02-Aug-2021 18:03 GMT.
UNDER CONSTRUCTION
Anonymous, there are 58 items in your selection [1 - 50] [51 - 58]
[News] WoaSE picturesANN.lu
Posted on 02-Nov-2002 18:36 GMT by Ian Stedman58 comments
View flat
View list
Just put some pictures of the World of Amiga show online at http://www.ianstedman.co.uk/woase/woase2002.htm Not put a show report yet but I have added comment text to each picture.

The show was good from my point of view, got a few bargains and got to play with a few Pegasos systems. All of the companies present were doing good trade and had good offers

Apparently AmigaOS 4 is 90% done, just the last few bits to do. The last 10% is always the tricky part. Ben Hermann's who presented had a sense of humour with a few jokes about Amiga user complaints! ;)

Some techie stuff on OS4 while I remember, exec re-written in C taking 42000 lines of code? Context switching times of 4 microseconds on AmigaOne and 40 microseconds on Blizzard PPC. Compare this to 400 microseconds or more for OS X. Apparently gives better real time performance than QNX momentics!

Amiga OS 4 has new TCP/IP stack, support for ReiserFS and the old FFS. USB built in. Hardware abstraction layer working well. HDToolbox replacement with media support. 68K emulation on a 600 MHZ G3 gives equivalent performance to a 68060@50MHz.

Unfortunately there was no demo of Amiga OS 4.

Thendic came well prepared with many Pegasos systems running MorphOS, MacOnLinux and SuSE Linux. I was impressed with these systems.

Eyetech had an AmigaOne in an ATX case, a motherboard and a system sort of running. I say sort of as I never saw anything more than a boot console.

That is all I can remember at the moment. Enjoy the pictures. Overall the show was enjoyable.

WoaSE pictures : Comment 1 of 58ANN.lu
Posted by André Siegel on 02-Nov-2002 17:45 GMT
Nice pictures, Ian!
WoaSE pictures : Comment 2 of 58ANN.lu
Posted by cOrpse on 02-Nov-2002 17:47 GMT
"Eyetech had an AmigaOne in an ATX case, a motherboard and a system sort of running. I say sort of as I never saw anything more than a boot console. "
What exactly is a *boot console*?
Not everyone runs X or even has it installed , and don't confuse a unix shell with msdos.
WoaSE pictures : Comment 3 of 58ANN.lu
Posted by cOrpse on 02-Nov-2002 17:53 GMT
"ReiserFS"
:)))
WoaSE pictures : Comment 4 of 58ANN.lu
Posted by SlimJim on 02-Nov-2002 17:59 GMT
Some techie stuff on OS4 while I remember, exec re-written in C taking 42000
lines of code? Context switching times of 4 microseconds on AmigaOne and 40
microseconds on Blizzard PPC. Compare this to 400 microseconds or more for OS
X. Apparently gives better real time performance than QNX momentics!
This does sound impressive. And the ExecSE is now working on AOne. That a great
stepping-stone (but didn't we know this already, coming to think of it?)
.
SlimJim
WoaSE pictures : Comment 5 of 58ANN.lu
Posted by takemehomegrandma on 02-Nov-2002 18:03 GMT
Nice pictures! BTW, you didn't happen to see a G4 card of any kind, beta or not?
WoaSE pictures : Comment 6 of 58ANN.lu
Posted by Don Cox on 02-Nov-2002 18:06 GMT
In reply to Comment 4 (SlimJim):
The switching time is impressive, and just what we need for audio and
video.
The emulation speed quoted must be wrong, surely? It should be around
equivalent of a 250MHz 68k on a 600MHz CPU.
It isn't clear from reports so far whether the Exec is running on the
A1 hardware, or just the PPC boot BIOS.
WoaSE pictures : Comment 7 of 58ANN.lu
Posted by cOrpse on 02-Nov-2002 18:12 GMT
In reply to Comment 6 (Don Cox):
>> Context switching times of 4 microseconds on AmigaOne and 40 microseconds on Blizzard PPC.
