Optimization drive

[DDTUNG]

OK I still see some XS D2OL members using non-optimized clients which cut their credit by up to 50%.

Go to the BOINC Rosetta stats page for XS, check on the list of computers used by someone on XS that you know well, and if you find any of the benchmarks are off, PM him/her and remind him/her to use the appropriate optimized clients.

That alone may nudge the team past DPC. We were only 4.5k down in today's production, or 4%.

DDTUNG:cool:

[WesM63]

Agreed. Everyone needs to run the optimized clinets!

[[XC] DragonOrta]

I'm running fully optimized

[DDTUNG]

I'm running fully optimized

Make sure all your friends are.:)

DDTUNG:cool:

[shmaa]

Make sure all your friends are.:)

DDTUNG:cool:

My friends are... but hypothetically someone like Rodzilla, who of course has no friends, should....?











Just kiddin Rod... :slapass:

We can't help people who have elected to hide their computers (for whatever reason :confused: )

[STEvil]

I think I am, but I could be wrong. I'll be checking all my machines tomarrow and fixing up the two that are having a hard time putting out (both A64's @ 2.6+).


Also, this is annoying! http://www.xtremesystems.org/forums/attachment.php?attachmentid=42278&stc=1&d=1137213619 :slap:

[WesM63]

stevil,
I got that too. After I let it update it changed R@H stats page.. they give u alot more info now.

[WindForce]

OK I still see some XS D2OL members using non-optimized clients which cut their credit by up to 50%.

Go to the BOINC Rosetta stats page for XS, check on the list of computers used by someone on XS that you know well, and if you find any of the benchmarks are off, PM him/her and remind him/her to use the appropriate optimized clients.

That alone may nudge the team past DPC. We were only 4.5k down in today's production, or 4%.

DDTUNG:cool:

Victor
In what stats page do you see whether the benchmarks are on/off for a machine?
Jaime

[DDTUNG]

Victor
In what stats page do you see whether the benchmarks are on/off for a machine?
Jaime

http://boinc.bakerlab.org/rosetta/

Click 'Your Account', then 'View Computers', then 'Computer ID' to check the benchmark scores for each individual computer under your account.

The thread titled Rosetta Benchmarks should give you an idea of whether your client is optimized.

DDTUNG:cool:

[DDTUNG]

A quick look through the list of DC rigs for XS_The_Machine shows one 4400+ and one 3800+ not running the optimized client. That's over 1200 points lost per day if run 24/7. I haven't checked the other rigs on that subteam.

Guys, let's get our act together.:slapass:

DDTUNG:cool:

PS Haltech and Mr_slinky also have non-optimized DCs.

[P0G0]

Don't hurt me but I won't have access to my machines untill atleast Wednesday to switch to optimized clients... But I'll make it a point to hit up all my machines next week (on my "weekend")

[DDTUNG]

Don't hurt me but I won't have access to my machines untill atleast Wednesday to switch to optimized clients... But I'll make it a point to hit up all my machines next week (on my "weekend")

Nah nobody is going to hurt you. Rod will just kill a couple of kittens.:rolleyes:

DDTUNG:cool:

[L0$t Pr0PhEt]

At the moment we have 8k more than DPC for this day :)

[nhenhophach]

At the moment we have 8k more than DPC for this day :)
And tomorrow? 8k*2 and.....:D

[L0$t Pr0PhEt]

And tomorrow? 8k*2 and.....:D

nhenhophach, I saw that all your computers are running without optimized client, please install the optimized clients and you will double your output!

[DDTUNG]

nhenhophach, I saw that all your computers are running without optimized client, please install the optimized clients and you will double your output!


Link to optimized clients:

http://www.guntec.de/Crunch3r/boincx86.html

DDTUNG:cool:

[DDTUNG]

At the moment we have 8k more than DPC for this day :)

DPC must be 'buffering'.:D:rolleyes:

[Martijn]

I still need to install the optimized client on my P3, but that won't make much of a difference, I guess :p:

[ex2cib]

also, check your benchmark scores occasionally, it might end up being lower if something else was running in the background during the time another program was taking CPU cycles

[ex2cib]

im looking through all of our computers as i type,

Dak, you have a computer thats not optimized 4400+
bench #'s
Measured floating point speed 2403.83 million ops/sec
Measured integer speed 4503.89 million ops/sec


