ATOM™

5nm Versatile Inference SoC

Versatile yet Energy Efficient
AI System-on-Chip

Description

Worldclass Inference for Edge and Cloud Computing

Standing on par with industry’s leading competitors, ATOM™ delivers top performance across various types of AI tasks like computer vision, natural language processing and recommendation models. Utilizing the silicon-proven neural core ION™ as the compute granule that scales up with perfect linearity, ATOM™ is the optimal chip for large-scale inference required in edge computing and data centers.

Bringing Latency-Critical AI Tasks to Hyperscale

Scalable, energy-efficient, and fast: ATOM™ is poised to be a pivotal enabler for server-level AI services, optimizing the AI-as-a-Service (AIaaS) stack. 
As a multi-core System-on-Chip (SoC), ATOM™ is built upon a dataflow architecture to deliver superior inference performance. Supported by SR-IOV-based user-level parallelization, our proprietary multi-level dependency control mechanism minimizes latency overhead for inter- and intra-chip data orchestrations. ATOM™’s high system utilization rate is enhanced by up to 40% through the seamless interplay of our hardware, software, and firmware.

Advanced Components

Samsung Foundry’s advanced EUV process node enables ATOM™ to deliver improved performance, reduced power consumption and enhanced energy efficiency. Geared with PCIe Gen5, GDDR6 and high-speed I/O, ATOM™ is the optimal AI accelerator to serve different markets, spanning from edge computer to data centers.

Performance

MLperf™

Inference v3.0, Single Stream

At MLPerf™, ATOM™ showed impressive results across both vision and language models, outperforming Qualcomm and NVIDIA, with 0.24ms in ResNet50 Single Stream and 4.30ms in BERT-Large Single Stream, respectively.

Vision : ResNet50

  • Rebellions ATOM™
    1.0x 0.239ms
  • Qualcomm Cloud AI100
    1.4x 0.336ms
  • NVIDIA A2 (Ampere)
    3.0x 0.713ms
  • NVIDIA T4 (Turing)
    3.4x 0.818ms

* Shorter the faster

Language : BERT-Large

  • Rebellions ATOM™
    1.0x 4.297ms
  • Qualcomm Cloud AI100
    1.8x 7.547ms
  • NVIDIA A2 (Ampere)
    2.0x 8.506ms
  • NVIDIA T4 (Turing)
    1.4x 6.093ms

* Shorter the faster

Key features

ION™ Core
Architecture

ATOM™ is built on the small yet powerful silicon-proven AI core ION™, which, with its Coarse-Grained Reconfigurable Architecture (CGRA), is flexible, programmable, and scalable. This design allows ATOM™ to handle networks with varying depths and complexities, from small applications to hyperscale services, all while maximizing energy efficiency.

System-on-chip
Architecture

ATOM™ is a multi-core System-on-Chip, consolidating all essential components onto a unified substrate. The communication overhead within the chip is masterfully handled via on-chip data dependency handling.

Supporting
Networks

ATOM™ can accelerate different types of neural networks, including Convolutional Neural Networks (CNNs), Long Short-Term Memory (LSTM), Bidirectional Encoder Representations from Transformers (BERT) and recent transformers-based models (e.g. T5, GPTs, etc.).

Multi-Instance

ATOM™ can be partitioned to support up to 16 separate jobs simultaneously, with each job isolated at both the hardware and software levels. ATOM™’s Multi-Instance feature allows higher resource usage and flexibility, optimizing utilization and performance.

System specs

Single Chip
FP16
32 TFLOPS
INT8
128 TOPS
On-chip SRAM
64 MB
Max Thermal Design Power (TDP)
60 - 130W (Configurable)
Memory
16 GB, GDDR6 (ECC enabled), 256 GB/s
Host / Chip-to-Chip Interface
PCIe Gen5, 64 GB/s
Multi-Instance NPU
HW isolation up to 16 independent tasks

See also...

System Solutions

Start Lean, Scale Green

Explore System Solutions
markets-ion

ION™

Poised to Revolutionize the High-Frequency Trading Market

In the high-frequency trading (HFT) market, a microsecond can make a huge difference. ION™ was specifically tailored for this finance industry. Within just one year of inception, Rebellions introduced this powerful chip to the market, marking a significant milestone in our pursuit of technological excellence.
Explore ION™