El ESWIN EIC7700X is a high-performance quad-core RISC-V architecture-based SoC designed for artificial intelligence applications. on the edge. It offers an impressive feature set that makes it an ideal choice for a wide range of applications, and with a powerful NPU to accelerate AI loads.
The EIC7700X It is perfect for industrial quality inspection tasks, such as the detection of defects in manufactured products. Its high performance and AI capabilities allow it to process large amounts of image and video data quickly and accurately, for machine vision equipment. Likewise, it also has the ability to identify at a facial level, to recognize human faces, something vital in user authentication applications, surveillance or tracking of people, etc.
LLMs (Large Language Models) are large language models that are trained on massive data sets of text and code. The EIC7700X supports high-precision LLM, making it an ideal tool for applications such as text generation, machine translation, and question answering.
It can also be used for recognize behavioral patterns in video data, such as human behavior in public environments. This makes it a valuable tool for security and surveillance applications. It can even be used to classify objects and data automatically, allowing you to automate, for example, updating a company's inventory.
On the other hand, it can function as an M.2 module or as a stand-alone SoC module. In both cases, you can have Ubuntu 18.04 or CentOS 7.4 operating systems, as well as Linux 5.17 or Linux 6.6 with its own SDK respectively. And, the integrated NPU is compatible with different frameworks such as Pytorch, Tensorflow, PaddlePaddle, ONNX, etc., and with high-precision LLM.
RISC-V chip technical specifications
The SoC, which is expected to appear in the coming months, has the following Technical specifications:
- CPU
- 4x high-performance SiFive P550 RV64GC RISC-V cores @ 1.4GHz (up to 1.8GHz) similar to Cortex-A75 performance
- 1KB L32 cache for instructions + 32KB for data (per core)
- 2KB L256 cache per core
- 3MB L4 cache shared by all cores
- Support for ECC (SECDED) in cache
- NPU
- DNN accelerator with up to 19,95 TOPS (INT8)
- DSP for vision
- DSP with support for 512 INT8 SIMD
- Multimedia Decoder/Encoder
- HEVC (H.265) and AVC (H.264)
- H.265 up to 8K @ 50fps or 32 channels of 1080p30 video decoding
- H.265 up to 8K @ 25fps or 13 channels of 1080p30 video encoding
- JPEG ISO/IEC 10918-1, ITU-T T.81, up to 32Kx32K
- Vision Engine
- HAE (2D Blit, Crop, Resize, Normalization)
- 3D GPU (with support for OpenGL-ES 3.2, EGL 1.4, OpenCL 1.2/2.1 EP2, Vulkan 1.2, Android NN HAL graphics APIs)
- OSD (3-layer)
- audio codec
- AAC-LC encoding
- Decodificación G.711/G.722.1/G.726/MP2L2/PCM/MP3/AAC-LC
- RAM
- Up to 32GB 64-bit LPDDR 4/4x/5
- Storage interface
- Support for eMMC 5.1 memory, 2x SDIO 3.0, SATA III (6Gb/s), SPI NOR flash
- video I/O
- HDMI 2.0 output with HDCP1.4/2.1 support and four MIPI-DSI TX lines
- MIPI DPHY v2.1 and CPHY v1.2 Sub LVDS/SLVS input or 6 camera inputs. As well as four MIPI D-PHY/2-Trio C-PHY lanes up to 2.5Gbps/lane, and another four LVDS/Sub-LVDS/HiSPi lanes up to 1.0Gbps/lane
- Peripheral interfaces
- 2x USB 3.0/2.0 (DRD) ports
- 4x PCIe 3.0 lanes (RC+EP)
- 2x GMAC with support for RGMII (GbE)
- 12x I2C @ 1Mbps, 5x UART, 2x SPI
- 3x I2S (slave + master)
- Security
- Support TEE, TRNG, ECDSA, RSA4096, AES, SM4, DES, HMAC, CRC32
- Dual core dedicated to security to speed up encryption, etc.
- 16KB OTP
- Energy consumption
- 8W with CNN
- Packaging available
- FC-CSP with 17x17mm
- FC-BGA with 23x23 mm
- Operating temperature range
- From -20°C to +105°C