Matt Schramm Ph.D

As someone with multiple cards currently on the reserve list (dual lands, wheels, and way too many forks), destroy the reserve list and reprint them all.

Another option, though I am biased:
github.com/nimbix/jarvi...

Scaling of the GROMACS Molecular Dynamics Code to 65k CPU Cores on an HPC Cluster We benchmarked the performance of the GROMACS 2024 molecular dynamics (MD) code on a modern high-performance computing (HPC) cluster with AMD CPUs on up to 65,536 CPU cores with input systems ranging...

Breaking from JCC: GROMACS reaches a new height in #HPC scalability! @mpibp.bsky.social @mpi-nat.bsky.social onlinelibrary.wiley.com/doi/full/10....

Microsoft word

Like this?

Ok. Not as bad as I was fearing.
Looks like the installation size is ~130 GB.
Looks like the podman build was just scaring me and double counting the installation.

Getting the initial base image built right now.
Looking like it will be ~230GB in size.

I download the DVD ISOs from the Ansys website to build the applications.
No idea if you can get physical disks.

I have some good news and some bad news for my #HPC friends. Ansys 2025 R1 is now available....
Bad news is that it is now twice the size of 2024 R2 (looking at the number of DVD ISOs compared to last release).
Best of luck to everyone...

I will need to look into this...
How does load balancing and ghost cell/atoms work?

Happy New Year everyone.

Remember to update all of your copyright notices from 2024 to 2025...

I didn't need that reminder... Lol

Guess his LLM is becoming too negative...

I really don't need another botnet device mining crypto...

Looks like OpenFOAM gave us a Christmas present...

Time to get this release available to our users.

Once I submitted my answer, I went back and converted the four numbers into signed chars and then packed the signed chars into an int...

Today's #AOC made me do something I'm not too proud of... Needed a hashmap but I wanted to use 4 it's as the key... There is no hash for arrays, so I converted the four numbers into a string...

The only time I use chrome now is for my Team meetings.
Also, PiHole FTW.

Well, day 14 part 2 of #AOC was something... Good luck LLMs...

Cat stares at Christmas tree.

Perf Support For 2,048 CPU Cores Is Becoming Not Enough - Patches Bump Kernel Limit Currently the Linux kernel's 'perf' performance monitoring subsystem has a limit on 2,048 CPU cores for its CPU map that is set by the MAX_NR_CPUS value

The Linux kernel is being updated to allow 4096 cores...
www.phoronix.com/news/Perf-Su...

Do you know if there is a section of video that goes into any updates associated with their EFA network?

It is not...
But the benchmark uses a spack recipe to create the executable. And a bash/python script is used to create the dat file.
I stick to creating as square of a parameter space as possible and not just maxing the flops as I care more about relative performance between clusters.

My question would be, why are you doing this? Are you looking for relative performance between systems? Or are you just looking for a big number?

I do have a hpl benchmark that has to be portable for all of our clusters.

So, a friend and my wife are also trying to save Christmas. My friend is kicking the rust off of their Python... While my wife decided to do it in Excel and got both stars...
#AOC

Got the tree up.
#caturday

Countdown to the start of Advent of Code

Who else is going to be helping to save Christmas this year?
#AOC

Mostly HPC related posts with the occasional cat pic.

👀

We do something like this. As an example, we may split up a 16 core 4 GPU machine into 4 4 core 1 GPU machines.