In other words, he's simply refusing to accept reality. His list already confused everyone (by providing wrong/irrelevant data) and really should be completely deleted and replaced by your much more competent list.
Personally, I find his (and his supporters') stance unacceptable. Making errors is not the problem, refusing to learn and do at least some damage control is.
OK, so I read your thread and it all makes sense ;-)
Now, is it possible to at least be near the results of your test, but without needing to do a 3 hours process for each file ?
Regarding the GS thread, I think it would be better to add other test results, but only if it's done for all the devices, but I can understand the fact of not adding it into the already started thread. There's a lot of information already, but it's better to keep it, even if a better test is found, as long as both test can be linked and known by everybody.
It would be too much information on the same thread, and it may just be better to start a new thread, because imagine there's later another method being better, what would you do with 3 tests results in the same thread ?
As I see it, the data posted in that thread is useful but only in an "engineering" sense: it is not directly related to sound quality and it's not proportional to anything meaningful, except to an engineer. A large error could simply mean that the filter isn't phase-linear, or it's frequency cut off is large, or the near-DC frequencies are filtered out. None of these are a major audible concern, but these can result in a huge, 30-40dB difference in the RMS null value. To a non-technical person looking at these numbers, some interfaces will look much, much better. The reality is, they could sound the same or worse as some of the better measuring units.
This is exactly what I thought first, testing a loop is not test an input ony or a output only, it's the combo of both and can't tell you if both ADC and DAC are performing average, or if the ADC is great and the DAC not so good for example... even with the best test results we could get, it will be specific to cases where you need to do a loop, and it's clearly at external analog processing during mixing and mastering for engineers not working all in the box.
With both ADC and DAC in one device, or two different device, you ideally need the best ADC-DAC combo
Did you find a method that would be at least between the GS process and the one that KSTR has developped ?
