Hi,
WebPL is already outdated I guess. It doesn't
show the versions of the other Prolog systems
it is using. While I had these results for
the primes example in the WebPL playground:
/* Trealla Prolog WASM */
(23568.9ms)
When I run the example here:
https://php.energy/trealla.html
I get better results:
/* trealla-js 0.27.1 */
?- time(test).
% Time elapsed 9.907s, 11263917 Inferences, 1.137 MLips
Bye
Hi,
Heap/Stack Prolog systems could solve some Prolog
String Problems, especially in connection with a FFI, but I am
not showing that. More a general design limitation of the common
take of WAM resp. ZIP. The new WebPL Prolog describes itself as a
merged Heap/Stack architecture Prolog system. And has a reference
in its escorting paper to an academic work by Xining Li (1999):
A new term representation method for prolog
Xining Li - 1999 https://www.sciencedirect.com/science/article/pii/S0743106697000629
Besides that Program Sharing (PS), as it is called in the paper,
is nothing new, WebPL also shows a more modern take, in that
it already uses compound data types from Rust. Can we
replicate some of the performance advantages of a PS system
versus the more traditional WAM resp. ZIP based systems? Here
is a simple test in the WebPL Playground, for Web PL without GC:
/* WebPL NoGC */
?- test2(10).
(1795.6ms)
?- test2(30).
(1785.5ms)
?- test2(90).
(1765.6ms)
Then SWI-Prolog WASM as found in SWI-Tinker:
/* SWI-Prolog WASM */
?- test2(10).
(1239.3ms)
?- test2(30).
(2276.1ms)
?- test2(90).
(5372.3ms)
https://webpl.whenderson.dev/
Bye
The test case:
data(10, [10, 9, 8, 7, 6, 5, 4, 3, 2, 1]).
data(30, [30, 29, 28, 27, 26, 25, 24, 23,
22, 21, 20, 19, 18, 17, 16, 15, 14, 13,
12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1]).
data(90, [90, 89, 88, 87, 86, 85, 84, 83,
82, 81, 80, 79, 78, 77, 76, 75, 74, 73,
72, 71, 70, 69, 68, 67, 66, 65, 64, 63,
62, 61, 60, 59, 58, 57, 56, 55, 54, 53,
52, 51, 50, 49, 48, 47, 46, 45, 44, 43,
42, 41, 40, 39, 38, 37, 36, 35, 34, 33,
32, 31, 30, 29, 28, 27, 26, 25, 24, 23,
22, 21, 20, 19, 18, 17, 16, 15, 14, 13,
12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1]).
test(N) :- between(1,1000,_), data(N,_), fail.
test(_).
test2(N) :- between(1,1000,_), test(N), fail.
test2(_).
between(Lo, Lo, R) :- !, Lo = R.
between(Lo, _, Lo).
between(Lo, Hi, X) :- Lo2 is Lo+1, between(Lo2, Hi, X).
Mild Shock schrieb:
Hi,
WebPL is already outdated I guess. It doesn't
show the versions of the other Prolog systems
it is using. While I had these results for
the primes example in the WebPL playground:
/* Trealla Prolog WASM */
(23568.9ms)
When I run the example here:
https://php.energy/trealla.html
I get better results:
/* trealla-js 0.27.1 */
?- time(test).
% Time elapsed 9.907s, 11263917 Inferences, 1.137 MLips
Bye
Hi,
Ok lets run the test case on the desktop,
and not on the web. What do we get? Its almost
constant for Trealla Prolog as well, in
WebPL it was perfectly constant, but here
its only almost constant:
/* Trealla Prolog 2.82.14 */
?- time(test2(10)).
% Time elapsed 0.188s, 3004002 Inferences, 16.014 MLips
true.
?- time(test2(30)).
% Time elapsed 0.210s, 3004002 Inferences, 14.321 MLips
true.
?- time(test2(90)).
% Time elapsed 0.228s, 3004002 Inferences, 13.147 MLips
true.
Scryer Prolog fails the test horribly. Which
is amazing, since it is a Rust Prolog system just
like WebPL. But they are too traditional in
following the stupid WAM design:
/* Scryer Prolog 0.9.4-599 */
?- time(test2(10)).
% CPU time: 0.714s, 7_049_076 inferences
true.
?- time(test2(30)).
% CPU time: 1.284s, 7_049_099 inferences
true.
