ClojureScript

Speed up js->clj on objs using forEach and transient volatile

Details

  • Type: Enhancement Enhancement
  • Status: Open Open
  • Priority: Minor Minor
  • Resolution: Unresolved
  • Affects Version/s: 1.9.908
  • Fix Version/s: None
  • Component/s: None
  • Labels:
  • Patch:
    Code

Description

It is possible to speed up js->clj on JavaScript objects by revising the implementation to employ goog.object/forEach, accumulating by bashing on a transient volatile map.

The speedups range from 1.5 to 2.1 over the current implementation.

Note: The current implementation uses js-keys. The optimization in CLJS-2340 could help js->clj, but it doesn't appear to provide much speedup in itself (perhaps 1.1?) compared to the change in implementation described above.

  1. CLJS-2341.patch
    30/Aug/17 9:19 AM
    2 kB
    Mike Fikes
  2. CLJS-2341-2.patch
    30/Jun/18 10:12 AM
    2 kB
    Mike Fikes

Activity

Hide
Mike Fikes added a comment -
Speedup Summary:

        Engine  small, medium, with sub-object
            V8: 1.62, 1.50, 1.13
  SpiderMonkey: 1.91, 1.74, 1.59
JavaScriptCore: 1.67, 1.74, 2.10
       Nashorn: 1.54, 2.13, 1.51
    ChakraCore: 1.71, 2.10, 1.95


Before:

Benchmarking with V8
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 55 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 63 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 93 msecs

Benchmarking with SpiderMonkey
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 155 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 87 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 94 msecs

Benchmarking with JavaScriptCore
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 45 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 33 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 42 msecs

Benchmarking with Nashorn
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 1123 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 1195 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 773 msecs

Benchmarking with ChakraCore
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 65 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 44 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 78 msecs


After:

Benchmarking with V8
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 34 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 42 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 82 msecs

Benchmarking with SpiderMonkey
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 81 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 50 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 59 msecs

Benchmarking with JavaScriptCore
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 27 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 19 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 20 msecs

Benchmarking with Nashorn
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 728 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 561 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 513 msecs

Benchmarking with ChakraCore
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 38 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 21 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 40 msecs
Show
Mike Fikes added a comment -
Speedup Summary:

        Engine  small, medium, with sub-object
            V8: 1.62, 1.50, 1.13
  SpiderMonkey: 1.91, 1.74, 1.59
JavaScriptCore: 1.67, 1.74, 2.10
       Nashorn: 1.54, 2.13, 1.51
    ChakraCore: 1.71, 2.10, 1.95


Before:

Benchmarking with V8
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 55 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 63 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 93 msecs

Benchmarking with SpiderMonkey
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 155 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 87 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 94 msecs

Benchmarking with JavaScriptCore
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 45 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 33 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 42 msecs

Benchmarking with Nashorn
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 1123 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 1195 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 773 msecs

Benchmarking with ChakraCore
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 65 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 44 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 78 msecs


After:

Benchmarking with V8
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 34 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 42 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 82 msecs

Benchmarking with SpiderMonkey
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 81 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 50 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 59 msecs

Benchmarking with JavaScriptCore
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 27 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 19 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 20 msecs

Benchmarking with Nashorn
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 728 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 561 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 513 msecs

Benchmarking with ChakraCore
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 38 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 21 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 40 msecs
Hide
Thomas Mulvaney added a comment -

Hi Mike. I don't think the volatile is required, see for example: PersistentHashMap.createWithCheck - the transient can be mutated in place. Although quite often in core.cljs it does seem to be passed as an arg in recursive calls such as in most of the .-fromArray methods of collections. Removing the volatile will likely make even more of an improvement.

Show
Thomas Mulvaney added a comment - Hi Mike. I don't think the volatile is required, see for example: PersistentHashMap.createWithCheck - the transient can be mutated in place. Although quite often in core.cljs it does seem to be passed as an arg in recursive calls such as in most of the .-fromArray methods of collections. Removing the volatile will likely make even more of an improvement.
Hide
Francis Avila added a comment -

To clarify what Thomas said: It's not part of the contract of transients that they can be mutated in place--you are supposed to use the return value of assoc! etc as you have done.

