REW clipping at very low levels

[Deadly Mix Studios]

Please take the issue with using the Marian Clara E Dantre PCIe card in order to let others know about the issue and maybe find a solution to whatever caused the "bug" or whatever coders would consider it. I know its an issue that's likely to affect professionals with big systems more than the average bedroom budget studio hobbyist, pros tend to use SMAART and TRACT more often, and we're a much smaller demographic to begin with, but we probably buy more licenses. Hobbyists always seem to want things more when they know we use them anyway lol so I'm sure there's plenty of benefit.
Speaking of licenses, this is a 7.2.4 Dolby Atmos mastering studio, so what I actually need is the multichannel upgrade. I just didn't wanna spend the money if I was gonna have to buy SMAART or TRACT due to the issue. I need to compare the tools between the 3 again, to make sure there isn't something one of the other 3 offers that I need and that REW doesn't have, but I'm leaning towards upgrading REW. Someone was telling me this morning that I need SMAART for something but they didn't explain what they were talking about and we have the rest of the Waves catalog so TRACT kind of makes sense, but I've used REW while designing a few studios and testing acoustic devices I've designed, so I'm not sure what feature they could be talking about that I need besides the multichannel "Pro" upgrade for REW.

 

[John Mulcahy]

You don't need the Pro upgrade, that's for capturing from multiple mics simultaneously. Install the V5.30 beta release though.

 

[Deadly Mix Studios]

Uh oh. Another bug, I think...
The last 2 times that I deleted all of the measurements, after making at least 1, I got this popup and had to restart the program.

 

[John Mulcahy]

As above, install the V5.30 beta.

 

[John Mulcahy]

Please take the issue with using the Marian Clara E Dantre PCIe card in order to let others know about the issue and maybe find a solution

Looking at your screenshots the Marian interface is using the Int32LSB24 format, which is 24-bit data in the lower bits of a 32-bit word. That's rare, not seen it used before. Looks like the handling of input data in that format is incorrect, not accounting for the fact the input data is not sign-extended. I'll fix it in the next V5.30 beta build.