Friday, 16 November 2018

Building Tensorflow GPU on Fedora Linux

First off, let's say that there are easy ways of configuring Tensorflow for GPU usage such as using one of the docker images.  However, I'm a bit old school for some things and having always done so I've recently got Tensorflow going on my machine using my GPU.  Tensorflow CPU support is quite easy to do and generally works quite nicely using the pip install method.  GPU support, I've always found, is quite a bit more difficult as there are a whole bunch of things that need to be at just the right level for everything to work i.e. it's quite brittle!

What follows are my notes (it's in the name of the blog) for how to build Tensorflow from scratch to enable GPU support and I do this on Fedora Linux.  If you want to know why it's worth bothering going to this effort, I've tested the Keras MNIST CNN example as a bench mark.  It takes:
  • 11 minutes 7 seconds on my CPU
  • 2 minutes 55 seconds on my GPU
That's just over 3.8 as fast on my GPU as per my CPU so for large jobs this will be huge!

Some info on my machine and config:
  • Lenovo P50 Laptop
  • Intel Core i7-6820HQ CPU @ 2.70GHz (4 core with hyper threading)
  • 32GB RAM
  • Nvidia Quadro M1000M (CUDA compute capability 5.0)
  • Fedora 28 running kernel 4.18.18-200.fc28.x86_64
Install Required Nvidia RPMs
You need to get everything Nvidia and CUDA installed on your machine first.   I quite like the Negativo17 repository for Nvidia on Fedora Linux and so I use this but you could also go with RPM Fusion or even download everything directly from Nvidia.  For me, right now, I have this little lot installed:
cuda-9.2.148.1-2.fc28.x86_64
cuda-cli-tools-9.2.148.1-2.fc28.x86_64
cuda-cublas-9.2.148.1-2.fc28.x86_64
cuda-cublas-devel-9.2.148.1-2.fc28.x86_64
cuda-cudart-9.2.148.1-2.fc28.x86_64
cuda-cudart-devel-9.2.148.1-2.fc28.x86_64
cuda-cudnn-7.2.1.38-1.fc28.x86_64
cuda-cudnn-devel-7.2.1.38-1.fc28.x86_64
cuda-cufft-9.2.148.1-2.fc28.x86_64
cuda-cufft-devel-9.2.148.1-2.fc28.x86_64
cuda-cupti-9.2.148.1-2.fc28.x86_64
cuda-cupti-devel-9.2.148.1-2.fc28.x86_64
cuda-curand-9.2.148.1-2.fc28.x86_64
cuda-curand-devel-9.2.148.1-2.fc28.x86_64
cuda-cusolver-9.2.148.1-2.fc28.x86_64
cuda-cusolver-devel-9.2.148.1-2.fc28.x86_64
cuda-cusparse-9.2.148.1-2.fc28.x86_64
cuda-cusparse-devel-9.2.148.1-2.fc28.x86_64
cuda-devel-9.2.148.1-2.fc28.x86_64
cuda-docs-9.2.148.1-2.fc28.noarch
cuda-gcc-7.3.0-1.fc28.x86_64
cuda-gcc-c++-7.3.0-1.fc28.x86_64
cuda-gcc-gfortran-7.3.0-1.fc28.x86_64
cuda-libs-9.2.148.1-2.fc28.x86_64
cuda-npp-9.2.148.1-2.fc28.x86_64
cuda-npp-devel-9.2.148.1-2.fc28.x86_64
cuda-nvgraph-9.2.148.1-2.fc28.x86_64
cuda-nvgraph-devel-9.2.148.1-2.fc28.x86_64
cuda-nvml-devel-9.2.148.1-2.fc28.x86_64
cuda-nvrtc-9.2.148.1-2.fc28.x86_64
cuda-nvrtc-devel-9.2.148.1-2.fc28.x86_64
cuda-nvtx-9.2.148.1-2.fc28.x86_64
cuda-nvtx-devel-9.2.148.1-2.fc28.x86_64
nvidia-driver-cuda-libs-410.73-4.fc28.x86_64

You might wonder about some of the above, particularly why you might need a back level version of GCC.  When Fedora 28 has a quite capable GCC version 8 why on earth would you want version 7?  The answer lies in my comment about things being difficult or brittle, it's quite simply that CUDA doesn't yet support GCC 8 so you do need a back level compiler for this

