Free tools Windows power users keep installed
One-click scans. No signup required.
Qualcomm introduced its IQ industrial processor family on October 9, 2024, at Embedded World North America. Now branded Qualcomm Dragonwing IQ9, IQ8 and IQ6, the platforms target robots, machine-vision systems, drones, industrial gateways, automation equipment and other products that need local AI plus industrial I/O and long-term support. The IQ9 family reaches up to 100 dense TOPS in applicable configurations; that is a vendor-rated peak, not a guarantee that every IQ9 device or application will achieve the same performance.
What Qualcomm announced
The IQ Series launch was Qualcomm’s attempt to package edge-AI acceleration with the characteristics industrial equipment designers need: operation in harsh environments, memory-error protection, real-time control resources, camera and networking interfaces, safety-oriented architecture and a longer product lifecycle. The announcement covered three performance tiers:
- IQ9: the highest-compute option for demanding robotics, multi-camera vision and edge-AI workloads.
- IQ8: a balanced high-performance tier for industrial equipment that needs substantial inference capacity without the IQ9’s maximum scale.
- IQ6: a lower-power, lower-complexity design for gateways, displays, monitoring and less demanding embedded AI.
Qualcomm also presented an IoT Solutions Framework combining chipsets, operating systems, software enablers, reference designs, SDKs, cloud services, containers, microservices and ecosystem partners. The intended markets include industrial automation, autonomous mobile robots, drones, inspection, machine vision, edge gateways, industrial HMIs, retail and security, medical devices and agricultural equipment.
This is an OEM and system-integrator proposition, not a consumer processor sold like a laptop CPU. Final availability, board choices and commercial terms normally come through Qualcomm, module makers, distributors and design partners.
#1 Best Overall
- ✅Powered by 26 Tera-Operations Per Second (TOPS) Hailo-8 AI Processor. 2.5W typical power consumption
- ✅Scalable, enabling simultaneous processing of multi-streams & multi-models
- ✅Enabling real-time, low latency and high-efficiency AI inferencing on the edge devices
- ✅Supports TensorFlow, TensorFlow Lite, ONNX, Keras, Pytorch frameworks
- ✅Supports Linux and Windows. Supports the temperature range of -40°C to 85°C
Although the announcement dates from 2024, Qualcomm’s current briefs use the Dragonwing IQ branding. A 2026 buyer should therefore treat the launch as the origin of the family, not as a new product release.
IQ9, IQ8 and IQ6 compared
The families are differentiated by more than an AI headline. Memory capacity, camera inputs, PCIe generation, real-time resources, temperature specifications and power envelopes can determine whether a design is practical.
| Feature | Dragonwing IQ9 | Dragonwing IQ8 | Dragonwing IQ6 |
|---|---|---|---|
| Maximum stated AI performance | Up to 100 dense TOPS, configuration-dependent; IQ-9075 documentation lists 50- or 100-TOPS configurations | Up to 40 dense TOPS | No equivalent headline TOPS figure is presented in the cited launch material; Qualcomm emphasizes the Hexagon DSP and vector extensions |
| CPU | Eight-core Qualcomm Kryo Gen 6 | Eight-core Qualcomm Kryo Gen 6 | Octa-core Kryo 460: two Gold cores at 1.9 GHz and six Silver-lite cores at 1.6 GHz in launch coverage |
| GPU | Adreno 663 | Adreno 623 | Adreno 612 |
| Memory | Up to 36 GB LPDDR5 with inline ECC | LPDDR5 with ECC support | Up to 8 GB LPDDR4x |
| Camera capability | Up to 16 concurrent cameras | Up to 12 concurrent cameras | MIPI CSI-2 camera connectivity; exact count depends on the part and design |
| PCIe | Gen 4 | Gen 4 | Gen 2 |
| Industrial temperature specification | –40°C to 115°C junction temperature | –40°C to 125°C junction temperature | Confirm the exact SKU’s product brief |
| Real-time and safety resources | Separate MCU-like subsystem with four real-time cores | Separate MCU-like subsystem with four real-time cores | Safety and real-time details require exact SKU documentation |
| Typical design fit | High-end robotics, multi-camera inspection and demanding edge AI | Compute-heavy industrial equipment and substantial vision workloads | Lower-power gateways, HMIs and less demanding vision or monitoring |
Primary specifications are in Qualcomm’s IQ9 brief, IQ8 brief and IQ6 brief. Family-level figures should not be treated as universal specifications for every SKU.
