@FrozstOk, thank you, but I think your issue was because you didn't install the chipset drivers first. Sadly, it didn't solve anything for me, but I think I've found the hardware that causes the problem.
@mackbolan777I've found out that the computer booted up after some Windows crashes: almost the same time passed until it was "ready" to boot, so I had an idea, today I tried first booting up the computer in BIOS and staying there for 3-4 minutes before booting into Windows, and it worked! I've booted successfully on first try. At this point I really think that something is wrong with the PSU, and if you think that too, I'll get a replacement, and also buy an UPS to avoid such problems in the future.
Also, remember that bzz sound inside the computer, wonder if that's really the PSU, but if it is something like the mobo or GPU, is it normal?
GPU can make that noise and it's called "coil whine". The PSU can do it as well. Best to listen carefully with the side off and all fans disconnected. If it's the PSU, replace it. The GPU, well it's pretty normal unless it's unbearably loud. I replaced my Corsair RM1000 that was from 2013 last week just to be "safe" since I bought it remanufactured back then. I had a buzzing sound and thought it was my former RX 5600 XT but it did turn out the PSU was it. If you're not doing so, run 2 PCIe power cords to the GPU. Running 1 cord was fine with older cards, these newer ones are power hungry and spike when booting or gaming at high FPS. A single cord is good for ~300W and 9A, an RX 6800 can pull that under a spike or high FPS game and that can cause a crash. The card pulls roughly 2.5W per frame and uses ~14A at full tilt, exceeding the load rating of the power cable.
Using a UPS vs. a power strip with surge protection is always best but if the PC is crashing, it won't save the OS from getting corrupted from frequent improper shut downs. Most UPS have "line conditioners" built in and that keeps the electricity "clean" or noise free, another plus. I use an APC 1500vA but you can get away with a far less product. It depends on how much run time you want in a power outage. I can run ~45 minutes with the A/V turned off and no gaming.
If no screws, 2 other things I've seen are forgotten, unused, standoff's shorting the board and damage to the back of the board from installing the bracket for certain CPU coolers. The latter takes only one scratch in the board to mess it up for good and make for crazy problems while still booting. Also, did you check all the USB ports and SATA's to be sure none got broken and might be shorting? I had a case with a broken USB drive me nuts until I found the one port with the plastic tab missing and 2 contacts touching!
Thank you for answering. I really think it is my GPU making that sound, but even though it is the PSU, doesn't matter since I'm going to replace it, since right now the computer only boots when I stay in the BIOS for at least 2 minutes. The sound is not even loud, it is just hearable when no other background noise is there.
I also think I'll buy a UPS to avoid such problems in the future (APC 500VA/700VA) and I learnt by myself that it is always good practice to have one.
I also correctly installed the standoffs, and I really think (and hope) I did not scratch the motherboard (I mean, it booted for 2 months and I did not really touch it recently), and the only case's USB header I have connected rn is the 3.1 one, since the usb 2.0 is taken by the wifi/bluetooth card.
As soon as I'll get my new PSU and I'll get to install it, I'll update this post and mark it as answered. I am really thankful to everyone who posted here, if it wasn't for you, I probably wouldn't have been able to isolate the problem and find out which really was the cause, and probably spent weeks trying to.
Are you using the latest BIOS F12? It seems odd that more than 2 minutes in BIOS would stop your problem. It takes a few seconds for capacitors to build a charge in a defective PSU and if you had a poor "power good" signal, you wouldn't boot right away, it would require multiple power button presses. At any rate, good luck on that PSU part. Now it sounds more like defective board, very possible with that model. Since the problem occurs with either SATA or NVME, it could be the onboard controller since you need to wait at least 2 minutes to boot. It sounds like a low level driver not loading timely, pre-OS meaning in the BIOS chip.
I am using the F11 BIOS that was latest a few days ago, anyways, I didn't test it properly, but I think at least 1-2 minutes are necessary, if using graphics drivers, otherwise it boots right away.
