This work aims to observe the effects the workload has on the reliability of a system in order to provide better testing methodologies for complex devices. The BeagleV-Fire was tested under heavy ions at LBNL under their 20 MeV/amu tune. No destructive effects nor single event latchup (SEL) were observed in the processor up to an LET of 46.16 MeV cm^2/mg. The processor's error correcting code in its L1 and L2 cache lead to four categories of data errors: data errors that can and cannot be corrected, and cache coherence errors that can and cannot be corrected. There did not appear to be any significant difference in the single-event functional interrupt (SEFI) reliability for the configuration performing matrix multiplication (MM) and the configuration that was idling (IDLE). This consistency implies that, in the case of SEFIs such as kernel panics and crashes, the critical memory space for the kernel is static regardless of the workload. For uncorrectable data errors, the idling configuration showed significantly higher cross sections, or lower reliability, than the matrix multiplication configuration. Correctable cache coherence and data were saturated at the lowest tested LETs and were not significantly different from each other. Finally, there were not enough samples of non-correctable coherence errors to draw any conclusions from, however all of these errors resulted in a SEFI.