Welcome, Guest. Please login or register.

Author Topic: Timberwolf Progress Update  (Read 44035 times)

Description:

0 Members and 6 Guests are viewing this topic.

Offline unusedunused

  • Sr. Member
  • ****
  • Join Date: Nov 2005
  • Posts: 479
    • Show only replies by unusedunused
Re: Timberwolf Progress Update
« Reply #89 from previous page: August 11, 2011, 07:35:52 PM »
Quote from: Fab;654064
Well, I provided OWB sources and  gave my help to port it to OS4. And if AROS people are willing to (and that Zune is up to the task), i have no problem doing the same for it.

As for a 68k port, considering the requirements of WebKit, I'm not sure it makes much sense, except for UAE users, but is it really needed? They can just use the browser from the host machine then. :)

there is also natami here.now 2 more developers get a board.peterk (i guess its peter keuneke)do btw some fpu stuff on winuae and i write some mails with him also with AFA or iconbefast, i dont think that he make vapor announces.so seem natami can get reality.

http://www.natami.net/knowledge.php?b=1¬e=41108

I am a lazy boy, but if there is a browser possible in a few hours for AOS, which can play videos and most site work, i like to use it on winuae.

You have speedup OWB, maybe it is faster now as netsurf.

Can somebody who have MOS OWB and a Peg with 1 GHZ CPU tell what render time this page need ?

http://www.amiga.org/forums/showpost.php?p=649724&postcount=146

Can also tell time when page is offline

Here are some values of 68k netsurf and OS4 netsurf on some machines.

@klx300r
>let the whiners whine and the doers do

The problem is the doers do nothing visible.they only do a big hype.A amizilla alpha was here since several years, so whats on OS4 side currently is no progress.

Only the Marketing about firefox have increase alot, since MOS OWB is better than OS4 OWB.

and people as me tell that they think this is only simular as a poker trick.
maybe this motivate the friedens to show that they are really able to do that and do not just simple poker tricks to get more money and users for OS4.

same is real3d and dopus magellan, candy factory, that should come only for OS4, i see that too as simple poker tricks to show users, OS4 have better future, so they buy it

if timberwolf never get to a stable version its waste time.

nobody want use a buggy firefox.and a buggy and old timberwolf does not attract more users i think

And there are many guys that dont like a system that is promote with announces that seems like simple poker tricks.guys think the devs have no morale.and maybe dont want OS4.

For me this many unrealistic announces are a main reason, wy i never ever want use OS4.luckily there are other solutions..
same is  on the X1000 too

Or can you explain wy in dec. 2009 Hyperion want release a full system before summer, and suddenly when the time is over  they call for 100 betatesters that should buy this Hardware with preorder.
wy have they not know that before release 100 betatesters are need to test the Hardware better ?

What do you think ?

on other side, when look on acube, there hardware is slow and expensive, but i think their marketing ok
« Last Edit: August 11, 2011, 07:45:15 PM by bernd_afa »
 

Offline Piru

  • \' union select name,pwd--
  • Hero Member
  • *****
  • Join Date: Aug 2002
  • Posts: 6946
    • Show only replies by Piru
    • http://www.iki.fi/sintonen/
Re: Timberwolf Progress Update
« Reply #90 on: August 11, 2011, 08:39:37 PM »
Quote from: bernd_afa;654100
You have speedup OWB, maybe it is faster now as netsurf.
You're comparing apples and oranges.
Quote
Can somebody who have MOS OWB and a Peg with 1 GHZ CPU tell what render time this page need ?
What you're doing has nothing to do with benchmarking. What you're mostly measuring is the network latencies (DNS and connect, data transmission) for specific user at the specific moment in time. Also the server load the specific point in time can affect the result. To get any meaningful results the test must be of loading a page offline, with cache fully cleared before each test. But of course you know all this as you have been told numerous times already.

