Page 2 of 2
Re: Speed of Pure & Spider comparsion
Posted: Fri Jun 15, 2018 2:07 pm
by StarBootics
The original test on my computer with DisableDebugger and compiling without the debugger give me 9.695 seconds
The User_Russian's test with DisableDebugger and compiling without the debugger give me 4.406 seconds
The NicTheQuick's test with DisableDebugger and compiling without the debugger give me 0.260 seconds
Best regards
StarBootics
Re: Speed of Pure & Spider comparsion
Posted: Fri Jun 15, 2018 2:50 pm
by marc_256
Hi,
I did some tests,
but I do not understand the results ?
I use PB 5.62 Final x64 on my old programming laptop ... I like this PC
If I start the program and check the taskmanager,
I see that PB is only using 23% of cpu time even when there is a lot of idle time ?
Why ?
Marc,
Re: Speed of Pure & Spider comparsion
Posted: Fri Jun 15, 2018 2:57 pm
by Bitblazer
marc_256 wrote:Hi,
I did some tests,
but I do not understand the results ?
I use PB 5.62 Final x64 on my old programming laptop ... I like this PC
If I start the program and check the taskmanager,
I see that PB is only using 23% of cpu time even when there is a lot of idle time ?
Why ?
Marc,
Sounds like a quadcore CPU with a software which utilitizes a single thread only. 3 of 4 cores cant participate so 25% would be the maximum you could get in the summary display. Switch do show each single core in a graph and see
Re: Speed of Pure & Spider comparsion
Posted: Fri Jun 15, 2018 3:12 pm
by Trond
The original code without the debugger (DisableDebugger shouldn't have an effect in this case):
13 758 ms
The original code, with the debugger and DisableDebugger:
13 916 ms
There is a small, but consistent difference. In other scenarios, there are actually big differences, especially when calling PB commands. When the debugger is turned off from the menu, the optimizer is turned on in the compiler (I think), and the library commands behave differently.
With debugger:
107 001 ms
Re: Speed of Pure & Spider comparsion
Posted: Fri Jun 15, 2018 4:32 pm
by djes
Bitblazer wrote:marc_256 wrote:Hi,
I did some tests,
but I do not understand the results ?
I use PB 5.62 Final x64 on my old programming laptop ... I like this PC
If I start the program and check the taskmanager,
I see that PB is only using 23% of cpu time even when there is a lot of idle time ?
Why ?
Marc,
Sounds like a quadcore CPU with a software which utilitizes a single thread only. 3 of 4 cores cant participate so 25% would be the maximum you could get in the summary display. Switch do show each single core in a graph and see
Maybe it's also the reason why JavaScript could be faster than a compiled program, parts of the js code being executed on different cores. Threading is then very important...
Re: Speed of Pure & Spider comparsion
Posted: Sat Jun 16, 2018 10:54 am
by the.weavster
Kiffi wrote:That makes me sceptical:
Yes, it would be interesting to see the JS code. JavaScript compilers use inference engines that ascertain the datatype of a variable when it's initialized. If you take FireFox for example, iirc there are three levels of execution:
Interpreter (Functions that are only run once)
Baseline compiler (Warm code - A quick compile to byte code)
IonMonkey compiler (Hot code - A further more optimized compile generated from the byte code of Baseline)
If you change the datatype of a variable in JS that's running in FireFox it can slow down execution a lot because that routine will get kicked back down to the interpreter even if it's hot code (i.e. a routine that's being called repeatedly during execution). SpiderBasic's type checking should prevent developers from falling into that trap.
Re: Speed of Pure & Spider comparsion
Posted: Sat Jun 16, 2018 11:36 am
by the.weavster
NicTheQuick wrote:It is impossible that Spiderbasic is faster than Go, C, C++, C# except your coding is bad.
I'm not saying it's the case in this instance but when Mozilla were experimenting with asm.js it did outperform C++ on a Binary Trees benchmark.
Details here
Re: Speed of Pure & Spider comparsion
Posted: Sat Jun 16, 2018 2:17 pm
by firace
Original code took 19s here (with debugger disabled from the menu).