?- time(test2(90)).
% CPU time: 2.984s, 7_049_099 inferences
true.
Bye
Mild Shock schrieb:
Hi,
Heap/Stack Prolog systems could solve some Prolog
String Problems, especially in connection with a FFI, but I am
not showing that. More a general design limitation of the common
take of WAM resp. ZIP. The new WebPL Prolog describes itself as a
merged Heap/Stack architecture Prolog system. And has a reference
in its escorting paper to an academic work by Xining Li (1999):
A new term representation method for prolog
Xining Li - 1999
https://www.sciencedirect.com/science/article/pii/S0743106697000629
Besides that Program Sharing (PS), as it is called in the paper,
is nothing new, WebPL also shows a more modern take, in that
it already uses compound data types from Rust. Can we
replicate some of the performance advantages of a PS system
versus the more traditional WAM resp. ZIP based systems? Here
is a simple test in the WebPL Playground, for Web PL without GC:
/* WebPL NoGC */
?- test2(10).
(1795.6ms)
?- test2(30).
(1785.5ms)
?- test2(90).
(1765.6ms)
Then SWI-Prolog WASM as found in SWI-Tinker:
/* SWI-Prolog WASM */
?- test2(10).
(1239.3ms)
?- test2(30).
(2276.1ms)
?- test2(90).
(5372.3ms)
https://webpl.whenderson.dev/
Bye
The test case:
data(10, [10, 9, 8, 7, 6, 5, 4, 3, 2, 1]).
data(30, [30, 29, 28, 27, 26, 25, 24, 23,
22, 21, 20, 19, 18, 17, 16, 15, 14, 13,
12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1]).
data(90, [90, 89, 88, 87, 86, 85, 84, 83,
82, 81, 80, 79, 78, 77, 76, 75, 74, 73,
72, 71, 70, 69, 68, 67, 66, 65, 64, 63,
62, 61, 60, 59, 58, 57, 56, 55, 54, 53,
52, 51, 50, 49, 48, 47, 46, 45, 44, 43,
42, 41, 40, 39, 38, 37, 36, 35, 34, 33,
32, 31, 30, 29, 28, 27, 26, 25, 24, 23,
22, 21, 20, 19, 18, 17, 16, 15, 14, 13,
12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1]).
test(N) :- between(1,1000,_), data(N,_), fail.
test(_).
test2(N) :- between(1,1000,_), test(N), fail.
test2(_).
between(Lo, Lo, R) :- !, Lo = R.
between(Lo, _, Lo).
between(Lo, Hi, X) :- Lo2 is Lo+1, between(Lo2, Hi, X).
Mild Shock schrieb:
Hi,
WebPL is already outdated I guess. It doesn't
show the versions of the other Prolog systems
it is using. While I had these results for
the primes example in the WebPL playground:
/* Trealla Prolog WASM */
(23568.9ms)
When I run the example here:
https://php.energy/trealla.html
I get better results:
/* trealla-js 0.27.1 */
?- time(test).
% Time elapsed 9.907s, 11263917 Inferences, 1.137 MLips
Bye
Hi,
Smarter Partial Strings would use Program
Sharing. Take the invention of Scryer
Prolog and think about it from a Program
Sharing prespective:
p --> "abc", q
Translates to with Partial Strings:
p(C, B) :- C = "abc"||A, q(A, B).
Unfortunately straight forward Program
Sharning of the partial string doesn't
work anymore, since it is not ground:
p(C, B) :- C = [a,b,c|A], q(A, B).
But we could translate the DCG also to:
p(C, B) :- '$append'([a,b,c],A,C), q(A, B).
Where '$append'/3 is a mode (+,-,-) specialization
of append/3. Could be natively implemented.
The mode (+,-,-) will be more clever
then the failed programm sharing. The program
sharing can share the string "abc", since with
'$append'/3, the DCG is basically:
p(C, B) :- '$append'("abc",A,C), q(A, B).
Now '$append'/3 would do a copying of the string,
if A is unbound, this is usually the "DCG used for
text generation" mode. But if A is bound, the
'$append'/3 would not do some copying, but it
would actually match the prefix. So it gives
a much better DCG for parsing, since this is
"DCG used for text parsing" mode.
