Build Your Own SLA/SLS

Hi, All!
Recently, I was trying to operate my LittleRP with the netbook (Intel Atom 450).
The results are horrible. Printing process is significantly slower(20-30%) and I got Z-axis problem . But when I switched back to my normal PC - everything OK.
CW versions are similar (1.0.0.36), FTDI drivers (2.8.24) are similar. So, maybe Intel Atom is not powerful enough? What do you think?

The Atom is too much of a lightweight for the way CW works right now. I think the minimum processor I've seen for a laptop is an i3 class chip.

The faster the CPU, the more RAM you have and the better graphics card will maximize your performance. Low performing peripherals will drag CW down with it, including fouling up the timing between layer exposure and delayed command submission to the printer.

Yeah, Atom N450 is weak. But we have Creation Workshop RPi Host on much weaker hardware.
Maybe, there are some issues with OS (I used Windows 7 professional) or .NET Framework that are drawing CW down.

Are you slicing and orienting the model with RPi host? No, it is just taking the CWS file with the slices saved in it and printing it directly. That is what my understanding of how RPi host works. You still do most of the heavy lifting on the PC and then upload the results to the Pi. If you had to do graphic manipulation and slicing on the Pi, you'd go eat lunch for an hour and come back to hope it is complete. Now Raspberry Pi 2 perhaps, but certainly not a B+.

Actually, someone did a benchmark and equated the 800 MHz RPi CPU performance to that of an Intel Pentium II 300 MHz from back in 1997. That is almost 20 year old technology.

The code design for CW does not appear to be optimal. Perhaps it is .NET or whatever they used to compile it, but it is known to not properly handle its internal memory management and as soon as the computer is forced to swap memory to the file, the performance goes into the toilet because it appears to be tightly coupled with the file system. Almost like a threading problem or a wait on resources. Keeping all of those slices in memory or doing on-demand slicing is too expensive, particularly on very large STL files. That is why closing CW and reopening it improves performance when it starts to slow down.

I am printing with Core i7/8 Gb DDR3 right now. CPU load by CW is 18%, memory load is 900 Mb. Now it is obvious, why Intel Atom/2Gb DDR2 was not able to handle it.
And CW are no longer supported by it's creators...
I think, NanoDLP + Rasberry Pi will be a good alternative to CW + hi-end PC. But maybe I need to edit firmware on LittleRP's Arduino Uno, not sure about it. It will be surely not so simple as CW.