LHC@home

>In fact I got 115
are you counting the zip files ore how do you (let) count, to get numbers?
even with 5 heavy workers, you have seen a huge amount of work units passing while the day is long.
==========

I'm taking the count right from the BOINC GUI, thats how many WU's it shows under the Work Tab ... Or sometimes from the BOINC LHC Projects Directory Folder it's self ...

It does seem like a lot of work to finish on time but the P4 HT CPU's will do a lot of work in one day, each one is capable of doing 4-5 of the normal running Tunescan WU's in one day and between 65-72 of the Regular running WU's in one day...I took a look at 1 of my PC's Messages and it showed 19 Regular WU's Uploaded between 12:00AM and 6:00 AM this morning, Projected over 24 Hr's thats 76 per day for the Regular ones ... :/ ... I need lots of WU's to keep them busy ... :)

Then of course there are so many of the WU's that take only a few seconds to run...I may do 1 WU in 40-50 minutes and the next 3 will be done in less than 5 minutes...

This is actually the first time I've been able to keep WU's in my Cache's for my faster PC for more than a couple of days. Usually I download the 100 WU limit and I would be out of work in less than 1 day on them but with the TuneScan WU's being released I've been able to stock up a little so I can keep them running ... :)

having about similar times, about 2 per hour. 53 57 54 Min.




now with the tunescanxy, the number is "relatively" low, but still able to bring them down in short delay

Yes, but you have to realize that the P4 with HT is doing 2 at a time, between 40 & 50 Minutes to do 2 normal WU's depending on the PC ...

> Yes, but you have to realize that the P4 with HT is doing 2 at a time, between
> 40 & 50 Minutes to do 2 normal WU's depending on the PC ...
>
Oh yes I'm aware of this. :-)

Perhaps wrong, I think it's a disadvantage for the Boincs design of today.

Let's take a HT having a benchmark of 1500 and 2000 per "cpu", just to put som e numbers. (total 3000 and 4000)

The CPU, non HT mode, lets say 3000 and 4000 to.

So, now the clients is asking for work.
The clients "capability" is calculated from the benchmarks result.
This "reflects" the how many work (in hour).

Due the benchmark is per cpu and not per client (running) we get have a less total amount than a non HT cpu, having the same theoretic benchmark.

I believe, for HT (dual core /CPUs) based clients, even when the limit is set to 1000 per host, on a one time download, the effectively clients capacity is never exhausted.
As mentioned, perhaps wrong.

BTW, don't do the same mistake as myself, building a fast node with the thinking behind, to remove the slower ones. In reality you will say, like me, take all you can get. Even the CPUs in the remote control & play stations, if there would be a compiled client for.

Well my 1 HT PC Benchmarks at 1600 with HT Enabled and 2100 with it Disabled, so the score is neither cut in Half with HT Enabled nor Doubled with it Disabled...The other ones are similar with their Benchmarking...

I Benchmark them with HT Enabled and thats the way I run them also...I know some of the guys are Benchmarking them with HT Disabled and then turning HT back on again, this can lead to a big discrepancy in the Credits asked for.

I was checking out a Result the other day when the Results where still accessible and ran across one guy Claiming 175 Credits for the same WU I was only claiming 75 Credits for (It was a TuneScan WU) we both had the same CPU and our times were within a couple of minutes of each other also.

So I know he Benchmarked with HT Disabled and then turned it back on again to Claim that much Credit. He will never get it though because it's to way out of line with the other results turned in and in fact I believe this leads to delays in Credit being Granted because of the big difference in Claimed Credit.

I used to do that and then got to thinking about it and came to the conclusion it's almost a form of cheating to do it that way so I stopped doing it. If your going to run your computer in HT Mode then I think you should Benchmark it that way and claim the correct credits. If you run with HT Disabled then Benchmark it that way also...

When I asssemble a new client, running a HT able CPU, the first thing verrified/changed is the memory timing and if HT is on. I only running with HT on and if possible(if work), 2 Workunits same time. so havent the with without HT support numbers

>1600 with HT Enabled and 2100 with

Intel says, about 25% gain versus a non HT based "same" cpu.

1600x2=3200 versus 2100, but we cant run more than 50% of a virtual cpu.

Also from the result "on" time, I remember, for one WU, the claim was 340 (!)


3 of my clients I had to let run slower, for seti ok, not ok for LHC,
on� is an AMD 2800, running at 183 MHz internal speed, had to reduce to 177 MHz, now running fine and stable.

The Prescott 3.2 (running at 3.528 is stable and running fast like a moth through a ramshackle carpet

I would like to talk more, but now I' out for a while in town, hoping no crashing clients.

greetings

ric

@PoorBoy

It looks like you made some scientific work for LHC...

7 PoorBoy 1013.29 21839.24

me too, watching bit the stats.

From about 72x to over 1000

The credit system is running fine and quick, at least for the propper returning
WUs.

good job! ;-)

Did the new parts arrived?

Grandma always spoked: It's not only a question of "fast" CPUs, it more a question to keep them busy..

