Comment # 20 on bug 1107617 from
(In reply to Martin Wilck from comment #19)
> My recommendation is still 2, with an increased constant minimum (K=32 in
> comment 0). This would provide more workers in the laptop case, while
> avoiding sky-rocketing numbers for big systems.
> 

Or maybe instead of introducing a new constant 'K', increase the existing one
whose value is currently 8 in:

  l = min(8 + C * c , m / M)

If the value of the constant is changed to 32, you get:

       |     SUSE     |    upstream   |
#CPUs  | C=64, K = 10 | C = 2, K = 10 |  C = 2,  32
-----------------------------------------------------
   1   |   72      9  |  10     1.3   |   34     4.0
   2   |  136     17  |  12     1.5   |   36     4.6
   4   |  262     33  |  16     2.0   |   40     5.1
   8   |  520     66  |  24     3.0   |   48     6.1
  16   | 1032    132  |  40     5.0   |   64     8.0


This constant as well as 'K' serve the same purpose and are supposed to be
useful for systems with few cpus.

> 
> Systems with ~1000 CPUs are not rare these days. Looking at the last row,
> you can see that the worker numbers get quite large with C=8. IMHO much more
> than 2000 workers can't be healthy on any system, no matter now big.

Then we should put a hard limit that can be exceeded unless "udev.children-max"
parameter specifies a bigger value.


You are receiving this mail because: