Watch out for cascaded counters! If the desired output clock, the one you want to drive out on the dedicated output pin, has a frequency that is too low, the Quartus ALTPLL MegaWizard will use counter cascading to generate it. You will see a note like this:
Info (176353): Automatically promoted node pll:pll|altpll:altpll_component|pll_altpll:auto_generated|wire_pll1_clk[0] (placed in counter C1 of PLL_2)
But the output from counter C1 cannot drive the dedicated pin, so then Quartus gives the very misleading warning:
Warning (15064): PLL "pll:pll|altpll:altpll_component|pll_altpll:auto_generated|pll1" output port clk[0] feeds output pin "test_clk~output" via non-dedicated routing -- jitter performance depends on switching rate of other design elements. Use PLL dedicated clock outputs to ensure jitter performance
There is NO MENTION of cascaded counters as a possible reason for it. I got this for a Cyclone IV device; my test design had only one input (a 40MHz clock) and one output (200kHz) with no pin assignments. One quick way to figure out if your PLL will require cascaded counters is to scroll down in the description window of c0 output clock in the ALTPLL MegaWizard. Here you will find the various settings required for the C0 output. If you see a section titled "Cascade tap settings: (For c0)" the output will not be on c0 and it will not be able to directly drive the PLL output pin.
If you need clock this slow, perhaps you don't care about the jitter and can ignore the warning. If you do care about the jitter, then a better implementation would be to use a higher frequency clock to create a registered version of the output clock.
I hope Altera will recognize this and add it to the possible causes of the warning. Or better yet, make a new warning specifically for this problem.
If they had to pick one output, I wonder why the Altera chip designers chose the c0 output to drive the dedicated pin, instead of say c5, which would have allowed cascading to be used with the dedicated output.