Milkyway@home (Projet optimisable - Calcul compatible Carte Graphique)
+9
overclockman
F5LCU 31
badafond-castanha
f11ksx
mjm47
kkkkkkkkkkk
polanx
Meteore31
Fm82290
13 participants
- Meteore31Occitan
- Nombre de messages : 1216
Age : 48
Localisation : 31 - Lanta
Date d'inscription : 07/05/2008
oui f11ksx a mis l'accélérateur
t'as mangé du lion ? tu crédites un max sur GPUGrid et Milky, ça sent la carte ATI ça...
t'as mangé du lion ? tu crédites un max sur GPUGrid et Milky, ça sent la carte ATI ça...
- f11ksxRégional
- Nombre de messages : 672
Age : 62
Localisation : PAMIERS - 09
Date d'inscription : 27/10/2008
Ah bon, il fallait pas en rajouter une?
Moi j'ai cru qu'il fallait parce que tu avais dit qu'on était que 11° au RAC
C'était pour aider un peu...
Moi j'ai cru qu'il fallait parce que tu avais dit qu'on était que 11° au RAC
C'était pour aider un peu...
- Meteore31Occitan
- Nombre de messages : 1216
Age : 48
Localisation : 31 - Lanta
Date d'inscription : 07/05/2008
t'as acheté quoi alors ? tu fais tourner une nvidia et une ATI en même temps sur Boinc ?
va falloir que je m'y mette aussi alors... j'attend de recevoir la facture EDF qui doit arriver en avril et on verra ce qu'on fait...
va falloir que je m'y mette aussi alors... j'attend de recevoir la facture EDF qui doit arriver en avril et on verra ce qu'on fait...
Je fais des émules super!!
- f11ksxRégional
- Nombre de messages : 672
Age : 62
Localisation : PAMIERS - 09
Date d'inscription : 27/10/2008
J'ai monté une ati sur une autre machine
VFO, tu es mon maître....
VFO, tu es mon maître....
- kkkkkkkkkkkRégional
- Nombre de messages : 500
Age : 70
Localisation : Rueil Malmaison
Date d'inscription : 08/06/2008
sur GPUgrid, avec ma 3850 je peux scorer aussi ?
- f11ksxRégional
- Nombre de messages : 672
Age : 62
Localisation : PAMIERS - 09
Date d'inscription : 27/10/2008
Hélas non, c'est réservé aux GTX
Si ça change je te ferais signe.
Allez si ça peut te remonter le moral...
Si ça change je te ferais signe.
Allez si ça peut te remonter le moral...
- mjm47Occitan
- Nombre de messages : 1923
Age : 54
Localisation : Villeneuve sur Lot
Date d'inscription : 01/05/2008
Cette année , c'est sur , je change mon pc de collec
On va tous les enrhumer
On va tous les enrhumer
- chris32Départemental
- Nombre de messages : 235
Age : 43
Localisation : Auch
Date d'inscription : 17/05/2008
perso je plafonne, milky est limité à 6 taches/processeur... et apparement ça a planté chez eux...
Ils veulent séparer le serveur en 2 (un pour CPU et un pour GPU avec des WU + longues) pour éviter la saturation. Enfin j'espère que c'est pas le crash tour.
Chris c'est normal la limite des 6 wus par cpu c'est le programmeur du projet qui l'impose. Il te faut régler le cache BOINC en fonction, mais si il te le dit c'est pas grave que tu en demandes +
Chris c'est normal la limite des 6 wus par cpu c'est le programmeur du projet qui l'impose. Il te faut régler le cache BOINC en fonction, mais si il te le dit c'est pas grave que tu en demandes +
- chris32Départemental
- Nombre de messages : 235
Age : 43
Localisation : Auch
Date d'inscription : 17/05/2008
oki, en epserant que leur mise a jour se passe bien, la je pars en vacances en laissant la becanne seule, faudrai pas que ça tourna à vide ;-)
- kkkkkkkkkkkRégional
- Nombre de messages : 500
Age : 70
Localisation : Rueil Malmaison
Date d'inscription : 08/06/2008
Je suis en manque de WUT Milky.
- chris32Départemental
- Nombre de messages : 235
Age : 43
Localisation : Auch
Date d'inscription : 17/05/2008
j'ai eu du bol alors, moi j'en ai reçu un bon paquet
rdv dans 1 semaine la je pars en vacances
rdv dans 1 semaine la je pars en vacances
- Meteore31Occitan
- Nombre de messages : 1216
Age : 48
Localisation : 31 - Lanta
Date d'inscription : 07/05/2008
f11ksx ou vfoa, pouvez-vous me dire ce que vous installer pour cruncher avec vos ATI 48xx ? ça peut me servir dans un futur proche, lol
j'ai trouvé ça c'est bon ?