No I haven't received the Parts yet, if fact I don't even know whats going on with them. I haven't received a Tracking # from either place I ordered from yet & if I don't receive one by today I guess I'll have to give them a call tomorrow to see whats going on with the orders...

Yes, I've been moving up the rankings slowly but steadily but a lot of the guys are not using all their fire power at this site so it makes it a little easier because of that...I know if some of the guys I passed were to devote all their CPU Power to this Site they would blow me away in no time at all... ;)

I don't have any problem keeping my PC's busy right now, if fact I kicked the Server in the Butt this Morning an got it to give me some more WU's, but I may have over done it and got to many...Hopefully enough of them are the short running ones & then I will be alright, if not there may be just a few I won't be able to get done in time...

I notice even with the TuneScana WU's some of them are just real short running, I may do 4 or 5 in the normal run times & then 3 or 4 or them will only take a few minutes, so I should be alright in getting them all done in time if that keeps happening with them... :)

@ ric

I had to re-order the CPU's, The place I ordered them from finally notified me yesterday that they were out of stock & that they didn't accept Back orders ... I'm kinda pissed at them for taking 3 days just to tell me that, they could have done that the day after or even when I ordered them I would think ... :/

Now I won't get the CPU's until next week sometime ... :(

WoW, I see you got to the #1 spot for Recent Credit ... Good Job ... :)

@PoorBoy


Argh understanding, thats really not nice, to not deliver a waiting for
CPU. UR mentioning CPUs in Plural. Is it a dual or quad system or several "nodes" comming to you? It looks like a cold winter is comming..

>Now I won't get the CPU's until next week sometime ...
Just send me your accountID and I will turn a client into your account till
your hardwarmer arrives.

Thanks for the "good job", same to U back ( and to every other contributor)
But if I'm honnest, the tanks has to go to Mr Prescott, to Mr. Northwood, Mme Willamate and the AMD family

It's not a secret, I do IT as Profession, to silly for something else.

I had the chance to earn a couple of grafic cards and older Harddisk.
The most part of my clients are only having Mainbord, ram, cpu, CPU COOLER, small HDD and a powersupply. Nothing more. So a client can be kept relatively cheap.

>WoW,..
And You? increasing your RAC in 2 days over 30%, thats great..

You know, "normaly" I'm sitting on several projects, including seti.
For a while I turning the small farm (yes small) to LHC and CPDN only (and pirates! of course)
With the 3 already attached clients, LHC had 20-25% or less work settings, was around place 420-490/533, with about 23xy credits.

I have seen the large numbers of other contributors, giving myself a pale face and let me sit down. BTW the same at SETI, there are boys using a "full enterprise network" , so we "normal" only home user would have "no chance".

Since I plugged all nodes to LHC, it took "only" 13 days to get to the "target" number. Have seen a couple of WU coming and going
Also have to say, about 2,8-3 Gigabytes of data downloaded in this time.. (amazing amount isn't it?

->For the total, no hope. And seti is calling also..

But indeed, me to, still suprised..
(for the team it's good ;-)

Do you know about your Work, on the number level, what you are doing?

If have the time, take a look to
http://www.setisynergy.com/stats/individual_lhc.php

Individual LHC Stats

tks 4 posting
wishing a soon delivery

and "full Queus"

It looks like there is enough work for everybody:)

Server Status

Up, 78979 workunits to crunch

friendly
ric

@PoorBoy
Argh understanding, thats really not nice, to not deliver a waiting for
CPU. UR mentioning CPUs in Plural. Is it a dual or quad system or several "nodes" comming to you? It looks like a cold winter is comming..
==========

It's 2 separate Pc's that I have a P4 3.4 coming for, to upgrade them...I got the Motherboards & Ram & aftermarket coolers for the CPU's today so now I just have to wait for the CPU's from another place I ordered them from. Then I have to find the time to do the Upgrades but if all goes well they should be upgraded by the end of next week I hope... :)

">Now I won't get the CPU's until next week sometime ...
Just send me your accountID and I will turn a client into your account till
your hardwarmer arrives."

hahaha...not to worry, I switched another P4 3.4 PC over to the LHC Site Last night so while I upgrade the 2 slower PC's to P4 3.4's I wouldn't slip any in the Rankings ... ;)

> hahaha...not to worry, I switched another P4 3.4 PC over to the LHC Site Last
> night so while I upgrade the 2 slower PC's to P4 3.4's I wouldn't slip any in
> the Rankings ... ;)
Welcome, there is place for all.

They is a bit of work to bring down,

As more work is done as better for the project people. everybody it's toy..
But even the 3.4 hot dog of today will be a "slow" one tomorrow.
I sure, at least one of them is an EE you got:)

So I'm to late to wish happy assempling an installing, alway a sight for sore eyes.

I hate to be right, But I guess, without balancing, slowly you will run into power supply problems...,the Prescots and the Northworth are power demanding.
And for the heap!! It's like you will have a running Ironstation beside you.
All the time.
But enjoy the trip.

The only person should have to worry, is the admin stuff.

Now there are sitting back, yes they and the SERVER(s) have done a good job.

