diff --git a/doc.zih.tu-dresden.de/docs/jobs_and_resources/rome_nodes.md b/doc.zih.tu-dresden.de/docs/jobs_and_resources/rome_nodes.md
index a6cdfba8bd47659bc3a14473cad74c10b73089d0..57ab511938f3eb515b9e38ca831e91cede692418 100644
--- a/doc.zih.tu-dresden.de/docs/jobs_and_resources/rome_nodes.md
+++ b/doc.zih.tu-dresden.de/docs/jobs_and_resources/rome_nodes.md
@@ -2,50 +2,48 @@
## Hardware
-- Slurm partiton: romeo
-- Module architecture: rome
-- 192 nodes taurusi[7001-7192], each:
- - 2x AMD EPYC CPU 7702 (64 cores) @ 2.0GHz, MultiThreading
+- Slurm partition: `romeo`
+- Module architecture: `rome`
+- 192 nodes `taurusi[7001-7192]`, each:
+ - 2x AMD EPYC CPU 7702 (64 cores) @ 2.0GHz, Simultaneous Multithreading (SMT)
- 512 GB RAM
- - 200 GB SSD disk mounted on /tmp
+ - 200 GB SSD disk mounted on `/tmp`
## Usage
-There is a total of 128 physical cores in each
-node. SMT is also active, so in total, 256 logical cores are available
-per node.
+There is a total of 128 physical cores in each node. SMT is also active, so in total, 256 logical
+cores are available per node.
!!! note
- Multithreading is disabled per default in a job. To make use of it
- include the Slurm parameter `--hint=multithread` in your job script
- or command line, or set
- the environment variable `SLURM_HINT=multithread` before job submission.
-Each node brings 512 GB of main memory, so you can request roughly
-1972MB per logical core (using --mem-per-cpu). Note that you will always
-get the memory for the logical core sibling too, even if you do not
-intend to use SMT.
+ Multithreading is disabled per default in a job. To make use of it include the Slurm parameter
+ `--hint=multithread` in your job script or command line, or set the environment variable
+ `SLURM_HINT=multithread` before job submission.
+
+Each node brings 512 GB of main memory, so you can request roughly 1972 MB per logical core (using
+`--mem-per-cpu`). Note that you will always get the memory for the logical core sibling too, even if
+you do not intend to use SMT.
!!! note
- If you are running a job here with only ONE process (maybe
- multiple cores), please explicitly set the option `-n 1` !
-Be aware that software built with Intel compilers and `-x*` optimization
-flags will not run on those AMD processors! That's why most older
-modules built with intel toolchains are not available on **romeo**.
+ If you are running a job here with only ONE process (maybe multiple cores), please explicitly
+ set the option `-n 1`!
+
+Be aware that software built with Intel compilers and `-x*` optimization flags will not run on those
+AMD processors! That's why most older modules built with Intel toolchains are not available on
+partition `romeo`.
-We provide the script: `ml_arch_avail` that you can use to check if a
-certain module is available on rome architecture.
+We provide the script `ml_arch_avail` that can be used to check if a certain module is available on
+`rome` architecture.
## Example, running CP2K on Rome
First, check what CP2K modules are available in general:
`module load spider CP2K` or `module avail CP2K`.
-You will see that there are several different CP2K versions avail, built
-with different toolchains. Now let's assume you have to decided you want
-to run CP2K version 6 at least, so to check if those modules are built
-for rome, use:
+You will see that there are several different CP2K versions avail, built with different toolchains.
+Now let's assume you have to decided you want to run CP2K version 6 at least, so to check if those
+modules are built for rome, use:
```console
marie@login$ ml_arch_avail CP2K/6
@@ -55,13 +53,11 @@ CP2K/6.1-intel-2018a: sandy, haswell
CP2K/6.1-intel-2018a-spglib: haswell
```
-There you will see that only the modules built with **foss** toolchain
-are available on architecture "rome", not the ones built with **intel**.
-So you can load e.g. `ml CP2K/6.1-foss-2019a`.
+There you will see that only the modules built with toolchain `foss` are available on architecture
+`rome`, not the ones built with `intel`. So you can load, e.g. `ml CP2K/6.1-foss-2019a`.
