docs: command line options "-c" is not documented.

[fingolfin]

I am not sure what is meant for -- other than what I can read in the code, that is: it modifies SyCacheSize, which is the "size of the data cache". And this affects gasman somehow. But it is unclear to me whether it has any real-world noticable impact; and whether anybody uses it.

It was apparently added before the initial CVS import in July 1996.

In summary: We should either document it, or remove it. My guess is that the latter option is the best choice, but perhaps somebody knows better?

[stevelinton]

The idea was to keep the "young generation" in GASMAN to a limited size so that it would fit in cache. This forces more, but shorter partial garbage collections.

I haven't tried it in many years, but I don't recall it ever producing measurable speedups. On the other hand that was in the days of 256KB caches. I don't know if it would be useful on a more modern CPU.

[frankluebeck]

I have made a few tests yesterday. You can easily see the difference by trying gap -g -g, and then for example ReadLib("../tst/testinstall.g");. Then run the same in a GAP called with gap -g -g -c 1m. On my hexa core pentium cpu with 12MB cache size the latter is about 30% slower (it is worse with -c 10m and the slowdown is about the same in the range -c 500k to -c 4m).

So, the default (that is -c 0k) seems always better.

How about adding a short description to the documentation, but also saying that in the current implementation a non-zero argument doesn't seem to be useful.