Catalyst Neurocore

Neuromorphic processor architecture — 128 cores, 131K neurons, full Loihi 2 feature parity, 90.7% SHD benchmark (beats Loihi 1).

Two generations of neuromorphic silicon. Validated on real FPGA hardware. Accessible via cloud API or dedicated dev boards.

---

What is Catalyst?

Catalyst is a neuromorphic processor family designed for energy-efficient spiking neural network inference and on-chip learning. The architecture runs hardware-accurate LIF neuron dynamics, programmable synaptic plasticity, and dendritic computation — all at a fraction of the power consumed by conventional GPUs.

Two ways to use it:

• Catalyst Cloud — REST API for neuromorphic simulation. No hardware, no install. Free tier available.
• FPGA Dev Boards — Physical hardware with the Catalyst bitstream. Deploy at the edge.

---

Architecture at a Glance

Feature	Catalyst N1	Catalyst N2
Cores	128	128
Neurons/core	1,024 (CUBA LIF)	1,024 (programmable microcode)
Total neurons	131,072	131,072
Synapses/core	131K (CSR compressed)	131K (CSR compressed)
State precision	24-bit fixed-point	24-bit fixed-point
Dendritic compartments	Yes (tree topology)	Yes (tree topology)
Spike traces	Dual (pre/post)	Dual (pre/post)
Graded spikes	Yes (Loihi 2 feature)	Yes
Programmable delays	1-63 timesteps	1-63 timesteps
Learning engine	16 registers, 14 opcodes	16 registers, 14 opcodes
STDP	Yes	Yes
3-factor reward learning	Yes	Yes
Homeostatic normalization	Yes	Yes
Stochastic threshold noise	Per-neuron LFSR	Per-neuron LFSR
Synapse encodings	Sparse, Dense, Population	Sparse, Dense, Population + Convolutional
Neuron models	CUBA LIF	CUBA, Izhikevich, Adaptive LIF, Sigma-Delta, Resonate-and-Fire
Embedded processors	Triple RV32IMF RISC-V	Triple RV32IMF RISC-V
Loihi parity	Loihi 1	Loihi 2

Catalyst matches or exceeds all Loihi functional features (neuron models, learning engine, compartments, graded spikes, delays, noise, synapse formats). The "parity" designation refers to architectural feature equivalence, not identical physical specifications.

---

Validation

Metric	Value
SDK test suite	3,091 tests
Feature coverage	155 total (152 FULL, 3 HW_ONLY)
FPGA validation	28/28 pass (AWS F2, Xilinx VU47P, 62.5 MHz)
RTL testbenches	25 (98 scenarios, 0 failures)
SHD benchmark	90.7% (adLIF) / 85.9% (LIF baseline)

---

Benchmarks

Full benchmark suite: catalyst-neuromorphic/catalyst-benchmarks — clone, train, deploy, reproduce.

Benchmark	Classes	Architecture	Neuron	Float Acc	vs Loihi
SHD	20	700→1024→20 (rec)	adLIF	90.7%	Beats Loihi 1 (89.0%)
SSC	35	700→1024→512→35 (rec)	adLIF	72.1%	Beats Loihi 2 (69.8%)
N-MNIST	10	Conv2D+LIF→10	LIF	99.2%	—
GSC KWS	12	40→512→12 (rec, S2S)	adLIF	88.0%	—
DVS Gesture	11	—	—	in progress	—

All models trained with surrogate gradient BPTT, deployed to Catalyst FPGA hardware with int16 quantization.

# Reproduce any benchmark
git clone https://github.com/catalyst-neuromorphic/catalyst-benchmarks.git
cd catalyst-benchmarks && pip install -e .
python shd/train.py --neuron adlif --hidden 1024 --epochs 200 --device cuda:0 --amp

Simulation Performance

Backend	1K neurons, 1K timesteps	32K neurons, 10K timesteps
CPU	0.8s	45s
GPU (RTX 3080)	0.01s	0.4s
FPGA (VU47P)	0.016s	0.5s

---

Feature Parity Scorecard

Full comparison against Intel Loihi 1 and Loihi 2:

Category	Loihi 1	Loihi 2	Catalyst
LIF neuron model	Yes	Yes	Yes
Dendritic compartments	Yes	Yes	Yes
Programmable learning	Yes	Yes	Yes
STDP	Yes	Yes	Yes
3-factor learning	Yes	Yes	Yes
Graded spikes	No	Yes	Yes
Stochastic rounding	No	Yes	Yes
Homeostasis	Yes	Yes	Yes
Programmable delays	Yes	Yes	Yes
Embedded processors	3x LMT	6x LMT	3x RV32IMF
Score	10/10	10/10	10/10

---

Try It Now

Catalyst Cloud API

Neuromorphic compute as a service. Define a network, submit a job, get spikes back.

import catalyst_cloud as cc

# Sign up (once)
account = cc.Client.signup("you@lab.edu")
print(account["api_key"])  # Save this

# Create a client
client = cc.Client("cn_live_...")

# Define a network
net = client.create_network(
    populations=[
        {"label": "input", "size": 100, "params": {"threshold": 1000}},
        {"label": "hidden", "size": 50},
    ],
    connections=[
        {"source": "input", "target": "hidden", "topology": "random_sparse",
         "weight": 500, "p": 0.3},
    ],
)

# Run simulation (blocking)
result = client.simulate(
    network_id=net["network_id"],
    timesteps=1000,
    stimuli=[{"population": "input", "current": 5000}],
)

print(result["result"]["firing_rates"])

pip install catalyst-cloud

• Free tier: 10 jobs/day, 1,024 neurons, no credit card
• Paid tiers: Higher limits, priority compute, dedicated support
• Interactive demo: HuggingFace Spaces

Cloud Pricing | API Docs

FPGA Dev Boards

Physical hardware with the Catalyst bitstream. Planned for Crowd Supply.

---

Support Development

Back independent neuromorphic silicon development via GitHub Sponsors. Sponsors get compute credits for the Catalyst Cloud API, priority access to hardware, and early access to new features.

---

Papers

Catalyst N1: A 131K-Neuron Open Neuromorphic Processor with Programmable Synaptic Plasticity and FPGA Validation

Henry Arthur Shulayev Barnes, University of Aberdeen

Catalyst N2: Full Loihi 2 Feature Parity in an Open Neuromorphic Processor with Programmable Neuron Microcode and Cloud FPGA Validation

Henry Arthur Shulayev Barnes, University of Aberdeen

N1 Paper (Zenodo) | N2 Paper (PDF)

---

Contact

• Email: henry@catalyst-neuromorphic.com
• Website: catalyst-neuromorphic.com
• Cloud API: catalyst-neuromorphic.com/cloud

---

Built by one person. 3,091 tests. Loihi 2 feature parity. Beats Loihi 1 on SHD, beats Loihi 2 on SSC.