Bck2BrwsrViaRegisters - Revision history

JaroslavTulach at 06:44, 16 December 2012

2012-12-16T06:44:14Z

←Older revision		Revision as of 06:44, 16 December 2012
Line 30:		Line 30:
	i1 = arg1;		i1 = arg1;
	i2 = arg2;		i2 = arg2;
-	~~return~~ i1 + i2;	+	i1 += i2;
		+	return i1;
	}		}
	</source>		</source>

JaroslavTulach at 12:25, 14 December 2012

2012-12-14T12:25:54Z

←Older revision		Revision as of 12:25, 14 December 2012
Line 36:		Line 36:
	Which is good enough for [[Chrome]]'s [[JavaScript]] virtual machine to swallow (and hence the ten times boost). [[Firefox]] probably does not optimize the unnecessary assignments well enough and thus the speed up was less dramatic.		Which is good enough for [[Chrome]]'s [[JavaScript]] virtual machine to swallow (and hence the ten times boost). [[Firefox]] probably does not optimize the unnecessary assignments well enough and thus the speed up was less dramatic.

-	PS: A thing to note is that originally (with the stack based approach) the [[Firefox]] was faster than [[Chrome]]. Looks like [[~~Chrome~~]] guys forgot to optimize for '''Array.push''' and '''Array.pop'''!	+	PS: A thing to note is that originally (with the stack based approach) the [[Firefox]] was faster than [[Chrome]]. Looks like [[V8]] guys forgot to optimize for '''Array.push''' and '''Array.pop'''!

	PPS: Subjective measurements showed that [[Bck2Brwsr]] on [[V8]] runs the benchmark only four times slower compared to [[HotSpot]] - the real [[Java]]'s virtual machine.		PPS: Subjective measurements showed that [[Bck2Brwsr]] on [[V8]] runs the benchmark only four times slower compared to [[HotSpot]] - the real [[Java]]'s virtual machine.

JaroslavTulach at 12:25, 14 December 2012

2012-12-14T12:25:28Z

←Older revision		Revision as of 12:25, 14 December 2012
Line 38:		Line 38:
	PS: A thing to note is that originally (with the stack based approach) the [[Firefox]] was faster than [[Chrome]]. Looks like [[Chrome]] guys forgot to optimize for '''Array.push''' and '''Array.pop'''!		PS: A thing to note is that originally (with the stack based approach) the [[Firefox]] was faster than [[Chrome]]. Looks like [[Chrome]] guys forgot to optimize for '''Array.push''' and '''Array.pop'''!

-	PPS: ~~Measurements~~ showed that [[Bck2Brwsr]] on [[V8]] runs the benchmark only four times slower compared to [[HotSpot]].	+	PPS: Subjective measurements showed that [[Bck2Brwsr]] on [[V8]] runs the benchmark only four times slower compared to [[HotSpot]] - the real [[Java]]'s virtual machine.

JaroslavTulach at 12:24, 14 December 2012

2012-12-14T12:24:43Z

←Older revision		Revision as of 12:24, 14 December 2012
Line 37:		Line 37:

	PS: A thing to note is that originally (with the stack based approach) the [[Firefox]] was faster than [[Chrome]]. Looks like [[Chrome]] guys forgot to optimize for '''Array.push''' and '''Array.pop'''!		PS: A thing to note is that originally (with the stack based approach) the [[Firefox]] was faster than [[Chrome]]. Looks like [[Chrome]] guys forgot to optimize for '''Array.push''' and '''Array.pop'''!
		+
	PPS: Measurements showed that [[Bck2Brwsr]] on [[V8]] runs the benchmark only four times slower compared to [[HotSpot]].		PPS: Measurements showed that [[Bck2Brwsr]] on [[V8]] runs the benchmark only four times slower compared to [[HotSpot]].

JaroslavTulach at 12:23, 14 December 2012

2012-12-14T12:23:45Z

←Older revision		Revision as of 12:23, 14 December 2012
Line 37:		Line 37:

	PS: A thing to note is that originally (with the stack based approach) the [[Firefox]] was faster than [[Chrome]]. Looks like [[Chrome]] guys forgot to optimize for '''Array.push''' and '''Array.pop'''!		PS: A thing to note is that originally (with the stack based approach) the [[Firefox]] was faster than [[Chrome]]. Looks like [[Chrome]] guys forgot to optimize for '''Array.push''' and '''Array.pop'''!
-	PPS: Measurements showed that [[Bck2Brwsr]] on [[V8]] runs the benchmark only four times slower compared to [[~~Hotspot~~]].	+	PPS: Measurements showed that [[Bck2Brwsr]] on [[V8]] runs the benchmark only four times slower compared to [[HotSpot]].

JaroslavTulach at 12:23, 14 December 2012

2012-12-14T12:23:29Z

←Older revision		Revision as of 12:23, 14 December 2012
Line 23:		Line 23:
	Congratulation Ľubomír!		Congratulation Ľubomír!