I think that hints that ExecSE is running on the AmigaOne ... unless that benchmark was a guess :\
It also shows the all round performance of the AmigaOne , its higher clock speed is only 3 -> 4 times that of the blizzard yet its switching time is 10 times faster :)
WoaSE pictures : Comment 8 of 58ANN.lu
Posted by SlimJIm on 02-Nov-2002 18:18 GMT
I want transcripts of the speeches!
.
SlimJim
WoaSE pictures : Comment 9 of 58ANN.lu
Posted by Christoph Gutjahr on 02-Nov-2002 18:19 GMT
In reply to Comment 6 (Don Cox):
It's ExecSG running on the AmigaOne:
http://www.amigart.com/amigaone/execsga1.jpg
WoaSE pictures : Comment 10 of 58ANN.lu
Posted by cOrpse on 02-Nov-2002 18:22 GMT
In reply to Comment 2 (cOrpse):
/Me is corrected :)
I thought they were running linux on the aOne not OS4 ! :D
WoaSE pictures : Comment 11 of 58ANN.lu
Posted by Steffen Haeuser on 02-Nov-2002 18:24 GMT
In reply to Comment 6 (Don Cox):
Hi!
This "Boot Console thing" is ExecSG running on the AmigaOne with a small console to be able to do some things already including (look at the
Screenshot at Amigart):
- Showing that the Multitasking works (ExecSG is fully completed)
- Showing that the Virtual Memory already works
- Showing that it actually works on a G4 machine
- Showing that PPCBoot is used and what sort of data it gives about the
machine
so this could be showed off before OS 4 is fully running - with all OS modules - on an AmigaOne.
This stuff is also running on "old-style" PPC Boards since a longer time
already, BTW.
Steffen Haeuser
WoaSE pictures : Comment 12 of 58ANN.lu
Posted by Some Farker on 02-Nov-2002 18:25 GMT
In reply to Comment 11 (Steffen Haeuser):
So what's the story, then? Decemberish? Januaryish?
WoaSE pictures : Comment 13 of 58ANN.lu
Posted by Andrea Maniero on 02-Nov-2002 18:28 GMT
In reply to Comment 9 (Christoph Gutjahr):
And this was all that was displayed of AmigaOS4? We had better luck at Pianeta Amiga in Italy, than!
Kind regards,
Andrea
WoaSE pictures : Comment 14 of 58ANN.lu
Posted by cOrpse on 02-Nov-2002 18:29 GMT
In reply to Comment 11 (Steffen Haeuser):
"so this could be showed off before OS 4 is fully running - with all OS modules - on an AmigaOne.
This stuff is also running on "old-style" PPC Boards since a longer time
already, BTW."
So is OS4 still going to be a stepped release ( cyberstorm version , amigaOne version then blizzard ) or will there be one unified release ?.
I ask this because I can't afford an AmigaOne right now and I have a blizzard but don't want to be left till last.
WoaSE pictures : Comment 15 of 58ANN.lu
Posted by Don Cox on 02-Nov-2002 18:50 GMT
In reply to Comment 13 (Andrea Maniero):
"And this was all that was displayed of AmigaOS4? We had better luck
at Pianeta Amiga in Italy, than! "
Well, getting the Exec to run on the new boards is the crucial step.
Once that is done, and the emulator is working, the hard part is over.
WoaSE pictures : Comment 16 of 58ANN.lu
Posted by Sam Smith on 02-Nov-2002 18:57 GMT
How about Quake 2? Any benchmarks?
---
Sam
WoaSE pictures : Comment 17 of 58ANN.lu
Posted by Mr Goatse.cx on 02-Nov-2002 19:30 GMT
LOOK OUT PEOPLE!
It's a woase.cx link!
;)
WoaSE pictures : Comment 18 of 58ANN.lu
Posted by Daniel Miller on 02-Nov-2002 19:35 GMT
The Pegasoses in those pictures look JAMMIN'! Super modern cases,
neat LCD monitors, sleek, lean, powerful and to the point all
around. This is the way we need to be going guys, keep it up!!!
WoaSE pictures : Comment 19 of 58ANN.lu
Posted by Georg Steger on 02-Nov-2002 19:37 GMT
In reply to Comment 4 (SlimJim):
> Context switching times of 4 microseconds on AmigaOne and 40
> microseconds on Blizzard PPC
Can we know how this was measured (how corresponding benchmark
code looks like)?