1. Download the Catalyst drivers - see where it says 'find a driver' against the ATI cards in zslip
2. Download the apps for ATI 0.19 from zslip
3. Switch off PC, fit the ATI HD card.
4. Power up PC, use the downloaded drivers to install the ATI card.
5. Install latest version of BOINC, if already not done so. Make sure the install does not have the protected install set on.
(protected off may be the default in 6.4.x and later, it's on by default in 6.2.x and earlier)
6. Run BOINC and Attach to MilkyWay. (No WUs will download)
7. Exit BOINC manager.
8. Copy the downloaded app files in (2) above to your MilkyWay project folder (eg, C:\ProgramData\BOINC\projects\milkyway.cs.rpi.edu_milkyway in Vista)
7. Start BOINC Manager.
j'ai trouvé ça c'est bon ?
1. Download the Catalyst drivers - see where it says 'find a driver' against the ATI cards in zslip
2. Download the apps for ATI 0.19 from zslip
3. Switch off PC, fit the ATI HD card.
4. Power up PC, use the downloaded drivers to install the ATI card.
5. Install latest version of BOINC, if already not done so. Make sure the install does not have the protected install set on.
(protected off may be the default in 6.4.x and later, it's on by default in 6.2.x and earlier)
6. Run BOINC and Attach to MilkyWay. (No WUs will download)
7. Exit BOINC manager.
8. Copy the downloaded app files in (2) above to your MilkyWay project folder (eg, C:\ProgramData\BOINC\projects\milkyway.cs.rpi.edu_milkyway in Vista)
7. Start BOINC Manager.
Vive EDF
Tu installes le driver 9.1 (sinon le 9.2, il faut renommer des fichiers dll comme indiqué dans le readme.txt de l'appli optimisée).
Tu fermes Boinc tu installes le 0.19e dans le répertoire projet/milky...
Tu édites le fichier app_config.xml (n3, w1.1 et f0 si tu veux un max de puissance mais des ressources pour bouger la souris).
Tu relances bonc.
Attention il faut que Boinc soit installé en administrateur sous vista, pas en service.
Tu fermes Boinc tu installes le 0.19e dans le répertoire projet/milky...
Tu édites le fichier app_config.xml (n3, w1.1 et f0 si tu veux un max de puissance mais des ressources pour bouger la souris).
Tu relances bonc.
Attention il faut que Boinc soit installé en administrateur sous vista, pas en service.
F11ksx comment tu fais pour avoir des wu aujourd'hui?
Moi nada grrrr le rac équipe baisse
Moi nada grrrr le rac équipe baisse
- Meteore31Occitan
- Nombre de messages : 1216
Age : 48
Localisation : 31 - Lanta
Date d'inscription : 07/05/2008
oui les wus sont rares en ce moment...
- Meteore31Occitan
- Nombre de messages : 1216
Age : 48
Localisation : 31 - Lanta
Date d'inscription : 07/05/2008
ok j'essaierai ça quand j'en aurai une...
f11ksx tu confirmes la manip ? comment fais-tu pour choper des wus alors qu'on peut pas en avoir ?
f11ksx tu confirmes la manip ? comment fais-tu pour choper des wus alors qu'on peut pas en avoir ?
- kkkkkkkkkkkRégional
- Nombre de messages : 500
Age : 70
Localisation : Rueil Malmaison
Date d'inscription : 08/06/2008
Idem la poule aux œufs d'or s'assèche
- Meteore31Occitan
- Nombre de messages : 1216
Age : 48
Localisation : 31 - Lanta
Date d'inscription : 07/05/2008
ils sont peut être en train de travailler sur les wus longues pour les GPUs...
- f11ksxRégional
- Nombre de messages : 672
Age : 62
Localisation : PAMIERS - 09
Date d'inscription : 27/10/2008
Je fais rien de spécial
je relance à la main de temps en temps.
Si ça passe, ça passe.
je relance à la main de temps en temps.
Si ça passe, ça passe.
- Meteore31Occitan
- Nombre de messages : 1216
Age : 48
Localisation : 31 - Lanta
Date d'inscription : 07/05/2008
ok, aujourd'hui ça passe pas trop apparemment...
- f11ksxRégional
- Nombre de messages : 672
Age : 62
Localisation : PAMIERS - 09
Date d'inscription : 27/10/2008
Oui, moi je suis en 8.12 pour les drivers et XP32 pour l'OS
Pour le cc_config.xml j'ai mis
ncpu 8 / ncpu
report_results_immediatly 1 / report_results_immediatly
(i7 oblige, je pense; le report pour envoyer et récupérer des UT plus souvent?)
