@damocles

i thought to include only amiga result...because could be to small to read data..

ok

chart with, Amiga and mobile.

Last edited by Seiya on 04-Apr-2012 at 11:27 PM.

_________________

SunSpider 0.9.1


V8 Benchmark Suite - version 7


for now..

Last edited by Seiya on 06-Apr-2012 at 12:43 AM.
Last edited by Seiya on 05-Apr-2012 at 05:53 PM.
Last edited by Seiya on 04-Apr-2012 at 11:37 PM.

_________________

I think in test results we should separate JIT results and non-JIT results.
It would reflect system performance differences better.

If we have both JIT and non JIT SW, it shows more the qualities of the SW application, not the HW/system.
When everyting is mixed, results become harder to interpret.

(for example, at best my SAM440 seems 80% as powerfull as 1.7Ghz Core2 laptop(with wind and IE8), when both use non JIT javascript tests )


With some hollywood benchmark we would at least have same application on every platform and not some JIT affect.

Last edited by KimmoK on 05-Apr-2012 at 06:55 AM.
Last edited by KimmoK on 05-Apr-2012 at 06:52 AM.

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

@KimmoK

yes, but in all of these tests what systems are jit and no jit?

_________________

@Seiya

Thx. That looks nice ;)

@nikosidis

I guess that means the Aros web experience is acceptable now then

_________________
Amiga 1000, 3000D Toaster, Checkmate A1500 Plus
http://www.checkmate1500plus.com/

@Seyia

V8 Benchmark Suite on A1 XE G4 1,26 Ghz and MUI-OWB :

Score: 155

Richards: 204
DeltaBlue: 168
Crypto: 91.5
RayTrace: 223
EarleyBoyer: 318
RegExp: 60.4
Splay: 490
NavierStokes: 52.1

_________________
PowerMac G5 2,7Ghz - 2GB - Radeon 9650 - MorphOS 3.14
AmigaONE X1000, 2GB, Sapphire Radeon HD 7700
FPGA Replay + DB 68060 at 85Mhz

if on aros owb we could use html5 or sfwdec, anythings will be more interesting :)
i'm very curious to see futurmark test on aros one day :)

_________________

@nikosidis


IE 9

============================================
RESULTS (means and 95% confidence intervals)
--------------------------------------------
Total: 226.0ms +/-

Chrome 18

============================================
RESULTS (means and 95% confidence intervals)
--------------------------------------------
Total: 343.6ms +/- 7.1%
--------------------------------------------


Opera 11.62

============================================
RESULTS (means and 95% confidence intervals)
--------------------------------------------
Total: 272.6ms +/- 0.8%
--------------------------------------------

All run on this machine
windows 7
Intel E6500
4 Gigs of ram

_________________
Support AROS sponsor a developer.

"AROS is prolly illegal ~ Evert Carton" intentionally quoted out of context for dramatic effect

@K-L

update che chart with your score :)

_________________

Quote:

Seiya wrote: (slower is better)

No it isn't!

@Seiya

interesting but:

- jit and non jit compare wasnt usefull

- also was useless to compare different owbversions since there was really big difference from 1.9 to latest 1.16...

as all benchmarks better compare same program version/features...

_________________
Simone"Tuxedo"Monsignori, Perugia, ITALY.

@Tuxedo

Was it not usefull ??

Seams pretty usefull to me.

Is it not very obious that JIT makes a huge difference. We are talking 5-10 times faster performance than without JIT.

@nikosidis

Quote:

Is it not very obious that JIT makes a huge difference.

That was obvious even before this benchmarks thread.


I think these benchmarks are useful - they can show where is room for improvements (eg. strange JS performance of SAM460).

v8 benchmarks, as before:

Samsung Galaxy SII (2x Cortex-A9 @ 1.2Ghz)
Android ICS with default browser


Richards: 1762
DeltaBlue: 2083
Crypto: 1707
RayTrace: 1709
EarleyBoyer: 2780
RegExp: 405
Splay: 411
NavierStokes: 805

Score 1189

_________________
Whyzzat?

@Tuxedo

Quote:

Tuxedo wrote: @Seiya interesting but: - jit and non jit compare wasnt usefull - also was useless to compare different owbversions since there was really big difference from 1.9 to latest 1.16... as all benchmarks better compare same program version/features...

That's like yelling "We have only 3 wheels!" on F1 race....

@Spectre660

Sam440ep-Flex 800mhz..
Debian Squeeze Epiphany Browser

============================================
RESULTS (means and 95% confidence intervals)
--------------------------------------------
Total: 16515.9ms 12.9%
--------------------------------------------

3d: 1955.2ms 23.5%
cube: 597.8ms 26.6%
morph: 723.6ms 23.6%
raytrace: 633.8ms 21.9%

access: 2108.8ms 18.5%
binary-trees: 162.5ms 21.1%
fannkuch: 1048.2ms 25.9%
nbody: 413.5ms 25.0%
nsieve: 484.6ms 16.9%

bitops: 1722.1ms 10.9%
3bit-bits-in-byte: 258.8ms 24.9%
bits-in-byte: 420.2ms 12.7%
bitwise-and: 399.3ms 14.8%
nsieve-bits: 643.8ms 8.0%

controlflow: 306.2ms 12.4%
recursive: 306.2ms 12.4%

crypto: 1149.8ms 12.9%
aes: 609.9ms 18.7%
md5: 191.8ms 7.5%
sha1: 348.1ms 12.2%

date: 1502.7ms 12.4%
format-tofte: 726.1ms 8.5%
format-xparb: 776.6ms 20.1%

math: 1794.1ms 15.0%
cordic: 486.9ms 33.8%
partial-sums: 949.2ms 11.6%
spectral-norm: 358.0ms 29.7%

regexp: 1917.0ms 25.9%
dna: 1917.0ms 25.9%

string: 4060.0ms 16.4%
base64: 562.9ms 17.7%
fasta: 467.8ms 7.9%
tagcloud: 672.6ms 5.4%
unpack-code: 1434.5ms 19.2%
validate-input: 922.2ms 32.1

