Running time performance is not consistent

[Seonmo Kim]

While I was running some benchmarks, I found out that the running time performance was not consistent. I installed the latest version and tried the command "$ ./approxmc --delta=0.4001 --epsilon=0.8 --gaussuntil=400 or-60-5-2-UC-30.cnf" multiple times. When I tested it, 2 out of 10 cases took around 20 seconds and the others took more than 2 hours. I have attached the benchmark. It seems like Gaussian elimination is not always working. I tried "--nodisablegauss" turning on but it didn't help. However, the option "--restart=static" helped to produce the result consistently.

[Kuldeep Meel]

Thanks Seonmo for reporting the issue. Let me look into this.