Slow Swift

Exploring Swift runtime performance. Singapore 2017. http://iOSConf.sg

Video: https://www.youtube.com/watch?v=G88qaR9R0v0

Transcript

1. 2017 Slow Swift
2. Marcin Krzyżanowski @krzyzanowskim PDFViewer.io pspdfkit.com github.com/krzyzanowskim CryptoSwift ObjectivePGP Natalie krzyzanowskim.com
4. 1.4x
5. –Waldi Ravens “A Objective- (red.) C program is like a fast dance on a newly waxed dance floor by people carrying razors.” Why Swift?
6. –Waldi Ravens “A Objective- (red.) C program is like a fast dance on a newly waxed dance floor by people carrying razors.” Why Swift?
7. Script Swift? #!/usr/bin/env xcrun -sdk macosx swift print(“iOSConf.sg Rocks!”)
8. Script Swift? #!/usr/bin/env xcrun -sdk macosx swift print(“iOSConf.sg Rocks!”) -Onone 500x to 1000x slower than default optimized build
9. Compiled Swift • Optimized builds • Build times ;-(
10. Compiled Swift • Optimized builds • Build times ;-( let arr: [Int] = [1] + [2] + [3] + [4] + [5] + [6]
11. Compiled Swift • Optimized builds • Build times ;-( let arr: [Int] = [1] + [2] + [3] + [4] + [5] + [6] Expression is too complex to be solved in reasonable time
12. Compiled Swift “Speedup AES.encrypt() compilation time from 68230.37ms to 678.34ms”
13. Source → Binary as seen…
14. Compilation SWIFT_OPTIMIZATION_LEVEL -Onone -O -O -whole-module-optimization
15. Compilation SWIFT_OPTIMIZATION_LEVEL -Onone -O -O -whole-module-optimization
16. Compilation SWIFT_DISABLE_SAFETY_CHECKS Disable runtime safety checks when optimizing.
17. Compilation SWIFT_ENFORCE_EXCLUSIVE_ACCESS full compile-time none (SE-0176)
18. Compilation GCC_GENERATE_DEBUGGING_SYMBOLS The shortcuts taken by optimized code may occasionally produce surprising results: some variables you declared may not exist at all
19. Source Optimization • &+ and &- and &* but + and - and * • discarding any overflow (~~SWIFT_DISABLE_SAFETY_CHECKS) • init(truncatingIfNeeded:) • init(truncatingBitPattern:) no longer public ;-) • private init(_truncatingBits:) uses Builtin.truncOrBitCast
20. Source Optimization In-Out Parameters –Swift Programming Language “An inout parameter has a value that is passed in to the function, is modified by the function, and is passed back out of the function to replace the original value.” func core(block: inout Array<UInt8>)
21. Source Optimization • Can be optimized to pass-by-value • Pointers are useful for performance In-Out Parameters
22. Source Optimization Obvious • struct over class • final class • private final class
23. Source Optimization non-Obvious 2.7s
24. Source Optimization non-Obvious
25. Source Optimization non-Obvious 1.2s 2.7s
26. Source Optimization • for-loop over map • for-loop over reduce • transformation over the dictionary with reduce may end up with O(n2)
27. Source Optimization 0.9s
28. Source Optimization 0.8s
29. Source Optimization Inlining
30. Source Optimization Inlining • Automatic inlining (SIL, LLVM)
31. Source Optimization Inlining • Automatic inlining (SIL, LLVM) • Inline all the things with @inline(__always)
32. Source Optimization Inlining • Automatic inlining (SIL, LLVM) • Inline all the things with @inline(__always) • Force inlining with @_transparent
33. Source Optimization Inlining • Automatic inlining (SIL, LLVM) • Inline all the things with @inline(__always) • Force inlining with @_transparent • Public interface @_inlineable
34. Source Optimization Inlining • Automatic inlining (SIL, LLVM) • Inline all the things with @inline(__always) • Force inlining with @_transparent • Public interface @_inlineable • Be careful with ABI and public API