If you want to benchmark browsers (or specific subsystems of them in most cases), see:
http://clients.futuremark.com/peacekeeper/
http://www.webkit.org/perf/sunspider/sunspider.html
http://v8.googlecode.com/svn/data/benchmarks/v6/run.html
http://dromaeo.com/
http://www.howtocreate.co.uk/csstest.html

Oh wait! Netsurf can't run most of those test because it doesn't do javascript...
« Last Edit: August 11, 2011, 09:11:03 PM by Piru »
 

Offline HotRod

  • Sr. Member
  • ****
  • Join Date: Mar 2002
  • Posts: 370
    • Show only replies by HotRod
Re: Timberwolf Progress Update
« Reply #91 on: August 11, 2011, 08:57:14 PM »
Quote from: jorkany;654063
Timberwolf is for OS4 only, there is no "keep another 4(?) Amiga camps happy". Also if you think the Friedens give a **** about anything other than OS4, you need to up the meds. They would be perfectly happy if real Amiga, MOS and AROS all sank to the bottom of the ocean.


I hate people who tells what other people thinks. How the h*ll do you know what they think? Aiming at becoming the troll-master, are you? ;-)
 

Offline takemehomegrandma

  • Hero Member
  • *****
  • Join Date: Oct 2002
  • Posts: 2990
    • Show only replies by takemehomegrandma
Re: Timberwolf Progress Update
« Reply #92 on: August 11, 2011, 09:08:56 PM »
Quote from: HotRod;654106
Quote from: jorkany;654063
Timberwolf is for OS4 only, there is no "keep another 4(?) Amiga camps happy". Also if you think the Friedens give a **** about anything other than OS4, you need to up the meds. They would be perfectly happy if real Amiga, MOS and AROS all sank to the bottom of the ocean.


I hate people who tells what other people thinks. How the h*ll do you know what they think? Aiming at becoming the troll-master, are you? ;-)


Oh, but the accumulated comments, public proclamations, actions, etc from the Hyperion pack over the last decade gives quite a clear picture you know... ;)
MorphOS is Amiga done right! :)
 

Offline jorkany

  • Hero Member
  • *****
  • Join Date: Sep 2006
  • Posts: 1009
    • Show only replies by jorkany
    • http://www.amigaos4.com
Re: Timberwolf Progress Update
« Reply #93 on: August 11, 2011, 09:18:04 PM »
Quote from: HotRod;654106
I hate people who tells what other people thinks. How the h*ll do you know what they think? Aiming at becoming the troll-master, are you? ;-)

From statements the Friedens have made in the past, unless they were lying. At least they haven't gotten to the point where they say they want people to die in a fire, like Rigo.

Edit: And like TMHG mentioned, actions. This Timberwolf browser is a good example - the Friedens don't plan to make it open source, and I'm sure they would be the first to say "We did something and you didn't so you don't deserve it, port it yourself". Meanwhile you've got people like Fab, kase1, and Karlos who actually do write code across platforms. Where do you think the split in the communities comes from? Primarily from the Friedens and Hyperion. I never see the AROS, MOS and Amiga guys arguing about each other, it's always OS4 vs. Everybody else - and it's all down to the crappy attitudes of the gang at Hyperion. Take off the blinders.
« Last Edit: August 11, 2011, 09:32:18 PM by jorkany »
 

Offline magnetic

  • Hero Member
  • *****
  • Join Date: Feb 2002
  • Posts: 2531
    • Show only replies by magnetic
Re: Timberwolf Progress Update
« Reply #94 on: August 11, 2011, 09:58:03 PM »
Once again a verbose thread on OS4 or major os4 software and not one post from any rep or dev from Hyperion themselves. its quite amusing and sad. Its amazing these guys inherited the "Amiga" legacy and the Morph guys are the bad guys when they have delivered miles more...

oh and JJ you are right "usb2" as a big feature is quite lame. Should be usb3 especially when you are selling a $3000 X1000! wtf you gonna have PATA and usb1? Running outdated browsers and SuperNes emus?