Does/did such a context switch also save/store FPU state. -->
In some (most?) OSes things like FPU state/MMX state etc.
is not saved/restored in each and every context switch but
only if FPU/MMX was used in the task's last ~"time slice"
(time between when the task got TS_RUN until it now gets
put to the ready/wait queue again).
WoaSE pictures : Comment 20 of 58ANN.lu
Posted by amigammc on 02-Nov-2002 19:54 GMT
In reply to Comment 18 (Daniel Miller):
Geez, Daniel, you can get the "ouside layer" for any machine, even an old P90 based computer.
WoaSE pictures : Comment 21 of 58ANN.lu
Posted by Daniel Miller on 02-Nov-2002 20:04 GMT
In reply to Comment 20 (amigammc):
Yeah, Luca, but the appearance is very important to the whole experience.
For instance I am sure you take pride in your home, and you want nice
furniture and wall-hangings and stuff. So having a stylish computer is
important too.
PS: Check out the picture where the cool Pegasos dude is having a good
conversation with that young woman as they admire the MorphOS setup. I
knew there was some reason that blue butterfly was made the mascot. It's
a babe magnet. ;)
WoaSE pictures : Comment 22 of 58ANN.lu
Posted by cOrpse on 02-Nov-2002 20:06 GMT
In reply to Comment 18 (Daniel Miller):
"The Pegasoses in those pictures look JAMMIN'! Super modern cases,
neat LCD monitors, sleek, lean, powerful and to the point all
around. This is the way we need to be going guys, keep it up!!!"
I thought they looked a bit gay* , and the whole purple theme of morphos looks a bit gay* aswell.
*gay in the comical , english , nothing to do with what you were thinking , old sense of the word , same as when people disribe the XP ui.
WoaSE pictures : Comment 23 of 58ANN.lu
Posted by cOrpse on 02-Nov-2002 20:08 GMT
In reply to Comment 21 (Daniel Miller):
" It's a babe magnet. ;) "
Women love my pierced lip , they couldn't give a shit about a freaking butterfly.
The places this BCR's been ;)
WoaSE pictures : Comment 24 of 58ANN.lu
Posted by LAD-knows-his-stuff on 02-Nov-2002 20:14 GMT
In reply to Comment 6 (Don Cox):
The performance numbers are a repeat of the embarassing stuff Ben was spewing at the other of the year, and also reminiscent of his claims for the VM.
A 4us context switch is ordinary, both Linux and Windows can beat that for the kind of toy benchmark I see running in these screenshots. QNX is about 10 times as fast. The reason you see much larger numbers (in the order of milliseconds) thrown around is because people are interested in /worst case latency/ for their full blown applications, the audio and video software that you're talking about.
For a hard disk recorder type of app the problem isn't "How quickly on average will my app be woken up?" but instead "Can I be sure that my app will ALWAYS be woken in time to fill/empty the buffers?". That's why e.g. a stock Red Hat Linux kernel just won't do for pro audio recording. Red Hat could publish numbers just like Ben's [only smaller] and it would mean nothing.
WoaSE pictures : Comment 25 of 58ANN.lu
Posted by Daniel Miller on 02-Nov-2002 20:25 GMT
In reply to Comment 23 (cOrpse):
cOrpse typed:
> Women love my pierced lip , they couldn't give a shit about
> a freaking butterfly.
Yeah, keep telling yourself that. ;)
WoaSE pictures : Comment 26 of 58ANN.lu
Posted by cOrpse on 02-Nov-2002 20:31 GMT
In reply to Comment 25 (Daniel Miller):
I don't need to tell myself , i'm reminded everyday by happy customers ;)
WoaSE pictures : Comment 27 of 58ANN.lu
Posted by Davy Wentzler on 02-Nov-2002 20:50 GMT
In reply to Comment 24 (LAD-knows-his-stuff):
> For a hard disk recorder type of app the problem isn't "How quickly on average will my app be woken up?" but instead "Can I
> be sure that my app will ALWAYS be woken in time to fill/empty the buffers?". That's why e.g. a stock Red Hat Linux kernel
> just won't do for pro audio recording.
