LHC@home

> > Up, 217431 workunits to crunch
> -update-
> Up, 223386 workunits to crunch
> >
It's going down now... somewhat 200000 workunits to crunch....

we'll see :-)

---

Nice to have a full LHC queue, but everyone else having very short CPU time on these? Or is it just my box?

---

<img src="http://www.boincstats.com/stats/banner.php?cpid=162d855da85b4ee993029a39cfeed63a">

> Nice to have a full LHC queue, but everyone else having very short CPU time on
> these? Or is it just my box?
>

Prediction time is : 50 - 62 minutes/WU
Stats-time from 420 WU's average CPU-time 0.6 hours ~36 minutes.
(Stats. based on 'boinclogX' - datas)
(Boxes: 3.4 GHz HT and 1.5 GHz M with 2MB L2)

greetz littleBouncer
BTW: I have to crunch all those 'reserve'-WU's first ... ;-)

> but everyone else having very short CPU time on
> these? Or is it just my box?
>

Well, on a P4 3.4GHz with 1GB of Memory & HT enabled, I can do 2 100000 turn WU's in :30 to :45 minutes.

My AMD Athlon XP 2400+ (2GHz) with 768mb of Memory does 1 WU every 1:00 to 1:15

My Dual Pentium III (1GHz) with 256mb of Memory does 2 WU every 1:30 to 1:50

So, yes I too am noticing shorter processing times, but it really depends on the type of processor and memory you are using.

> > but everyone else having very short CPU time on
> > these? Or is it just my box?
> >
>
> Well, on a P4 3.4GHz with 1GB of Memory & HT enabled, I can do 2 100000
> turn WU's in :30 to :45 minutes.

My 244 Opteron (1.8 GHz) does them in about 38 min average... :-)

Some wu's finish in less than 4 seconds... but most do the 38 min average....

---

>
> Some wu's finish in less than 4 seconds... but most do the 38 min average....
>

Ah, these were the ones I was worring aboot, just wondered if anyone else was getting these off and on. So I guess its like a S@H WU, it aborts the running due to some inside problem with the WU? Or its an actual 5 minute say CPU time, somewhat concering.

---

<img src="http://www.boincstats.com/stats/banner.php?cpid=162d855da85b4ee993029a39cfeed63a">

>
> >
> > Some wu's finish in less than 4 seconds... but most do the 38 min
> average....
> >
>
> Ah, these were the ones I was worring aboot, just wondered if anyone else was
> getting these off and on. So I guess its like a S@H WU, it aborts the running
> due to some inside problem with the WU? Or its an actual 5 minute say CPU
> time, somewhat concering.

The point of LHC@Home is to test conditions. Usually the conditions are ok and the WU runs the simulation through to the end, which means that the stuff did not run into the walls.

If the simulation goes unstable, it is the same thing as the particles running into the wall ... which is one of the things we are testing for. Does the system fall apart under conditions as specified in the work unit.

What is still up in the air, as far as I know, is if the tests really do mimic the real world in the collider ...

---

> The point of LHC@Home is to test conditions. Usually the conditions are ok
> and the WU runs the simulation through to the end, which means that the stuff
> did not run into the walls.
>
> If the simulation goes unstable, it is the same thing as the particles running
> into the wall ... which is one of the things we are testing for. Does the
> system fall apart under conditions as specified in the work unit.
>
> What is still up in the air, as far as I know, is if the tests really do mimic
> the real world in the collider ...
>
Thanks for the info. We'll have to wait for the collider to start working to find out whether LHC@Home = Real world :-) Fine for me.

The Number of available WU's is going down slowly.... This kind of reflects the daily throughput with the current user and host base... :-) Interesting.

cheers,

sysfried

---