And all this Gallium talk.. ya right!  Just like all those awesome "SNAP drivers" lmfao! bafa has points with Candy Factory, Aladdin, Gorky 17, etc etc ad naseum...
bPlan Pegasos2 G4@1ghz
Quad Boot:Reg. MorphOS | OS4.1 U4 |Ubuntu GNU-Linux | MacOS X

Amiga 2000 Rom Switcher w/ 3.1 + 1.3 | HardFrame SCSI | CBM Ram board| A Squared LIVE! 2000 | Vlab Motion | Firecracker 24 gfx

Commodore CDTV: 68010 | ECS | 9mb Ram | SCSI -TV | 3.9 Rom | Developer EPROMs
 

Offline Iggy

  • Hero Member
  • *****
  • Join Date: Aug 2009
  • Posts: 5348
    • Show only replies by Iggy
Re: Timberwolf Progress Update
« Reply #95 on: August 12, 2011, 03:50:18 AM »
Quote from: magnetic;654123
Once again a verbose thread on OS4 or major os4 software and not one post from any rep or dev from Hyperion themselves. its quite amusing and sad. Its amazing these guys inherited the "Amiga" legacy and the Morph guys are the bad guys when they have delivered miles more...

oh and JJ you are right "usb2" as a big feature is quite lame. Should be usb3 especially when you are selling a $3000 X1000! wtf you gonna have PATA and usb1? Running outdated browsers and SuperNes emus?

And all this Gallium talk.. ya right!  Just like all those awesome "SNAP drivers" lmfao! bafa has points with Candy Factory, Aladdin, Gorky 17, etc etc ad naseum...

Well Mag, look at it this way. We'll soon will have Powerbook support, one day we may have G5 support, and they (may) get a really expensive computer.
It's a "real" Amiga. yay! An Amiga for the Special Ed crowd!
"Not making any hard and fast rules means that the moderators can use their good judgment in moderation, and we think the results speak for themselves." - Amiga.org, terms of service

"You, got to stem the evil tide, and keep it on the the inside" - Rogers Waters

"God was never on your side" - Lemmy

Amiga! "Our appeal has become more selective"
 

Offline Fab

  • Full Member
  • ***
  • Join Date: Jun 2009
  • Posts: 217
    • Show only replies by Fab
Re: Timberwolf Progress Update
« Reply #96 on: August 12, 2011, 11:17:36 AM »
Quote from: bernd_afa;654100

Can somebody who have MOS OWB and a Peg with 1 GHZ CPU tell what render time this page need ?

http://www.amiga.org/forums/showpost.php?p=649724&postcount=146


As Piru and many more explained, this is wrong to benchmark this way, but if you want a number, i can give you one anyway. With my not so good inet link, OWB loads the site in about 4 seconds, and reloads in 2 seconds. Of course, i disabled Javascript and Flash to be in the "same" conditions as netsurf.

And let me tell you again, the inet latency in this case *is* important. It might not be that much important on 68k where there's also much more time spent in displaying/parsing/decoding, but once you have a faster machine, inet latency becomes the bottleneck.
 

Offline unusedunused

  • Sr. Member
  • ****
  • Join Date: Nov 2005
  • Posts: 479
    • Show only replies by unusedunused
Re: Timberwolf Progress Update
« Reply #97 on: August 12, 2011, 03:59:26 PM »
Quote from: Fab;654176
As Piru and many more explained, this is wrong to benchmark this way, but if you want a number, i can give you one anyway. With my not so good inet link, OWB loads the site in about 4 seconds, and reloads in 2 seconds. Of course, i disabled Javascript and Flash to be in the "same" conditions as netsurf.

And let me tell you again, the inet latency in this case *is* important. It might not be that much important on 68k where there's also much more time spent in displaying/parsing/decoding, but once you have a faster machine, inet latency becomes the bottleneck.

Are this test on a Peg 1 GHZ ?

You know, for 68k ixemul is gprof profiler here and you can see in the profile how much inet influence a page have, or what functions cost most time.
Too bad that MOS and OS4 have no gprof, so you can verify better what code cost most time and internet have no influence, the response times of inet access are really fast.

If you think inet access slow down, you can ofcourse do the test and show the page from harddrive.
I know this benchmark is not perfect to compare, but i dont want buy a OS4 system with all hardware and benchmark it myself.

