nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

Siemens Community Catalyst Program

The Siemens Community Catalyst program was co-created with our community to acknowledge technology leaders who consistently contribute to the Siemens Community. Nominations are accepted on a rolling basis.

Nominate Now

A question about configuring EDEM batch processing scripts in HyperStudy.

Jin1

I'm currently facing a scenario. I have already created a .bat file for coupling EDEM with other multibody dynamics software, successfully registered it with HyperStudy, passed the model test, and completed a full DOE analysis. Now I would like to upgrade my batch processing script, but I'm not sure if this can be achieved within HyperStudy. I have no clue where to start, so I'd like to hear some advice from experts.

My current idea is to add parallel computing. I'm encountering two issues at the moment. The first issue is shown in Figure 1: for example, I want to run four coupled simulations simultaneously, with each case using a different GPU. After these four computations are completed, the subsequent tasks should continue to use the same four GPUs that were just used. I have asked the EDEM and HyperStudy connector team about this before, and it seems that this cannot be implemented either – this is a rather tricky problem.

The second issue is shown in Figure 2. Suppose I am running two coupled cases simultaneously. Normally, the EDEM instance in the M1 folder should couple with the corresponding software in the M1 folder. However, during multiple concurrent runs, it feels like the EDEM in the RUN001 folder ends up coupling with the other software in the RUN002 folder instead. I can manually resolve this by starting only one computation at a time and waiting until they are coupled before starting the next one. However, I'm concerned that sometimes they may finish at the same time, causing the subsequent cases to start simultaneously again, which could lead to coupling errors.

Here is the EDEM code from my batch script. I'm completely stuck and have no idea where to go from here.Here, %1 denotes the .DEM file, while %2 denotes the .dfg file.

Find more posts tagged with

EDEM

HyperStudy

Accepted answers

Stephen Cole

Hi,

There are some possible solutions but may need the latest version of HyperStudy and EDEM.

In HyperStudy 2026 there is a new EDEM connection which allows you to run on different GPU's.

https://support.sw.siemens.com/en-US/product/269200794/knowledge-base/AL33527

There maybe ways to customise your option, if you run nvidia-smi command this shows the load on each GPU. If you could run this first, check the GPU load, determine the GPU with lowest load (e.g. not running a job) and then submit a command to hyperstudy to run on the specified GPU device (using the -D flag in EDEM, -D 0 is running on the first GPU device, -D 1 is the second etc.). However this is essentially implemented in the 2026 version.

On the second issue you can specify to start EDEM via command prompt listening on a specific port:

--cs-port Enable the EDEM Coupling Server using the specified port in range 32768-65535.

https://help.altair.com/edem/topics/coupling_interface_programming_guide/starting_and_stopping_the_coupling_server_t.htm

I'm unsure if the latest MotionSolve coupling has an option for specifying the connection port also, I'll check on that.

Regards

Stephen

All comments

Stephen Cole

Hi,

There are some possible solutions but may need the latest version of HyperStudy and EDEM.

In HyperStudy 2026 there is a new EDEM connection which allows you to run on different GPU's.

https://support.sw.siemens.com/en-US/product/269200794/knowledge-base/AL33527

On the second issue you can specify to start EDEM via command prompt listening on a specific port:

--cs-port Enable the EDEM Coupling Server using the specified port in range 32768-65535.

https://help.altair.com/edem/topics/coupling_interface_programming_guide/starting_and_stopping_the_coupling_server_t.htm

I'm unsure if the latest MotionSolve coupling has an option for specifying the connection port also, I'll check on that.

Regards

Stephen

Jin1

I would like to express my highest respect to you. I remember that I previously mentioned the matter of HyperStudy and EDEM multi-GPU computing, and it was you who replied to me. I didn't expect it to be developed so quickly—this is truly amazing. I will definitely study it. In addition, does this connector now support RecurDyn? Because the coupling between EDEM and MotionView does not allow flexible bodies to interact with particles, so I have to use RecurDyn. However, RecurDyn and EDEM cannot be used together in HyperStudy for DOE analysis and optimization, which is why I came up with the idea of writing my own .bat file. Currently, this is achievable, but the batch command execution still has the two issues I mentioned. If the EDEM and HyperStudy connector could support RecurDyn in the future, or if MotionView could support flexible bodies being transferred to EDEM, I wouldn't have to worry so much. Thank you for your reply, and thanks again.

Jin1

When you say the latest version, do you mean 2026.0 or 2026.1?"

Jin1

I just tried using the connector between EDEM and HyperStudy 2026.0, but it didn't achieve the effect I expected. My idea was that I should be able to specify the number of Jobs—for example, if I set Job=4, then in HyperStudy, four DEM cases would run simultaneously on GPU0 through GPU3. When these four jobs finish, new jobs would start and still use GPU0 through GPU3. However, that's not what actually happened. If I use the -D 0 command, every case runs on GPU 0. Even after four cases finish, the next four new cases still run on GPU 0. This doesn't achieve automation. If I have to manually assign different GPUs each time, and when calculations finish overnight, no one is there to manually specify which GPU number the new cases should continue running on, the computational efficiency remains very low.

My idea is this: have the Job=n parameter, where n can be captured by the EDEM connector. Then, based on this n value, it would automatically assign D=0~n to Job 1 through 4. For example, once Job 4 finishes, it would check the extraction status. As soon as the extraction status is complete, a new Job 5 would start, and this Job 5 would inherit and reuse the GPU index that was just used by the completed Job 4.

My original .bat script code already had the D=0 option, but it just couldn't achieve the state I wanted. I noticed that this connector seems to be of the same type as the .bat script I wrote myself—it simply assigns all calculations to use the same graphics card by setting D=some specific value. I'm not sure if I'm understanding this correctly?

Stephen Cole

Hi Jin,

Thanks for your comments. I should have said it's 2026.1 where the new functionality is added, this is added directly into HyperStudy 2026.1 so we don't need to also install the HyperStudy connector, just EDEM 2026.1 and HyperStudy 2026.1.

In the tutorial there is a note about setting the GPU device indexes, this you can set manually and then it would allow the multi-execution to place the job on the apropriate GPU

https://docs.sw.siemens.com/documentation/external/PL20260206385499731/en-US/html/topics/tutorials/hst/hs_1692_edem_model.htm

Regards

Stephen