-----------------------------------------------------------------------------------------------------------------------------
Quote:

Spectre660 wrote: @nikosidis Sam440ep-Flex 800 Mhz MUI-OWB 1.9 ============================================ RESULTS (means and 95% confidence intervals) -------------------------------------------- Total: 29862.8ms +/- 0.9% -------------------------------------------- 3d: 6410.8ms +/- 1.0% cube: 1546.2ms +/- 4.8% morph: 2954.9ms +/- 1.1% raytrace: 1909.7ms +/- 1.3% access: 3477.4ms +/- 3.4% binary-trees: 187.3ms +/- 41.0% fannkuch: 854.9ms +/- 3.6% nbody: 2148.9ms +/- 3.0% nsieve: 286.3ms +/- 18.4% bitops: 1666.1ms +/- 5.0% 3bit-bits-in-byte: 257.1ms +/- 17.6% bits-in-byte: 224.2ms +/- 18.1% bitwise-and: 263.8ms +/- 18.7% nsieve-bits: 921.0ms +/- 7.5% controlflow: 192.3ms +/- 27.4% recursive: 192.3ms +/- 27.4% crypto: 2303.1ms +/- 4.6% aes: 616.1ms +/- 10.9% md5: 750.1ms +/- 7.8% sha1: 936.9ms +/- 3.3% date: 2041.1ms +/- 1.8% format-tofte: 974.7ms +/- 1.4% format-xparb: 1066.4ms +/- 3.0% math: 4746.6ms +/- 6.4% cordic: 1561.4ms +/- 16.1% partial-sums: 1652.2ms +/- 6.4% spectral-norm: 1533.0ms +/- 8.0% regexp: 1912.9ms +/- 0.8% dna: 1912.9ms +/- 0.8% string: 7112.5ms +/- 1.8% base64: 1208.3ms +/- 6.5% fasta: 1345.3ms +/- 6.4% tagcloud: 908.6ms +/- 0.9% unpack-code: 2125.0ms +/- 1.6% validate-input: 1525.3ms +/- 6.6%

_________________
Sam460ex : Radeon Rx550 Single slot Video Card : SIL3112 SATA card

@nikosidis

Amiga 4K running 4.1 on 603e and OWB.

============================================
RESULTS (means and 95% confidence intervals)
--------------------------------------------
Total: 28901.7ms +/- 5.5%
--------------------------------------------

3d: 3754.6ms +/- 6.4%
cube: 1097.2ms +/- 10.9%
morph: 1379.4ms +/- 4.1%
raytrace: 1278.0ms +/- 6.1%

access: 4017.1ms +/- 9.4%
binary-trees: 366.4ms +/- 17.0%
fannkuch: 1729.0ms +/- 0.6%
nbody: 980.1ms +/- 2.0%
nsieve: 941.6ms +/- 33.6%

bitops: 2798.1ms +/- 1.5%
3bit-bits-in-byte: 555.6ms +/- 3.5%
bits-in-byte: 599.8ms +/- 3.3%
bitwise-and: 598.3ms +/- 1.2%
nsieve-bits: 1044.4ms +/- 1.0%

controlflow: 392.4ms +/- 1.4%
recursive: 392.4ms +/- 1.4%

crypto: 1922.5ms +/- 17.6%
aes: 980.1ms +/- 32.8%
md5: 477.6ms +/- 3.8%
sha1: 464.8ms +/- 1.9%

date: 2605.5ms +/- 11.2%
format-tofte: 1554.3ms +/- 16.9%
format-xparb: 1051.2ms +/- 6.1%

math: 3358.4ms +/- 0.8%
cordic: 1009.3ms +/- 0.7%
partial-sums: 1521.0ms +/- 0.8%
spectral-norm: 828.1ms +/- 1.8%

regexp: 3928.5ms +/- 1.0%
dna: 3928.5ms +/- 1.0%

string: 6124.6ms +/- 5.3%
base64: 811.9ms +/- 6.8%
fasta: 822.5ms +/- 5.4%
tagcloud: 1277.9ms +/- 2.8%
unpack-code: 2015.6ms +/- 5.3%
validate-input: 1196.7ms +/- 18.7%

@nikosidis

the problem was that comparing an x86 jit browser(with also differnet core since the OS4.1 was 1.9 against 1.16), with a ppc one(so different cpu with different optimisations and different cores of the browser) not have a real sense...

The best benchmrk was made on SAME code withou opts to test it on different CPUs...

opts and so on can be testd only with same hw(cpu/gfx) to have the real performance gain imho...

A code compiled for a different cpu with different hw, different opts and different code core not have any sense imho...

We can se where it works faster ok, but the becnhmark nature wasnt that...benchmarks was made to lok where the same works better so giving us wich hw was better in same conditions, testing different things wasnt a benchmark..only that.

But no problems at all with that tests only like to be precise on that point("what is a benchmark") nothing more nothing less...

_________________
Simone"Tuxedo"Monsignori, Perugia, ITALY.

@Tuxedo

X1000 users now have Debian, with all cpu optimization that they and you want..
Do you seen some X1000 linux benchmark?

This comparison is all that we have...
as soon as i hope that it will be update with Debain tests...

Last edited by Seiya on 27-Apr-2012 at 07:24 PM.
Last edited by Seiya on 27-Apr-2012 at 07:24 PM.

_________________