Bad batch of ACEMD 3: fails on Linux after ~20 sec, with:

ERROR: read error for file "input.coor", byte number 4: number of atoms (1880162304) != (107863) expected
ERROR: /home/user/mambaforge/conda-bld/acemd_1704215649797/work/src/mdsim/forcefield.cpp line 300: Cannot read BINCOORD file: input.coor

Tasks 35376930, 35377052, 35377128.

aucun probleme chez moi sous linux mint.

https://www.gpugrid.net/results.php?userid=563937

no problems with my linux mint.

https://www.gpugrid.net/results.php?userid=563937

---

aucun probleme chez moi sous linux mint.

https://www.gpugrid.net/results.php?userid=563937

no problems with my linux mint.

https://www.gpugrid.net/results.php?userid=563937

You just haven't been sent one of the bad ones yet.

https://www.gpugrid.net/result.php?resultid=35372532
https://www.gpugrid.net/result.php?resultid=35376959

Host https://www.gpugrid.net/show_host_detail.php?hostid=462662

Bad batch of ACEMD 3: fails on Linux after ~20 sec, with:

ERROR: read error for file "input.coor", byte number 4: number of atoms (1880162304) != (107863) expected
ERROR: /home/user/mambaforge/conda-bld/acemd_1704215649797/work/src/mdsim/forcefield.cpp line 300: Cannot read BINCOORD file: input.coor

Same here:
https://www.gpugrid.net/workunit.php?wuid=28923983

All of the ACEMD 3 2.28 tasks I received have failed, as well for the other people who were running them. So I'd venture with it being a fresh release, there are some issues with this batch.

I recently started back with the project after a fairly long hiatus, so excuse me if this was obvious.

---

Darn lag in posting!

---

mine are still failing

Client state Compute error
Exit status -112 (0xffffffffffffff90) ERR_XML_PARSE

---

<core_client_version>8.0.2</core_client_version>
<![CDATA[
<message>
(unknown error) (317) - exit code 4294967184 (0xffffff90)</message>
]]>

Is there anything I can do on my end to resolve this issue?

Ryzen 4500 6core
GTX1060
32GB RAM
Win11
All drivers updated.

Hi, same here .....
Nome e51s3_e39s1p0f34-ADRIA_Explor_srcpp1_e2t_25ns_allcontacts4_10us_b2-0-1-RND8030_3
Lavoro 28934849
Creato 15 Jul 2024 | 8:36:59 UTC
Mandato 15 Jul 2024 | 8:37:22 UTC
Ricevuto 15 Jul 2024 | 8:38:05 UTC
Stato server Terminato
Risultato Errore di elaborazione
Stato client Errore di computazione
Stato di uscita -112 (0xffffffffffffff90) ERR_XML_PARSE
ID Computer 400693
Scadenza del report 20 Jul 2024 | 8:37:22 UTC
Tempo di elaborazione 2.66
Tempo CPU 0.00
Stato di validazione Invalido
Crediti 0.00
versione dell'applicazione ACEMD 3: molecular dynamics simulations for GPUs v2.28 (cuda1121)
Output su Stderr

<core_client_version>7.24.1</core_client_version>
<![CDATA[
<message>
(unknown error) (317) - exit code 4294967184 (0xffffff90)</message>
]]>

---

Dreams do not always come true. But not because they are too big or impossible. Why did we stop believing.
(Martin Luther King)

bonsoir
pour ne plus avoir d'erreur je vous conseille de passer sous linux si vous etes sous windows.
linux parait compliqué mais c'est simple.En cherchant on trouve toutes les réponses sur google.
Je détestais linux mais ça marche mieux que windows.
je n'ai plus d'erreur sauf celles que je provoque ou les annulations par le serveur de gpugrid a cause de la lenteur de ma connexion internet.
J'étais sous windows depuis des années et ai mis linux depuis 3 ou 4 mois,je ne remettrais plus windows dans ma tour.Surtout ne pas mettre de gpu amd sous linux car c'est compliqué a installer mais les gpu nvidia c'est une merveille de simplicité.

good evening
to have no more error I advise you to pass under linux if you are under windows.
linux seems complicated but it’s simple. By searching we find all the answers on google.
I hated linux but it works better than windows.
I have no more errors except those I cause or cancellations by the gpugrid server because of the slowness of my internet connection.
I was on windows for years and put linux for 3 or 4 months, I would not put windows back in my tower. Especially do not put amd gpu under linux because it is complicated to install but nvidia GPUs it is a marvel of simplicity.

linux mint-
i5 11400f-
rtx 4060-
rtx a2000-
32 gigas de ram
ssdf 1 téra-
msi z590 wifi

---

Hi, thanks Pascal but "mia mamma usa Windows" (my mom like/use Windows).
K.

---

Dreams do not always come true. But not because they are too big or impossible. Why did we stop believing.
(Martin Luther King)

e peccato.anch'io useva windows e non volevo conoscere linux ma adesso e il contrario.ciao
Buongiorno a tutti gli amici italiani

---

Hello, it does appear the the Windows app version 2.28 is now broken. 2.27 worked. We are investigating.

It is now fixed, one of the files was corrupted.

Hi, thanks a lot.
K.

---

Dreams do not always come true. But not because they are too big or impossible. Why did we stop believing.
(Martin Luther King)

Have started getting the ACEMD 3: molecular dynamics simulations for GPUs v2.30 tasks for Windows.

It's good even if it's only a trickle, which has been 3 tasks today. Anything is welcome!

Hi,
I was playing with zluda and got following error.

Error compiling program: nvrtc: error: invalid value for --gpu-architecture (-arch)

http://www.gpugrid.net/show_host_detail.php?hostid=603175
http://www.gpugrid.net/result.php?resultid=35643503


Hopefully helpful
coproc_info.xml

<cudaVersion>12020</cudaVersion>

<major>8</major>
<minor>8</minor>

I'm thinking cuda toolkit min/maj is outside compatibility of the app. Could we get v8?
I was not successful in changing to different cuda version by installing different cuda toolkits tried
(11.2, 11.6, 12.2, 12.4; boinc detects it as 12.2), and i don't have enough tasks to really validate my findings as i got only 1 task in an hour.

I think, at this project, that the application is more likely to be having a problem with your GPU hardware than the driver version.

Hi,
I was playing with zluda and got following error.

Error compiling program: nvrtc: error: invalid value for --gpu-architecture (-arch)

http://www.gpugrid.net/show_host_detail.php?hostid=603175
http://www.gpugrid.net/result.php?resultid=35643503


Hopefully helpful
coproc_info.xml

<cudaVersion>12020</cudaVersion>

<major>8</major>
<minor>8</minor>

I'm thinking cuda toolkit min/maj is outside compatibility of the app. Could we get v8?
I was not successful in changing to different cuda version by installing different cuda toolkits tried
(11.2, 11.6, 12.2, 12.4; boinc detects it as 12.2), and i don't have enough tasks to really validate my findings as i got only 1 task in an hour.

ZLUDA will only work with PTX code, code that's agnostic to CC version.

the acemd3 app is not compiled with PTX code, it's compiled with discrete CC compatibility values (6.0, 7.5, 8.6, etc). ZLUDA gives you a cc of 8.8, which is not a real CC from nvidia, and as such it is not possible to compile non-PTX code for this CC.

---