http://pastebin.com/pCL3CBx2

First of all, you need to know (or be able to figure out somehow), the domain ID of the VM(s) you want to plot the vCPUs of. Let’s say we know it, and it was domain 16. We then do:
$ xenalyze –scatterplot-pcpu trace_file.bin | grep ^16v0 > d16v0.txt
$ xenalyze –scatterplot-pcpu trace_file.bin | grep ^16v1 > d16v1.txt
Each file will contain something like this:
$ cat d16v0.txt
16v0 0.125842420 1
16v0 0.125983122 1
16v0 0.125983122 1
16v0 0.126210870 1
16v0 0.126210870 1
…
Now, to plot this, the following gnuplot script should do:
$ cat d16_vcpus.gp
set term png enhanced size 1024, 768 font ‘Times-Roman’
set output ‘d16_vcpus.png’
set title ‘vCPU on pCPU execution’
set grid
set key outside bottom center horizontal
set xrange [0:]
set yrange [0:16]
set xtics 100
set ytics 16
set xlabel ‘pCPU’
set ylabel ‘Time’
plot \
‘< paste d16v0.txt d16v1.txt’ u 2:3 ti ‘d16v0′ w p lw 2, \
‘< paste d16v0.txt d16v1.txt’ u 5:6 ti ‘d16v1′ w p lw 2
$ gnuplot d16_vcpus.gp && eog ./d16_vcpus.png

Adding more VMs and/or vCPUs should be trivial at this poit… Just be careful that, the last two lines of the gnuplot script _must_ start with TABs, or gnuplot will complain about it (or so I recall ).