MPG 321URX QD-OLED: FW.021 rejects slightly off-standard refresh rates (FW regression)

M

msi.g00u15eb02e9

New member
Joined
Jul 31, 2026
Messages
5
Model: MPG 321URX QD-OLED
Firmware: FW.021 (build 20260112)
Input: HDMI 2.1 port (source device is HDMI 2.0 / TMDS)

Summary: Since updating to firmware FW.021 (from a much older version that I quite frankly do not remember), the monitor rejects any HDMI input whose refresh rate is not one of the standard CEA rates (59.94 / 60.000 / 119.88 / 120.000 Hz etc.). Inputs that are only very slightly off, which the monitor accepted on earlier firmware, are now refused: instead of a picture, the monitor shows its full-screen red/green/blue/white/black colour cycling (indicating a problem with the input).

My setup: I use a retro-gaming FPGA device* with an upscaler** that recreates the video timing of classic consoles exactly. Old consoles don't run at exactly 60 Hz, for example the NES and SNES run at 60.0988 Hz. The upscaler outputs modern resolutions (1080p, 1440p, 4K) but keeps that original refresh rate (or doubles it to 120.198 Hz) to have zero-lag, perfectly smooth scrolling. While a niche hobby, this is a common and well-established way to play retro games on modern displays among retro enthusiasts.

Behaviour on FW.021: 60.0988 Hz (and its 120.198 Hz double) produce no picture, just the full-screen red/green/blue/white/black colour cycling, in every HD/UHD mode I tested: 4K60, 1440p60, 1440p120, 1080p60, 1080p120, and several custom timings with lower resolutions. The exact same modes with standard rates (59.94, 60.000, 119.88, 120.000) work perfectly, so the source device and cable are fine.

Interesting detail: 480p60 and 480p120 do accept 60.0988 / 120.198 Hz without any issue. So the panel itself is fully capable of syncing to these rates, only the HD/UHD signal-validation path refuses them is my best guess. The deviation is +0.16%, well within the ±0.5% tolerance that VESA DMT allows a source to emit.

This is a regression: 4K at 60.0988 Hz produced a stable picture on this monitor for as long as I've owned it, across earlier firmware versions. After the single update to FW.021 the exact same source and settings are refused. In addition: the doubled rate, 120.198 Hz, has never been accepted on any firmware, so the update didn't introduce the strictness from nothing, it extended it to rates that previously worked. I've ruled out everything else I could: Adaptive-Sync on/off, FreeSync and VESA VRR modes, DSC on/off, factory reset, full power drain, no difference.

I also note the FW.021 changelog contains one line: "Fixed RGB Rolling Pattern appearing during PS5 game is running. (GT7)." That fix touches the same signal-validation area that produces the colour cycling, so I suspect it tightened the timing check as a side effect. But this is just an educated guess, not fact.

Request to the MSI firmware gods: please restore the previous timing tolerance for slightly off-standard refresh rates in HD/UHD modes such as 1440p and 4k to allow 4k/60.0988Hz and 1440p/120.198Hz for example, or add it as an option (in the service menu if you want to hide it from regular users if you must). The hardware demonstrably supports it (480p still accepts these rates, and older firmware accepted 4k/60.0988 Hz happily). I know retro gaming is a tiny niche within the gaming crowd, but these monitors are marketed and sold as gaming monitors, and for us, those slightly off-standard refresh rates are gaming. So: could we please get them back? We'd be very grateful! :)

* For those interested, this is a MiSTer FPGA device.
** And this would be a Retrotink-4K Pro set to Sync Lock: Gen mode (lowest latency, the holy grail of retro gaming).
 
Sort by date Sort by votes
plutomate said:
When you say earlier firmware, are you referring to FW.020 from MSI website?
Click to expand...
Hi plutomate, thanks for your engagement.

To answer your question: no - as stated in the first line in the Summary portion => I jumped straight from a much older firmware that I couldn't recall to FW.021 (but I have since dug up some more information, see below).

I've looked up my purchase details for when I bought this monitor as I remember immediately updating the firmware on delivery day to whatever was newest at the time. My delivery date was 22 Nov. 2024 and the newest firmware at that time was FW.012. After that I never installed a firmware update again because I simply forgot about it until May 2026.

So my firmware jump was from FW.012 (released 13 Nov. 2024) => FW.021 (released 04 March 2026).

To re-iterate:
1. On FW.012 this panel had no problem accepting 4k/60.0988 Hz. It did however have a problem accepting 1440p/120.198 Hz, that never worked properly (and was also a curiosity at the time, I somewhat lazily expected this to be fixed in a future firmware upgrade).
2. On FW.021 this panel no longer accepts 4k/60.0988 Hz (FW Regression) nor 1440p/120.198 Hz (pre-existing problem). Both result in an input signal error (red/green/blue/white/black colour cycling).
3. I have never tested an 'in between' firmware.