Its always hard from OS4 users to get some benchmark results.And i am not a guy that ask users again and again to make benchmarks.
The results of classic, SAM and pegasus are all from same user done on same day and simular time.so inet access you can compare good.

that inet access is not important in time, you can also see because Pegasos  G4 1 GHZ is 2.3* faster as SAM 667 MHZ with netsurf

this fit the blender benchmark, here Peg is too around 2* faster.And because the G4 have same much better performance /MHZ in the netsurf test and blender test, can be 99% sure inet is no speedbrake.


maybe you have a SDL OWB somewhere that use no Cairo, and can test what values it reach.

because i remember firefox was on my PC very fast, and as soon they add Cairo support it go lots slower.I have a PCIe gforce 6600 Card.so not the slowest.

I hope you not understand my test wrong.I dont want say 68k and SDL is the best system, i only want test with that test, if the port run at optimal speed.

most intresting is for me the speed of Acorn risc OS Version.This use no Cairo too.
maybe here is somebody who can do the test on a Acorn Risc Machine.

@Piru
>Oh wait! Netsurf can't run most of those test because it doesn't do javascript...

this tests have nothing to do with real world pages.No page do extensive java script access or this what the tests test.when i look in profiler, most time cost CSS layout, text and image render.Is there a benchmark here that test CSS layout speed ?

this links are synthetic benchmarks.
same as landmark mips or some else for CPU.

I prefer real world benches on often used pages.
so my test you can see simular to blender speed test.
« Last Edit: August 12, 2011, 04:12:04 PM by bernd_afa »
 

Offline Piru

  • \' union select name,pwd--
  • Hero Member
  • *****
  • Join Date: Aug 2002
  • Posts: 6946
    • Show only replies by Piru
    • http://www.iki.fi/sintonen/
Re: Timberwolf Progress Update
« Reply #98 on: August 12, 2011, 06:29:04 PM »
Quote from: bernd_afa;654259
The results of classic, SAM and pegasus are all from same user done on same day and simular time.so inet access you can compare good.
You think the internet connection is working constantly all the time? Are you that uninformed? Latencies can vary from second to second due to multitude of reasons, some client and some server side. Benchmarking page load on live inet connection can never produce any reasonable benchmark results, even if you run the tests minutes apart.

Quote
that inet access is not important in time, you can also see because Pegasos  G4 1 GHZ is 2.3* faster as SAM 667 MHZ with netsurf

this fit the blender benchmark, here Peg is too around 2* faster.And because the G4 have same much better performance /MHZ in the netsurf test and blender test, can be 99% sure inet is no speedbrake.
Correlation does not imply causation

This is just one of the fallacies you seem to be blind to.

@Piru
Quote
Is there a benchmark here that test CSS layout speed ?
Yes.
 

Offline unusedunused

  • Sr. Member
  • ****
  • Join Date: Nov 2005
  • Posts: 479
    • Show only replies by unusedunused
Re: Timberwolf Progress Update
« Reply #99 on: August 12, 2011, 07:48:23 PM »
Quote from: Piru;654289
You think the internet connection is working constantly all the time? Are you that uninformed? Latencies can vary from second to second due to multitude of reasons, some client and some server side. Benchmarking page load on live inet connection can never produce any reasonable benchmark results, even if you run the tests minutes apart.


Correlation does not imply causation

This is just one of the fallacies you seem to be blind to.

@Piru

Yes.

I tell the users that they should do every test several times and only if the test result differ not more than 1-2 sec they should use it.

I tell the users to load also a page offline and test speed.

But i always dont understand wy guys only tell this results are total wrong INSTEAD POST values from show the page from ram disk or maybe harddrive.

I can save a page complete, to my 120 GB SSD with sandforce controller.Its fast.
I notice only that 1.1 sec the page need more time to load from inet.

I pay for cheapest DSL 6000 connection, because slower is not cheaper.
But I have a DSL 3000 Connection, because cable is not good for DSL 6000.So my inet is lot slower.

my firefox results are with java enable.so more slow, because more data need load.
I am too lazy to disable it.i stop time until the reload button come again, this indicate on firefox page is complete load.

My results are 2.6 sec load from inet with firefox 5
1.5 sec load from harddrive.

But with faster DSL it get faster so inet time is maybe 0.5- 0.6 sec.

EDIT:

I have now test it too, times are same.

osnews is a fast and constant response server.


