next »

[H-kon]

Due to heat issues because of record heat, i have been forced to take a few computers offline for the time being. Seems like the two work machines have been taking offline too as of July 18th so our output is just 2K a day now.

I will leave my E8400 and my fileserver online when on vacation and hope to recoup the loss during autumn if i land a new job

[themystic]

Will probably be given a blade rack with 16 servers soon. Should be fun to test out what they can do. dual 1GHz P3s.

[H-kon]

Put those to work asap Someone turned off my machines while i am away so no work units have been sent in since i went on holiday.
Will put them back up as soon as i get home

[H-kon]

Everything is running as normal again. My Recent average has plummeted though so the forecast is way off. Going to take a few weeks to fix that.

Be sure to download the optimized client on page 1 of this thread. It halves the time you crunch the units.  My E8400 finishes a Work Unit in little over 30 minutes now.

Thanks Teal for helping our efforts. I see you are close to 60K now. Mystic. When are you putting that 16 Pentium III's to work?

[H-kon]

Nice. In the race for 1.000.000 credits, we edged out the team that was in 10th place by a couple of thousand.  Well actually, i dumped 12 thousand earlier today from my E8400 He was producing around 2000 a day. We are up to 4500 and climbing a bit more on the average.

Current score according to BOINC is 1.009.000 credits. We're also in the top 10 list. Congrats people!

The stats are a bit behind, but the BOINC manager is up to date.

#9 position is at 1,500.000 - So we need some big guns if anyone has them.

[H-kon]

1100.000 credits passed today. One E2200 has been sold so i only got one E2200 and one E8400 on it now.

Hope to break 1,5 mill before christmas though

[H-kon]

Some benchmarks:

AMD Athlon XP 1800+, 1.53GHz, 768MB RAM

1423 floating point MIPS (Whetstone) per CPU
2360 integer MIPS (Dhrystone) per CPU


Intel Pentium 4, 2.4GHz,768MB RAM

966 floating point MIPS (Whetstone) per CPU
1836 integer MIPS (Dhrystone) per CPU


Intel Celeron, 1GHz, 768MB RAM

930 floating point MIPS (Whetstone) per CPU
1573 integer MIPS (Dhrystone) per CPU


Intel Pentium 3 (katmai), 500MHz, 320MB RAM

456 floating point MIPS (Whetstone) per CPU
749 integer MIPS (Dhrystone) per CPU


Intel Core2Duo, 2.66GHz clocked at 3.2GHz, 3.5GB RAM

3106 floating point MIPS (Whetstone) per CPU
6806 integer MIPS (Dhrystone) per CPU

Just added a Celeron 2,66/400fsb to the mix. Running BOINC to check the stabillity of it. Its running Mandriva One Linux
Interesting to see that it performs on par with the Celeron 1Ghz you had there Mystic

858 Floating Point MIPS
2487 Integer MIPS

BOINC is heavilly floating point dependant, so it means that your Celeron is actually faster.
Now checking Pentium 4 2.0 Ghz in the next post.

[H-kon]

Pentium 4 2.0Ghz 400/512k

666 Floating Point MIPS
1651 Integer MIPS

Weird. I really thought it would be faster, not slower.

Could be Linux, but i am not sure.

Edit: Comparing it to my E8400 3.0Ghz running at 3,6Ghz:

3749 Floating Point MIPS
7925 Integer MIPS

Meaning that it is just not worth the electricity to keep the Socket 478 on at all.

Hmm. going to test with a Pentium 4 2,8Ghz 800 fsb that will only run at half speed.

[H-kon]

Never got the P4 to work at more than 1,8Ghz. Linux wont give me any reliable info either.

My main E8400 took a dump today and corrupted my entire drive which i attributed to one stick of memory. Had to reinstall and did so with XP x64 bit version since i now have 6GB of memory installed.

Installed the SSE4,1 version of the optimized client. Going to be interesting to see if it cuts some additional time off every work unit. I used the SSE3 client before.

Which you can find here:

http://calbe.dw70.de/

[H-kon]

It shaves of a few minutes it seems.