IQ9: maximum compute for demanding perception
IQ9 is the platform for systems where several cameras, large models or high-throughput perception pipelines make memory and accelerator capacity central design constraints. Qualcomm lists a Kryo Gen 6 eight-core CPU, an Adreno 663 GPU and two Hexagon Tensor Processors with vector and matrix acceleration. The brief allows up to 16 concurrent cameras and as much as 36 GB of LPDDR5 with inline ECC.
Connectivity includes PCIe Gen 4, USB 3.1 Gen 2 and Ethernet with time-sensitive networking (TSN). In the listed IQ-9075 configuration, Qualcomm specifies up to two 2.5-GbE TSN interfaces. A dedicated MCU-like subsystem has four real-time cores for monitoring, self-tests, real-time functions and possible offload to an external MCU.
The IQ-9075 brief lists an SoC power range of 3.8 W to 20 W, depending on configuration and workload. That spread is a reminder that a part number’s thermal design cannot be inferred from its maximum TOPS number. Qualcomm also states more than 10 years of product-longevity support for IQ9, while warning that longevity dates can change.
IQ9’s 100-TOPS qualification
The launch announcement says the family reaches up to 100 TOPS. The IQ9 brief is more specific: the IQ-9075 material lists configurations rated at 50 dense TOPS or 100 dense TOPS. Therefore, “IQ9 delivers 100 TOPS” is too broad. The accurate statement is that an applicable IQ9 configuration is rated for up to 100 dense TOPS.
IQ8: a balanced industrial-AI tier
IQ8 keeps the eight-core Kryo Gen 6 CPU architecture but pairs it with an Adreno 623 GPU and up to 40 dense TOPS of AI performance. Qualcomm lists up to 12 concurrent cameras, LPDDR5 with ECC support, PCIe Gen 4, USB 3.1 and Ethernet with TSN.
Recommended Free Tools
Rank #2
- A USB accessory that brings machine learning inferencing to existing systems. Works with Raspberry Pi and other Linux systems
- Performs high-speed ML inferencing: the on-board edge TPU Coprocessor is capable of performing 4 trillion operations (tera-operations) per second (tops), using 0.5 watts for each tops (2 tops per watt). For example, it can execute state-of-the-art mobile vision models such as mobilenet V2 AT 400 FPS, in a power efficient manner
- Works with Debian Linux: connects to any debian-based Linux system with an included USB 3.0 Type-C cable
- Supports tensorflow Lite: no need to build models from the ground up. Tensorflow Lite models can be compiled to run on the edge TPE
- Supports automl vision edge: easily build and deploy fast, high-accuracy custom image classification models to your device with automl vision edge
Its brief gives a –40°C to 125°C junction-temperature range and describes a dedicated MCU-like subsystem with four high-performance real-time cores. Qualcomm Linux and Ubuntu support are listed, as is more than 10 years of product-longevity support subject to change.
IQ8 is a reasonable starting point when an inspection cell, robot controller or industrial appliance needs significant vision and inference capacity but does not need IQ9’s maximum camera count, memory ceiling or accelerator rating. The decision still depends on the exact model, camera pipeline, latency target and thermal enclosure.
IQ6: a different, lower-power design point
IQ6 should not be viewed as merely a slower IQ9. It is a lower-complexity platform built around an 11-nm FinFET process, a Kryo 460 octa-core CPU, Adreno 612 graphics and a Spectra 230 image-signal processor. Launch coverage identifies two Gold cores at 1.9 GHz and six Silver-lite cores at 1.6 GHz.
The platform supports up to 8 GB of LPDDR4x, 4K60 video decode and 1080p60 encode, PCIe Gen 2, USB 3.1, Ethernet and MIPI CSI-2 camera connections. A Hexagon DSP with dual Hexagon Vector Extensions handles signal-processing and embedded-AI tasks. Optional Wi-Fi 6E and Bluetooth 5.3 are available through the relevant connectivity design.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Qualcomm’s launch materials do not give IQ6 an equivalent 40- or 100-TOPS headline. Buyers should instead validate the actual model and pipeline on the chosen IQ6 part. Typical fits include an industrial display with local vision, a compact gateway, condition monitoring or a lower-camera-count system where power and bill of materials matter more than maximum throughput.
What “up to 100 TOPS” really tells you
TOPS means trillions of operations per second. “Dense TOPS” is Qualcomm’s terminology for the IQ9 AI figure, but it remains a peak accelerator metric rather than an application benchmark.
Observed frame rate, latency and energy per inference depend on:
- Model architecture, precision and quantization.
- Which operators the accelerator and compiler support.
- Memory bandwidth and the size of model weights and intermediate tensors.
- Camera preprocessing, image-signal-processor work and data movement.
- Concurrent CPU, GPU, DSP and networking activity.
- Thermal limits and any throttling inside the final enclosure.
