... | @@ -2,27 +2,13 @@ |
... | @@ -2,27 +2,13 @@ |
|
|
|
|
|
On a recent-enough operating system, installing nabu boils down to `pip install nabu` (along with `pip install pycuda && pip install scikit-cuda` for the Cuda backend).
|
|
On a recent-enough operating system, installing nabu boils down to `pip install nabu` (along with `pip install pycuda && pip install scikit-cuda` for the Cuda backend).
|
|
|
|
|
|
Unfortunately, ESRF infrastructure is far from this ideal setting, sometimes for understandable reasons (old hardware, dedicated beamlines machines where upgrading is risky, etc). In this case, conda is often needed to bring a decent version of Python.
|
|
However the configuration is not so homogeneous (amd64, power9, Cuda10, Cuda11, ...).
|
|
|
|
The [automatix project](https://gitlab.esrf.fr/tomotools/automatix) aims at making deployment simple.
|
|
However, conda environments sometimes cannot be used between hosts when GPU drivers differ. For example, one node might have Debian 8.11 and a Kepler GPU, while another has Debian 8.11 with a Fermi GPU. In this case, a conda environment must be created for each. See (1) in notes for more details.
|
|
|
|
|
|
|
|
## Conda environments classification
|
|
|
|
|
|
|
|
The servers can be classified as follows:
|
|
|
|
- "Ideal Case": recent operating system (Debian >= 10, Ubuntu >= 20.04). Plain venv is enough. Examples: scisoft 10,11,14,15
|
|
|
|
- "venv + PyQt5 patch": venv is almost enough, but some tinkering with `PyQt5` needs to be done to have GUI working. Examples: p9-gpu slurm partition
|
|
|
|
- Conda
|
|
|
|
- Fermi driver
|
|
|
|
- Kepler driver
|
|
|
|
- ...
|
|
|
|
|
|
|
|
|
|
|
|
## ESRF machines
|
|
## ESRF machines
|
|
|
|
|
|
Last updated: 12/12/2022
|
|
Last updated: 16/12/2022
|
|
|
|
|
|
**Debian 8 is not supported anymore**.
|
|
|
|
The OS is now Ubuntu 20.04 on almost all machines. Therefore the "OS" column has been removed.
|
|
|
|
|
|
|
|
| **Machine name/partition/number** | **CPU, GPU** | **Cuda information** | **Notes** |
|
|
| **Machine name/partition/number** | **CPU, GPU** | **Cuda information** | **Notes** |
|
|
|-----------------------------------|---------------------------------|-------------------------------------|-----------------------------|
|
|
|-----------------------------------|---------------------------------|-------------------------------------|-----------------------------|
|
... | @@ -42,7 +28,7 @@ The OS is now Ubuntu 20.04 on almost all machines. Therefore the "OS" column has |
... | @@ -42,7 +28,7 @@ The OS is now Ubuntu 20.04 on almost all machines. Therefore the "OS" column has |
|
| crunch (Debian 8!) | Xeon X5690, Titan black | Cuda 6.5, driver 384 | Hopelessly outdated |
|
|
| crunch (Debian 8!) | Xeon X5690, Titan black | Cuda 6.5, driver 384 | Hopelessly outdated |
|
|
| scisoft10,scisoft11 | Xeon E5-2643, Quadro P2000 | Cuda 10.1/11.4, driver 470.129 | Used for CI |
|
|
| scisoft10,scisoft11 | Xeon E5-2643, Quadro P2000 | Cuda 10.1/11.4, driver 470.129 | Used for CI |
|
|
| scisoft12 | Xeon E5-2667, RTX A2000 | Cuda 11.2 | Not stable yet |
|
|
| scisoft12 | Xeon E5-2667, RTX A2000 | Cuda 11.2 | Not stable yet |
|
|
| scisoft13 (Debian 11) | Xeon E5-2643, 1080 Ti | Cuda 11.2, driver 460.91 | Debian 11 |
|
|
| scisoft13 (Debian 11) | Xeon E5-2643, 1080 Ti, RTX A4500| Cuda 11.2, driver 460.91 | Debian 11 |
|
|
| scisoft14 (Debian 11) | Xeon Gold 6134, Titan V, P6000 | Cuda 11.2, Driver 460.91 | Debian 11 |
|
|
| scisoft14 (Debian 11) | Xeon Gold 6134, Titan V, P6000 | Cuda 11.2, Driver 460.91 | Debian 11 |
|
|
| scisoft15 | Power9, Tesla V100 | Cuda 10.1, driver 418.126 | ppc64le architecture |
|
|
| scisoft15 | Power9, Tesla V100 | Cuda 10.1, driver 418.126 | ppc64le architecture |
|
|
|
|
|
... | @@ -61,4 +47,22 @@ However, it seems that the `cudatoolkit` conda package is not the way to go. Ins |
... | @@ -61,4 +47,22 @@ However, it seems that the `cudatoolkit` conda package is not the way to go. Ins |
|
|
|
|
|
## See also
|
|
## See also
|
|
|
|
|
|
http://www.esrf.eu/Infrastructure/Computing/NICE/Implementation |
|
http://www.esrf.eu/Infrastructure/Computing/NICE/Implementation
|
|
\ No newline at end of file |
|
|
|
|
|
|
|
|
|
## Notes on the usage of conda
|
|
|
|
|
|
|
|
Some ESRF machines use a very outdated operating system (less and less true as most machines are almost all Ubuntu 20.04).
|
|
|
|
In this case, conda is often needed to bring a reasonably-up-to-date version of Python.
|
|
|
|
|
|
|
|
However, conda environments sometimes cannot be used between hosts when GPU drivers differ. For example, one node might have Debian 8.11 and a Kepler GPU, while another has Debian 8.11 with a Fermi GPU. In this case, a conda environment must be created for each. See (1) in notes for more details.
|
|
|
|
|
|
|
|
### Conda environments classification
|
|
|
|
|
|
|
|
The servers can be classified as follows:
|
|
|
|
- "Ideal Case": recent operating system (Debian >= 10, Ubuntu >= 20.04). Plain venv is enough. Examples: scisoft 10,11,14,15
|
|
|
|
- "venv + PyQt5 patch": venv is almost enough, but some tinkering with `PyQt5` needs to be done to have GUI working. Examples: p9-gpu slurm partition
|
|
|
|
- Conda
|
|
|
|
- Fermi driver
|
|
|
|
- Kepler driver
|
|
|
|
- ... |
|
|
|
\ No newline at end of file |