Bye
Mild Shock schrieb:
Hi,
Ok lets run the test case on the desktop,
and not on the web. What do we get? Its almost
constant for Trealla Prolog as well, in
WebPL it was perfectly constant, but here
its only almost constant:
/* Trealla Prolog 2.82.14 */
?- time(test2(10)).
% Time elapsed 0.188s, 3004002 Inferences, 16.014 MLips
true.
?- time(test2(30)).
% Time elapsed 0.210s, 3004002 Inferences, 14.321 MLips
true.
?- time(test2(90)).
% Time elapsed 0.228s, 3004002 Inferences, 13.147 MLips
true.
Scryer Prolog fails the test horribly. Which
is amazing, since it is a Rust Prolog system just
like WebPL. But they are too traditional in
following the stupid WAM design:
/* Scryer Prolog 0.9.4-599 */
?- time(test2(10)).
% CPU time: 0.714s, 7_049_076 inferences
true.
?- time(test2(30)).
% CPU time: 1.284s, 7_049_099 inferences
true.
?- time(test2(90)).
% CPU time: 2.984s, 7_049_099 inferences
true.
Bye
Mild Shock schrieb:
Hi,
Heap/Stack Prolog systems could solve some Prolog
String Problems, especially in connection with a FFI, but I am
not showing that. More a general design limitation of the common
take of WAM resp. ZIP. The new WebPL Prolog describes itself as a
merged Heap/Stack architecture Prolog system. And has a reference
in its escorting paper to an academic work by Xining Li (1999):
A new term representation method for prolog
Xining Li - 1999
https://www.sciencedirect.com/science/article/pii/S0743106697000629
Besides that Program Sharing (PS), as it is called in the paper,
is nothing new, WebPL also shows a more modern take, in that
it already uses compound data types from Rust. Can we
replicate some of the performance advantages of a PS system
versus the more traditional WAM resp. ZIP based systems? Here
is a simple test in the WebPL Playground, for Web PL without GC:
/* WebPL NoGC */
?- test2(10).
(1795.6ms)
?- test2(30).
(1785.5ms)
?- test2(90).
(1765.6ms)
Then SWI-Prolog WASM as found in SWI-Tinker:
/* SWI-Prolog WASM */
?- test2(10).
(1239.3ms)
?- test2(30).
(2276.1ms)
?- test2(90).
(5372.3ms)
https://webpl.whenderson.dev/
Bye
The test case:
data(10, [10, 9, 8, 7, 6, 5, 4, 3, 2, 1]).
data(30, [30, 29, 28, 27, 26, 25, 24, 23,
22, 21, 20, 19, 18, 17, 16, 15, 14, 13,
12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1]).
data(90, [90, 89, 88, 87, 86, 85, 84, 83,
82, 81, 80, 79, 78, 77, 76, 75, 74, 73,
72, 71, 70, 69, 68, 67, 66, 65, 64, 63,
62, 61, 60, 59, 58, 57, 56, 55, 54, 53,
52, 51, 50, 49, 48, 47, 46, 45, 44, 43,
42, 41, 40, 39, 38, 37, 36, 35, 34, 33,
32, 31, 30, 29, 28, 27, 26, 25, 24, 23,
22, 21, 20, 19, 18, 17, 16, 15, 14, 13,
12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1]).
test(N) :- between(1,1000,_), data(N,_), fail.
test(_).
test2(N) :- between(1,1000,_), test(N), fail.
test2(_).
between(Lo, Lo, R) :- !, Lo = R.
between(Lo, _, Lo).
between(Lo, Hi, X) :- Lo2 is Lo+1, between(Lo2, Hi, X).
Mild Shock schrieb:
Hi,
WebPL is already outdated I guess. It doesn't
show the versions of the other Prolog systems
it is using. While I had these results for
the primes example in the WebPL playground:
/* Trealla Prolog WASM */
(23568.9ms)
When I run the example here:
https://php.energy/trealla.html
I get better results:
/* trealla-js 0.27.1 */
?- time(test).
% Time elapsed 9.907s, 11263917 Inferences, 1.137 MLips
Bye
| Sysop: | DaiTengu |
|---|---|
| Location: | Appleton, WI |
| Users: | 1,064 |
| Nodes: | 10 (0 / 10) |
| Uptime: | 148:00:29 |
| Calls: | 13,691 |
| Calls today: | 1 |
| Files: | 186,936 |
| D/L today: |
33 files (6,120K bytes) |
| Messages: | 2,410,927 |