Install NVidia NCCL
This library isn't available through an RPM installation or the Negativo17 repository and so you must:
  • Go to the Nvidia NCCL home page 
  • Click the link to download NCCL (requires an Nvidia developer login account)
  • Agree to the Terms and Conditions
  • Download the NCCL zipped tar file that matches your CUDA version (9.2 for this blog post)

At the time of writing the file required is nccl_2.3.7-1+cuda9.2_x86_64.txz

I simply untar this file into /usr/local and create a symbolic link as follows:
  • cd /usr/local
  • sudo tar -xf /path/to/file/nccl_2.3.7-1+cuda9.2_x86_64.txz
  • sudo ln -s nccl_2.3.7-1+cuda9.2_x86_64.txz nccl


Install the Bazel Build Tool
You're going to need a build tool called Bazel which isn't directly available in the Fedora repositories (that I know of at least) but fortunately there's a version in a copr repository you can use as documented run the following commands:
  •  dnf copr enable vbatts/bazel
  •  dnf install bazel
Get a Copy of Tensorflow Source
For this it's just as easy to use git as it is anything else.  You can directly clone the 1.12 release of Tensorflow into a new directory by running:
  • git clone --single-branch -b r1.12 https://github.com/tensorflow/tensorflow tensorflow-r1.12
  • cd tensorflow-r1.12
Simply replace r1.12 in the above commands if you want to use a different Tensorflow release.

Run the Tensorflow Configure Script
This step is actually quite simple but you'll need the answers to some questions to hand, simply run:
  • ./configure
I accept all the default options with the exception of:
  • "location of python" set to /usr/bin/python3 since Fedora still uses Python 2.7 as the default version at /usr/bin/python
  • "build TensorFlow with CUDA support" set to Yes
  • "CUDA SDK version" set to 9.2 (this value should match the cuda version you have installed and at the time of writing 9.2 is the current version from the Negativo17 repository)
  • "location where CUDA 9.2 toolkit is installed" set to /usr
  • "cuDNN version" set to 7.2 (similar to the cuda version above, this value should match the cuda-cudnn package version and 7.2 is the current version from the Negativo17 repository)
  • "NCCL version" set to 2.3
  • "location where NCCL 2 library is installed" set to /usr/local/nccl
  • "Cuda compute capabilities you want to build with" set to 5.0 (but this value should match the CUDA compute capability of the GPU in the machine you're building for)
  • "which gcc" set to /usr/bin/cuda-gcc (to use the back level GCC version 7)


Fix Bazel Config
The above config command writes a file but the location isn't compatible with the latest version of Bazel.  Presumably this issue will be fixed at some point in the future, it's not an issue with Bazel 0.18 and below as far as I'm aware, but has just become an issue on 0.19.  Simply copy the config to the correct place:
  • cat tools/bazel.rc >> .tf_configure.bazelrc
Build Tensorflow with GPU Support
This took around an hour to complete on my machine:
  • bazel build --config=opt --config=cuda //tensorflow/tools/pip_package:build_pip_package
  • bazel-bin/tensorflow/tools/pip_package/build_pip_package /tmp/tensorflow_pkg
The first step is the long one for the build, the second simply builds the python wheel file.

Install Tensorflow with GPU Support
You've got your wheel file so simply install and enjoy:
  • pip3 install tensorflow-1.12.0-cp36-cp36m-linux_x86_64.whl 
Run Some Code
The first time I attempted to run some code to test I got an error:
  • failed call to cuInit: CUDA_ERROR_UNKNOWN
This seems to be some sort of permissions issue and running the following simple script to output the GPUs available on my machine but as root seems to have fixed the above issue i.e. put the following into a script, run that script as root, then any time you want to run code as an unprivileged user the above issue is fixed and the code will work:
from keras import backend as K
K.tensorflow_backend._get_available_gpus()

If the above works then you can try out the Keras MNIST CNN example code.