Hmm.. not necessarily the app, but the drivers. Usually, some sort of double/triple
buffering method is used, so it tends not to be too critical for the app, unless
really small buffers are used.
Oh well, if it works good enough on a 'classic' Amiga, then no doubt will it run
flawlessly on the A1. When synchronizing apps on the same system, it helps to
have fast switching times though.
Davy Wentzler
Audio Evolution
WoaSE pictures : Comment 28 of 58ANN.lu
Posted by strobe on 02-Nov-2002 21:26 GMT
These "context switching" benchmarks are pretty meaningless. It doesn't say what is the ful context of the switch (which registers, how many threads, overall load, etc), and more importantly if the kernel implements real-time threading. Note that real-time threading doesn't mean 'faster', rather it means a thread will be given time at a specific interval, pre-empting normal threads. This is more important than how quick can a single context switch potentially happen.
WoaSE pictures : Comment 29 of 58ANN.lu
Posted by takemehomegrandma on 02-Nov-2002 22:05 GMT
In reply to Comment 23 (cOrpse):
> Women love my pierced lip , they couldn't give a shit about a freaking
> butterfly.
Have you tried stickin a butterfly through your lip? That would do it, I promise!
WoaSE pictures : Comment 30 of 58ANN.lu
Posted by cOrpse on 02-Nov-2002 22:24 GMT
In reply to Comment 29 (takemehomegrandma):
You can have flesh tunnels upto and beyond 20mm ... You could probably get a butterfly thru that (rolled up ofcourse).
I have a standard peircing as i'm not insane ... But even little ones promote untold amounts of senstivity so its great for <snipped> - refrences of cheesey taste and pubic hair.
/cOrpse - giving something back ;)
WoaSE pictures : Comment 31 of 58ANN.lu
Posted by smithy on 02-Nov-2002 22:29 GMT
Just got back from the show today. Last year Amithlon stole the show, but this year it was MorphOS. The stand was very professional-looking, the hardware looked swish and advanced, and the software looked sexy and modern. Even the posters looked smart, and the stand had about 6 people (all very smartly dressed) to answer questions and to persuade people to play on one of the dozen Pegasos machines that were at the show. At last, someone with some professionalism is back at the reigns!
The rest of the show was pretty much the same as last year. The presentations were good. First up, was Ben Hermans who made a charming speech about OS4, then Alan Redhouse, then Fleecy Moss, who announced the now familliar new big deal for Amiga Inc.
It's a shame Thendic didn't do a presentation. Apart from that, and the foul weather, and having to pay 79UKP because I missed my train, it was great!
WoaSE pictures : Comment 32 of 58ANN.lu
Posted by cOrpse on 02-Nov-2002 22:46 GMT
In reply to Comment 31 (smithy):
"Just got back from the show today. Last year Amithlon stole the show, but this year it was MorphOS. The stand was very professional-looking, the hardware looked swish and advanced, and the software looked sexy and modern. Even the posters looked smart, and the stand had about 6 people (all very smartly dressed) to answer questions and to persuade people to play on one of the dozen Pegasos machines that were at the show. At last, someone with some professionalism is back at the reigns!"
Which reigns would thendic be holding ? Personally I think they care a bit too much about the amiga market , and I don't think its because they want it grow and advance.
"It's a shame Thendic didn't do a presentation. Apart from that, and the foul weather, and having to pay 79UKP because I missed my train, it was great!"
IIRC Thendic haven't produced anything , so giving a technical speech would be a little bit silly. Thendic are PR , nothing more and they're only needed because of the sour attitude of the developers.
WoaSE pictures : Comment 33 of 58ANN.lu
Posted by LAD-back-from-pub on 02-Nov-2002 23:26 GMT
In reply to Comment 27 (Davy Wentzler):
The drivers aren't a problem unless you're really overloaded and the user put the sound card on a low priority line. The driver gets an interrupt and has a chance to move data almost immediately in the fast IRQ handler, if I understand what you mean correctly (I've never written a sound card driver from scratch, although my next work may soon change that).
Yes, I was talking about pro software, sorry I didn't make that clear. You can run a Windows Sound Recorder type app (even one with multi-channel support) on any hardware and OS using huge buffers. Pro audio is a moving goal though, and I don't know what your software achieves but 5ms (256 samples @ 44.1) latency is certainly considered "annoying lag" even by semi-pro users.