Now. It will take time, only a amount of time to bring the work queued "down".

Your set of 3.4 GHz Monsters will add to nail the work down soon.
(*evil grins*)


Server Status
Up, 157430 (at 18:39 22.10.2004 utc-2)

First guess, you will pass 1800 of RAC, after 3 days all node "on".

Further goodluck!!

ric

But even the 3.4 hot dog of today will be a "slow" one tomorrow.
I sure, at least one of them is an EE you got:)
==========

Your right ric, within 6 months to a year the P4 3.4 will just be a ho hum CPU but they will still be quite the crunching machine...IMO

I have 1 P4 3.4 EE and 1 Regular P4 3.4 right now. The only reason the P4 3.4 EE is faster right now is because I have it overclocked where as I'm just running the Regular P4 3.4 at the stock speed. If I set the Overclocked EE Chip back to it's original CPU Speed it's no faster than the Regular 3.4 ...

For the added $700 - $800 for the EE Chip it's not worth it for just crunching BOINC WU's, it may come in handy for other things like playing Games or something but not for BOINC ... IMO

Thats why I just ordered the Regular P4 3.4 NorthWood CPU's for the 2 upgrades...They should be here Monday as I got the shipping notice on them yesterday ... :)

One of the todays last adventures must be called boinc (-LHC).
We passed the DL and other "instabillities".
Now it looks like the server or parts of him is are a bit in use.
Even posting CAN be (now) an adventure.

The setiates forces in LHC takes over :(

Last week, at friends, we had a good discussion, the conclusion was,
the highest risk for the "computer" is the user him/herself.
Some thinking about how this could be changed to.

One Idea was to install a gun system for selfprotecting the "computer"
(including server, Workstations and *home* systems")

might this idea will be included in further server hardware updates


At least I try to put this into the forum.

-----------------------------------------------------------

Good new! the CPUs are on the way to you.:)

EE versus E

This is advanced funny. no gain in speed with the 2 MB internal (3rd level?) cache?
Was thinking at least 5-10% must be seen.
You are right, this money is bad invested.
for the money an additonal host, maybe more, could be set up.

good to know!

The last 3GHz wanted to add, no more Northwood available here, only the expensive EE
so I took the prescott 2.8 MHz on a regulary intel board, no oclock:(, but build in VGA and stable.
(one day will change the MB, to more explore the possibilities of the prescotts..)

So it's not to late to wish fun and goodluck with assembling the parts and ADD the host to
you favorit project.


As more host U are running as more you will be see a "bad thing".
Your pending credits!!!
Don't be surprised abount the amount you will more generate as today..

myself have now (07:18 24.10.2004)

Pending credit: 7147.76

(basically I could power off the clients and watch the stats growing..)
When I (or other users ofcourse) get more than 10'000 pending, we should discuss this

some other "data" of *my* account
Pending credit: 6746.41 10:15 23.10.2004
Pending credit: 6722.89 10:43 23.10.2004
Pending credit: 6763.22 11:12 23.10.2004
Pending credit: 6872.05 11:22 23.10.2004
Pending credit: 6773.10 12:00 23.10.2004
Pending credit: 6351.66 15:53 23.10.2004
Pending credit: 6482.75 17:06 23.10.2004
Pending credit: 6906.27 03:38 24.10.2004
Pending credit: 6994.02 05:36 24.10.2004
Pending credit: 7009.33 05:43 24.10.2004

Oh yet, the credits ARE spoken, but it looks like, the flow is not smouth
to avoid mis(s)understandings, I do NEVER complain, just put some Datas to share, because
I'm lust curious what's (not) going on there

keep on!

ric

________________________________________________

here your stats, we will see in a week, when your "upgrades" brings back the fruits
and you are sailing the stats...




Thumbnail, click to open

EE versus E

This is advanced funny. no gain in speed with the 2 MB internal (3rd level?) cache?
Was thinking at least 5-10% must be seen.
You are right, this money is bad invested.
for the money an additional host, maybe more, could be set up.
==========

The EE Chip is faster most definitely, the Benchmarks show that. And testing at different sites on various other applications show it to be about 10% faster also. But for running BOUNC WU I don't see where its any faster other than the Benchmark Scores...

re
BTW, don't do the same mistake as myself, building a fast node with the thinking behind, to remove the slower ones. In reality you will say, like me, take all you can get. Even the CPUs in the remote control & play stations, if there would be a compiled client for.

if you modify an xbox its posible to run linux on it, had suse8 on one but found the 640*480 screen ristrictive for xwindows
there cheep enough to cluster having a ethernet built in,and should have little problem rining boinc on them

heard that the same was possible on the ps2 but you would have to add a hd to it

:)
Dave

> if you modify an xbox its posible to run linux on it, had suse8 on one but
> found the 640*480 screen ristrictive for xwindows
> there cheep enough to cluster having a ethernet built in,and should have
> little problem rining boinc on them

We had XBox Linux running LHC@Home at Grid Cafe on CERN Open Day, in an X environment with resolution something like 1200x900. But I must admit that I don't know details on how this was done.

Markku Degerholm
LHC@home Admin