if you start with a 1644 and upsample it, you will probably not get the same resolution than if you start with a higher res source recording. My masters begin at 2496 and then the generations go down from there. So even if you start with a 2444 from me, you are handicapped by being a generation down from the 2496 original. So, your first obstacle is to consider the provenance.

Your second obstacle is that it turns out that the quality of the chip-based upsamplers in DACs is inferior to the quality of a no-compromise software upsampler like the Weiss Saracon. Studies conducted by TC Electronic concluded that chip-based DACs sound and perform better when some of those stages can be eliminated, or when they don't have to work as hard, and one of the ways is to feed them a higher sample rate source to begin with. When TC built their own state of the art upsampler in front of their DAC, it sounded better than if they let the DAC do the upsampling, because the internal resolution, filtering and quality of their own upsampler was much better. This was a "discrete" upsampler using a programmed DSP chip as opposed to using the chip upsampler.

There are DACs, such as the Bricasti and the DCS, which have high quality custom-programmed upsampling filters that give the Saracon a run for the money. But these are expensive DACs. So, in general it seems if you feed a chip-based DAC with a high quality upsampled signal to begin with, you are going to get better sound. My own listening tests seem to show this to be the case. To repeat, the sources you are likely to have which you are going to upsample with your DAC are by necessity going to be a generation or two down from the 2496 original, so you can still do better playing the earlier-generation source, even if you have a Bricasti or DCS-quality DAC.

Read more at http://www.innerfidelity.com/content/katzs-corner-episo ... BH3dfFL.99

This whole "upsampling" thing has always been a huge mystery, from when it was first introduced over 15 years ago. The bottom line is that one is simply inserting another digital filter in front of the existing digital filter. This can only result in an "improvement" if the existing digital filter is poorly implemented, as the first filter in the chain dominates the overall response.
We don't use digital filters that are poorly implemented.