35. Memory
36. Memory
37. Memory • What if…. raw chunk of memory (Unmanaged) • allocate buffer • deallocate buffer • no ARC, no COW, nothing
38. Memory
39. Array<UInt8>() turns out to be nearly 10% slower compared to UnsafeMutablePointer<UInt8>.allocate(capacity:) Memory
40. Memory Stack free space Heap Static data Literals Instructions 0x000000 0xFFFFFF
41. Memory Stack free space Heap Static data Literals Instructions • Stack allocation - fast (%rsp)
42. Memory Stack free space Heap Static data Literals Instructions • Stack allocation - fast (%rsp) • Heap allocation - slow (slower) (syscalls)
43. Memory Stack free space Heap Static data Literals Instructions • Stack allocation - fast (%rsp) • Heap allocation - slow (slower) (syscalls) • Preallocation
44. Memory Stack free space Heap Static data Literals Instructions • Stack allocation - fast (%rsp) • Heap allocation - slow (slower) (syscalls) • Preallocation • array.reserveCapacity(1024)
45. Quiz
46. Allocation Stack free space Heap Static data Literals Instructions
47. Allocation Stack free space Heap Static data Literals Instructions (lldb) register read rsp rsp = 0x00007fff5fbff630 (lldb) frame variable -L 0x00007fff5fbff658: ([Int]) array = 3 values { 0x00007fa5391de610: [0] = 0 0x00007fa53e6bf3f0: [1] = -1 0x00007fa53e6a91d0: [2] = 1 }
48. Allocation Stack free space Heap Static data Literals Instructions
49. Allocation Stack free space Heap Static data Literals Instructions (lldb) register read rsp rsp = 0x000070000ec06c00 (lldb) fr variable -L 0x000070000ec06c18: (@lvalue [Int]) array = 0x0000000100d002b0: { 0x0000000100d002b0: &array = 2 values { 0x00007fd7df41ae10: [0] = 0 0x00007fd7df485c90: [1] = -1 0x00007fd7df4829b0: [2] = 1 }
50. Allocation Stack free space Heap Static data Literals Instructions
51. Allocation Stack free space Heap Static data Literals Instructions (lldb) register read rsp rsp = 0x000070000ec06c00 (lldb) fr variable -L 0x00007fff5fbff680: (@lvalue MyStruct) s = 0x0000000100f06820: { 0x0000000100f06820: &s = { 0x0000000100f06820: value = "test" } }
52. Generics
53. –who knows “Generic code enables you to write flexible, reusable functions and types that can work with any type, subject to requirements that you define.”
55. Generics
56. Generics 0.20s
57. Generics
58. Generics 0.06s 0.20s
59. Generics
60. Generics
61. Generics
62. Generics & Optimization • Automatic specialization
63. Generics & Optimization • Automatic specialization • in the same module
64. Generics & Optimization • Automatic specialization • in the same module • otherwise use @_specialize(exported: true, where T == Int)
65. Generics & Optimization • Automatic specialization • in the same module • otherwise use @_specialize(exported: true, where T == Int) • Whole Module Optimization won’t help
66. Generics & Optimization • Automatic specialization • in the same module • otherwise use @_specialize(exported: true, where T == Int) • Whole Module Optimization won’t help • Sometimes it won’t optimize for the same module
67. Generics & Optimization • Automatic specialization • in the same module • otherwise use @_specialize(exported: true, where T == Int) • Whole Module Optimization won’t help • Sometimes it won’t optimize for the same module • Avoid generics in public API
68. Let's Recap…
69. Let's Recap… Debug is slow
70. Pre-allocate memory Let's Recap…
71. Let's Recap… Stack vs Heap
72. Let's Recap…Copy is bad
73. Let's Recap… Use carefully
74. Let's Recap… Less Swift is faster Swift
75. @krzyzanowskim krzyzanowskim.com Thank you!