-	Of course the magic in in eliminating the stack and using normal local variables. At the end the code looks like:	+	Of course the magic is in eliminating the stack and using normal local variables. At the end the code looks like:

	<source lang="javascript">		<source lang="javascript">
Line 37:		Line 37:

	PS: A thing to note is that originally (with the stack based approach) the [[Firefox]] was faster than [[Chrome]]. Looks like [[Chrome]] guys forgot to optimize for '''Array.push''' and '''Array.pop'''!		PS: A thing to note is that originally (with the stack based approach) the [[Firefox]] was faster than [[Chrome]]. Looks like [[Chrome]] guys forgot to optimize for '''Array.push''' and '''Array.pop'''!
		+	PPS: Measurements showed that [[Bck2Brwsr]] on [[V8]] runs the benchmark only four times slower compared to [[Hotspot]].

JaroslavTulach at 12:21, 14 December 2012

2012-12-14T12:21:56Z

←Older revision		Revision as of 12:21, 14 December 2012
Line 21:		Line 21:
	[[Image:Bck2BrwsrRegistersChrome.png]]		[[Image:Bck2BrwsrRegistersChrome.png]]

-	Congratulation Ľubomír! Of course the magic in in eliminating the stack and using normal local variables. At the end the code looks like:	+	Congratulation Ľubomír!
		+
		+	Of course the magic in in eliminating the stack and using normal local variables. At the end the code looks like:

	<source lang="javascript">		<source lang="javascript">

JaroslavTulach at 12:21, 14 December 2012

2012-12-14T12:21:21Z

←Older revision		Revision as of 12:21, 14 December 2012
Line 11:		Line 11:
	</source>		</source>

-	This is almost one to one translation of [[Java]]'s [[bytecode]], but obviously, this is not really fast. Instead of a single operation, we are asking the [[JavaScript]] engines to ~~manipulate~~ with stack six times!	+	This is almost one to one translation of [[Java]]'s [[bytecode]], but obviously, this is not really fast. Instead of a single operation, we are asking the [[JavaScript]] engines to deal with the stack six times!

	On Dec 14, 2012 Ľubomír made a significant break through: after rewrite to register based system (thanks to analysis of stackmap information), our matrix benchmark got at least twice as fast on [[Firefox]]:		On Dec 14, 2012 Ľubomír made a significant break through: after rewrite to register based system (thanks to analysis of stackmap information), our matrix benchmark got at least twice as fast on [[Firefox]]:

JaroslavTulach at 12:10, 14 December 2012

2012-12-14T12:10:46Z

←Older revision		Revision as of 12:10, 14 December 2012
Line 15:		Line 15:
	On Dec 14, 2012 Ľubomír made a significant break through: after rewrite to register based system (thanks to analysis of stackmap information), our matrix benchmark got at least twice as fast on [[Firefox]]:		On Dec 14, 2012 Ľubomír made a significant break through: after rewrite to register based system (thanks to analysis of stackmap information), our matrix benchmark got at least twice as fast on [[Firefox]]:

-		+	[[Image:Bck2BrwsrRegistersFirefox.png]]

	and about 10x(!) faster on [[Chrome]]:		and about 10x(!) faster on [[Chrome]]:
		+
		+	[[Image:Bck2BrwsrRegistersChrome.png]]
		+
		+	Congratulation Ľubomír! Of course the magic in in eliminating the stack and using normal local variables. At the end the code looks like:
		+
		+	<source lang="javascript">
		+	function plus(arg1, arg2) {
		+	var i1, i2;
		+	i1 = arg1;
		+	i2 = arg2;
		+	return i1 + i2;
		+	}
		+	</source>
		+
		+	Which is good enough for [[Chrome]]'s [[JavaScript]] virtual machine to swallow (and hence the ten times boost). [[Firefox]] probably does not optimize the unnecessary assignments well enough and thus the speed up was less dramatic.
		+
		+	PS: A thing to note is that originally (with the stack based approach) the [[Firefox]] was faster than [[Chrome]]. Looks like [[Chrome]] guys forgot to optimize for '''Array.push''' and '''Array.pop'''!

JaroslavTulach: New page: Original implementation of Bck2Brwsr virtual machine used Java classical stack based operations. E.g. when one needed to sum two numbers we had to manipulate the stack:
2012-12-14T12:01:00Z

New page: Original implementation of Bck2Brwsr virtual machine used Java classical stack based operations. E.g. when one needed to sum two numbers we had to manipulate the stack: <source la...

New page
Original implementation of [[Bck2Brwsr]] virtual machine used [[Java]] classical stack based operations. E.g. when one needed to sum two numbers we had to manipulate the stack:

<source lang="javascript">
function plus(arg1, arg2) {
var stack = new Array();
stack.push(arg1);
stack.push(arg2);
stack.push(stack.pop() + stack.pop());
return stack.pop();
}
</source>

This is almost one to one translation of [[Java]]'s [[bytecode]], but obviously, this is not really fast. Instead of a single operation, we are asking the [[JavaScript]] engines to manipulate with stack six times!

On Dec 14, 2012 Ľubomír made a significant break through: after rewrite to register based system (thanks to analysis of stackmap information), our matrix benchmark got at least twice as fast on [[Firefox]]:

and about 10x(!) faster on [[Chrome]]: