@nikosidis

Of course JIT makes a big difference. What could be interesting is to ask deadwood to build a binary with jit disabled to compare properly.

Though you could also just compare with the older build, but it was also slower due to a less efficient memory allocator.

Writing it for PPC wouldn't be that much work, actually, but that's for someone knowing a minimum PPC asm, which is not my case. Someday, maybe... :)
That being said, WebKit guys should just write it for PPC as well. They even wrote it for SH4... :)

@nikosidis

hope now is enough of those benchmark results here, otherwise please close that thread, it starts to be tiresome..

i7 2600, 8 GB DDR3 1600 Corsair Vengeance, HD5870, Gigabyte GA-P67A-UD3-B3, Win7 ultimate 64bit

Firefox 11.0
============================================
RESULTS (means and 95% confidence intervals)
--------------------------------------------
Total: 168.0ms +/- 3.2%
--------------------------------------------

V8 Benchmark Suite

Score: 9446

Richards: 13164
DeltaBlue: 14189
Crypto: 19410
RayTrace: 4561
EarleyBoyer: 11458
RegExp: 1685
Splay: 10887
NavierStokes: 18235


PeaceKeeper
2703

Last edited by Seiya on 02-Apr-2012 at 05:32 PM.

_________________

I7-2600 / 16 GB RAM / AMD RADEON HD 6450 / Windows 7 Home Premiun / Internet Explorer 9


==========================================
RESULTS (means and 95% confidence intervals)
--------------------------------------------------------------------------
Total: 140.4ms +/- 0.4%
--------------------------------------------------------------------------
3d: 19.8ms +/- 2.3%
cube: 8.7ms +/- 4.0%
morph: 2.1ms +/- 10.8%
raytrace: 9.0ms +/- 0.0%
access: 19.4ms +/- 1.9%
binary-trees: 2.4ms +/- 15.4%
fannkuch: 7.0ms +/- 0.0%
nbody: 8.0ms +/- 0.0%
nsieve: 2.0ms +/- 0.0%
bitops: 13.0ms +/- 0.0%
3bit-bits-in-byte: 1.0ms +/- 0.0%
bits-in-byte: 3.0ms +/- 0.0%
bitwise-and: 3.0ms +/- 0.0%
nsieve-bits: 6.0ms +/- 0.0%
controlflow: 1.0ms +/- 0.0%
recursive: 1.0ms +/- 0.0%
crypto: 9.5ms +/- 4.0%
aes: 3.4ms +/- 10.9%
md5: 3.0ms +/- 0.0%
sha1: 3.1ms +/- 7.3%
date: 16.0ms +/- 0.0%
format-tofte: 6.0ms +/- 0.0%
format-xparb: 10.0ms +/- 0.0%
math: 11.0ms +/- 0.0%
cordic: 0.0ms +/- NaN%
partial-sums: 8.0ms +/- 0.0%
spectral-norm: 3.0ms +/- 0.0%
regexp: 5.0ms +/- 0.0%
dna: 5.0ms +/- 0.0%
string: 45.7ms +/- 1.1%
base64: 3.0ms +/- 0.0%
fasta: 11.1ms +/- 2.0%
tagcloud: 14.6ms +/- 2.5%
unpack-code: 10.0ms +/- 0.0%
validate-input: 7.0ms +/- 0.0%

_______________________________________________________
V8 Benchmark Suite

Score: 3832

Richards: 4049
DeltaBlue: 3778
Crypto: 5611
RayTrace: 2922
EarleyBoyer: 5630
RegExp: 3055
Splay: 3075
NavierStokes: 3504

Last edited by Fernecho on 02-Apr-2012 at 07:15 PM.

_________________
AmigaONE X1000, Radeon HD 7950
Amiga 1200 / Blizzard 1260 / Mediator 1200 / Voodoo 3000 / 64 Mb / AmigaOS 3.9

@Fab

We really need one and it's crazy that the fastest hardware got it but not us and the rest. I think it makes a big difference, how big I don't know though. Any opinions? I'm thinking that you might have tried TenFourFox which is supposed to have a JIT compiler for javascript.

@hotrod

It all depends on how heavy java-scripts are used on web-page.

Try this one:

http://helldesign.net/

Click on one of the pages and see how fast you can move the magnifying glass around.

I guess without JIT page will not be very smooth.

Last edited by nikosidis on 02-Apr-2012 at 07:52 PM.

@nikosidis

it's fast on i7-2600 with firefox 11.0, but to me seems no much smooth to change page..

Last edited by Seiya on 02-Apr-2012 at 08:16 PM.

_________________

@itix

For completeness result from PowerBook:

Quote:

Total: 3933.7ms +/- 1.5%

G4 1.67 GHz, MorphOS 3, OWB 1.11

_________________
Amiga Developer
Amiga 500, Efika, Mac Mini and PowerBook

@itix

Quote:

G4 1.67 GHz, MorphOS 3

Some users here see such line more than torturing...

@Seiya

Try to click on one of the first pages. Then hold your pointer a few sec.. and this magnifying glass will show up. Move it around. That is quite fast even with ACER aspire one, but without jit it is quite slow.

@nikosidis

ah, it's very much ok :)

_________________

@itix

Quote:

G4 1.67 GHz, MorphOS 3, OWB 1.11

Pfff...newest OS and oldest browser...

_________________
X1000|II/G4|440ep|2000/060|2000/040|1000

Had to try my laptop at work.

Intel Pentium Dual CPU T2370 1.73GHz 2GB mem.

Latest Chrome and Vista

Total: 555.0ms +/- 2.0%

haha, slower than my weaker AROS laptop with Odyssey

Again AROS prooves to be faster than Windows :D

Last edited by nikosidis on 03-Apr-2012 at 11:05 AM.

tests run in multitasking on AMD64x2 3800+ & pinguOS & firefox:
v8: score 2911 Sunspider: 481.0ms +/- 4.0%
(CPU load was not 100%, firefox was a little bit unresponsive, no hickups in simultaneous .avi playback)

In chromium web browser:
v8: Score 3634 Sunspider: 676.4ms +/- 3.1%
(again in multitasking, browser unresponsive but videopleyback ok, CPU load total 20...80%)

Interesting how much better linux performs. CPU load is more even between CPUs. But the test is pretty pathetic when it does not load CPUs fully.

Rerun in chromium with less multitasking (only the videoplayback in parallel):
v8: 3539 sunspider: 661.3ms +/- 3.2%

Firefox:
v8: 2673 Sunspider: 477.2ms +/- 2.2%

(funny that cromium responsiveness is signifficantly affected when the test is run on firefox, even when CPU load is not 100%)

Peacekeeper in chromium: 1578
Peacekeeper in firefox: 906

1.7Ghz Sempron, Bodhi Linux and midori browser:
Sunspider: 1719.0ms +/- 1.6%
V8 Score: 751
Nice to see the difference between 64x2 and sempron. 64x2 is 3x+ faster even when it's clock frequency is only 300Mhz higher and only one core is in stressed on 64x2. Bigger caches and 6x faster RAM makes big difference. And at the same time ... it shows how pathetic my Core2duo laptop is at work.
With browser that would have no JIT (IE8?), this HW might have hard time in beating SAM440ep in this test.
(if I had time, this would be my AROS/Aeros PC)

Last edited by KimmoK on 06-Apr-2012 at 03:14 PM.
Last edited by KimmoK on 06-Apr-2012 at 03:05 PM.
Last edited by KimmoK on 03-Apr-2012 at 03:23 PM.
Last edited by KimmoK on 03-Apr-2012 at 03:06 PM.
Last edited by KimmoK on 03-Apr-2012 at 03:00 PM.

_________________
- KimmoK
// For freedom, for honor, for AMIGA
//
// Thing that I should find more time for: CC64 - 64bit Community Computer?

