hfss 10
http://www.ansoft.com/hfworkshop05/a...s_bios.cfm#a17
I would like to add that the most important change will be in the mesh generation engine:
1) It will not create too many tetrahedrons. As you might know, the HFSS mesh generator sometimes gets confused and creates more than enough tetrahedrons.
2) The mesh generator will decide how much to refine the mesh automatically during the adaptive refinement. Of course the existing method of giving it some constant value, say 30 percent, will still exist.
V10 is to be released in march.
As to V11, its going to have at least two cool features:
1) Capability to decompose the total domain to smaller regions and solve all through an adaptive procedure. As such it will be able to solve very big problems without resorting to huge amount of RAM.
2) The solver engine will have an iterative solver. They will probably use some type of multigrid method.
All the abovementioned info are based on the recent workshop of Ansoft.
Just a little update:
1) Distributed computing will be in version 10.
h**p://w*w.dac.com/data2/42nd/42exhibitorArea.nsf/0/B54C902559032C2B87256E2E007BF1B5
2) Release date for version 10 has been mentioned to be around the MTTS time.
i have used hfss version 10 which was released to a company i worked for. the biggest change i saw (and it was big), was the way memory swapping was occuring for large problems. for instance instead of crashing because hfss ran out of RAM to use it started swapping up to a gig (maybe even more i cant remember) in main memory. this slowed down the process, as you can imagine but it still allowed for large problems to be solved.
another change i saw which i dont believe was available in 9, was when one creates multiple models within the same file name, there is then an option to simulate all. kind of like a built in batch command. this was quite handy when wanting to run multiple models.
Dear ghb,
Memory swapping is available in version 9 too. I've done simulations with more than 3 GB swapping on hard disk and more than 1 GB in RAM. But if you mean something like what Feko can do, i.e. addressing tens of GB of memory through swapping, that would be great. As to speed issue, I recommend using 15000 RPM SCSI hard disks.
Regards