Monday, 5 November 2018

VueJS Example for IBM App ID

I was recently working on a project in VueJS that needed an authorisation layer added to it.  It turns out there aren't any existing examples of how to do this anywhere, unusually not even on Stack Overflow.  So I set about writing one and thought I would share it.  My work was based upon some other useful examples and information, particularly a blog post from the IBM Cloud blog.

Before I go any further, the code samples are available and documented on GitHub as follows:

  1. IBM App ID API Server
  2. App ID VueJS Client

The code is deliberately split into two such that:
  1. the API Server is used to demonstrate how to secure an API on the server side.  This is done with the WebAppStrategy of App ID which is simply an implementation of a strategy package for passportjs.  The code here isn't anything particularly new over existing examples you can find on the web but it's necessary in order to fully demonstrate the capabilities of the client code.
  2. the VueJS Client is used to demonstrate two things:
    1. how to secure a VueJS route for which I can currently find no example implementations on the web
    2. how to call an API that has been secured by App ID by passing credentials through from the client application to the API server
The API Server should be relatively trivial to get up and running as it's a standard NodeJS API implementation using Express.  If you refer to the WebAppStrategy and the blog post I mention above then you'll see the sample code I've come up with is broadly the same i.e. an amalgamation of the two.

The VueJS Client code can be simple to get up and running as well but it's probably more important to understand how it was created such that you can apply the same principles in your own application(s).  For this then, the explanation is a little longer...

Start by running the VueJS command line client (cli) to create a bare project and for the sample to make sense you will need to add VueX and Router components using the tool:
vue create vue-client
Then understand the 3 modifications you need to make in order to have a working set of authenticated routes.

1. A store for state. 
It doesn't really matter how you achieve this in VueJS, you can use any form of local state storage.  The example code I have come up with uses VueX and a modification to the store.js code you get from the client above.  The idea of this is such that the client application can cache whether the user has already authenticated themselves.  If they have not then the client must request authentication via the server.  If they have, then all the credentials required for making an authenticated call to a server-side API are already available in the browser.  Essentially, this is a speed-up mechanism that stops the client from requesting client credentials on each API call since the session store for the authentication actually lives on the server side when using App ID.

2. A new VueJS Component
This is the component whose route is to be protected via authentication.  In the case of the example code below the standard vue cli "About" component has been used and modified slightly to include an authenticated call to the server API.  The thing to note here is that the credentials from the client side must be sent over to the server with each API call.  Using the fetch API as per the below to implement your GET request means you have to add the credentials: 'include' parameter.

<template>
  <div class="about">
    <h1>This is a protected page</h1>
    <h2>hello: {{ hello }}</h2>
  </div>
</template>

<script>
export default {
  data: function () {
    return {
      hello: undefined
    }
  },
  computed: {
    user () {
      return this.$store.state.user
    }
  },
  methods: {
    getProtectedAPI () {
      fetch('http://localhost:3000/protected/get-some-info',{
            credentials: 'include',
          }).then(res => res.text())
          .then(body => {
            console.dir(body)
            this.hello = JSON.parse(body).hello
          })
    },
  },
  created() {
    this.getProtectedAPI()
  }
} 
</script>

3. A VueJS Navigation Guard
You need to write a function that will be added as a VueJS middleware upon each route change.  The middleware is inserted automatically by the VueJS route code when using the beforeEnter call on a route.  This is known in VueJS as a Navigation Guard.

function requireAuth(to, from, next) {
  // Testing authentication state of the user
  if (!store.state.user.logged) {
    // Not sure if user is logged in yet, testing their login
    const isLoggedUrl = "http://localhost:3000/auth/logged"
    fetch(isLoggedUrl, {credentials: 'include'}).then(res => res.json()).then(isLogged => {
      if (isLogged.logged) {
        // User is already logged in, storing
        store.commit("setUser", isLogged)
        next()
      } else {
        // User is not logged in, redirecting to App ID
        window.location.href=`http://localhost:3000/auth/login?redirect=${to.fullPath}`
      }
    }).catch(e => {
      // TODO: do something sensible here so the user sees their login has failed
      console.log("Testing user login failed - D'oh!")
    })
  } else {
    // User already logged in
    next()
  }
}

The requireAuth function does the following in plain English:

  1. Using the VueJS client side cache, test if the user is already logged in
  2. If they are not. then ask the server if the user is already logged in
    1. If they are not, then redirect them to the server login page
    2. If they are, then cache the information and load the next piece of middleware
  3. If they are, then simply load the next piece of middleware


Each route you want to protect with the above function must have a beforeEnter: requireAuth parameter specified on the route.  When this is done, VueJS will call the requireAuth function before the component specified by the route is loaded.