>>Is there a benchmark here that test CSS layout speed ?
>Yes.

what test it is ?
maybe it is possible to run on netsurf.
« Last Edit: August 12, 2011, 07:51:50 PM by bernd_afa »
 

Offline Piru

  • \' union select name,pwd--
  • Hero Member
  • *****
  • Join Date: Aug 2002
  • Posts: 6946
    • Show only replies by Piru
    • http://www.iki.fi/sintonen/
Re: Timberwolf Progress Update
« Reply #100 on: August 12, 2011, 09:16:38 PM »
Quote from: bernd_afa;654305
I tell the users that they should do every test several times and only if the test result differ not more than 1-2 sec they should use it.

That doesn't help. The inet connection can be slow for long period of times. Or not. It's just impossible to tell. The inet connection is the largest random variable and thus is must be eliminated to get any sensible data.

From rest of you reply I see that you still are unable to understand the problem. I give up. But if you bring your flawed benchmarks up again I will keep on pointing out that they're completely pointless.

Quote
what test it is ?

The one that says "csstest" perhaps?
 

Offline Duce

  • Off to greener pastures
  • Hero Member
  • *****
  • Join Date: Jul 2009
  • Posts: 1699
    • Show only replies by Duce
    • http://amigabbs.blogspot.com/
Re: Timberwolf Progress Update
« Reply #101 on: August 12, 2011, 09:25:46 PM »
Better off testing speeds via a LAN setup, entirely off the Internet for some consistent benchmarks.
 

Offline wawrzon

Re: Timberwolf Progress Update
« Reply #102 on: August 12, 2011, 09:51:51 PM »
@bernd:
i dont know how difficult it is to test with locally saved page? you can even send the same to other users to test to have dependable results. why argue about that all so much?
 

Offline unusedunused

  • Sr. Member
  • ****
  • Join Date: Nov 2005
  • Posts: 479
    • Show only replies by unusedunused
Re: Timberwolf Progress Update
« Reply #103 on: August 13, 2011, 11:03:45 AM »
Quote from: wawrzon;654320
@bernd:
i dont know how difficult it is to test with locally saved page? you can even send the same to other users to test to have dependable results. why argue about that all so much?

as i have written before, i tell the users to test it offline but nobody do.I also accept if 68k do the test on internet and the fast platforms can do that with local saved pages.68k SDL Version is over 3* faster, so save internet access does also not help much on other platform i am 99.99% sure.

But fab give me time values of OWB only for inet acces and no time for local saved page.
or are MOS /OS4 browsers not able to save a page local with all images and other stuff ?

only this inet values i get.and here can see that G4 1 GHZ Peg  is around 2* faster as a 667 MHZ SAM, same faster as blender bench, which do btw not load data from internet

So my feeling is, these guys search only for a excuse that ixemul SDL build is not fastest per 100 MHZ, because it doesnt fit the bash without Facts again SDL and ixemul programs some do.also the bash that  68k Compiler produce slow code, seem with that benchmark results wrong, when such a slow small cache and memory performance system as a classic with 040 or 060 can reach this good values / MHZ

And i am sure, when do tests with local files, then there is no such easy excuse possible, so maybe nobody do that.
« Last Edit: August 13, 2011, 11:13:32 AM by bernd_afa »
 

Offline Piru

  • \' union select name,pwd--
  • Hero Member
  • *****
  • Join Date: Aug 2002
  • Posts: 6946
    • Show only replies by Piru
    • http://www.iki.fi/sintonen/
Re: Timberwolf Progress Update
« Reply #104 on: August 13, 2011, 11:16:28 AM »
Quote from: bernd_afa;654430
as i have written before, i tell the users to test it offline but nobody do.
How about if you'd provide the test kit with instructions how to use it? If you don't, everyone will do random thing x and you will get totally random results. I know I know, you're 99% sure everyone is doing it correctly...

The only way to get reliable results would be to run the tests on the same system, with offline content. Anything else is just pointless random guessing about the testing environment for any specific result. Drawing any kind of conclusions from such a random result set is just idiotic.

« Last Edit: August 13, 2011, 11:24:30 AM by Piru »