CHANGES

* 2019.01.31 - updated all platform.* from cKnowledge.org

* 2018.11.19 - updated platforms

* 2018.09.26 - added new platforms participated in crowd-benchmarking and crowd-tuning

* 2018.07.14 - added platform.npu

* 2018.06.23 - new set of platforms (with the release of ACM ReQuEST proceedings and new Android app)

* 2018.03.13 - new set of platforms

* 2017.11.11 - new set of NN platforms

* 2017.11.04 - new set of platforms participating in collaborative AI/SW/HW co-design

* 2017.09.04 - new set of platform descriptions during collaborative optimization

* 2017.07.16 - new set of platform descriptions during crowd-tuning and SW/HW co-design (AI, compilers, OpenCL, CUDA, etc)

* 2017.05.07 - new set of platforms for collaborative optimization

* 2017.04.20 - new set of platforms which also participated in DNN crowd-benchmarking
               on mobile devices with constraint resources, desktops and HPC servers
               (Caffe and TensorFlow engines)

* 2017.01.15 - added costs of various hardware as features 
               (see our CPC'15 paper on performance and cost-aware computer engineering
                https://arxiv.org/abs/1506.06256)

* 2017.01.14 - added meta of new platforms

* 2016.12.18 - added new platforms (CPU, GPGPU, GPU, OS, etc)

* 2016.11.02 - added new platforms 

* 2016.10.13 - added new platforms after crowd-tuning and new ARM WA crowd-benchmarking

* 2016.09.15 - added new platforms after crowd-tuning and crowd-benchmarking

* 2016.08.17 - added new platforms after crowd-tuning and crowd-benchmarking

* 2016.06.26 - added Samsung Galaxy S7

* 2016.05.29 - added new platforms collected in CK repo (now in Azure)
             - added Nvidia GeForce GTX 980 (Maxwell architecture) in Dell PowerEdge R730

* 2016.05.13 - added new GPGPU platforms (preparing OpenCL/CUDA crowdtuning)

* 2016.04.30 - added new platforms from GCC/LLVM and OpenCL crowdtuning

* 2016.03.25 - added new platforms from GCC/LLVM crowdtuning

* 2016.03.08 - added new platforms info from collaborative program optimization using Android-based mobile devices

* 2016.02.25 - added new platforms from GRID5000 (during GCC 6.0.0 / LLVM 3.9.0 crowdtuning)

* 2016.02.17 - changed platform.accelerator to platform.gpu (to reflect what is detected correctly)