If you bash in place, you will get this:

(let [t (transient {})]
  (dotimes [i 32]
    (assoc! t i i))
  (persistent! t))
=> {0 0, 1 1, 2 2, 3 3, 4 4, 5 5, 6 6, 7 7} ;; broken!

(t is initially a transient array-map; assoc! returns a transient hash-map after 8 items, which you never see because of the in-place bashing.)

So your patch is good in this respect: you should definitely not bash transients in place!

However, particular transient types may be implemented in such a way that the only thing that their edit functions (assoc! dissoc! etc) ever return is the same (identical) object passed in to them. This is true for TransientHashMap's assoc!, which is why PersistentHashMap.createWithCheck is not broken. (It's liable to be brittle though if its implementation ever changes--I hope createWithCheck has good test coverage.)

Show
Francis Avila added a comment - To clarify what Thomas said: It's not part of the contract of transients that they can be mutated in place--you are supposed to use the return value of assoc! etc as you have done. If you bash in place, you will get this:
(let [t (transient {})]
  (dotimes [i 32]
    (assoc! t i i))
  (persistent! t))
=> {0 0, 1 1, 2 2, 3 3, 4 4, 5 5, 6 6, 7 7} ;; broken!
(t is initially a transient array-map; assoc! returns a transient hash-map after 8 items, which you never see because of the in-place bashing.) So your patch is good in this respect: you should definitely not bash transients in place! However, particular transient types may be implemented in such a way that the only thing that their edit functions (assoc! dissoc! etc) ever return is the same (identical) object passed in to them. This is true for TransientHashMap's assoc!, which is why PersistentHashMap.createWithCheck is not broken. (It's liable to be brittle though if its implementation ever changes--I hope createWithCheck has good test coverage.)
Hide
Mike Fikes added a comment -

So, TL;DR is that, with some risk of brittleness, we could bash in place and potentially get more perf.

Show
Mike Fikes added a comment - So, TL;DR is that, with some risk of brittleness, we could bash in place and potentially get more perf.
Hide
Francis Avila added a comment -

No, wrong TLDR: in this particular patch you cannot bash (transient {}) in place, because this starts out as an array-map and returns a transient hash-map once assoc! would return a map with a count greater than 8.

PersistentHashMap.createWithCheck (the example Thomas cited) can bash because it is written very carefully and knows all the types and their implementations.

If you only use hash-maps e.g. with (transient (hash-map)) you can bash in place, but then you won't have array-map for small maps.

(Really, don't bash transients in place.)

Show
Francis Avila added a comment - No, wrong TLDR: in this particular patch you cannot bash (transient {}) in place, because this starts out as an array-map and returns a transient hash-map once assoc! would return a map with a count greater than 8. PersistentHashMap.createWithCheck (the example Thomas cited) can bash because it is written very carefully and knows all the types and their implementations. If you only use hash-maps e.g. with (transient (hash-map)) you can bash in place, but then you won't have array-map for small maps. (Really, don't bash transients in place.)
Hide
Thomas Mulvaney added a comment -

Francis is right, ignore me. Bashing things in place that don't change type works but possibly might break if implementations change. Docs clearly say its a bad idea - just because I've seen it done doesn't make it ok!

Show
Thomas Mulvaney added a comment - Francis is right, ignore me. Bashing things in place that don't change type works but possibly might break if implementations change. Docs clearly say its a bad idea - just because I've seen it done doesn't make it ok!
Hide
Mike Fikes added a comment -

Since CLJS-844 has been applied, the previous patch and benchmarks are no longer applicable. Attaching re-basedlined patch. It gives mixed results relative to the code now on master. Including updated benchmarks here so that a decision can be taken on whether it is worth applying or not.

