Some issues on Nektar++ calculations
******************* This email originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list https://spam.ic.ac.uk/SpamConsole/Senders.aspx to disable email stamping for this address. ******************* Dear Nektar++ users, I am writing to report a problem that I am currently experiencing with Nektar++ that I am using. I am a PhD student at Peking University and I am currently working on a computational project involving a quasi-3D separated flow over a flat plate. The initial condition is a pre-computed instantaneous flow field file (format .chk), where HomModesZ is 32. I am running the calculation with HomModesZ of 128 on a supercomputer with 16 nodes and a total of 896 cores. The npz is set to 64. After the calculation is completed, the instantaneous flow field files are generated with 896 fld files. However, two of the fld files are problematic. Specifically, the fld file with the number P0000171 is missing the <NEKTAR> tag at the beginning and the content is disordered. The fld file with the number P0000195 is an empty file. Both prevent me from converting the entire ".chk" folder to the .plt file format using the FieldConvert command. This issue has occurred multiple times in several computations, so it does not appear to be an isolated incident. The disk space usage during the computation did not reach full capacity, so I suspect there may be a bug in the calculation process. Additionally, in the similar calculations about 2 months ago, I didn’t find problems. It means that there’s probably nothing wrong with my case. Could you please provide any suggestions on how to resolve this issue? Thank you for your attention to this matter. Best regards, Du Zengzhi
Dear Du, Many thanks for sharing the issue you are experiencing. I am wondering if the issue happens only for 128 planes or you have seen this for other number of Fourier planes? Could you please open an issue on gitlab and if possible kindly provide us a small test case failing so we can debug and isolate the bug if any? Many thanks Mohsen From: nektar-users-bounces@imperial.ac.uk <nektar-users-bounces@imperial.ac.uk> on behalf of duzz@pku.edu.cn <duzz@pku.edu.cn> Date: Friday, 5 May 2023 at 10:23 To: nektar-users@imperial.ac.uk <nektar-users@imperial.ac.uk> Subject: [Nektar-users] Some issues on Nektar++ calculations ⚠ External sender. Take care when opening links or attachments. Do not provide your login details. This email from duzz@pku.edu.cn originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list<https://spam.ic.ac.uk/SpamConsole/Senders.aspx> to disable email stamping for this address. Dear Nektar++ users, I am writing to report a problem that I am currently experiencing with Nektar++ that I am using. I am a PhD student at Peking University and I am currently working on a computational project involving a quasi-3D separated flow over a flat plate. The initial condition is a pre-computed instantaneous flow field file (format .chk), where HomModesZ is 32. I am running the calculation with HomModesZ of 128 on a supercomputer with 16 nodes and a total of 896 cores. The npz is set to 64. After the calculation is completed, the instantaneous flow field files are generated with 896 fld files. However, two of the fld files are problematic. Specifically, the fld file with the number P0000171 is missing the <NEKTAR> tag at the beginning and the content is disordered. The fld file with the number P0000195 is an empty file. Both prevent me from converting the entire ".chk" folder to the .plt file format using the FieldConvert command. This issue has occurred multiple times in several computations, so it does not appear to be an isolated incident. The disk space usage during the computation did not reach full capacity, so I suspect there may be a bug in the calculation process. Additionally, in the similar calculations about 2 months ago, I didn’t find problems. It means that there’s probably nothing wrong with my case. Could you please provide any suggestions on how to resolve this issue? Thank you for your attention to this matter. Best regards, Du Zengzhi
Hi Mohsen, I discussed this problem with Zengzhi. This is a 3DH1D simulation and the npz parameter has been correctly set as half of the number of Fourier modes. The computation node has 56 cores (2 Intel Xeon 6258R) and 16 nodes are used. Xml-type fld is used for IO. After our investigation, two threads output wrong fld files, with one file (171) missing Nektar tag (see attached png) and another file (195) being just empty. I was wondering if there is a bug in the FieldIOXml. I suggested Zengzhi use Hdf5 format at the moment. Cheers Ankang -- Research Associate Department of modern mechanics University of Science and Technology of China -----Original Messages----- From:"Mohsen Lahooti" <Mohsen.Lahooti@newcastle.ac.uk> Sent Time:2023-05-05 17:31:40 (Friday) To: "duzz@pku.edu.cn" <duzz@pku.edu.cn>, "nektar-users@imperial.ac.uk" <nektar-users@imperial.ac.uk> Cc: Subject: Re: [Nektar-users] Some issues on Nektar++ calculations Dear Du, Many thanks for sharing the issue you are experiencing. I am wondering if the issue happens only for 128 planes or you have seen this for other number of Fourier planes? Could you please open an issue on gitlab and if possible kindly provide us a small test case failing so we can debug and isolate the bug if any? Many thanks Mohsen From: nektar-users-bounces@imperial.ac.uk <nektar-users-bounces@imperial.ac.uk> on behalf of duzz@pku.edu.cn <duzz@pku.edu.cn> Date: Friday, 5 May 2023 at 10:23 To: nektar-users@imperial.ac.uk <nektar-users@imperial.ac.uk> Subject: [Nektar-users] Some issues on Nektar++ calculations ⚠ External sender. Take care when opening links or attachments. Do not provide your login details. | This email from duzz@pku.edu.cn originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list to disable email stamping for this address. | Dear Nektar++ users, I am writing to report a problem that I am currently experiencing with Nektar++ that I am using. I am a PhD student at Peking University and I am currently working on a computational project involving a quasi-3D separated flow over a flat plate. The initial condition is a pre-computed instantaneous flow field file (format .chk), where HomModesZ is 32. I am running the calculation with HomModesZ of 128 on a supercomputer with 16 nodes and a total of 896 cores. The npz is set to 64. After the calculation is completed, the instantaneous flow field files are generated with 896 fld files. However, two of the fld files are problematic. Specifically, the fld file with the number P0000171 is missing the <NEKTAR> tag at the beginning and the content is disordered. The fld file with the number P0000195 is an empty file. Both prevent me from converting the entire ".chk" folder to the .plt file format using the FieldConvert command. This issue has occurred multiple times in several computations, so it does not appear to be an isolated incident. The disk space usage during the computation did not reach full capacity, so I suspect there may be a bug in the calculation process. Additionally, in the similar calculations about 2 months ago, I didn’t find problems. It means that there’s probably nothing wrong with my case. Could you please provide any suggestions on how to resolve this issue? Thank you for your attention to this matter. Best regards, Du Zengzhi
participants (3)
-
ankanggao＠ustc.edu.cn
-
duzz＠pku.edu.cn
-
Mohsen Lahooti