Down to 36 minutes pr work unit now on each core.

[H-kon]

I am really talking to myself here, but going to Windows XP x64 increased the performance quite a lot.
I haven't got a reliable time yet but my cpu benchmark is now:

Floating Point  3625
Integer: 10594
This is per core.

Meaning the E8400 is about 6 times faster than a Pentium 4 2,0 Ghz.

[Elf]

It seems like you are a one man team.

It's interesting that Windows XP 64bit would increase the performance by a lot. Don't know the difference is between the two XP's.

Sorry, I have been tied up with folding lately.

I do have one computer AMD 1800+ that I only fold part time with. The points that I get with it isn't worth extra power that I pay for having it fold. I wonder how boinic would go?

[H-kon]

The floating point performance is very good for the 1800+, it was the strong point compared to the Pentium IV at it's time.

Chris posted a result from his machine back a while:

AMD Athlon XP 1800+, 1.53GHz, 768MB RAM

1423 floating point MIPS (Whetstone) per CPU
2360 integer MIPS (Dhrystone) per CPU

Thats actually pretty good. I think it will do even better with paired PC3200. Chris had PC2700.

Pentium 4 2.0Ghz 400/512k

666 Floating Point MIPS
1651 Integer MIPS
 
My E2180 2.0 Ghz at 2,7 Ghz: 
1583 Floating Point
7552 integer  (Integer good because of 64 bit Ubuntu 8.10 beta 3)

(One WU in 1:20 hours)

You might be able to run the SSE2 optimized client as well which will speed things up. The install is just dropping 3 files into a folder and BOINC takes care of the rest.

You should be able to overtake Doc in not so long with that machine, and he isn't running an optimized client. Check with cpu-z to see what it supports, and i'll go and find the correct file if you're interested.

[Elf]

The 1800+ is a Palimono Core and only has 3DNow and SSE instruction sets.
It is also on a old motherboard that uses PC133 ram.

I will let it finish the current WU for folding @ home which should be in a few hours.

Chris posted a result from his machine back a while:

AMD Athlon XP 1800+, 1.53GHz, 768MB RAM

1423 floating point MIPS (Whetstone) per CPU
2360 integer MIPS (Dhrystone) per CPU

Thats actually pretty good. I think it will do even better with paired PC3200. Chris had PC2700.

It is the same as Chris's machine but with 768MB PC133 ram.

You might be able to run the SSE2 optimized client as well which will speed things up. The install is just dropping 3 files into a folder and BOINC takes care of the rest.

You should be able to overtake Doc in not so long with that machine, and he isn't running an optimized client. Check with cpu-z to see what it supports, and i'll go and find the correct file if you're interested.

Interesting......
CPU-Z 1.48 said it has support for MMX and 3D Now.
CPU World said it has support for MMX, 3D Now and SSE.
Everest said it has support for MMX, 3D Now and SSE.
AMD's Data Sheet said it has support for MMX, 3D Now and SSE.

So I guess cpu-z missed out missed the sse instruction set.

[H-kon]

I don't think it will hurt it much that it runs SDRAM, but do post the benchmarks from BOINC.

Here is a direct link to the SSE client:

http://calbe.dw70.de/2.4V/2.4V_Windows_x32_SSE.zip

Instructions: Start - Run - "net stop boinc" (if it is running)

Then unzip the files to a folder somewhere, and take the files inside "files to install" and copy them to (Win XP)

C:\Documents and Settings\All Users\Application Data\BOINC\projects\setiathome.berkeley.edu

Start - Run - "Net start boinc".

Thats it. BOINC will use the optimized files instead, you will see the platform as "anonymous" in the message tab and you will see that "2.4_Windows_x32" is running in the task manager, it takes some seconds before it shows up the first time.

Viewing the work units, you'll also see that it tries to figure out how much time it will use to complete a unit. Running an optimized client will take some time before it gets it right as seconds are shaved off after each tick.  The optimized client doesn't affect the benchmarks either.

[H-kon]

And then i passed one million cobblestones

On to the next!

