Hi everybody !

I tried again to reach the promised 1.7GHz through this omap5.dtsi modification :

 operating-points = <
                                /* kHz    uV */
                                1000000 1060000
                                1500000 1250000
+                                1700000 1290000
                        >;

Found those values here :
https://github.com/omapconf/omapconf/blob/master/arch/arm/mach-omap/omap5/opp54xx.c

I tested on both EVM and IGEP, 4.7.4 vanilla, and it now works (failed one year ago with IGEP's 3.8.x) :
cpufreq stats: 1000 MHz:9.10%, 1.50 GHz:7.29%, 1.70 GHz:83.61%

$ time echo "scale=5000; a(1)*4" | bc -l
@1.0 GHz : 1m26.008s
@1.5 GHz : 0m56.667s
@1.7 GHz : 0m50.420s

Under long make -j3 , the average temperature is only 2~4 °C more, but I have glued far better heatsinks on both my boards.
No crash so far.

Thought this info may help here to expand ways of testing.

To add redundant information, I fuel the EVM with 12V and never encountered any crash @1.5GHz .

Cheers,
LS