Had to reset my bios last night on G436. Now performance is lacking

My apologies for the late response, I left the office for the evening and ran into some PC issues of my own (leak in my custom loop). Couldn't get it resolved by the end of the night either which was a bit of a bummer, but I am here now. Looking over the previous posts, it looks like you may have been able to get the XMP profile to load. Just to double check that to be certain, can you get me a screenshot of the Memory Tab & SPD tab of your CPU-Z application? I want to make sure the primary timings are matching what should be loading with that memory.

@TSDavey walked you through reinstalling the GPU driver which is great as that rules out any weird complications on that end. Once we get confirmation on the memory speed, we can move on with other areas from there.

That might explain the performance issues, you'll definitely need to reinstall the latest driver after running DDU. I also agree with you on the BIOS, it's why we try to customize them right the first time as they can be finicky to deal with. My board at home is also an MSI board (B550 Unify X) and it has... quirks. If I manually overclock my memory under the OC section, settings under AMD's CBS sub-menu won't change with my overclock settings, so I have to change them manually or else I won't post. I love the board, but these quirks can definitely be frustrating to work through. The things we do for performance, lol.

As for my loop, I am hoping to get to the bottom of it. I have a small hand-pump that lets me pressurize the loop to find leaks. It's not leaking anywhere in the front of the loop which is good, as it means all of my various blocks are fine, but something in the back is leaking, likely an o-ring near one of my radiators. Leaks are not common, but I built this loop in a hurry and it's possible I didn't chamfer the tubing as nicely as I should have, so I get to pay the price now.

Either way, I hope that I can get my problem solved by then so I can be on in time to answer any questions you may have. If not, I'll see if I can find time to check the forum from my phone every now and then.

Everything here looks normal, definitely confirms XMP has been loaded correctly. A DRAM frequency of 1600mhz = 3200mhz effective (double data rate) and those primary timings are exactly what we should be seeing.

Just a quick question, are you running through the same areas of the game each time or using an integrated benchmarking tool? It's also possible that different areas of the game will hit your hardware differently and result in differences in performance. Some areas can be densely populated or have intense physics that hit the CPU really hard which can bottleneck the GPU. Having the data from Afterburner/RTSS could help us determine if this is happening, so if we can see that data, we can figure out what is going on here.

Thank you for that video, that is definitely helpful to better understand what is going on here.

I wanted to show one very specific part that I noticed:

This was around the 16:55 mark of the video, and was by far the worst dip I was able to see.

Looking at the resource usage, none of your individual CPU threads peaked at 100%, meaning we know that the engine itself wasn't bound by a single thread during this scene. We also see that the GPU is still pegged at 99% usage and boosting to 2085mhz, meaning it's utilizing all of the GPU core resources that the card has available to it and isn't thermal throttling as indicated by the GPU thermals and boost behavior. You are using 13GB of system memory in a system that ships with 32GB, meaning you have plenty of memory overhead left over before swapping to your drive, so we know that isn't the limiting factor here.

So this begs the question, what can cause a frame dip in a situation where a GPU is being 100% utilized and the CPU/memory utilization isn't abnormal? For this title in particular, it's likely going to be a bottleneck in the RT cores. You had RayTracing enabled at medium settings and most of the performance dips occurred when entering vehicles where traditional path lighting effects were occurring from the interior on top of the ray tracing effects occurring from the exterior to light the scene. This is likely extremely expensive in terms of computational resources for those RT cores. This is further aided by the fact that your performance saw a significant uplift when you adjusted DLSS to Performance, making use of those Tensor cores, alleviating some of that computational burden.

It doesn't explain why you were seeing higher numbers before tinkering with the BIOS from earlier, unless you are playing entirely different parts of the game. If that is true, we may need to simulate an earlier save and try to run through earlier parts to see if we can mimic a similar environment in-game to what you were playing before you made your BIOS adjustments and see if the performance matches up.

I just don't see anything out of the ordinary based on the resource usage, as nothing appears to be bottlenecking the GPU unless the CPU clock speed is dipping, as that is the only sensor I do not see monitored throughout the run, but given the usage is quite low, that shouldn't make much of a difference either way. Not sure if CD Project Red's engine scales all that well with CPU frequency in the first place.

1. This one sounds like an odd memory training issue. Typically when your memory trains, it attempts to reboot the system 1-3 times until it likes the timing combination or times out. The fact that it stops doing so and returns no video signal is definitely odd. The next time you encounter this, try powering down completely, pulling the power cable from the system, then holding the power button down for 60 seconds. This will drain any residual power from the capacitors in the system (including the memory) and plug it back in and power back on. See if it ever occurs again.
2. Can you confirm if the fans are spinning during idle? On these modern cards, their stock fan curve is designed not to spin until the cards reach roughly 60C. It's odd that you were able to idle in the low-30's before as any sort of 3D load, even mundane HTML videos on Youtube would cause the GPU to work so long as Hardware Acceleration is enabled for your browser. You can always use MSI Afterburner to impose a very low fan speed at idle to keep those temps down, though it really doesn't matter as long as the load temps look good (and based on your previous video, I'd say they look great).
3. This sounds like air may have found its way into the pump chamber causing cavitation or that the pump may be running dry completely. Definitely keep an eye on your CPU temps and if they exceed 85C under gaming load across all cores, it would be worth bringing in for diagnostics under warranty. This is assuming the CPU hasn't been overclocked and that voltage is still on stock/auto. Overclocking will definitely exceed this temp threshold. If you are comfortable trying to fix this yourself, you can remove the radiator from the top of the chassis, sit the system down on its side, and hold the rad up until the tubes are perfectly straight towards the pump. Tilt the radiator sideways 90 degrees to work the fluid back down into the pump chamber and to get the air back up into the radiator. There is typically a recessed chamber on the radiator where the excess air is stored, and the goal is to get the air back into that location. The biggest problem with shipping AIO's in general is that depending on how they are handled, air can often find its way into the pump before the system is powered on, and it's very difficult for the low-pressure AIO pumps to work the air out on its own unlike that of your DDC or D5 custom loop pumps.

Here is a video somewhat demonstrating that process:

NOTE: This is for Asetek pumps (Cooler Master does not use Asetek pumps in their AIO) however the principles are the same for ALL AIO pumps as the limitations imposed by physics is the same throughout all designs.

I hope this helps!

It's interesting that the cards thermals are that high at idle while the fans are spinning, yet your temps & boost clock speeds looked fantastic during that gaming video you sent. Were you gaming with the side panel off during that video?

As for the chassis, this board is a standard ATX board and will fit in any chassis that supports standard ATX size motherboards. The Corsair 5000D Airflow would indeed be compatible, just make sure you hold on to your original chassis should you need any warranty claims as it will be needed in the event an issue arises.