initial run on isambard cascadelake for paper

[dcaseGH]

Set up:
. ~/miniconda3/bin/activate
conda create -n "benchmarking-paper" python=3.11
conda activate benchmarking-paper
export SPACK_ROOT="/home/mox-dcase/spack"
source "${SPACK_ROOT}/share/spack/setup-env.sh"
export RFM_CONFIG_FILES="${PWD}/benchmarks/reframe_config.py"
pip install -e .

Run intel:
reframe -c benchmarks/apps/hpcg -r -n HPCG_Intel --system isambard-macs:cascadelake --performance-report

Run 'out' HPCG stuff (but no intel):
reframe -c benchmarks/apps/hpcg -r -n HPCG_ -x HPCG_Intel --system isambard-macs:cascadelake --performance-report -S build_locally=false

First results:
PERFORMANCE REPORT

[HPCG_IntelSKX_Isambard /e4309d0d @isambard-macs:cascadelake:default]
num_cpus_per_task: 1
num_tasks_per_node: 40
num_tasks: 40
performance:
- flops: 38.6863 (r: 0 l: -inf% u: +inf%)
[HPCG_IntelAVX2_Isambard /ae3f0e20 @isambard-macs:cascadelake:default]
num_cpus_per_task: 1
num_tasks_per_node: 40
num_tasks: 40
performance:
- flops: 39.0427 (r: 0 l: -inf% u: +inf%)
[HPCG_IntelAVX_Isambard /efe9bac8 @isambard-macs:cascadelake:default]
num_cpus_per_task: 1
num_tasks_per_node: 40
num_tasks: 40
performance:
- flops: 38.7652 (r: 0 l: -inf% u: +inf%)

[HPCG_LFRic /cde082ad @isambard-macs:cascadelake:default]
num_cpus_per_task: 1
num_tasks_per_node: 40
num_tasks: 40
performance:
- flops: 18.5399 Gflops/seconds (r: 1 Gflops/seconds l: -inf% u: +inf%)
[HPCG_Original /bde436dc @isambard-macs:cascadelake:default]
num_cpus_per_task: 1
num_tasks_per_node: 40
num_tasks: 40
performance:
- flops: 24.027 Gflops/seconds (r: 1 Gflops/seconds l: -inf% u: +inf%)
[HPCG_Stencil /21b95432 @isambard-macs:cascadelake:default]
num_cpus_per_task: 1
num_tasks_per_node: 40
num_tasks: 40
performance:
- flops: 51.007 Gflops/seconds (r: 1 Gflops/seconds l: -inf% u: +inf%)

[dcaseGH]

Additional result from ARCHER2. (I know not directly related to code, but I've put it here as we're compiling the manuscript and it has instructions).

python3 -m venv ./localenv
source ./localenv/bin/activate
module load cray-python
pip install -e .
export RFM_CONFIG_FILES="${PWD}/benchmarks/reframe_config.py"
export SPACK_ROOT="/work/n02/n02/dcase/spack"
source "${SPACK_ROOT}/share/spack/setup-env.sh"
reframe -c benchmarks/apps/hpcg -n HPCG_ -x HPCG_Intel -r --performance-report --system archer2 -J'--account=n02-cms' -J'--time=0:20:0' -J'--qos=standard'

PERFORMANCE REPORT

[HPCG_LFRic /cde082ad @Archer2:compute-node:default]
num_cpus_per_task: 1
num_tasks_per_node: 128
num_tasks: 128
performance:
- flops: 55.9839 Gflops/seconds (r: 1 Gflops/seconds l: -inf% u: +inf%)
[HPCG_Original /bde436dc @Archer2:compute-node:default]
num_cpus_per_task: 1
num_tasks_per_node: 128
num_tasks: 128
performance:
- flops: 39.2325 Gflops/seconds (r: 1 Gflops/seconds l: -inf% u: +inf%)
[HPCG_Stencil /21b95432 @Archer2:compute-node:default]
num_cpus_per_task: 1
num_tasks_per_node: 128
num_tasks: 128
performance:
- flops: 124.197 Gflops/seconds (r: 1 Gflops/seconds l: -inf% u: +inf%)

[dcaseGH]

I have added this branch and this PR to detail the changes for the paper, and instructions (and results) as used to calculate the table in that section.
@JamieJQuinn is probably away whilst I'm here, but @christophermaynard and I should talk about code review
@tkoskela might want to think about the timescale of merging - but there should be time - and @giordano may want things tidied up

You can all put comments below and I'll try to address them. Thanks