newLISP in a browser

Notices and updates
HPW
Posts: 1390
Joined: Thu Sep 26, 2002 9:15 am
Location: Germany
Contact:

Re: newLISP in a browser

Post by HPW »

Hello,

Another test on an aged WIN 1o Laptop

Core 2 duo P8400 2.26 Ghz 4GB
Windows 10 Pro 1703 32bit

Edga 40 JS

>>>>> total time: 3759
>>>>> Performance ratio: 6.32 (1.0 on MacOSX 10.9, 2.3GHz Intel Core i5, newLISP v10.6.0-64-bit)

Edge 40 WA with enabled experimental javascript

ERR: regular expression in function find-all : "error -4 when executing"
called from user function QA:(qa)
called from user function (module "qa-bench")

So we see that WebAssembly is not yet everywhere. Best performance boost on chrome.

Regards
Hans-Peter

HPW
Posts: 1390
Joined: Thu Sep 26, 2002 9:15 am
Location: Germany
Contact:

Re: newLISP in a browser

Post by HPW »

Hello,

Another note for WIN XP:

Neither Chrome 49 or Firefox ESR 52.2.0 work with WebAssembly.
But the JS-flavour works.

Regards
Hans-Peter

HPW
Posts: 1390
Joined: Thu Sep 26, 2002 9:15 am
Location: Germany
Contact:

Re: newLISP in a browser

Post by HPW »

Hello,

CAD-Workstation:
Dell Precision T3610
CPU E5-1620 v2 3.7 GHZ 16GB
Windows performance index 7.7
WIndows 7 PRO SP1

http://www.ferrydb.nl/newlisp/
http://www.ferrydb.nl/newlisp-wa/

(module "qa-bench")


Chrome 59 JS

>>>>> total time: 1832
>>>>> Performance ratio: 3.18 (1.0 on MacOSX 10.9, 2.3GHz Intel Core i5, newLISP v10.6.0-64-bit)

Chrome 59 WA

>>>>> total time: 796
>>>>> Performance ratio: 1.50 (1.0 on MacOSX 10.9, 2.3GHz Intel Core i5, newLISP v10.6.0-64-bit)


Firefox 54 JS

>>>>> total time: 1032
>>>>> Performance ratio: 1.88 (1.0 on MacOSX 10.9, 2.3GHz Intel Core i5, newLISP v10.6.0-64-bit)

Firefox 54 WA

>>>>> total time: 990
>>>>> Performance ratio: 1.81 (1.0 on MacOSX 10.9, 2.3GHz Intel Core i5, newLISP v10.6.0-64-bit)

Regards
Hans-Peter

fdb
Posts: 66
Joined: Sat Nov 09, 2013 8:49 pm

Re: newLISP in a browser

Post by fdb »

Hi,

i've now als run some test on my iPad Pro, running beta of IOS 11 and also Safari 11 which should be able to run web assembly, results below:

With javascript:
>>>> total time: 3139
>>>>> Performance ratio: 5.31 (1.0 on MacOSX 10.9, 2.3GHz Intel Core i5, newLISP v10.6.0-64-bit)

With web-assembly:
>>>>> total time: 1633
>>>>> Performance ratio: 2.86 (1.0 on MacOSX 10.9, 2.3GHz Intel Core i5, newLISP v10.6.0-64-bit)

So almost twice as fast with web assembly.

HPW
Posts: 1390
Joined: Thu Sep 26, 2002 9:15 am
Location: Germany
Contact:

Re: newLISP in a browser

Post by HPW »

Hello,

Latest emscripten: Emscripten 1.37.27 (December 29, 2017)

Firefox Quantum 57.0.3 64 bit with new impressive Performance on my tablet
Asus Eee Slate
i5 U470 1.33 Ghz 4 Gb
Windows performance Index 3.1
Tested on Ferry's pages with 1.37.14

With javascript:
http://ferrydb.nl/newlisp/

>>>>> total time: 1891
>>>>> Performance ratio: 3.28 (1.0 on MacOSX 10.9, 2.3GHz Intel Core i5, newLISP v10.6.0-64-bit)

With web-assembly:
http://ferrydb.nl/newlisp-wa/

>>>>> total time: 1539
>>>>> Performance ratio: 2.71 (1.0 on MacOSX 10.9, 2.3GHz Intel Core i5, newLISP v10.6.0-64-bit)

Much faster than the last test with Firefox 54 on that Hardware.

Regards
Hans-Peter

HPW
Posts: 1390
Joined: Thu Sep 26, 2002 9:15 am
Location: Germany
Contact:

Re: newLISP in a browser

Post by HPW »

Hello,

Also news on Edge compatibility:

On the WIN 10 laptop with latest updates and Edge 41

(module "qa-bench")

With javascript:
http://ferrydb.nl/newlisp/

>>>>> total time: 3550
>>>>> Performance ratio: 5.98 (1.0 on MacOSX 10.9, 2.3GHz Intel Core i5, newLISP v10.6.0-64-bit)

With web-assembly:
http://ferrydb.nl/newlisp-wa/

>>>>> total time: 3968
>>>>> Performance ratio: 6.66 (1.0 on MacOSX 10.9, 2.3GHz Intel Core i5, newLISP v10.6.0-64-bit)

Note: the only browser where web-assembly is slower than javascript.
But it works now with wa. A little faster in javascript than the last test.