ID #126753 is not optimized, athlon 64 3200+
bench #'s
Measured floating point speed 1855.07 million ops/sec
Measured integer speed 3433.51 million ops/sec

dak, another 2.8 pentium4 you have i would check, just looks low to me
bench #'s
Measured floating point speed 2130.03 million ops/sec
Measured integer speed 2365.43 million ops/sec

this could take a while, all guys with pentium's, just check to make sure you have the optimized clients.

heh, mikee, your wife's PC is faster than yours ;)

had to put this one their, for a 3.4p4, this looks low
# 135711
Bench #s
Measured floating point speed 1516.87 million ops/sec
Measured integer speed 1408.79 million ops/sec

bullet when you get yours stable, check to make sure you have optimized client, your scores are very low it seems

basically it seems the optimized bench scores of most athlon 64/754semprons should be in the 8000/9000+ range, im still not sure of pentiums, someone with a client they know is optimized might could tell us all of that

[L0$t Pr0PhEt]

I think a P4C @ 3GHz should get about 4500-5000 (whetstone+drystone).

[bullet2urbrain]

Yeah im responsible for the 3800 x2 titled Bullet2urbrain...but TCCD + mem controller is giving me trouble.. i will work with it Sunday or Monday whenever i get back to school (Again!)

also real quick. what should an x2 benchmark look like.. i thought ~10000 was all right for the 2nd number.

[Movieman]

I still need to install the optimized client on my P3, but that won't make much of a difference, I guess :p:
I installed the optimised client on my OLD dual P3 xeon and it showed a huge difference. It only does app 250-300 a day but I'll take whatever it gives!:D

[ex2cib]

10000 is good for teh second number

[STEvil]

fixed my dual933, it was unoptimized before.. woops ;)

[P0G0]

This may sound stupid but that page has either p4/Athlon64 or P3/XP... which should I use for a celeron?

[STEvil]

is your celeron P2, P3, P3-M, PT-M, P-M, P4, or P4 based?

[[XC] DragonOrta]

Do you have have CPUZ? if you do run it. If it shows that you have SSE2, get the P4/A64 one, if you only have SSE, get the P3/AXP one.

[Smurfy]

Are Xeon Machines rather bad in this project?

My Xeon machine with running @ 3.2Ghz seem to do around 600 a day, if the X2's are putting out a 1000 a day i may have to consider changing my 2 comps over to X2's or some 165's

[DDTUNG]

Are Xeon Machines rather bad in this project?

My Xeon machine with running @ 3.2Ghz seem to do around 600 a day, if the X2's are putting out a 1000 a day i may have to consider changing my 2 comps over to X2's or some 165's

A dedicated DX @3.2 should give you around 800 points a day, with the optimized client 5.2.13.

DDTUNG:cool:

[Smurfy]

maybe im slightly under estimating there daily output, seems to be around 700 most days, and thats with me watching tv and downloading on it.

[STEvil]

thats about right.

[Fr3ak]

Is a 1.5Gb drive enough to install win2k and rosetta on it ?

[Martijn]

Yes, but there won't be much left...

[Fr3ak]

I don't need anything left tbh. I can get a usb2.0 1.5gb microdrive for 23€ shipped. Its 0db, slow as banana, so I guess it will be perfect for a cruncher as long as 1.5gb is enough to run rosetta...

[thienloi]

I'm running fully optimized :cool:

[Martijn]

I don't need anything left tbh. I can get a usb2.0 1.5gb microdrive for 23€ shipped. Its 0db, slow as banana, so I guess it will be perfect for a cruncher as long as 1.5gb is enough to run rosetta...

It should work, yes, but you lose a lot of time when loading your wu's :p:

[Paladin]

Took a quick peek. MrSlinky's fleet of DC Opterons look to all be using the stock/slow client. Rod's 2.0 Xeon seems low unless it's only running @2.0.

[Movieman]

Are Xeon Machines rather bad in this project?

My Xeon machine with running @ 3.2Ghz seem to do around 600 a day, if the X2's are putting out a 1000 a day i may have to consider changing my 2 comps over to X2's or some 165's
I'm getting app 1000 a day out of my DX machine. Will test and give you an exact number.

[bullet2urbrain]

just went on an optimizing binge..

optimized 3 P4's.. and checked on my last remaining d2ol machine... 800 buffered :D

my P4's @ 1.6 are now doing 1350/2390 with the improved client.

:woot:

[NickS]

optimised on my PIII rig and this rig :)

[tekman]

my P4 630 (3.0GHz) have R@H benchmark scores:
1959 Whetstone
2886 Dhrystone
Is it OK ? (although I have installed optimized client).