All on the same engine (AMD Athlon II X2 250 Processor × 2 ):

AROS: old OWB

Total: 2949.1ms +/- 1.4%

Ubuntu: Firefox (9.0.1)

Total: 288.9ms

Ubuntu: Firefox (11)

Total: 236.6ms +/- 1.6%

WinXP: Firefox (11.0)

Total: 254.2ms +/- 1.2%

Last edited by Arko on 03-Apr-2012 at 11:22 PM.
Last edited by Arko on 03-Apr-2012 at 08:02 PM.
Last edited by Arko on 03-Apr-2012 at 07:55 PM.
Last edited by Arko on 03-Apr-2012 at 07:50 PM.

_________________
AmigaONE. Haha. Just because you can put label on it does not make it Amiga.

I borrowed this comments from here (#27 & #28):
http://amigaworld.net/modules/newbb/viewtopic.php?topic_id=38873&forum=2&start=20&order=0

Samsung Galaxy SII (2x Cortex-A9 @ 1.2Ghz)

Android ICS with default browser:

1873.9ms +/- 2.1%

_________________
Whyzzat?

@minator

Galaxy S2 using Opera

============================================
RESULTS (means and 95% confidence intervals)
--------------------------------------------
Total: 1593.3ms +/- 1.0%
--------------------------------------------

@nikosidis

*********** SunSpider:

============================================
RESULTS (means and 95% confidence intervals)
--------------------------------------------
Total: 7949.2ms +/- 0.8%
--------------------------------------------

3d: 1872.2ms +/- 2.1%
cube: 723.3ms +/- 4.8%
morph: 679.0ms +/- 6.0%
raytrace: 469.9ms +/- 2.5%

access: 1421.1ms +/- 2.5%
binary-trees: 62.9ms +/- 2.5%
fannkuch: 345.4ms +/- 0.9%
nbody: 895.7ms +/- 3.9%
nsieve: 117.1ms +/- 1.3%

bitops: 583.1ms +/- 1.8%
3bit-bits-in-byte: 112.9ms +/- 1.4%
bits-in-byte: 100.3ms +/- 1.6%
bitwise-and: 108.4ms +/- 2.3%
nsieve-bits: 261.5ms +/- 2.5%

controlflow: 86.2ms +/- 3.4%
recursive: 86.2ms +/- 3.4%

crypto: 405.8ms +/- 1.2%
aes: 131.5ms +/- 1.4%
md5: 133.5ms +/- 1.8%
sha1: 140.8ms +/- 2.0%

date: 250.8ms +/- 1.0%
format-tofte: 135.7ms +/- 1.5%
format-xparb: 115.1ms +/- 1.9%

math: 1355.8ms +/- 2.0%
cordic: 514.8ms +/- 2.2%
partial-sums: 557.6ms +/- 4.5%
spectral-norm: 283.4ms +/- 0.9%

regexp: 974.1ms +/- 0.4%
dna: 974.1ms +/- 0.4%

string: 1000.1ms +/- 0.9%
base64: 133.5ms +/- 1.9%
fasta: 178.6ms +/- 1.1%
tagcloud: 218.1ms +/- 1.0%
unpack-code: 310.9ms +/- 1.3%
validate-input: 159.0ms +/- 3.4%


************** V8 Bench:

Score: 123
Richards: 168
DeltaBlue: 144
Crypto: 83.5
RayTrace: 185
EarleyBoyer: 277
RegExp: 53.2
Splay: 235
NavierStokes: 39.7



************* PeaceKeeper:

score 182

All tests done on Pegasos2@1131 and MUI-OWB 1.9 on AmigaOS4.1.4

_________________
Simone"Tuxedo"Monsignori, Perugia, ITALY.

maybe it's time make another chart :)

_________________

@Seiya

Quote:

maybe it's time make another chart :)

Hopefully you will add in the ARM based systems numbers in as well.

_________________
Dammy