Hi Paul,

 

Just wanted to add something else into the mix.

 

I’ve tried connecting to another site using NoMachine – they are running the venerable fsleyes 1.1.1, and with best appearance selected for rendering mode, the system runs as if you were on console – instantly responsive. I.e. much faster than the latest version with best performance put back in.

 

Not sure if this is helpful? Let me know if you want more information about the setup or indeed if you want to see it in action?

 

Cheers, Jon

 

 

Dear Paul,

 

 

Thanks for this. Seems better, but still not as fast as with previous version 1.3.3(?) – the default as was distributed with fsl 6.0.5.1 as per couple of months ago.

 

 

Followed your instructions to download using git.

 

 

I needed to “load” fsl in our normal way i.e. module load fsl

 

Otherwise fslpython isn’t found.

 

 

(base) [gnx20mmu@uwwbichead01 ~/fsleyes]$ cd fsleyes

 

(base) [gnx20mmu@uwwbichead01 ~/fsleyes]$ export PYTHONPATH=$(pwd)

 

(base) [gnx20mmu@uwwbichead01 ~/fsleyes]$ fslpython -m fsleyes -p 1

 

 

(40 seconds elapse before splash appears, fslpython showing ~7% CPU usage during this time).

 

 

/gpfs/home/gnx20mmu/fsleyes/fsleyes/colourmaps.py:572: UserWarning: Trying to register the cmap 'pink' which already exists.

 

  mplcm.register_cmap(key, cmap, override_builtin=True)

 

/gpfs/home/gnx20mmu/fsleyes/fsleyes/colourmaps.py:572: UserWarning: Trying to register the cmap 'hot' which already exists.

 

  mplcm.register_cmap(key, cmap, override_builtin=True)

 

/gpfs/home/gnx20mmu/fsleyes/fsleyes/colourmaps.py:572: UserWarning: Trying to register the cmap 'cool' which already exists.

 

  mplcm.register_cmap(key, cmap, override_builtin=True)

 

/gpfs/home/gnx20mmu/fsleyes/fsleyes/colourmaps.py:572: UserWarning: Trying to register the cmap 'copper' which already exists.

 

  mplcm.register_cmap(key, cmap, override_builtin=True)

 

/gpfs/home/gnx20mmu/fsleyes/fsleyes/colourmaps.py:572: UserWarning: Trying to register the cmap 'hsv' which already exists.

 

  mplcm.register_cmap(key, cmap, override_builtin=True)

 

 

(__main__.py:150232): Gtk-CRITICAL **: 21:09:22.638: gtk_window_resize: assertion 'width > 0' failed

 

/gpfs/software/ada/fsl/6.0.5.1/fslpython/envs/fslpython/lib/python3.8/site-packages/xnat/__init__.py:27: DeprecationWarning: the imp module is deprecated in favour of importlib; see the module's documentation for alternative uses

 

  import imp

 

/gpfs/software/ada/fsl/6.0.5.1/fslpython/envs/fslpython/lib/python3.8/site-packages/notebook/utils.py:280: DeprecationWarning: distutils Version classes are deprecated. Use packaging.version instead.

 

  return LooseVersion(v) >= LooseVersion(check)

 

 

(__main__.py:150232): Gdk-WARNING **: 21:10:33.640: gdkdrawable-x11.c:952 drawable is not a pixmap or window

 

 

Then fsleyes window appears and I’m able to poke around in the standard brain with a (noticeable) 100ms lag between cursor click and screen update.

 

 

Relevant?:

 

    I’m running through NoMachine connecting via VPN to server

    The server (CentOS) uses “module” to load fsl (set paths, envars) e.g. $ module load fsl

    When I start fsleyes python (not fslpython) shows about 10% CPU usage and the splash takes about 40 seconds to appear, and fsleyes is unusable – the screen redraws are incredibly slow (2 seconds between each one).

    It’s perfectly possible to watch a youtube video remotely through NoMachine connection – and glxgears runs smoothly i.e. I don’t think it’s the connection or the server’s performance

 

 

So it’s definitely better (x20), but still not great… and not as good as before.