[bullet2urbrain]

seems low for the second number.. but im not sure depending on your RAM situation

my x2 does ~103xx
while my P4 1.6 willies do 2300 DhryStone

[L0$t Pr0PhEt]

my P4 630 (3.0GHz) have R@H benchmark scores:
1959 Whetstone
2886 Dhrystone
Is it OK ? (although I have installed optimized client).

My P4B 2.66GHz does about 2000/4000, so yes yours seems low.

[thienloi]

My P4B 2.66GHz does about 2000/4000, so yes yours seems low.

Is it default?

[L0$t Pr0PhEt]

Is it default?

Yes, I still noticed that only half of VNM's computers are running optimized.
You guys have a 2.8ghz celeron that's almost doing 2500/4500 or so and then a P4C 3Ghz that's doing 1500/3000. Check it out

[SoF]

Here's my comparison:

http://www.xtremesystems.org/forums/attachment.php?attachmentid=42461&d=1137523833

That's it :)

[STEvil]

Yes, I still noticed that only half of VNM's computers are running optimized.
You guys have a 2.8ghz celeron that's almost doing 2500/4500 or so and then a P4C 3Ghz that's doing 1500/3000. Check it out

Did the P4 have hyperthreading? HT results will be approximately 1/2 of "pure" results.

[ex2cib]

some of the ones' dak is running arn't optimized i dont think,
@ least they dont look optimized

[[XC]thewildblue]

Seems that although I had put the right files in on my clients it hadnt picked up correctly (prolly me doing something wrong) but after a bit of fiddling I think Im there now.

[Jeff]

DPC must be 'buffering'.:D:rolleyes:

:lol: Yeah, I'm sure they are. :toast:

(note, they could also cheat in this project if they wanted too... being open source and not having any checks or balances with claimed credit is a disaster waiting to happen...)

[Frederik]

:lol: Yeah, I'm sure they are. :toast:

(note, they could also cheat in this project if they wanted too... being open source and not having any checks or balances with claimed credit is a disaster waiting to happen...)
small correction: EVERYBODY can cheat at R@H, in fact some are busy cheating for the moment(check the rigs that have the highest RAC)
check this rig (http://boinc.bakerlab.org/rosetta/show_host_detail.php?hostid=126650) for example:
rac=957,43 for a P4 3,06 GHz
benchmarks:
Measured floating point speed 12798.33 million ops/sec
Measured integer speed 29492.81 million ops/sec
if thats true, the pope has a daughter...:stick:
Who's machine is it? Is it from the cows? Is it from FDC? is it from XS, XPC, TA,... Nobody knows!
Got my point?

[Movieman]

small correction: EVERYBODY can cheat at R@H, in fact some are busy cheating for the moment(check the rigs that have the highest RAC)
check this rig (http://boinc.bakerlab.org/rosetta/show_host_detail.php?hostid=126650) for example:
rac=957,43 for a P4 3,06 GHz
benchmarks:
Measured floating point speed 12798.33 million ops/sec
Measured integer speed 29492.81 million ops/sec
if thats true, the pope has a daughter...:stick:
Who's machine is it? Is it from the cows? Is it from FDC? is it from XS, XPC, TA,... Nobody knows!
Got my point?
Good Catch! This is getting really frustrating! I was just looking at those numbers, comparing them to my DX 3600 system..There's no way in hell that a 3.06 is outproducing my 3.6 but those numbers say he is blowing away my system! It's time these companies build in some safeguards.
I'd like to see them trace that account.

[heater918]

They seem to be working on it. http://boinc.bakerlab.org/rosetta/forum_thread.php?id=794

[[XC] DragonOrta]

my benchmarks didn't do much of anyhing till I rebooted on my P4. I don't remember if I rebooted my A64 machine to get the right scores or not.

[DDTUNG]

Guys, let's not overreact. You can cheat in almost anything, but in this case it's so obvious that they get caught quickly.

BTW, after you copy the files for an optimized client into the BOINC folder and restart BOINC Manager, make sure you do a manual benchmark run to let the server record the new benchmarks.

DDTUNG:cool:

[Frederik]

Guys, let's not overreact. You can cheat in almost anything, but in this case it's so obvious that they get caught quickly.

BTW, after you copy the files for an optimized client into the BOINC folder and restart BOINC Manager, make sure you do a manual benchmark run to let the server record the new benchmarks.

DDTUNG:cool:
the first time the manager is opened after you put the optimized client in the folder, the manager automaticaly recognizes the new client and performs an benchmark automatically, so I don't think it is needed to perform another benchmark(unless you think the second time it gives higher points).

