And so it happened again! The PowerPC/MorphOS platform lay down one more platform in the RC5-72 space race. This time it was the Alpha/DEC OSF/1 platform that hit the ground.

And it went down with a loud *BLAM*, because at the same time the PowerPC/MorphOS platform sets another personal best when it comes to the number of blocks crunched in a single day:

* 3,836 * blocks!

http://www.morphos-news.de/guides/rc5-72/dnet-mos-stats.php

The MorphOS now puts itself at the #14 position from the top, and we are actually closing in on the "mainstream" platforms. The curve now gets steeper and the opponents gets tougher, but it's still not at all impossible (at least not the closest two opponents, which we probably will beat before the summer is over, perhaps even with some margins).

However, this will require more effort from all of us (if running the client in the background and letting your computer be switched on when you leave it can be considered an *effort*(?)). While some new team members has been added to the dnet-mos-stats page above, I have also noticed that some of you who helped crunching keys in the beginning now has stopped for some reason. Come back, we need you!

If everyone listed on the dnet-mos-stats page above starts crunching keys again, at least for a little while every day, then it's my belief that we easily can go up to about some 4,500 keys per day without any greater effort (every group of only ten users with G4 will bring in more than 700 blocks every 24h of non-stop crunching (72-96 per user/day)). And if only *a few* additional of you MorphOS users who is reading this, but has not yet joined the effort, pull yourselves together and really joins, then we would reach over 5,000 daily blocks *very easily* (could be a lot more). And then, when MorphOS1.5 is released (at some undefined future date), this figure could very well double over night thanks to the Altivec support in the client (with Altivec, the crunching is almost three times as fast on the same CPU, but not everyone has G4's)!

So this is a "call to arms"! Come and join us!

Look here below:

Quote:

I agree!!

I know a lot of us knew that Mos1.5 release was not contingent on
whenever we beat the _m_g_ ppc team. For those like myself, it was
the pure competitive aspect of gaining ground and watching our
progress daily. When the dnet stats server went down a few months
ago, it took away our ability to see our progress, and in turn many
stopped our activity. Now the server is up and running, WE have no
excuse anymore... my goal is to do 70 units a day. If all 92 users
averaged 40 units a day we could start with a base near 3600 per
day!! Then those that wish to contribute more could easily put us
over 4000.

..also of note when Mos1.5 comes, a base # of 3600 blocks per day
could very easily become 10,000 blocks per day!!! And that would put
our daily output as the sixth fastest os. Think about it...


Katos1

Ok I will join today :)

For other interested MorphOS users, here is the client:
ftp://ftp.distributed.net/pub/dcti/prerelease/rc/dnetc491-morphos-ppc.lha

and here the install docs:
http://www.morphos-news.de/guides/rc5-72/index.html

happy crunching!

Hi Ladies and Gentlemen!

Just do not miss this one:

The new SECRET WEAPON!

If you think the Pegasos II is good now, just wait...

R&B
bbrv@bplan-gmbh.de

@ BBRV



BTW, how many computers is that?

ONE!

Well, I looked through the stats of all 92 people, and only 40 submited blocks on 6-12-2004.
So, if 40 = 3836 blocks, lets see what 92 can do!! (7000+)

@bbrv
>ONE!
What is it a dual G5??

nels

Grrr... slow down ... I cant keep up earning the money to buy all
stuff you put up

Papyrus,BurnItDVD,Divine Divi#?,Northland,Pagestream and now the
secret weapon

But hey ... go on ... guess my bank is happy when I'm in their debt

@ BBRV

No, you have got to be kidding me here! Surely, that can't be correct?

I mean, the 9th of June it peaked at 711 blocks - that is almost 10 times as fast as my current 1GHz G4 (without Altivec) performs! And if I am not too mistaken, that would also be about 3 times as fast as current 1GHz G4's **WITH** Altivec support?!

One computer you say? WOW, what kind of monster IS that?

/me sits with an open mouth and stares with amaze ...

Dual G4 @ 1.5 GHz? G5?

Dual AMD 64 ???

Actually it's several (3 to my knowlege) G4@1000 machines.

Not even dual 970FX can crunch that many keys/s (...Or actually it might, not quite sure. Considering the clockrate and two cpus, it might be close. On the other hand G5 altivec is not as good as G4 one).