 

 

Cheers, Jon

 

 

 

Hi Jon,

 

 

I have an experimental version of FSLeyes with the performance setting added back in. Would you be able to test it on your system to see if it improves performance? You should be able to use it without modifying your existing FSL/FSLeyes installation by following these steps:

 

 

# Clone the git repository, and check out the development branch:

 

git clone https://git.fmrib.ox.ac.uk/paulmc/fsleyes.git

 

cd fsleyes

 

git checkout mnt/revert-remove-performance

 

 

# Use the PYTHONPATH variable to override the installed FSLeyes version

 

export PYTHONPATH=$(pwd)

 

 

# Run fsleyes through the fslpython interpreter, with low performance enabled

 

fslpython -m fsleyes -p 1

 

 

Paul

 

 

On Thu, 7 Jul 2022 at 21:26, Jon Brooks (PSY - Staff) <[log in to unmask]> wrote:

 

Hi Paul,

 

 

Thanks for getting back to me – I’m not sure what the issue is, but when using fsleyes on our headnode (CentOS7) rendering performance in orthoview is abysmal. We access the server via NoMachine Linux Terminal Server edition (v6), and the server is using hardware acceleration.

 

 

In the 1.3.x version we would change performance from Best Appearance to (Fast or Fastest) and it was like night and day. On the former fsleyes renders like a download from the 1980s over a dial-up modem, with Fastest it’s like being sat in front of my mac, instantly responsive, with no apparent loss of visual quality.

 

 

So that was a very long winded way of saying: yes please, can we have the option back 😉

 

 

Cheers, Jon

 

 

--

 

Jon Brooks (PSY - Staff)

 

 

Associate Professor & Head of MRI

 

University of East Anglia Wellcome Wolfson Brain Imaging Centre (UWWBIC)

 

School of Psychology, Room 01.56, Lawrence Stenhouse Building, Chancellors Drive,

 

Norwich Research Park, Norwich, NR4 7TJ, United Kingdom

 

Tel: +44(0)1603 591396

 

Int: 1396

 

A picture containing silhouette Description automatically generatedsignature_4163963870

 

 

UK 14th for Research Quality in Psychology, Psychiatry, and Neuroscience

 

(Times Higher Education rankings for the Research Excellence Framework 2021)

 

World Top 200 (Times Higher Education World University Rankings 2022)

UK Top 30 (The Times/Sunday Times 2022 and Complete University Guide 2022)

UK Top 20 for research quality (Times Higher Education Rankings for the Research Excellence Framework 2021)

World Top 50 for research citations (Times Higher Education World University Rankings 2022)

World Top 50 (Times Higher Education Impact Rankings 2022)

Athena SWAN Silver Award Holder (since 2019) in recognition of advancement towards gender equality for all (Advance HE)

 

 

 

From: Jon Brooks (PSY - Staff) <[log in to unmask]>

Date: Wednesday, 6 July 2022 at 16:07

To: [log in to unmask] <[log in to unmask]>

Cc: [log in to unmask] <[log in to unmask]>

Subject: Update to fsleyes 1.4.6

 

Hi,

 

 

We had been having some problems with slow rendering performance on orthoview, then discovered the rendering performance settings (best appearance, fast, fastest) – which solved it all…. until now.

 

 

We just updated to the latest version (1.4.6) and the rendering performance option under “view settings” (spanner icon) has gone.

 

 

Any ideas where to start with this one?

 

 

Cheers,

 

 

Jon

 

 

FYI

 

1.       we’re running on a remote server using NoMachine, with hardware acceleration turned on

 

2.       glxgears runs fine

 

3.       other info below:

 

 

FSLeyes version: 1.4.6

 

FMRIB Centre, Oxford, UK

 

Paul McCarthy

 

[log in to unmask]

 

FSL version: None

 

FSL directory: /gpfs/software/ada/fsl/6.0.5.1

 

OpenGL version: 3.1

 

OpenGL compatibility: 2.1

 

OpenGL renderer: softpipe

 

 

FSLeyes was developed at the FMRIB Centre, Nuffield Department of Clinical Neurosciences, Oxford University, United Kingdom.