-Then, when writing your batch script, you have to specify the **romeo**
-partition. Also, if e.g. you wanted to use an entire ROME node (no SMT)
-and fill it with MPI ranks, it could look like this:
+Then, when writing your batch script, you have to specify the partition `romeo`. Also, if e.g. you
+wanted to use an entire ROME node (no SMT) and fill it with MPI ranks, it could look like this:
```bash
#!/bin/bash
@@ -73,27 +69,26 @@ and fill it with MPI ranks, it could look like this:
srun cp2k.popt input.inp
```
-## Using the Intel toolchain on Rome
+## Using the Intel Toolchain on Rome
-Currently, we have only newer toolchains starting at `intel/2019b`
-installed for the Rome nodes. Even though they have AMD CPUs, you can
-still use the Intel compilers on there and they don't even create
-bad-performing code. When using the MKL up to version 2019, though,
-you should set the following environment variable to make sure that AVX2
-is used:
+Currently, we have only newer toolchains starting at `intel/2019b` installed for the Rome nodes.
+Even though they have AMD CPUs, you can still use the Intel compilers on there and they don't even
+create bad-performing code. When using the Intel Math Kernel Library (MKL) up to version 2019,
+though, you should set the following environment variable to make sure that AVX2 is used:
```bash
export MKL_DEBUG_CPU_TYPE=5
```
-Without it, the MKL does a CPUID check and disables AVX2/FMA on
-non-Intel CPUs, leading to much worse performance.
+Without it, the MKL does a CPUID check and disables AVX2/FMA on non-Intel CPUs, leading to much
+worse performance.
+
!!! note
- In version 2020, Intel has removed this environment variable and added separate Zen
- codepaths to the library. However, they are still incomplete and do not
- cover every BLAS function. Also, the Intel AVX2 codepaths still seem to
- provide somewhat better performance, so a new workaround would be to
- overwrite the `mkl_serv_intel_cpu_true` symbol with a custom function:
+
+ In version 2020, Intel has removed this environment variable and added separate Zen codepaths to
+ the library. However, they are still incomplete and do not cover every BLAS function. Also, the
+ Intel AVX2 codepaths still seem to provide somewhat better performance, so a new workaround
+ would be to overwrite the `mkl_serv_intel_cpu_true` symbol with a custom function:
```c
int mkl_serv_intel_cpu_true() {
@@ -108,13 +103,11 @@ marie@login$ gcc -shared -fPIC -o libfakeintel.so fakeintel.c
marie@login$ export LD_PRELOAD=libfakeintel.so
```
-As for compiler optimization flags, `-xHOST` does not seem to produce
-best-performing code in every case on Rome. You might want to try
-`-mavx2 -fma` instead.
+As for compiler optimization flags, `-xHOST` does not seem to produce best-performing code in every
+case on Rome. You might want to try `-mavx2 -fma` instead.
### Intel MPI
-We have seen only half the theoretical peak bandwidth via Infiniband
-between two nodes, whereas OpenMPI got close to the peak bandwidth, so
-you might want to avoid using Intel MPI on romeo if your application
-heavily relies on MPI communication until this issue is resolved.
+We have seen only half the theoretical peak bandwidth via Infiniband between two nodes, whereas
+OpenMPI got close to the peak bandwidth, so you might want to avoid using Intel MPI on partition
+`rome` if your application heavily relies on MPI communication until this issue is resolved.
diff --git a/doc.zih.tu-dresden.de/wordlist.aspell b/doc.zih.tu-dresden.de/wordlist.aspell
index c8b2c530bf1ddad273c373899e9c40387e24216b..744930ee80302a6f7192744e4f2030c035ec41e2 100644
--- a/doc.zih.tu-dresden.de/wordlist.aspell
+++ b/doc.zih.tu-dresden.de/wordlist.aspell
@@ -6,6 +6,7 @@ Amdahl's
analytics
anonymized
APIs
+AVX
BeeGFS
benchmarking
BLAS
@@ -22,6 +23,7 @@ Chemnitz
citable
conda
CPU
+CPUID
CPUs
css
CSV
@@ -56,6 +58,7 @@ FFTW
filesystem
filesystems
Flink
+FMA
foreach
Fortran
Gaussian
@@ -130,6 +133,7 @@ mpifort
mpirun
multicore
multithreaded
+Multithreading
NAMD
natively
NCCL
@@ -175,6 +179,7 @@ PowerAI
ppc
Preload
preloaded
+preloading
PSOCK
Pthreads
pymdownx
@@ -236,6 +241,8 @@ Theano
tmp
todo
ToDo
+toolchain
+toolchains
tracefile
tracefiles
transferability