Additional testing since my last post: I repeated the test bypassing HDMI entirely, using an active (externally powered) HDMI to USB-C adapter (Club3D CAC-1336, DisplayPort Alt Mode) into the monitor's Type C input. Standard rates (4K/60.000 Hz, 1440p/120.000 Hz, etc.) display fine through this adapter, but 4K/60.0988 Hz and 1440p/120.198 Hz produce the exact same colour cycling error as on HDMI. This suggests the signal validation is not specific to the HDMI input path but happens in the scaler firmware for all inputs.
 
Upvote 0 Downvote
Do you find any benefit from the new firmware? Or better just roll back to previous firmware. Perhaps not the earliest one but try v.20 or v.16.
The problem you describe look way beyond what user like us can help, better contact MSI directly for help.
 
Upvote 0 Downvote
plutomate said:
Do you find any benefit from the new firmware? Or better just roll back to previous firmware. Perhaps not the earliest one but try v.20 or v.16.
The problem you describe look way beyond what user like us can help, better contact MSI directly for help.
Click to expand...
Thanks again for your answer.

I prefer to stay on the latest firmware. The changelog between FW.012 and FW.021 contains many fixes that are relevant to me personally, and the retro gaming use case, highlighted in this thread, is only a small part of what I use my MSI monitors for.

I have opened an MSI support ticket that mirrors this forum post almost 1:1. Despite all the information provided, the only response I'm getting is "please downgrade your firmware and do the testing/troubleshooting for us, as we no longer have this monitor available" (paraphrased). Support seems more concerned with closing the ticket quickly than with reading the evidence and recognizing that they have a bug, or at least a regression, in their scaler firmware. Given what these panels cost, I don't think it's unreasonable to expect MSI to do this kind of regression testing themselves rather than asking customers to reflash firmware on their behalf.

To be frank, I don't really expect MSI to release a firmware update for retro gaming enthusiasts using niche refresh rates (though I'd be very happy to be wrong). I mainly created this thread and the ticket to document the problem. Others have run into it and more will, and now search engines will index it.

What frustrates me is that the panel is demonstrably capable of these rates: 4K/60.0988 Hz worked for 18 months on FW.012, and even on FW.021 the 480p path still accepts both 60.0988 and 120.198 Hz today. The rejection is purely a validation rule, a piece of code in the scaler firmware, written by MSI or Novatek: a software limitation, not a hardware one.

Meanwhile, a certain other display manufacturer one country to the north, that shall remain unnamed, has been accepting these very signals on far cheaper panels for years, so I'm confident this is not exactly an unsolved or impossible engineering problem. The MSI panel is one firmware patch away from retro gaming greatness. It's just a damn shame is what I'm trying to say.
 
Last edited:
Upvote 0 Downvote
Addendum: curiosity go the better of me, so I performed the FW downgrade to FW V0.20 and it didn't change a thing (also communicated this to MSI support). Back on FW V0.21 now.
It won't let me downgrade any further. FW V0.16 gets rejected by Gaming Intelligence with the message: "Please download correct firmware file version. FW.0XX to update scaler firmware."

So it seems I'm now stuck on this FW V0.20/21 train.
 
Upvote 0 Downvote
Nichrome said:
Just don't brick your monitor 😓
Also let us know if MSI responds with anything usable. I'm interested on what's happening here.
Click to expand...
I'm trying, haha. So far the flashing and re-flashing of FW doesn't seem like a high-risk thing, as long as you don't touch the bootloader/recovery block (which isn't where this artificial 'gate' lives). I've done 15+ flashes in the past day because I can't let go of things, and we live in the age of AI where, for better or worse, it can attempt to reverse-engineer firmware. Yes, this voids the warranty. I'm 100% aware of that, and no, on the off-chance that this eventually works, I would never 'release' it. I'm not stupid.

I also have to correct my earlier, somewhat negative statement where I said that "Support seems more concerned with closing the ticket quickly than with reading the evidence...". They've clarified since that they simply don't have a device that outputs the same signal as my scaler does. That's not an unreasonable answer. The only device on the planet that will produce NES/SNES frequencies at 1440p/4K is a retro gaming scaler, so it's niche (still gaming though! on a gaming monitor!!).

MSI support has been asking basic questions like "try FW V0.20" and "try DSC on/off", which I've been politely answering as they land in my mailbox. Nothing usable yet. I've asked whether it could be escalated to a firmware team of some kind, but I think I'm still at first-line support at the moment. Again: I don't really expect a fix or patch. This has turned into a hobby side-project.
 
Upvote 0 Downvote
Back
Top