It is quite hard to produce a truly stress-free sample. Stress relief annealing of AlSi10Mg is tricky as you might change the hardness of your samples, see Mertens, et al 2015 - THERMAL TREATMENTS OF AlSi10Mg PROCESSED BY LASER BEAM MELTING (Hardness was reduced by ~ 10 % due to the 250°C/2h heat treatment)

Non precipitation hardenable alloys are typically annealed at 200°C till 300°C and then slowly cooled e.g. in the furnace (GDA Merkblatt W7). Whereby, a higher temperature should be able to achieve a lover stress level. If your samples have the capability of a precipitation hardening and your residual stress is related to a fast cooling steep, "up-hill quenching" (e.g. -196°C and then heating up in a steam bath) might be suitable. (see ASM Handbook vol. 4 Heat Treatment)

The proper temperature depends on the condition (amount of Mg and Si in solution or/and the precipitations) of your samples.

Every machining (even EDM) is able to introduce residual stresses close to the surface. Therefore, you might be better of with a chemical enchant or rather electrolytic polishing before the testing and after your stress relief treatment.

Heating of a sample to a suitable temperature, holding at this temperature for a period of time, followed by slow cooling. So, By heat treatment at about 250c and holding for two hours.

As Jürgen mentioned, it is almost impossible to produce absolutely stress-free specimens. However, residual stress and the associated stored energy of cold work can be reduced considerably by anisothermal or isothermal annealing. The temperature and time required to release most of the stored energy of cold work depend on the material. For Cu, this information can be found, for example, in Hill and Abbaschian, Physical Metallurgy Principles, 1991 (pp 229-230).