Fullrun,
Sorry I haven't followed up, and glad to see my comments helped.
I don't necessarily claim one "strategy" is better that the other, though.
One "strategy" is to optimize the server and use a DAC that is very good and server-sensitive. I believe most DACs fall into this category, and it allows the use of JPlay or Audionirvana+ in Apple with which users are very happy.
Another strategy is to spend most of your budget on a DAC that is in-sensitive to the feeding computer - the catch with those is that over the years a number of companies have claimed such independence but few, if any, have yet achieved it.
About 9 months ago AudioPhil, the maker of the Audio Optimizer, told me a laptop with WS2012 and Optimizer would get me to 60-70% of where a fully optimized server would. Not bad for a rather simple solution! In my view, buying a very good DAC and feeding it with a laptop with WS2012+Optimizer would be a very valid solution. And you could later decide if you want to upgrade the server.
Sure, the other strategy is also valid - as long as the DAC is indeed insensitive to the source! The only DAC I know of that might fit this bill and is below $4k is exaSound, but I have not come to a conclusion yet with mine (have not had the time). exaSound president told me the previous owner of my DAC sold it out of frustration he couldn't leverage what he had spent on his optimized server into sound improvements. Encouraging from my point of view, but clearly disappointing to the previous owner.
Here's another curved ball :-) We all know the room has the highest impact on sound of all "components". Many have lately spoken highly about some digital room correction units and software packages. See the thread "Is DEQX a game changer?" here. It seems that after room correction and time alignment different hardware sound more alike than different and they all sound VERY good (within certain logical quality ranges, of course).
What if we used a powerful computer running DRC/DSP and feeding an insentive-to-the-source DAC (now the heavy processing at the server wouldn't translate into jitter fed into the DAC)? Maybe the shortcomings of a not as good volume section in the DAC would be more than compensated by DRC/DSP vs. our beloved preamps?
Indeed, that is part of my hidden agenda. ;-)
In reality I am most interested in the exaSound e28, that has 8 channels. My idea is to drive a pair of 3-way speakers plus 2 subs using digital crossovers at the server and using an amp to directly drive each speaker driver (no passive XO). Hence a 2 channel preamp won't work for me. So all of this is fascinating IF the exaSound is INDEED insensitive to the fed stream.
This opens a fascinating door. I have a hard time letting go of my preamp, though. So I bought the e22 to experiment and I'm also in the process of trying Acourate DSP/DRC software package.
Sorry for the extra long post to ramble over something you didn't ask. Bottom line is both strategies can lead to easier solutions to implement or to quite sophisticated ones. A dual-PC setup using JPlay would be the extreme sophistication along the lines of the first strategy. See JPlay/Computer Audio forum for this.
Finally, do read the long thread about DEQX. I have exchanged with a user who claimed to have replaced a $30k DAC with DEQX HDP-4 and was very happy. Bifiwyne in the mentioned thread tried one at home, was thrilled, and is coping with the idea it might sound the same as without his Audio Research Ref 5 preamp in the chain - a paradigm shift for him. The DEQX might even be a good solution for you!
I know this post is coming back without a clear suggestion of what I'm proposing. I just propose you think about the options, I try to provide food for thought, and let you decide what sounds more appealing to you. Personally, I currently don't have a point of view...I might be transitioning a paradigm shift myself!