{
  path: '/protected',
  name: 'protected',
  beforeEnter: requireAuth,
  component: Protected
}

Note: there are methods by which you don't have do call window.location.href to redirect the user to the login page (which does seem like a bit of a nasty hack.  However, these methods require the modification of the webpack configuration and so were kept out of scope of this example for the purposes of being simple.

Monday, 4 June 2018

South Downs Way Walk

I've just finished a pretty extraordinary journey, both physically and mentally, here's the story...

The Short Version - 100 Miles, 4 Days, 2 Charities

This bit of the blog post is a summary for the TL;DR brigade...
  • Starting in Eastbourne on 31st May
  • More or less 4 marathons in 4 days
  • Finishing in Winchester on 3rd June
  • Raising funds for:
    • The National Eczema Society
    • Parkinson's UK
  • You can still donate (please do):
  • We raised over £4500 (more than £5500 including gift aid)

I have made all the pictures available on Flickr in my South Downs Way Walk Album.


The Longer Version

My best mate decided he wanted to challenge himself as a consequence of coming to terms with his 40th birthday.  He called a few of us to the pub a few months back.  We settled on the idea of walking the South Downs Way in 4 days for 2 charities.  I was nuts enough to go along with this idea and I'm writing the morning after successfully finishing the entire route with him.

All done and in the restaurant at the finish
In the car on the way to get started...
The people you need to know about in this story are:
  • Matt Wettone - team leader
  • Me (obviously)
  • Andy McGrath - team super hero
  • Pete and Phil - team members
  • Stephen Warwick - honorary team member
  • Tim - Matt's dad, team logistics
  • Linda - Matt's mum, team chef

All done and in the restaurant at the finish
...all done and in the restaurant at the finish

The reason for doing this has already been much better explained by Matt than I'll manage to come up with so if you're interested in that then head over to our donation page or I've saved a copy of his text at the bottom of this post.  My personal reasons are fairly simple in as much as I'd find it hard to turn Matt down, doing something for charity is always worthwhile and challenging yourself every now and then has to be worth something as well.

It turns out that it certainly was a challenge both physically and mentally.  I'll try to be brief below with a summary of each day...

Day 1 - Eastbourne to Housedene Farm

Vital statistics: 27.4 miles in 8 hours 26 minutes moving time (3.25 MPH, moving average), about 58,000 steps and 3968 feet of elevation gain.


We planned Day 1 to be 24 miles long but ended up doing 27.4 miles.  This was due to starting further back than the documented start of the South Downs Way, we started at Eastbourne Pier, another 1.5 miles along the coast.  Also, the distance charts for the Eastern half of the walk seem to be about 10% out in terms of accuracy so we picked up another couple of miles we weren't expecting.

Matt, Andy and I left Eastbourne in heavy rain.  I think we were all excited and apprehensive but keen to get started.  After the first couple of miles, we climbed up onto the Seven Sisters and for the next 10 miles or so found ourselves repeatedly going up and down each of the 7 cliffs.  That was really hard work, especially in poor weather, and set the tone for the rest of the day as being a really tough day.

We pretty much dodged a bullet with the weather though, the rain cleared up by around 11am and it became overcast and warm for the remainder of the day, perfect walking weather.  The rest of the country seemingly got drowned in thundery showers that day so we were really lucky.

We finished the day in good spirits but with Andy starting to struggle a bit with a knee problem and having picked up a blister along the route somewhere.

Day 2 - Housedene Farm Camp Site to Amberley

Vital statistics: 29.7 miles in 9 hours 5 minutes moving time (3.23 MPH, moving average), about 62,000 steps and 3186 feet of elevation gain.



This route was also longer than the planned 26.5 miles we had in mind for day 2.  We picked up the trail where we left off the previous evening at Housedene Farm and walked to Amberley.  Pete joined us, meeting at the start point in the morning and intending to walk the next 2 days.