Could you post the SpiderBasic code too?
Your test results are very surprising (Java and Go faster than C++??)
Note: Nicthequick's version completed in 0.2s.
Re: Speed of Pure & Spider comparsion
Posted: Mon Jun 18, 2018 8:26 am
by zxretrosoft
firace wrote:Could you post the SpiderBasic code too?
Yes, of course...
Code: Select all
Global p.i
Global x.i=-1
Global a1.i,a2.i,a3.i
Global ps1.s,ps2.s,ps3.s
Global h1.s,h2.s,h3.s
Global h.i
Declare RenderFrame()
h1="L"
h2="W"
h3="K"
h=3724
RenderFrame()
Procedure RenderFrame()
Repeat
x+1
p+1
If x>9999
x=0
a3+1
If a3>25
a3=0
a2+1
If a2>25
a2=0
a1+1
EndIf
EndIf
EndIf
Select a1
Case 0
ps1="A"
Case 1
ps1="B"
Case 2
ps1="C"
Case 3
ps1="D"
Case 4
ps1="E"
Case 5
ps1="F"
Case 6
ps1="G"
Case 7
ps1="H"
Case 8
ps1="I"
Case 9
ps1="J"
Case 10
ps1="K"
Case 11
ps1="L"
Case 12
ps1="M"
Case 13
ps1="N"
Case 14
ps1="O"
Case 15
ps1="P"
Case 16
ps1="Q"
Case 17
ps1="R"
Case 18
ps1="S"
Case 19
ps1="T"
Case 20
ps1="U"
Case 21
ps1="V"
Case 22
ps1="W"
Case 23
ps1="X"
Case 24
ps1="Y"
Case 25
ps1="Z"
EndSelect
Select a2
Case 0
ps2="A"
Case 1
ps2="B"
Case 2
ps2="C"
Case 3
ps2="D"
Case 4
ps2="E"
Case 5
ps2="F"
Case 6
ps2="G"
Case 7
ps2="H"
Case 8
ps2="I"
Case 9
ps2="J"
Case 10
ps2="K"
Case 11
ps2="L"
Case 12
ps2="M"
Case 13
ps2="N"
Case 14
ps2="O"
Case 15
ps2="P"
Case 16
ps2="Q"
Case 17
ps2="R"
Case 18
ps2="S"
Case 19
ps2="T"
Case 20
ps2="U"
Case 21
ps2="V"
Case 22
ps2="W"
Case 23
ps2="X"
Case 24
ps2="Y"
Case 25
ps2="Z"
EndSelect
Select a3
Case 0
ps3="A"
Case 1
ps3="B"
Case 2
ps3="C"
Case 3
ps3="D"
Case 4
ps3="E"
Case 5
ps3="F"
Case 6
ps3="G"
Case 7
ps3="H"
Case 8
ps3="I"
Case 9
ps3="J"
Case 10
ps3="K"
Case 11
ps3="L"
Case 12
ps3="M"
Case 13
ps3="N"
Case 14
ps3="O"
Case 15
ps3="P"
Case 16
ps3="Q"
Case 17
ps3="R"
Case 18
ps3="S"
Case 19
ps3="T"
Case 20
ps3="U"
Case 21
ps3="V"
Case 22
ps3="W"
Case 23
ps3="X"
Case 24
ps3="Y"
Case 25
ps3="Z"
EndSelect
Until h1=ps1 And h2=ps2 And h3=ps3 And h=x
Debug(p)
EndProcedure
Re: Speed of Pure & Spider comparsion
Posted: Mon Jun 18, 2018 5:59 pm
by firace
I confirm, SpiderBasic completes it in about 2 seconds.
Really surprising.
@Fred, if you could comment on this, it would be great
PS: I'm using Google Chrome on Win 10 x64.
Re: Speed of Pure & Spider comparsion
Posted: Tue Jun 19, 2018 6:49 am
by Cyllceaux
sounds like the string-"problem"
viewtopic.php?f=3&t=58892