[EDIT]Just did some math. 2 x G4 @ 1.5GHz would do 644 blocks/day. So indeed, dual 970FX machine should be able to do this much work alone.[/EDIT]

3021? Come on guys we can do better! :)
I leave my PegII on crunching 2day...

My G3/600 feels so poor against this BPlan monster. But anyway - I will keep on crunching but I don't leave the computer crunching on through the night or when I am at work (It's just a waste of electric energy - even with this low energy consumption the Peg has). But this BPlan thing is really kind of scary - I bet it's a dual 7447A @ 1.5 Ghz with Altivec on. Impressing! Seems to be time to consider to order a PegII (I guess my PegI will not like such a CPU card).

Are you crazy? =) The G3 uses 12 times less energy than your average 60w lightbulb ;)

Okay, energy consumption (even it eats more than your described 5W, there is a hd spinning, a gfx card to be powerd as well as the board itself, all in all it seems to be around 20W) is not the 1st reason, only the 2nd. The 1st reason is just the organization of my day. During night I power the Peg down (yes, it is silent, but there is still some noise which is avoidable when powering it of (could use a passive PSU, but..). When I wake up in the morning I am happy to find my path to the shower and mostly hurry up to work and am just not in the mood to power the computer on. This happens mostly around 6 pm when returning from work. Then the Peg is invited to crunch until I go to sleep.

Attention!

It is a cluster of three boards with one G4 each working as a single unit and running MorphOS. The surprise is that the computational power is both enhanced by AltiVec and that the three boards are working together with three times the ability of one board. There is no degradation in the performance when they are configured as a cluster. It could be four boards or forty (or alot more).

Keep an eye on those stats...

R&B

Quote:

A cluster on MorphOS?!?? Three computers working as one unit? Wow, I would have expected that from a Linux setup only! But how does it work?

Quote:

That sounds good!

I'm sorry if I get carried away with my dreams of dual CPU's once in a while ...

BTW, are there any plans of CPU cards with dual CPU's, in a not too distant future?

Quote:

<impressed mode on>

That rocks! A funky MOS cluster!

<impressed mode off - oh wait! not off - still on>

Very cool to hear that morphos has that capability. Reminds me of linux using openmosix.

Killer indeed

Is this clustering issue regarded to Quark? What I mean is, does Quark do the resource management and track 'only some cpu cycles' to the ABox and the ABox actually is not aware that it uses three processors..?

Quote:
That's three boards running MorphOS, running one instance of the distributed.net client on each board?

@ Zylesea

(OK, I did such a fine job when speculating about dual G4 CPU's earlier, based only on my own clueless fantasies, so why not continue hyping up some new speculations? ):

You mean, kind of like the way QNX does it, where applications can run on multi-CPU machines (even on CPU's on multiple machines ) without any need for hardcoding it into the application? Kind of like:

https://morph.zone/modules/newbb/viewtopic.php?topic_id=1741&forum=9#13127

If this is handled in a *even slightly similar* way as QNX does it, then I must upgrade my view on MorphOS from "a really cool OS" to "THE most incredible OS throughout all time"! It will then have everything I liked with the Amiga, PLUS some kind of memory protection capabilities for Quark applications (which might surface gradually over time), PLUS multi-CPU capabilities, PLUS distributed computing/cluster capabilities, and this goes even for the OLD Amiga programs running inside the A-box (since they don't have to know, or even *care*, about all this)!

I mean:

But then again, this would be TOOO good to be true, so it simply can't be ...

Well. I don't know of any totally detailed definition of "clustering",
so presumably you could just set up a hard drive with RC5, connect
three netbooting Peg2's to that, and call it a "cluster" as for this
particular purpose it would pretty much fit the general description of
a "cluster". Making a cluster for RC5 crunching is obviously quite
simple considering this is what the RC5 clients are designed for, so
you shouldn't need any special Q-box stuff for that.

Anyway I wouldn't really call it "cluster" without such kernel support
features.

Yes Grandma, that is what I was thinking of - like in QNX. At least I understood it in that way when I was working on the last German version of the 'Morphos in Detail' article.
Quark just schedules cpu resources, the task does not need to care about that.

And since those deep things are not allowed within the ABox (me thinks about the discussion with Laire about a Mac on MorphOS chances with ABox only) it's unlikely that it is handled within A, but it seems to be a Q feature...