project-rig / nengo_spinnaker

87%

master: 81%

LAST BUILD ON BRANCH change-fixpoint

branch: change-fixpoint

CHANGE BRANCH

x

Reset

• change-fixpoint
• alternative-ptn-removal-order
• async_mode
• atexit
• atomic_filters
• auto-allocate-key-ids
• auto-allocate-key-ids-advanced
• better-filter-usage
• builder
• c-cont-arrays
• c-optimisations
• calloc-sdram
• config-setup-script
• deque-to-hold-sinks
• dev-0.3.0
• dev0217.2
• dev0716
• dev1016
• dev1016-with-packet-queue
• dev1016-with-packet-queue2
• dev1116
• ethernet_io
• faster-graph-colouring
• faster-netlist-build
• filter-when-making-signals
• fix-#15
• fix-#18
• fix-#22
• fix-#4
• fix-#47
• fix-#59
• fix-#6
• fix-#70
• fix-None-filter-test
• fix-building
• fix-cache-use
• fix-clique-finding
• fix-divide-slices
• fix-ensemble-sample
• fix-input-to-node
• fix-linear-filter-packing
• fix-malloc-or-die
• fix-multiple-filters-ensemble
• fix-parallel-filter
• fix-params
• fix-placement
• fix-readme
• fix-signal-parameter-identity
• fix-simulation-time-bug
• fix-simulation-time-bug2
• fix_ensemble
• fix_recording_sliced
• full-realtime-sim
• function-of-time-nodes
• further-simulation-time-fix
• graph-colouring-cluster-ids
• improve-app-ptr-table-construction
• improve-build-time2
• improve-filter-routing
• interposer-changes-rebased
• interposer_pipelining
• interposer_pipelining-with-jitter
• interposers-after-ensembles
• intrinsics
• list-region
• master
• mcplayer
• mega_hacking
• merge-trees
• merged_spinnaker_tools
• minimise-ensemble-filter-routing-regions
• minimise-filter-routing-regions2
• more-filter-parallelism
• more-filter-parallelism-with-FIQ-hack
• multisource-nets
• neaten-booting
• neaten-f_of_t-tests
• neaten-filter-operator
• neaten-scp-packet
• netlist-optimisations
• neural-model
• neuron-limit
• new-ensemble
• new-filter-opt
• new-filters
• new_builder
• new_ensemble-new_learning
• no-repeat-constants
• no-zero-packets
• nodes-to-global-inhibition
• numpy_1_8_fix
• optimize-connections
• packet-jitter
• packet-queue-performance-improvements
• packet-queues
• parallel-filter
• parallel-filter2
• parallel-probe
• partition-probes
• partitioning-fix
• pes
• pes-squashed
• pickle
• probe_spikes
• processes
• profiling
• profiling-utils-fix
• profiling_new
• prune-filters-with-no-output
• py3-hash-fix
• realtime-simulator
• record-voltages
• recording_buffer_reset
• reduce-ensemble-transmission-pars-memory
• reduce-key-allocation-mem-usage
• reduce-matrix-costs
• reduce-value-source-bursts
• refactor-transmission-parameters
• reinject-dropped-packets
• remove-interposers-with-no-input
• remove-passnodes
• remove-passnodes-3
• remove-passthrough-nodes
• remove-sourceless-ptns
• remove_slots
• revert-83-calloc-sdram
• revert-87-new-ensemble
• rig-1.0.0
• slight-rework-of-filters
• spalloc-fix
• speedier-connection-map
• spikes
• testing
• the-spaun-branch
• the-spaun-branch3
• timescaling
• update-passthrough-node-removal
• update-rig-api
• use-less-keyspace
• use-nengo-builder
• use-new-rig-api
• use-sark-tag-ptr
• use-spalloc
• use_rig_cpp_key_allocation
• v0.2.0
• v0.2.1
• v0.2.2
• v0.2.3
• v0.2.4
• v0.2.5
• v0.3.0
• v0.4.0
• voja
• zero-dependency-netlist
• zero-tau-synapse

Build # 873

Build Type

push

travis-ci

Committed by mundya

Commit Message

Change to S12.19 fixed point representation

This changes all the fixed point representations to S12.19 format. The
extra 4-bits fractional representation should help some larger neuron
models at the cost of very large decoders becoming saturated.

All the regression tests continue to pass, so this range seems
reasonable.

Neuron voltages are still recorded in U1.15 format, hopefully the ARM
barrel shifters should mean that there's not an additional cycle cost in
recording voltages.

Run Details

4488 of 5137 relevant lines covered (87.37%)

0.87 hits per line