Dans le app_info.xml j'ai mis
avg_ncpus 0.5 / avg_ncpus
max_ncpus 1 / max_ncpus
cmdline n3 / cmdline
VFO, peux-tu expliciter le W1.1 et f0 s'il te plaît?
Vos commentaires seront bienvenus
Pour le cc_config.xml j'ai mis
ncpu 8 / ncpu
report_results_immediatly 1 / report_results_immediatly
(i7 oblige, je pense; le report pour envoyer et récupérer des UT plus souvent?)
Dans le app_info.xml j'ai mis
avg_ncpus 0.5 / avg_ncpus
max_ncpus 1 / max_ncpus
cmdline n3 / cmdline
VFO, peux-tu expliciter le W1.1 et f0 s'il te plaît?
Vos commentaires seront bienvenus
- Meteore31Occitan
- Nombre de messages : 1216
Age : 48
Localisation : 31 - Lanta
Date d'inscription : 07/05/2008
J'ai trouvé ça qui peut aider :
One can configure the app through some parameters in the supplied app_info.xml.
This is done by editing the line
"cmdline/cmdline"
There are several options one can put in there, the order does not matter. If the same argument
is given twice, the last one counts. Remember that for editing the app_info.xml one should take
the same precautions as for the installation of the application, i.e. one should stop BOINC
completely and restart it afterwards.
Options:
kernel frequency: f (default 30)
The application determines the size of the work packages sent to the GPU in a dynamic manner. It
tries to get the number of executed packages per second above the value specified here by
splitting them to smaller ones. The OS is completely blocked from accessing the graphics card for
the duration of one package leading to a somewhat sluggish behaviour of the user interface.
Limiting the size and therefore reducing the execution time per package creates more opportunities
for the OS to slip in between two work packages and to react to user input. The default value of
30 Hz is tuned for usability of the system. Reducing it could increase the throughput of the app
slightly. A value of 0 enables the behaviour of version 0.19d (no limiting of the package size).
Example:
cmdlinef10/cmdline
The app will try to run 10 work packages per second (limits the execution time to 100ms or shorter).
Number of concurrent WUs: n (default 3)
It is now possible to define the maximum number of concurrently running WUs per GPU. Default value
is 3. If the BOINC Client schedules more WUs to run, they are put to a wait state and not crunched
until a GPU finishes another WU. Running more than a single WU can raise the efficiency of the GPU
as the remaining CPU calculations for one WU can be hidden during the wait for the GPU of another
WU. More than 2 WUs don't help anymore and running too many WUs concurrently first slow down the
calculation and finally crash if there is not enough memory. As a rule of thumb 256MB on a graphics
card are sufficient for 5 WUs. But as mentioned, more than 2 won't help either way. If the high GPU
load leads to a very sluggish behaviour of the user interface or even VPU recover events, limiting
the app to one WU (n1) could improve the situation (look also into the w and f options).
Example:
cmdlinen2/cmdline
The app will run a maximum of two WUs per GPU.
Wait factor: w (default 1.0)
The app tries to release the CPU during the the GPU computations. It does so by predicting
the runtime for the GPU calls and send the CPU thread to sleep for that time. One can manually
correct the prediction of the application with this factor. When raising it, the CPU thread
sleeps longer, decreasing the value will lead to some faster polling. Setting this value to 0.0
turns off the release of the CPU. Default is of course a value of 1.0.
If you see a low GPU load, you can try to decrease it (if you have a load of 80%, set it to 0..
If you see a high CPU load, a slight increase of this value may help. Increase it too much and it
will affect the crunching speed. You could use this for throttling of the GPU in case the high
GPU load leads to a very sluggish behaviour of the user interface or even VPU recover events.
Setting w1.1 could improve the situation (see also the n and f options).
Example:
w1.3
This would increase the sleep time by 30% in relation to the prediction.
Exclusion of GPUs: x (default none)
If a certain GPU should not be used by the app, one can exclude it. The GPU number has to be in
the range 0..31. The first GPU in the system has the number 0. This option can be used multiple
times. Per default, all double precision capable ATI GPUs are used. You don't have to exclude
cards supporting only single precision, it is done automatically.
Example:
x1 x2
The app will not use the second and the third GPU in the system.
