You would need to know a few things, relating to
radiation transport / shielding and absorbtion at the
X-ray energy in question. The literature for rad hard
microelectronics tends to center on 1MeV gammas
(which are very penetrating) while the 10keV X-rays
produced by Aracor machines can't even get past the
package lid / overmold.
The amount of power "on the chip" is not at all the
same as what's hitting the top face. Some diverges
and misses altogether; some doesn't make it that far,
some goes right on through, some drops energy and
does "TID stuff".
The most efficient approach might be dosimetry. But
this again is subject to the material discrepancies
between dosimeter and "victim". If you have a similarly
packaged and constructed photodiode, whose volume
is known and whose volumetric efficiency as a
photocurrent source is also known or surmisable,
the raw photocurrent reading could give you rad(Si)/sec
received dose rate from the tube, and time shot, the
total dose.
Irradiator manufacturers such as Hopewell Designs
(big X-ray irradiators), Aracor (who have ditched the
semiconductor test business in favor of homeland
security gravy train), maybe the specific mfr of your
setup, might be able to offer at least penetration /
absorbtion vs keV curves.