Jingjia Micro JM9271
Product Overview
The JM9271 is the higher-positioned model in Changsha Jingjia Microelectronics' (SZSE 300474) JM9 series. Under the company's 2018 private-placement funded project plan, the JM9 series comprises two chips, the JM9231 and the JM9271: the JM9231 targets domestic office PCs, portable computers, mid-to-low-end entertainment devices, and high-end embedded systems; the JM9271 focuses on strengthening scientific computing and data processing capabilities, aimed at high-end application areas with demanding compute speed requirements such as artificial intelligence, security surveillance, speech recognition, deep learning, and cloud computing.
On 2021-11-16, Jingjia Micro announced that the JM9 series graphics processing chips had completed tape-out, packaging, and initial testing. In May 2022, the company announced that the second JM9 series graphics processing chip had completed development, a model targeting desktops, servers, industrial PCs, self-service terminals, and similar equipment — the industry generally maps this "second chip" to the JM9271, but Jingjia Micro did not use the JM9271 name in its announcements, nor did it separately disclose the complete specifications of that model.
In terms of actual market deployment, a 2026 industry review article describes the JM9271 as a representative product of Jingjia Micro's JM9 series, stating that it uses the self-developed JM9 graphics architecture, supports graphics APIs such as Vulkan / OpenGL and 4K ultra-HD display output, has graphics rendering capability in the domestic first tier, and was paired with KylinSec in a fully domestic GPU virtualization solution deployed in critical infrastructure such as power grid dispatch. As of March 2026, Jingjia Micro confirmed that the JM9 series has achieved volume delivery, the fourth-generation JM11 series has launched with small-volume shipments, and the company continues to advance R&D on "high-performance general-purpose GPU chips" and next-generation graphics processing chips.
⚠️ Nature of the Specs on This Page (Must Read) Jingjia Micro has never officially released a complete specification sheet for the JM9271 in the form of a formal announcement. The fields marked "rumored" in the core specifications below come from 2019—2021 media reports (MyDrivers / Tom's Hardware) and securities research reports, representing early design targets or secondhand accounts, with contradictions among different sources (see the "Source Conflicts" subsection). Jingjia Micro representatives have publicly responded that the JM9 series and the GTX 1080 "differ in market position, application characteristics, and focus, and are not comparable", so "comparable to the GTX 1080" should be treated as a media framing rather than an official positioning.
Core Specifications
| Parameter | Value |
|---|---|
| Architecture | JM9 self-developed GPU architecture (Jingjia Micro's third-generation GPU) |
| Process Node | Not disclosed (media reports say 14nm; unconfirmed officially) |
| FP16 / BF16 Compute | Not disclosed |
| INT8 Compute | Not disclosed |
| FP32 Compute | 8 TFLOPS (rumored, not official) |
| Pixel Fill Rate | > 128 GPixel/s (rumored) |
| Core Clock | > 1,800 MHz Boost (rumored) |
| Memory Capacity | 16 GB (rumored; a 2026 source claims 8GB, see conflict note below) |
| Memory Type | HBM (rumored; another source says GDDR6) |
| Memory Bandwidth | 512 GB/s (rumored) |
| TDP | 200 W (rumored, not officially disclosed) |
| Interconnect | Not disclosed (no public multi-card interconnect solution) |
| Interface | PCIe 4.0 x16 (rumored; some early sources wrote PCIe 4.0 without specifying lane count) |
| Graphics APIs | OpenGL 4.5, OpenCL 2.0 (rumored); Vulkan / OpenGL (2026 industry framing) |
| Video Interfaces | HDMI 2.0, DisplayPort 1.3 (rumored) |
| Video Codec | H.265 / 4K@60fps (rumored) |
| Virtualization | GPU virtualization supported (joint solution with KylinSec, deployed in power grid dispatch) |
| Launch | 2021 (JM9 series tape-out and packaging announcement); second chip of the series completed development in May 2022 |
| Mass Production/Availability | Volume delivery of the JM9 series officially confirmed in March 2026 |
Source Conflicts
| Metric | Source A (2019—2021 media/research reports) | Source B (2026-08 industry review) |
|---|---|---|
| Memory Capacity | 16 GB | 8 GB |
| Memory Type | HBM | GDDR6 |
| Memory Bandwidth | 512 GB/s | 512 GB/s (another contemporaneous article) |
| FP32 | 8 TFLOPS | 8 TFLOPS |
⚠️ Uncertainty: The memory configuration has two mutually contradictory circulating accounts: 16GB HBM and 8GB GDDR6. The HBM option was reported by multiple English-language media in 2019, but given HBM's high cost and constrained supply, it is a reasonable inference that the actual mass-production version switched to GDDR6. In the absence of an official specification sheet, neither account can be confirmed.