Andy put in a heroic effort just to be ready for the start.  From the previous day, all of us thought there was a real danger he may have to pull out for at least day 2, but somehow he managed to be ready and start walking with us again.

The weather was decent enough for walking in as much as it wasn't sunny and the temperature was probably mid teens.  However, we basically spent the entire day walking through cloud.  The most we saw all day was dew drops forming in our hair from all the moisture in the air.  The dampness did make it a little tougher but we found the biggest drawback of this weather was the lack of reward for climbing some enormous hills.  For example, we walked the length of the south rim of Devil's Dyke and none of us have the foggiest (pun intended) idea what it looks like.  Visibility was between 20 and 100 yards all day!

At about 18 miles Pete wasn't able to continue, having aggravated an old knee injury.  We decided it was best for him to leave the trail and get picked up early so he headed to a nearby road for rescue.  At that point, Andy took the tactical decision to accompany Pete on the journey home to give himself more time and a better chance of recovery to finish the route, leaving just Matt and I.  This was definitely one of the low points for all of us in the team that day.

Matt and I soldiered on through the cloud for another 11-12 miles after that.  We were due to be picked up at the agreed point but due to a traffic accident and resulting traffic chaos, Tim was unable to meet us until 40 minutes or so after our agreed meet time.  Matt and I decided to add some more miles to make day 3 a bit shorter and easier.


Day 3 - Amberley to Queen Elizabeth Country Park

Vital statistics: 25.2 miles in 8 hours 3 minutes moving time (3.13 MPH, moving average), about 54,000 steps and 3287 feet of elevation gain.


Day 3 was a huge mental struggle for me.  We had lost Pete the day before, Andy was with us again and struggling through.  But the main problem was having eaten fish and chips for dinner the night before and not slept a wink resulting from the indigestion that followed, I was both extremely tired and feeling rather nauseous as well.  I kept going, putting faith in my physical ability (physically I felt fine), force fed myself as many calories as I could (I really didn't want to eat) and by around 2pm was starting to feel a little better.

There were a huge number of highlights on Day 3 that really helped:

  • We were joined in a fairly last minute addition to the plan by Stephen Warwick.  It was lovely to have someone new in the team with fresh legs and a different dynamic.  Stephen's used to lengthy exercise having walked some trails in the Himalayas and also ridden from Lands End to John O'Groats.  He really helped me get my head in the right place to soldier on with some good team talks and deliberately poor maths.  "We're half way" he said at 11 miles.  "No we're not, go back and re-sit your GCSE" I replied knowing full well that we'd be the wrong side of 25 miles by the time we'd finished the day.
  • The weather was much, much better.  We could see.  It was warm, not hot.  That was lovely.
  • We'd previously walked from Buriton to Didling during training, about the last third of Day 3.  So there was a point when we reached the part of the trail we knew.  That was a huge moment for all of us.  We stopped there and celebrated with a banana each (we know how to live).  Matt and I had now walked to the point where we knew we'd walked the entire length of the remaining trail into Winchester during training and most of that done in a day.  We started to believe this might actually be possible.
  • Amy, Matt's sister, joined us at Harting Down and walked the last 8 miles or so with us.  That was another huge boost to us - thank you, Amy!
  • Linda also joined us for the last 2 miles into Queen Elizabeth Country Park.  Since she's suffering with Parkinson's, I thought that it was particularly poignant for her to join us.  I'm really glad she did. 
  • Donations were still coming in, we'd had a particularly good donation day on Day 2, and for those people who decided to wait until we started walking - many thanks - you helped keep us going and our spirits up.
The last few miles were really tough for Andy.  Carrying his injury, he hobbled along to the finish, doubtful about whether he could complete Day 4.  Testament to Andy though, he sat down in a disabled parking space in the car park, causing much hilarity.  Other than in quite a lot of pain, I'm not sure exactly how he was feeling, but I imagine very proud of getting this far but also worried about the potential disappointment of not being able to finish the whole thing on Day 4.


Day 4 - Queen Elizabeth Country Park to Winchester

Vital statistics: 24.1 miles in 7 hours 12 minutes moving time (3.35 MPH, moving average), about 51,000 steps and 1972 feet of elevation gain.