I suppose it could be the PSU since the GPU drivers will call for power to the card when Windows posts. We'll need to see.
If the card doe the same thing, RMA it. If not, follow below.
You're at the checklist area of troubleshooting at that point. Boot the PC and with drivers and see what codes it puts in the Event Viewer that are written even after a "hard crash", usually a few minutes or seconds after the exact time of crash. They won't be right at the time of crash, usually slightly before the will be a "AMDKDG" type error or "Processor Hierarchy", something along those lines with a code 41 under "System". This might help narrow down the failure area. If there's a lot of CPU related errors, RMA that. See if it only happens when the driver is installed. If you haven't, roll back to the first BIOS that supports the Ryzen 3000 series. Moving on.
I would make sure all other hardware was tested like RAM form start to finish with 800% coverage to be sure. MemTest Manual (hcidesign.com) is the one I use. Try running just one stick of RAM, if it stays stable, there's most likely a board/CPU problem. Look for obvious damage on the motherboard you might have missed, bulging capacitors, cracked PCIe slot, etc. Test the SSD with the OEM test most manufacture's have on their sites. SeaTools by Seagate works on most SSD's that are more or less generic like ADATA. Pull the CPU and look for bent/missing pins, use a magnifier to be sure.
If all else fails, it could be a defective motherboard. That model wasn't Gigabyte's finest work. I almost used it due to how little it cost but skipped it for quality concerns after reading many poor reviews about it. The fact it's a B450 chipset running a Ryzen 3000 series is sketchy at best. That chipset wasn't made for that CPU series, it was forced by a BIOS upgrade to work. It's the lowest on the "totem pole" considered to be able to "handle" the 3000 series.
My gut is if there's nothing wrong with the card or the RAM, the CPU looks fine, replace the motherboard first, CPU is probably ok because there hasn't been a huge amount of complaints on that model. If you want to be certain, take the whole thing into a well qualified PC shop and let them test the entire PC. Anything we do here is flow chart based and some shooting from the hip. Truth be told, if it were in my hands, I'd be able to tell you for sure in about an hour what it is. So a qualified shop is your safest bet for any money spent.
I did already Memtest the whole ram, and drive too, also I found some errors in the event log regarding BLTHUSB, but I guess that was because it was missing drivers (indeed it disappeared), also I am using a B550M not a B450
My bad I didn't realize Gigabyte made a B550 version, my opinion was based off the B450 version of the same board. The B550 is much better but still on the low end scale. However, reliability did increase quite a bit when they moved this board to the B550 chipset. Not many complaints at all. That's not to say this one doesn't have an issue. Your problem is either a bad GPU, CPU or board. The board we can almost cross off since the RAM passed, I'm assuming multiple passes? Now the missing driver for Bluetooth USB is interesting in that do you have such a device connected? That refers to a USB Wi-Fi adapter or onboard chip/slot for a Wi-fi card. I'm pretty sure your board doesn't have onboard Wi-Fi.
When you install the OS are you using a USB drive with just Windows 10 on it? Is it the most recent 20H2 and made via the Windows Media Creation Tool? Just trying to make sure there's no OS corruption happening. If none of that is at issue, you're back to pulling the CPU to check for damage, RMA'ing the GPU, or taking the whole thing to a shop to have someone look at it. There's the option of swapping the card into another PC just to see as well.
To be clear, it won't run right on the 20.8.3 or 20.9.2 driver? The driver is available via this link anywhere in the world :AMD Radeon™ RX 5500 XT Previous Drivers | AMD
Use DDU in safe mode: Display Driver Uninstaller Download version 188.8.131.52 (guru3d.com)
If you can't locate the driver off that link, search it out in your country, it has to be available. Worst case is let Windows Update pick the driver or let AMD's site automatically install the newest one out and see. It just came out the other day.