[H-kon]

I am reconfiguring a couple of my machines these days, which is why the output is so low these days.
I also got one computer helping Ars Technica taking the second place in ABC@home to the end of November which will go to our BOINC totals. (16th in Norway)

[themystic]

Thrown in some resources in the mix. It's not much, but at least I'm back

My dreambuild would be a quad with water cooling.. The main issue I have today is how much noise my systems produce when BOINC'ing..

[H-kon]

 i am helping another team taking 1st in ABC@home for a few weeks before i put my little DC farm back on H-kon.net which now consists of:

E 8400@ 3,6 Ghz
E 5200@ 3,35 Ghz
E 2180@ 2,66 Ghz
E 6400@ 2,13 Ghz (not full time)

With a little push from more people, we're gonna climb pretty fast in the total team rankings. (1885 total at the moment)
 

[themystic]

I'm currently contributing with about 3500 points a day. At the moment I have no more computers to put on the project..

[H-kon]

Thats very nice numbers:) Thank you for the push. When my machines comes back online, we'll do 12-15K a day.

Are you using the optimized client?

A quad is being purchased pretty soon as well.-

[Elf]

I am having trouble running the optimized client. I downloaded from the link you posted and followed the instructions.

I finished a wu using the non optimized client. Installed the optimized client and it wont run.

[H-kon]

Check the task manager. It should be listed there.

If you didn't exit boinc when dropping the files into the setiathome folder,

Start - Run - "net stop boinc"

Start - run - "net start boinc"

without the quotes.

[themystic]

Are you using the optimized client?

Of course On all three machines. My compy at work is putting out about 200 points a day.. So it's barely worth the effort, but the two others are doing alright

[H-kon]

I will try to run seti off anything that i can get my hands on at work,

[Teal]

My main machine went down about a month ago and is just now coming back to life.
How are you guys getting such high points every day? I think my avrage is only about 80!!

[H-kon]

My main machine went down about a month ago and is just now coming back to life.
How are you guys getting such high points every day? I think my avrage is only about 80!!

The optimized clients will do that The Core2 Duo's are real screamers when it comes to Seti@home as well, and most other projects.

I'll bring my 20+ Ghz back online in the coming week

[H-kon]

Just a day or so of ABC@home work units left before they all start on seti@home again. I am converting one machine to Ubuntu 64-bit, seems to help a lot in production.

Thank you, Thank You, Thank you all for the help. We'll be the second biggest producer in Norway when my machines comes back online.

[H-kon]

The ramp has started. All abc@home units are gone now (got 220K there in just three weeks, second best in the project with 4 hosts).

tomorrow should be an interesting day

[Elf]

You are about to overtake number 9 in the Norway stats.

Still can't get the optimized client working. I followed the instructions bellow. There are only 3 files in the archive and I copied all of them to the "setiathome.berkeley.edu" folder. When I restart the client I get a error message box with the "send" or "dont send' buttons. I can't remember the exact wording. I will post that when I have some free time.

At the moment I am running a standard client. I may post some benchmarks to see if it is close to Chris's 1800+ computer.

I don't think it will hurt it much that it runs SDRAM, but do post the benchmarks from BOINC.

Here is a direct link to the SSE client:

http://calbe.dw70.de/2.4V/2.4V_Windows_x32_SSE.zip

Instructions: Start - Run - "net stop boinc" (if it is running)

Then unzip the files to a folder somewhere, and take the files inside "files to install" and copy them to (Win XP)

C:\Documents and Settings\All Users\Application Data\BOINC\projects\setiathome.berkeley.edu

Start - Run - "Net start boinc".

Thats it. BOINC will use the optimized files instead, you will see the platform as "anonymous" in the message tab and you will see that "2.4_Windows_x32" is running in the task manager, it takes some seconds before it shows up the first time.

Viewing the work units, you'll also see that it tries to figure out how much time it will use to complete a unit. Running an optimized client will take some time before it gets it right as seconds are shaved off after each tick.  The optimized client doesn't affect the benchmarks either.