Experimental techniques in nuclear physics often call for arithmetic operations on pulses whose amplitudes convey the experimental data. An instrument capable of pulse height multiplication, division, exponentiation, as well as pulse height addition and subtraction is described. These operations are performed in about 0.5 &mgr;sec with a precision of ±0.1% of full scale over a pulse height range of 20:1. At room temperature the computer exhibits a drift of about 0.05%/C° or ±0.3%/day. With an input rate of 10 000 counts/sec the pulse height distribution shifts 0.2%, and with 25 000 counts/sec it shifts 1%. Multiplication, division, and exponentiation are performed by using logarithms and antilogarithms. The log and antilog function generators are based on the fact that the emitter‐base voltage of a silicon planar transistor is proportional to the logarithm of its collector current. Addition and subtraction are performed by using linear operational amplifiers. Results of experiments are shown in which the computer was used for identification of charged particles, pulse shape discrimination, measurement of mass yield distributions in fission, as well as for other on‐line data reduction techniques.