Anyone reading this who's interested and has the equipment here's an experiment to try. Connect a mic and headphones to a decent desk with built-in delay effects etc. Disable any "bypass" monitoring and arrange for the output of the delay to be sent to the headphones. Set the delay for 0.5s or so and begin reading something. Hard isn't it! You are interrupting yourself and you're socially conditioned to stop talking. Now keep reading while steadily reducing the delay. You should find that at a certain point your mind accepts the delayed voice as "my voice" and you'll be able to read at a normal pace but you will still make a lot of errors because the aural feedback is delayed.
Now try with a guitar! It's like typing blind, your hands try to do the right thing but your brain is confused. Some people I've spoken to say they can't tolerate more than 0.5ms of software delay (the ADC and DAC cause additional delay) when playing a guitar with software FX.
WoaSE pictures : Comment 34 of 58ANN.lu
Posted by takemehomegrandma on 02-Nov-2002 23:36 GMT
In reply to Comment 32 (cOrpse):
> Personally I think they care a bit too much about the amiga market
Of course they care about the amiga market, but IMHO they can't care about it enough!
> and I don't think its because they want it grow and advance.
Of course they want!
The MorphOS is a reimplementation of AmigaOS. It has been that from the very beginning. And it plays that roll perfect! Look at it as another OS distribution of the platform we all know. They have plans for it. The Pegasos is only a beginning. It's not branded "AmigaOS", but it sure is ... at least a different flavour of it. And it tastes good!
WoaSE pictures : Comment 35 of 58ANN.lu
Posted by takemehomegrandma on 02-Nov-2002 23:41 GMT
In reply to Comment 33 (LAD-back-from-pub):
> Set the delay for 0.5s
I propably make a fool of myself now (but I'm tired, kind of drunk, and have never been very good at this stuff), but is half a second really the same as 5 milli seconds?
WoaSE pictures : Comment 36 of 58ANN.lu
Posted by Bill Hoggett on 03-Nov-2002 00:01 GMT
In reply to Comment 35 (takemehomegrandma):
> I propably make a fool of myself now (but I'm tired, kind of drunk, and
> have never been very good at this stuff), but is half a second really the
> same as 5 milli seconds?
More like 500.
WoaSE pictures : Comment 37 of 58ANN.lu
Posted by Ami603 on 03-Nov-2002 00:16 GMT
In reply to Comment 33 (LAD-back-from-pub):
Sorry, i don`t understand all of your conversation,but in the first statement they are talking
about Microseconds,not miliseconds.
Bests regards
WoaSE pictures : Comment 38 of 58ANN.lu
Posted by Joe Vidueira on 03-Nov-2002 00:19 GMT
In reply to Comment 31 (smithy):
What WAS the new big deal for Amiga, Inc. that Fleecy announced?
WoaSE pictures : Comment 39 of 58ANN.lu
Posted by strobe on 03-Nov-2002 00:47 GMT
In reply to Comment 34 (takemehomegrandma):
They seem to care more about the Amiga market than Amiga Inc. which seems to be more concerned about the AmigaDE.
WoaSE pictures : Comment 40 of 58ANN.lu
Posted by Anonymous on 03-Nov-2002 01:24 GMT
>68K emulation on a 600 MHZ G3 gives equivalent performance to a 68060@50MHz.
With JIT or without?
The Pegasoses (Pegasi?) look interesting, it's a Coolermaster ATC-600 aluminium
case which can use two 5 1/4" drives side-by-side. The 610 might be cool too,
but I miss a funky status LCD display (and an IR receiver) if they want to make
it a mediabox. The drawback is the high price of this case and to find a PSU
without a fan to have a noiseless system.
What does the "G4" poster in the pegasos-maclinux-linux.jpg mean? Did Thendic
show or announce anything?
Refering to the execsga1.jpg picture...why does OS4 have an idle.task?