Speedup Summary:

        Engine  small, medium, with sub-object
            V8: 1.04, 1.05, 0.92
  SpiderMonkey: 1.09, 1.00, 1.08
JavaScriptCore: 1.00, 1.31, 1.35
       Nashorn: 0.72, 0.93, 0.90
    ChakraCore: 1.07, 1.00, 1.09
       GraalVM: 0.98, 0.99, 0.84
	   
	   
Before:

Benchmarking with V8
;;; js->clj
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 25 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 22 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 45 msecs
Benchmarking with SpiderMonkey
;;; js->clj
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 49 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 30 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 39 msecs
Benchmarking with JavaScriptCore
;;; js->clj
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 15 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 17 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 23 msecs
Benchmarking with Nashorn
;;; js->clj
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 330 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 411 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 343 msecs
Benchmarking with ChakraCore
;;; js->clj
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 31 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 11 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 25 msecs
Benchmarking with GraalVM
;;; js->clj
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 243 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 318 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 419 msecs

After:

Benchmarking with V8
;;; js->clj
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 24 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 21 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 49 msecs
Benchmarking with SpiderMonkey
;;; js->clj
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 45 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 30 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 36 msecs
Benchmarking with JavaScriptCore
;;; js->clj
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 15 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 13 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 17 msecs
Benchmarking with Nashorn
;;; js->clj
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 461 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 440 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 381 msecs
Benchmarking with ChakraCore
;;; js->clj
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 29 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 11 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 23 msecs
Benchmarking with GraalVM
;;; js->clj
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 247 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 320 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 500 msecs
Show
Mike Fikes added a comment - Since CLJS-844 has been applied, the previous patch and benchmarks are no longer applicable. Attaching re-basedlined patch. It gives mixed results relative to the code now on master. Including updated benchmarks here so that a decision can be taken on whether it is worth applying or not.
Speedup Summary:

        Engine  small, medium, with sub-object
            V8: 1.04, 1.05, 0.92
  SpiderMonkey: 1.09, 1.00, 1.08
JavaScriptCore: 1.00, 1.31, 1.35
       Nashorn: 0.72, 0.93, 0.90
    ChakraCore: 1.07, 1.00, 1.09
       GraalVM: 0.98, 0.99, 0.84
	   
	   
Before:

Benchmarking with V8
;;; js->clj
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 25 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 22 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 45 msecs
Benchmarking with SpiderMonkey
;;; js->clj
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 49 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 30 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 39 msecs
Benchmarking with JavaScriptCore
;;; js->clj
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 15 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 17 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 23 msecs
Benchmarking with Nashorn
;;; js->clj
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 330 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 411 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 343 msecs
Benchmarking with ChakraCore
;;; js->clj
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 31 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 11 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 25 msecs
Benchmarking with GraalVM
;;; js->clj
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 243 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 318 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 419 msecs

After:

Benchmarking with V8
;;; js->clj
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 24 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 21 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 49 msecs
Benchmarking with SpiderMonkey
;;; js->clj
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 45 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 30 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 36 msecs
Benchmarking with JavaScriptCore
;;; js->clj
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 15 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 13 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 17 msecs
Benchmarking with Nashorn
;;; js->clj
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 461 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 440 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 381 msecs
Benchmarking with ChakraCore
;;; js->clj
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 29 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 11 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 23 msecs
Benchmarking with GraalVM
;;; js->clj
[obj (js-obj "a" 1 "b" 2)], (js->clj obj), 1000 runs, 247 msecs
[obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6)], (js->clj obj), 1000 runs, 320 msecs
[sub-obj (js-obj "g" 7 "h" 8 "i" 9 "j" 10) obj (js-obj "a" 1 "b" 2 "c" 3 "d" 4 "e" 5 "f" 6 "s" sub-obj)], (js->clj obj), 1000 runs, 500 msecs

People

Vote (9)
Watch (2)

Dates

  • Created:
    Updated: