77C
• GPU Power max: 365.4W

Performance increase is negligible when overclocking memory based on my results. The added temperature is probably not worth it but just thought I would present my results in case a memory overclock is important to you. Additional memory overclock could probably be achieved with less or negative core overclock.
 
At this point I wanted to see what undervolting could do to reduce temperature and noise from the system. Knowing what the potential GPU overclock is I was able to roughly assess how low of an voltage I could use while maintaining the same performance. This card is advertised as having a boost clock of 1725MHz but in my testing I have found that it typically reaches 1950-1965MHz max regardless of how much additional power I provide. Max voltage is typically 1081 mV using the above overclock settings. Knowing this my goal was to adjust the Voltage/Frequency curve in MSI Afterburner so that the peak frequency was set to 1950-1965MHz while stepping down the voltage until the required overclock at lower voltages did not exceed the +155MHz core overclock setting. This put me in the 937mV+ range. I found setting the frequency plateau at 1950MHz was best suited to a 950mV voltage setting. The MSI Afterburner Voltage/Frequency curve is a bit quirky depending on what the current GPU temperature is. From an idle/cool state I typically just drag the curve down so the plateau is below 1950MHz frequency and then dragged the 950mV data point up to 1950MHz and applied. Note that my undervolting tests did not include memory overclocks and the core overclock is determined by the Voltage/Frequency curve set as described above so I did not expect it to perform similarly in 3DMark. That being said, here are the results:
 
Undervolting Performance:

• 3DMark Port Royal Score: 12,887 (+3% over stock)
• Core Overclock: Curve
• Power Limit: 104%
• Temp Limit: 91C
• GPU Temp Max: 75C
• GPU Power Max: 366.1W

End result is that my undervolting didn't really reduce temperatures, noise, or power consumption as I was hoping. This is probably due to the restricted case size and case airflow. As a result I will likely remove the undervolting settings and apply the core/memory overclocks as tested above for maximum performance.
 
Please let me know if there is anything you recommend to reduce temperatures/noise without hobbling performance significantly.