 

 

FSLeyes is a Python application which leverages the following open-source software libraries:

 

 

- fsleyes-props [1.7.3] (https://git.fmrib.ox.ac.uk/fsl/fsleyes/props)

 

- fsleyes-widgets [0.12.3] (https://git.fmrib.ox.ac.uk/fsl/fsleyes/widgets)

 

- fslpy [3.10.0] (https://git.fmrib.ox.ac.uk/fsl/fslpy)

 

- indexed_gzip [1.6.13] (https://github.com/pauldmccarthy/indexed_gzip/)

 

- IPython [8.4.0] (https://ipython.org/)

 

- jinja2 [3.1.2] (http://jinja.pocoo.org)

 

- Jupyter notebook [6.4.12] (https://jupyter.org)

 

- matplotlib [3.5.2] (http://www.matplotlib.org)

 

- nibabel [4.0.1] (http://nipy.org/nibabel)

 

- numpy [1.22.4] (http://www.numpy.org)

 

- pillow [9.2.0]  (http://python-pillow.org/)

 

- pyopengl [3.1.6] (http://pyopengl.sourceforge.net)

 

- pyparsing [2.4.7] (http://pyparsing.wikispaces.com/)

 

- scipy [1.8.1] (http://www.scipy.org)

 

- six [1.16.0] (https://pythonhosted.org/six/)

 

- trimesh [3.12.0] (https://github.com/mikedh/trimesh)

 

- wxpython [4.1.1] (http://www.wxpython.org)

 

- wxnatpy [0.4.0] (https://github.com/pauldmccarthy/wxnatpy/)

 

- xnatpy [0.4.2] (https://bitbucket.org/bigr_erasmusmc/xnatpy)

 

 

Cubic/spline interpolation routines used in FSLeyes are provided by Daniel Ruijters and Philippe Thévenaz, described at http://www.dannyruijters.nl/cubicinterpolation/.

 

 

The GLSL parser is based on code by Nicolas P . Rougier, available at https://github.com/rougier/glsl-parser, and released under the BSD license.

 

 

Some of the icons used in FSLeyes are derived from the Freeline icon set, by Enes Dal, available at https://www.iconfinder.com/Enesdal, and released under the Creative Commons (Attribution 3.0 Unported) license.

 

 

DICOM to NIFTI conversion is performed with Chris Rorden's dcm2niix (https://github.com/rordenlab/dcm2niix).

 

 

The "brain_colours" colour maps were produced and provided by Cyril Pernet

 

(https://doi.org/10.1111/ejn.14430).

 

 

FSLeyes is released under Version 2.0 of the Apache Software License. Source code for FSLeyes is available at https://git.fmrib.ox.ac.uk/fsl/fsleyes/fsleyes

 

 

Copyright 2016-2019 University of Oxford, Oxford, UK.

 

 

--

 

Jon Brooks (PSY - Staff)

 

 

Associate Professor & Head of MRI

 

University of East Anglia Wellcome Wolfson Brain Imaging Centre (UWWBIC)

 

School of Psychology, Room 01.56, Lawrence Stenhouse Building, Chancellors Drive,

 

Norwich Research Park, Norwich, NR4 7TJ, United Kingdom

 

Tel: +44(0)1603 591396

 

Int: 1396

 

A picture containing silhouette Description automatically generatedsignature_4163963870

 

 

UK 14th for Research Quality in Psychology, Psychiatry, and Neuroscience

 

(Times Higher Education rankings for the Research Excellence Framework 2021)

 

World Top 200 (Times Higher Education World University Rankings 2022)

UK Top 30 (The Times/Sunday Times 2022 and Complete University Guide 2022)

UK Top 20 for research quality (Times Higher Education Rankings for the Research Excellence Framework 2021)

World Top 50 for research citations (Times Higher Education World University Rankings 2022)

World Top 50 (Times Higher Education Impact Rankings 2022)

Athena SWAN Silver Award Holder (since 2019) in recognition of advancement towards gender equality for all (Advance HE)

 

 

 


To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=FSL&A=1