Regards
Hans-Peter

HPW
Posts: 1390
Joined: Thu Sep 26, 2002 9:15 am
Location: Germany
Contact:

Re: newLISP in a browser

Post by HPW »

Hello,

Firefox Quantum 58.0 getting even better:
Asus Eee Slate
i5 U470 1.33 Ghz 4 Gb

With javascript:
http://ferrydb.nl/newlisp/
>>>>> total time: 1591
>>>>> Performance ratio: 2.79 (1.0 on MacOSX 10.9, 2.3GHz Intel Core i5, newLISP v10.6.0-64-bit)

With web-assembly:
http://ferrydb.nl/newlisp-wa/
>>>>> total time: 1405
>>>>> Performance ratio: 2.49 (1.0 on MacOSX 10.9, 2.3GHz Intel Core i5, newLISP v10.6.0-64-bit)

Regards
Hans-Peter

HPW
Posts: 1390
Joined: Thu Sep 26, 2002 9:15 am
Location: Germany
Contact:

Re: newLISP in a browser

Post by HPW »

Hello,

In last test with Google Chrome 67 I noticed that WA is now about 8x slower than JS.
In current Firefox WA is still faster than JS.

I also test a chromium based broser which shows similar Performance like Chrome. Also slow WA.

I wonder if it could be a problem with the used emscripten version.

Latest emscripten: Emscripten: 1.38.4 (June 02, 2018)

Regards
Hans-Peter

HPW
Posts: 1390
Joined: Thu Sep 26, 2002 9:15 am
Location: Germany
Contact:

Re: newLISP in a browser

Post by HPW »

Hello,

Did a bug report about chromes webassembly speed:

https://bugs.chromium.org/p/chromium/is ... ?id=849098

Regards
Hans-Peter

HPW
Posts: 1390
Joined: Thu Sep 26, 2002 9:15 am
Location: Germany
Contact:

Re: newLISP in a browser

Post by HPW »

Hello,

Seems that they found the bug. Waiting for the fix.

Regards
Hans-Peter

HPW
Posts: 1390
Joined: Thu Sep 26, 2002 9:15 am
Location: Germany
Contact:

Re: newLISP in a browser

Post by HPW »

Hello,

Chrome Version 68.0.3440.75 (Offizieller Build) (64-Bit) is back on full speed with webassembly.

Firefox 61.0.1 (64-Bit) still a bit faster.

Regards
Hans-Peter

HPW
Posts: 1390
Joined: Thu Sep 26, 2002 9:15 am
Location: Germany
Contact:

Re: newLISP in a browser

Post by HPW »

Hello Lutz,

In the index.html is some code to Show a progressbar for downloading.
I was wondering that it never Show up. Only the Status text changed.

Code: Select all

       if (!Module.setStatus.last) Module.setStatus.last = { time: Date.now(), text: '' };
          if (text === Module.setStatus.text) return;
          var m = text.match(/([^(]+)\((\d+(\.\d+)?)\/(\d+)\)/);
            alert (text);
            alert (m);
The m variable keeps always null so the later if gets never true.

Regards
Hans-Peter

Lutz
Posts: 5289
Joined: Thu Sep 26, 2002 4:45 pm
Location: Pasadena, California
Contact:

Re: newLISP in a browser

Post by Lutz »

The original index.html was taken from some other application dealing with compiled JavaScript. This is probably some left-over unused or only partly used code.

HPW
Posts: 1390
Joined: Thu Sep 26, 2002 9:15 am
Location: Germany
Contact:

Re: newLISP in a browser

Post by HPW »

Hello,
After browser updates on major browsers

CAD-Workstation:
Dell Precision T3610
CPU E5-1620 v2 3.7 GHZ 16GB
Windows performance index 7.7
WIndows 7 PRO SP1

http://www.ferrydb.nl/newlisp/
http://www.ferrydb.nl/newlisp-wa/

(module "qa-bench")


Chrome 69 JS

>>>>> total time: 1099

Chrome 69 WA

>>>>> total time: 860


Firefox 62 JS

>>>>> total time: 822

Firefox 62 WA

>>>>> total time: 679

Regards
Hans-Peter

HPW
Posts: 1390
Joined: Thu Sep 26, 2002 9:15 am
Location: Germany
Contact:

Re: newLISP in a browser

Post by HPW »

Hello,

Latest obserations with chrome 72:

http://www.ferrydb.nl/newlisp/
http://www.ferrydb.nl/newlisp-wa/

For both url's chrome 72 shows a warning sign in front of the url "Nicht sicher" (german for not secure)
Performance seems slower on both than older versions.
Not nice.

Regards
Hans-Peter

HPW
Posts: 1390
Joined: Thu Sep 26, 2002 9:15 am
Location: Germany
Contact:

Re: newLISP in a browser

Post by HPW »

Hello,

Oops, nothing special with Chrome and newlisp in the browser.
The "Not secure" Display is for every Non-https site.

Regards
Hans-Peter

Locked