The options can be combined in arbitrary order. Therefore the following argument list is also valid:
x1 n4 w0.9 x3 f20
(exclusion of the 2nd and the 4th GPU, 4 WUs per GPU, wait factor 0.9, frequency 20Hz)
Known issues:
- not running with a protected mode (service) installation of BOINC under Vista
- not running when machine is accessed via a remote desktop connection (use VNC instead)
- graphics driver stops to respond (leeding to 2s freezes and/or VPU recover events) on some
OS/driver combinations if certain GPU monitoring tools are running, these include GPU-Z and the
Catalyst Control Center
- appears to run only with Catalyst 8.12 driver on some systems
- somehow sluggish reactions of the Windows user interface, gets worse with longer WUs and better
with faster cards
One can configure the app through some parameters in the supplied app_info.xml.
This is done by editing the line
"cmdline/cmdline"
There are several options one can put in there, the order does not matter. If the same argument
is given twice, the last one counts. Remember that for editing the app_info.xml one should take
the same precautions as for the installation of the application, i.e. one should stop BOINC
completely and restart it afterwards.
Options:
kernel frequency: f (default 30)
The application determines the size of the work packages sent to the GPU in a dynamic manner. It
tries to get the number of executed packages per second above the value specified here by
splitting them to smaller ones. The OS is completely blocked from accessing the graphics card for
the duration of one package leading to a somewhat sluggish behaviour of the user interface.
Limiting the size and therefore reducing the execution time per package creates more opportunities
for the OS to slip in between two work packages and to react to user input. The default value of
30 Hz is tuned for usability of the system. Reducing it could increase the throughput of the app
slightly. A value of 0 enables the behaviour of version 0.19d (no limiting of the package size).
Example:
cmdlinef10/cmdline
The app will try to run 10 work packages per second (limits the execution time to 100ms or shorter).
Number of concurrent WUs: n (default 3)
It is now possible to define the maximum number of concurrently running WUs per GPU. Default value
is 3. If the BOINC Client schedules more WUs to run, they are put to a wait state and not crunched
until a GPU finishes another WU. Running more than a single WU can raise the efficiency of the GPU
as the remaining CPU calculations for one WU can be hidden during the wait for the GPU of another
WU. More than 2 WUs don't help anymore and running too many WUs concurrently first slow down the
calculation and finally crash if there is not enough memory. As a rule of thumb 256MB on a graphics
card are sufficient for 5 WUs. But as mentioned, more than 2 won't help either way. If the high GPU
load leads to a very sluggish behaviour of the user interface or even VPU recover events, limiting
the app to one WU (n1) could improve the situation (look also into the w and f options).
Example:
cmdlinen2/cmdline
The app will run a maximum of two WUs per GPU.
Wait factor: w (default 1.0)
The app tries to release the CPU during the the GPU computations. It does so by predicting
the runtime for the GPU calls and send the CPU thread to sleep for that time. One can manually
correct the prediction of the application with this factor. When raising it, the CPU thread
sleeps longer, decreasing the value will lead to some faster polling. Setting this value to 0.0
turns off the release of the CPU. Default is of course a value of 1.0.
If you see a low GPU load, you can try to decrease it (if you have a load of 80%, set it to 0..
If you see a high CPU load, a slight increase of this value may help. Increase it too much and it
will affect the crunching speed. You could use this for throttling of the GPU in case the high
GPU load leads to a very sluggish behaviour of the user interface or even VPU recover events.
Setting w1.1 could improve the situation (see also the n and f options).
Example:
This would increase the sleep time by 30% in relation to the prediction.
Exclusion of GPUs: x (default none)
If a certain GPU should not be used by the app, one can exclude it. The GPU number has to be in
the range 0..31. The first GPU in the system has the number 0. This option can be used multiple
times. Per default, all double precision capable ATI GPUs are used. You don't have to exclude
cards supporting only single precision, it is done automatically.
Example:
The app will not use the second and the third GPU in the system.
The options can be combined in arbitrary order. Therefore the following argument list is also valid:
(exclusion of the 2nd and the 4th GPU, 4 WUs per GPU, wait factor 0.9, frequency 20Hz)
Known issues:
- not running with a protected mode (service) installation of BOINC under Vista
- not running when machine is accessed via a remote desktop connection (use VNC instead)
- graphics driver stops to respond (leeding to 2s freezes and/or VPU recover events) on some
OS/driver combinations if certain GPU monitoring tools are running, these include GPU-Z and the
Catalyst Control Center
- appears to run only with Catalyst 8.12 driver on some systems
- somehow sluggish reactions of the Windows user interface, gets worse with longer WUs and better
with faster cards
Permission de ce forum:
Vous ne pouvez pas répondre aux sujets dans ce forum
|
|