Key Features
- High-end positioning within the JM9 series: compared with the JM9231, strengthened data processing and bulk information handling, able to take on basic data analysis, geographic information computation, and similar workloads
- Self-developed JM9 graphics architecture: not a modified IP-licensed design; Jingjia Micro controls the underlying logic/physical design and driver stack
- GPU hardware virtualization: a fully domestic GPU virtualization solution launched with KylinSec, already deployed in critical infrastructure such as power grid dispatch
- 4K ultra-HD output: 4K display output and H.265 hardware codec support
- Domestic first-tier graphics rendering (per the 2026 industry review)
- Military-grade reliability: continues the company's high-reliability design accumulation from demanding aerospace environments
- Professional industry deployment: professional surveying and mapping, geological information processing, industrial computer-aided design, command and dispatch
- Limitations: AI compute (FP16/BF16/INT8) never disclosed, cannot be used for large-model training; no public high-speed interconnect; official specifications long opaque
Vendor Information
| Parameter | Value |
|---|---|
| Company | Changsha Jingjia Microelectronics Co., Ltd. (Changsha Jingjia Microelectronics Co., Ltd.) |
| Headquarters | Changsha, Hunan Province |
| Founded | 2006 |
| Listing | Listed on the SZSE ChiNext in March 2016, code 300474 |
| Positioning | The first company in China to successfully develop a domestic GPU and achieve large-scale engineering application |
| 2025 Revenue | RMB 720 million |
| 2025 R&D Spend | RMB 428 million (59.39% of revenue) |
| R&D Platforms | National-level enterprise technology center; GPU and Intelligent Computing Systems Innovation Center |
| Ongoing R&D | High-performance general-purpose GPU chip R&D and industrialization project; General-Purpose GPU Advanced Architecture R&D Center; AI SoC endpoint chip successfully taped out |
| Official Website | https://jingjiamicro.com |
Use Cases
- ✅ Xinchuang workstation / desktop / server graphics card domestic substitution
- ✅ Cloud desktops and GPU virtualization (government clouds, power grid dispatch)
- ✅ Professional surveying and mapping, geological information processing, geographic information systems (GIS)
- ✅ Industrial computer-aided design (CAD), BIM, industrial digital twins
- ✅ Security surveillance, video structuring (the company's planned direction)
- ✅ Basic data analysis and scientific computing
- ❌ Large-model AI training (no public AI compute figures, no high-speed interconnect)
- ❌ Large-model inference clusters (memory and bandwidth configuration unclear and on the low side)
- ❌ Procurement decisions requiring precise specifications (official specs are not public; request the datasheet directly from the vendor)
- ❌ Direct migration from the CUDA ecosystem
Related Cards
- Jingjia Micro JM9 Series — Base model of the same series, including the 2021 official announcement's measured figures
- Innosilicon Fenghua 1 — Contemporary domestic GPU with server/data center positioning, 5 TFLOPS FP32
- Moore Threads MTT S5000 — Domestic full-function GPU, fellow member of the "Five Tigers of Domestic GPUs"
- MetaX C600 — Domestic high-end GPU, HBM3e route comparison
- Alibaba Hanguang 800 — Domestic dedicated AI acceleration route comparison
References
- Changsha Jingjia Microelectronics announcement on the R&D progress of its JM9 series graphics processing chips (2021-11-16, SZSE)
- Changsha Jingjia Microelectronics Co., Ltd. (Baidu Baike, with March 2026 volume delivery and ongoing project progress)
- Chinese Vendor Developing PCIe 4.0 GPU With 16GB HBM and GTX 1080-Like Performance (Tom's Hardware, 2019-08-22, original source of the rumored specs)