Automatically stopping P-1 to benchmark? Why?

[Uncwilly]

I am running P-1 on an exponent in the 332M range. Twice so far in the middle of the night Prime95 has stopped to run some benchmark, then continued. This was during stage 1. It was already over 10% in when it stopped the first time.

It said that the benchmarks were needed. Why?

[Main thread Feb 20 14:23] Starting worker.
[Main thread Feb 21 05:57] Benchmarking multiple workers to tune FFT selection.
[Main thread Feb 21 05:57] Timing 18432K FFT, 4 cores, 1 worker.  Average times: 64.03 ms.  Total throughput: 15.62 iter/sec.
[Main thread Feb 21 05:57] Timing 18432K FFT, 4 cores, 1 worker.  Average times: 64.69 ms.  Total throughput: 15.46 iter/sec.
[Main thread Feb 21 05:58] Timing 18432K FFT, 4 cores, 1 worker.  Average times: 66.41 ms.  Total throughput: 15.06 iter/sec.
[Main thread Feb 21 05:58] Timing 18432K FFT, 4 cores, 1 worker.  Average times: 69.15 ms.  Total throughput: 14.46 iter/sec.
[Main thread Feb 21 05:58] Timing 18432K FFT, 4 cores, 1 worker.  Average times: 66.76 ms.  Total throughput: 14.98 iter/sec.
[Main thread Feb 21 05:58] Timing 18432K FFT, 4 cores, 1 worker.  Average times: 66.90 ms.  Total throughput: 14.95 iter/sec.
[Main thread Feb 21 05:59] 
[Main thread Feb 21 05:59] Throughput benchmark complete.
[Main thread Feb 22 02:57] Benchmarking multiple workers to tune FFT selection.
[Main thread Feb 22 02:57] Timing 18432K FFT, 4 cores, 1 worker.  Average times: 68.87 ms.  Total throughput: 14.52 iter/sec.
[Main thread Feb 22 02:57] Timing 18432K FFT, 4 cores, 1 worker.  Average times: 65.24 ms.  Total throughput: 15.33 iter/sec.
[Main thread Feb 22 02:58] Timing 18432K FFT, 4 cores, 1 worker.  Average times: 66.72 ms.  Total throughput: 14.99 iter/sec.
[Main thread Feb 22 02:58] Timing 18432K FFT, 4 cores, 1 worker.  Average times: 108.14 ms.  Total throughput:  9.25 iter/sec.
[Main thread Feb 22 02:58] Timing 18432K FFT, 4 cores, 1 worker.  Average times: 78.49 ms.  Total throughput: 12.74 iter/sec.
[Main thread Feb 22 02:58] Timing 18432K FFT, 4 cores, 1 worker.  Average times: 67.80 ms.  Total throughput: 14.75 iter/sec.
[Main thread Feb 22 02:59] 
[Main thread Feb 22 02:59] Throughput benchmark complete.

[Prime95]

It is trying to figure out which FFT implementation is the fastest for your upcoming LL test.

[Uncwilly]

But, I am only running P-1, not an LL. I was picking off one of the poorly done P-1 from James's website.