WoaSE pictures : Comment 41 of 58ANN.lu
Posted by Bernd Meyer on 03-Nov-2002 01:40 GMT
In reply to Comment 24 (LAD-knows-his-stuff):
Just to put numbers into perspective, here is what a simple main(){while(1) sched_yield();} type of benchmark will give on linux (run two instances, then use "vmstat 1" to view context switches per second):
Celeron300A: 3.0us; Celeron566: 1.2us; AthlonXP-1533: 0.53us.
These machines are using various 2.4.x kernels, i.e. are not yet using the new O(1) scheduler. And, of course, those are just _average_ numbers, i.e. completely meaningless when talking about realtime or lack thereof.
WoaSE pictures : Comment 42 of 58ANN.lu
Posted by Anonymous on 03-Nov-2002 02:55 GMT
Has Quake2 been ON SALE??
WoaSE pictures : Comment 43 of 58ANN.lu
Posted by amorel on 03-Nov-2002 06:19 GMT
In reply to Comment 12 (Some Farker):
If ever, not sooner than 2005. But personally I wouldn`t count on it ever being released in a workable state.
WoaSE pictures : Comment 44 of 58ANN.lu
Posted by amorel on 03-Nov-2002 06:35 GMT
In reply to Comment 31 (smithy):
"Fleecy Moss, who announced the now familliar new big deal for Amiga Inc."
Ofcourse the pros and cons of having floocy around are debatable, but:
What big deal? Could you be more specific? :-)
WoaSE pictures : Comment 45 of 58ANN.lu
Posted by Anonymous on 03-Nov-2002 07:11 GMT
In reply to Comment 44 (amorel):
The BIIIG Microsoft deal they talked about at Amiga.org?
WoaSE pictures : Comment 46 of 58ANN.lu
Posted by Ole-Egil on 03-Nov-2002 07:28 GMT
In reply to Comment 41 (Bernd Meyer):
Another thing to remember is that in QNX you sacrifice low averages against low maximums. What you can NEVER do in a real time system is miss a deadline. But that doesn't mean the average has to suck. It's just that you can't go out in public and say "we have this much latency on average", because average is NOT important. So you must stick with your maximum latency numbers. In Linux, the maximum latency can be _very_ high. In Windows, it can be forever (unless your process is running as the highest priority alone. If it's got several threads, you'll have to stick to the two highest priorities, because the VM swapper is on the third pri, and if this kicks in for some reasonon a free millisecond, your plane has crashed WELL before the real time process get's back control ;-) )
WoaSE pictures : Comment 47 of 58ANN.lu
Posted by Neko on 03-Nov-2002 07:29 GMT
Context switches of 4ms and 40ms?
Where does Hermans think this is relevant? That 4ms context switches (a
benchmark at best) are faster than QNX's context switches doesn't take into
account that QNX is a fair few times more complex and capable than exec.
The real beef would be when they tell us what the exec quantum is, so we can
see how long apps get to run before they are forced to switch. Running for 40ms
then a 40ms switch would drop system performance considerably. I believe an
exec 68k quantum is 20ms and the standard value is 4 - so 80ms, then 40ms would
be comparable times to AmigaOS (dunno how fast the switch is there) - maybe they
let tasks run longer on BlizzardPPC than on the AmigaONE too to eke out more
performance.
But two ms values mean dick. :)
WoaSE pictures : Comment 48 of 58ANN.lu
Posted by Rebel on 03-Nov-2002 08:05 GMT
In reply to Comment 6 (Don Cox):
>> The emulation speed quoted must be wrong, surely? It should be around
equivalent of a 250MHz 68k on a 600MHz CPU.
You correct of course :-)
The quoted 68060@50MHz emuluation speed was for the blizzard/604@266MHz. No figures were given for the G3@600MHz, but Ben said that would be several times faster. (See the figures for switching times to get an idea of the magnitude)
<Wait for the audio/video of from the show for confirmation.>
WoaSE pictures : Comment 49 of 58ANN.lu
Posted by Timothy De Groote on 03-Nov-2002 08:06 GMT
In reply to Comment 42 (Anonymous):
Yes, it was, and it was received well.
WoaSE pictures : Comment 50 of 58ANN.lu
Posted by Ole-Egil on 03-Nov-2002 08:52 GMT
In reply to Comment 47 (Neko):
_micro_seconds.
Anonymous, there are 58 items in your selection [1 - 50] [51 - 58]
Back to Top