The start of day 4 was met with mixed feelings.  We had all stayed at home the previous night, our first night home for the previous 3 nights.  Seeing family and sleeping in your own bed was great.  However, Andy was unable to recover from the previous day in time and we had a message saying that he wasn't going to be able to join us for the final day.  That must have been a very hard decision to take but you'd never know it, having been delivered with the sense of humour and good grace Andy had shown throughout the walk.  We'd miss him but we did have Phil joining me and Matt with his fresh legs for the final day.

Matt and I were buoyed by the fact we'd walked the entire length of this day during training and done so in a single day.  That feeling of confidence and the fact we knew where we were going and exactly what lay ahead at each stage of the walk certainly helped.  It made things a lot easier when we were getting tired during the day.  It's probably due to a mixture of factors but we walked a little faster on day 4 too.

Andy parked in Winchester and walked backwards along our route to meet us and walk into town the last few miles.  Still struggling, that must have been a hard walk for him too but it was great to cross the line with the main team of 3 of us and Phil who joined for the final day.

We arrived in Winchester a bit earlier than any of us might have expected.  But with due warning, friends and family were dotted throughout the last couple of miles to meet and greet us.  That felt amazing, picking up more people as we got closer to the end point.  We were clearly going to make it, feeling extremely tired, but unstoppable at the same time.

We went for food and drink.  Left the building to be driven, a long way home but a shorter distance than we'd walked each day, that felt weird.

We're done.  This is now history.  But please consider a donation, if you haven't already.


Matt's Original Text

"Thank you so much for visiting my page. My name is Matt and this year I will be turning the grand age of 40. I was inspired to take on a challenge to mark the occasion and hopefully do some good while I'm at it. 
On 31st May I will be walking the South Downs Way. This is a walk that is 100 miles long and I will be attempting to cover it in 4 days. This means covering something close to a marathon each day. To help me do this I have put together a team of close friends. They will be joining me for either all (Graham & Andy) or part (Phil and Pete) of the walk. 
Together we are raising money for The National Eczema Society and Parkinsons UK. Here's why... 
I was diagnosed with atopic eczema aged just 3 months and I still struggle with the disease to this day. With the help of my parents, my family, close friends, some incredible doctors and some equally astounding nurses, my skin is in a far better condition that it ever was as a child. In July of 2017 my wife and I welcomed our second child into the world. Sadly he too has been suffering with eczema. 
The work that the National Eczema Society do is vitally important in the development of new treatments, drugs and support for sufferers of this condition. I am passionate to find something that not only helps my life but helps the generations to come (including my son). 
In September 2017 my Mum was diagnosed with Parkinson's. Prior to the diagnosis as a family we had seen her deteriorating. She was walking much slower than normal, she dragged her feet, her right arm developed a tremor, her speech was much quieter and she wasn't herself. 
My sister got married in early September and while helping to setup the day before Mum tripped on a kerb and broke her arm. Having recognised she wasn’t herself, Mum had been having various tests. A cardiologist she’d been sent to contacted the GP to say he thought Mum should be referred to a neurologist because he thought she had Parkinson’s disease. It is very likely that not picking her feet up (due to Parkinson's) caused her to trip and fall. 
Since then she has been given a medication, seen a Parkinsons nurse and a physiotherapist. This has meant a significant improvement and she is now standing much taller (quite a significant given how 'tall' she is), she is speaking more freely, she is interacting with all her grandchildren with a vibrancy that was missing. 
The work that Parkinsons UK do is vital for people with this disease. They provide invaluable support and research that could not be possible without the support of donors. 
I would be so grateful if you would like to sponsor my team and I as we embark on this challenge and help raise some money for some fabulous charities. 
Don’t forget to gift aid your donation so we can get even more from the Taxman."

Last, but very much not least, a great debt of thanks to everyone who has supported us.  Those of you donating your hard-earned cash to our two fantastic causes, thank you.  To our wives and families who we've abandoned to fulfil our own personal challenge, you're tired too from holding the fort while we've been gone, thank you.  Finally, particular thanks to Linda and Tim for the logistical support, you were a part of the team as well and we wouldn't have completed the journey without you.