[DDTUNG]

the first time the manager is opened after you put the optimized client in the folder, the manager automaticaly recognizes the new client and performs an benchmark automatically, so I don't think it is needed to perform another benchmark(unless you think the second time it gives higher points).

I've noticed that the default Rosetta client is now named 5.2.13, the same as the optimized client that some of us are using. On at least one instance I had to run a manual benchmark to ensure the new benchmarks are recognized.

DDTUNG:cool:

[Frederik]

I've noticed that the default Rosetta client is now named 5.2.13, the same as the optimized client that some of us are using. On at least one instance I had to run a manual benchmark to ensure the new benchmarks are recognized.

DDTUNG:cool:
never thought about that...

[Fr3ak]

From what I have noticed Boinc runs the benchmark by itself from time to time. I have a bit lower scores on the website right now. The reason for this is: I benchmark worse the longer the pc runs.. Have to reboot from time to time :)

[KaptainBlaZzed]

Good Catch! This is getting really frustrating! I was just looking at those numbers, comparing them to my DX 3600 system..There's no way in hell that a 3.06 is outproducing my 3.6 but those numbers say he is blowing away my system! It's time these companies build in some safeguards.
I'd like to see them trace that account.


i would have to say that even though the benchmarks for the 3.06 are WAY to high. your bmarks for your 3.6 are way to low. here are the marks for my dX 1.6LV @ 3.2

Measured floating point speed 1950.31 million ops/sec
Measured integer speed 2781.05 million ops/sec

no cheating just using the optomized clients, I tried many optimizations before i settled on these since the give me the highest score. I am currently using the 5.3.2 with my DX rigs and the 5.2.13 with my DC AMD's

[Movieman]

i would have to say that even though the benchmarks for the 3.06 are WAY to high. your bmarks for your 3.6 are way to low. here are the marks for my dX 1.6LV @ 3.2

Measured floating point speed 1950.31 million ops/sec
Measured integer speed 2781.05 million ops/sec

no cheating just using the optomized clients, I tried many optimizations before i settled on these since the give me the highest score. I am currently using the 5.3.2 with my DX rigs and the 5.2.13 with my DC AMD's
Any suggestions, I'll be more than happy to try them?
Mine taken just a couple mins ago:
1/18/2006 3:36:02 PM||Benchmark results:
1/18/2006 3:36:02 PM|| Number of CPUs: 4
1/18/2006 3:36:02 PM|| 1904 double precision MIPS (Whetstone) per CPU
1/18/2006 3:36:02 PM|| 2556 integer MIPS (Dhrystone) per CPU
running the optimised client 5.2.13 on winXP Pro w/sp1
system specs:
Supermicro X6DA8-G2 MB
2X3600/2mb/800 Irwindales
2X1 gig Viking DDR2-400 ECC Reg @3.3.3.7
Leadtek 7800GTX 256mb
2X36 gig WD Raptors in Raid 0
Antec 550 watt EPS12 PS

[NickS]

1/18/2006 4:03:33 PM||Suspending computation and network activity - running CPU benchmarks
1/18/2006 4:03:33 PM|rosetta@home|Pausing result NO_RAND_WTS_1n0u_249_2636_1 (removed from memory)
1/18/2006 4:03:35 PM||Running CPU benchmarks
1/18/2006 4:03:35 PM||request_reschedule_cpus: process exited
1/18/2006 4:04:34 PM||Benchmark results:
1/18/2006 4:04:34 PM|| Number of CPUs: 1
1/18/2006 4:04:34 PM|| 3076 double precision MIPS (Whetstone) per CPU
1/18/2006 4:04:34 PM|| 9283 integer MIPS (Dhrystone) per CPU
1/18/2006 4:04:34 PM||Finished CPU benchmarks
1/18/2006 4:04:35 PM||Resuming computation and network activity


Looks good. Specs in sig.

[P0G0]

I've got all of my computer optimized now:toast:

[Martijn]

DPC still has many members not running optimized, that can mean trouble in the coming months, as they've found out how to install it :rolleyes:

[Rodzilla]

I've been optimized from day one! :up:

On cheating...
I honestly don't care if the entire DPC team is cheating! Really I don't!!

All that truly matters is that we aren't! When it comes down to it we are here to volunteer our CPU time to advance the science of R@H. The stats are a pleasant side affect.

That said... as far as I'm concerned I don't care about cheaters as far as stats go... but it irks me as far as the science goes. If cheaters aren't returning valid results it will only hurt the science and diminish all of our accomplishments. For this reason cheaters should be crushed!