[ruby-core:93710] [Ruby master Feature#15997] Improve performance of fiber creation by using pool allocation strategy.

Issue #15997 has been updated by ioquatix (Samuel Williams).


> which size (xx KB, etc)?

#define RUBY_VM_FIBER_VM_STACK_SIZE           (  16 * 1024 * sizeof(VALUE)) /*   64 KB or  128 KB */
#define RUBY_VM_FIBER_VM_STACK_SIZE_MIN       (   2 * 1024 * sizeof(VALUE)) /*    8 KB or   16 KB */
#define RUBY_VM_FIBER_MACHINE_STACK_SIZE      (  64 * 1024 * sizeof(VALUE)) /*  256 KB or  512 KB */
#if defined(__powerpc64__)
#define RUBY_VM_FIBER_MACHINE_STACK_SIZE_MIN  (  32 * 1024 * sizeof(VALUE)) /*   128 KB or  256 KB */
#else
#define RUBY_VM_FIBER_MACHINE_STACK_SIZE_MIN  (  16 * 1024 * sizeof(VALUE)) /*   64 KB or  128 KB */
#endif

Assuming page size of 4K, each fiber needs 4KB + 512KB + 16KB stack.

However, this is only part of the picture. Using the given `count.rb` we can divide VSZ/RSZ by number of fibers, to give "actual" usage:

```
2.7.0-fiber-pool

2.28s to allocate 200000 fibers
  PID    VSZ   RSZ
16105 129500736 2606936

-> 13.03KB of physical memory per fiber, 647.5KB address space per fiber
```

```
2.7.0-master

82.67s to allocate 200000 fibers
  PID    VSZ   RSZ
22398 128513292 2396224

-> 11.98KB of physical memory per fiber, 642.5KB address space per fiber
```

There is no significant difference, and it also looks like there might be room for improvement.


----------------------------------------
Feature #15997: Improve performance of fiber creation by using pool allocation strategy.
https://bugs.ruby-lang.org/issues/15997#change-79339

* Author: ioquatix (Samuel Williams)
* Status: Open
* Priority: Normal
* Assignee: ko1 (Koichi Sasada)
* Target version: 
----------------------------------------
https://github.com/ruby/ruby/pull/2224

This PR improves the performance of fiber allocation and reuse by implementing a better stack cache.

The fiber pool manages a singly linked list of fiber pool allocations. The fiber pool allocation contains 1 or more stack (typically more, e.g. 512). It uses N^2 allocation strategy, starting at 8 initial stacks, next is 8, 16, 32, etc.

```
//
// base = +-------------------------------+-----------------------+  +
//        |VM Stack       |VM Stack       |                       |  |
//        |               |               |                       |  |
//        |               |               |                       |  |
//        +-------------------------------+                       |  |
//        |Machine Stack  |Machine Stack  |                       |  |
//        |               |               |                       |  |
//        |               |               |                       |  |
//        |               |               | .  .  .  .            |  |  size
//        |               |               |                       |  |
//        |               |               |                       |  |
//        |               |               |                       |  |
//        |               |               |                       |  |
//        |               |               |                       |  |
//        +-------------------------------+                       |  |
//        |Guard Page     |Guard Page     |                       |  |
//        +-------------------------------+-----------------------+  v
//
//        +------------------------------------------------------->
//
//                                  count
//
```

The performance improvement depends on usage:

```
Calculating -------------------------------------
                     compare-ruby  built-ruby 
  vm2_fiber_allocate     132.900k    180.852k i/s -    100.000k times in 0.752447s 0.552939s
     vm2_fiber_count       5.317k    110.724k i/s -    100.000k times in 18.806479s 0.903145s
     vm2_fiber_reuse      160.128     347.663 i/s -     200.000 times in 1.249003s 0.575269s
    vm2_fiber_switch      13.429M     13.490M i/s -     20.000M times in 1.489303s 1.482549s

Comparison:
               vm2_fiber_allocate
          built-ruby:    180851.6 i/s 
        compare-ruby:    132899.7 i/s - 1.36x  slower

                  vm2_fiber_count
          built-ruby:    110724.3 i/s 
        compare-ruby:      5317.3 i/s - 20.82x  slower

                  vm2_fiber_reuse
          built-ruby:       347.7 i/s 
        compare-ruby:       160.1 i/s - 2.17x  slower

                 vm2_fiber_switch
          built-ruby:  13490282.4 i/s 
        compare-ruby:  13429100.0 i/s - 1.00x  slower
```

This test is run on Linux server with 64GB memory and 4-core Xeon (Intel Xeon CPU E3-1240 v6 @ 3.70GHz). "compare-ruby" is `master`, and "built-ruby" is `master+fiber-pool`.

Additionally, we conservatively use `madvise(free)` to avoid swap space usage for unused fiber stacks. However, if you remove this requirement, we can get 6x - 10x performance improvement in `vm2_fiber_reuse` benchmark. There are some options to deal with this (e.g. moving it to `GC.compact`) but as this is still a net win, I'd like to merge this PR as is.





-- 
https://bugs.ruby-lang.org/

Unsubscribe: <mailto:[e-mail]?subject=unsubscribe>
<http://lists.ruby-lang.org/cgi-bin/mailman/options/ruby-core>