- Whether the model runs entirely on the intended neural accelerator or falls back to another engine.
The IQ9 brief includes Qualcomm examples involving Llama 2 token generation. Those are configuration- and workload-specific demonstrations, not a general promise for every language model. Object detection, defect inspection, segmentation, pose estimation and sensor fusion can produce very different results from language-model inference. Benchmark the actual model, input resolution, camera count, precision and latency target on an evaluation platform.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Rank #3
- High-Performance AI Processing: The MX3 is designed to handle the most demanding AI computer vision workloads, delivering exceptional performance and efficiency.
- Flexible Integration: The MX3 can be easily integrated into your existing systems via its M.2 M-key form factor and support for Linux operating systems.
- Energy Efficient: The MX3 is designed to provide high performance while minimizing power consumption.
- Comprehensive Software Development Kit (SDK): The MX3 is supported by a comprehensive SDK that simplifies development and deployment.
- Hardware compatability: The MX3 is compatible with the PCI-SIG M.2 M-key 2280 Specification. It can be used with the Raspberry Pi 5 with a M-key 2280 HAT.
Why industrial buyers care about reliability and control
ECC memory
Error-correcting code (ECC) can detect and correct certain memory errors. That is valuable for continuously operating equipment and electrically noisy environments, but ECC is not complete fault tolerance and does not make a finished system immune to hardware, software or power failures.
Real-time and safety subsystem
IQ9 and IQ8 include a physically and electrically separated MCU-like subsystem with dedicated real-time cores. Qualcomm describes functions such as monitoring, error detection, self-tests and real-time workloads, including the possibility of offloading work to an external MCU. This architecture can simplify partitioning between Linux applications and deterministic control, but the product team still has to design, verify and certify the complete safety function.
SIL-3 support is not system certification
Qualcomm’s launch announcement says the IQ family includes built-in safety features, including SIL-3 support with an integrated safety controller. “Support” does not mean every robot, controller or medical device built with the processor is SIL-3 certified. Certification depends on the complete hardware and software, safety lifecycle, diagnostics, documentation and application assessment.
Temperature ratings need context
The published IQ9 and IQ8 figures are junction-temperature specifications: –40°C to 115°C for IQ9 and –40°C to 125°C for IQ8. They are not automatically the ambient range of an assembled product. Heatsinks, airflow, enclosure geometry, board layout, workload, vibration and electromagnetic compatibility all affect system qualification. IQ6’s allowable range must be checked in the exact part brief.
Industrial networking, cameras and software matter as much as compute
PCIe, TSN Ethernet, USB, camera interfaces and memory determine how efficiently a product can move sensor data through the system. A design that needs synchronized multi-camera capture may be constrained by sensor links, ISP capacity or preprocessing before it reaches the TOPS limit. A robot that needs deterministic traffic may require TSN and a carefully selected real-time architecture, not simply a faster neural accelerator.
Qualcomm lists Qualcomm Linux and Ubuntu support for IQ9 and IQ8. The IoT Solutions Framework adds SDKs, reference designs, cloud and container tooling and partner support. Before committing, confirm the specific release’s drivers, camera stack, compiler, model-conversion workflow, supported operators and container support. A model that runs well on a development board may require quantization or Qualcomm-specific optimization to use the intended accelerator efficiently.
How to choose a tier
- Define the real workload. List models, input sizes, frame rate, latency, precision, concurrency and any non-AI CPU work. A simple classifier or sensor-fusion task may not justify IQ9.
- Count cameras and data paths. Include resolution, synchronization, ISP processing, storage and network traffic. Multi-camera inspection and robotic perception can make I/O the bottleneck.
- Separate control from inference. Determine whether TSN, CAN-FD, an external MCU, RTOS functions or the integrated real-time subsystem are required.
- Size memory and thermal design. Large models and multiple streams can be memory-bandwidth limited. Compare the enclosure’s cooling capacity with the selected SKU’s power range; the IQ-9075 IQ9 brief lists 3.8 W to 20 W depending on configuration and workload.
- Check SKU-level connectivity. Verify PCIe lanes and generation, Ethernet speed and TSN, USB, camera links, wireless companions and storage interfaces for the exact part.
- Confirm lifecycle terms in writing. Qualcomm advertises more than 10 years for IQ9 and IQ8, but its briefs say longevity dates can change. Plan approved sources, last-time-buy actions and distributor support.
- Validate the software path. Run the actual model and camera pipeline on an evaluation platform, measuring latency, throughput, thermals and power rather than relying on TOPS.
Common selection mistakes
Choosing by TOPS alone
Peak arithmetic throughput cannot overcome a camera, memory, preprocessing or thermal bottleneck. A lower-tier part may deliver the required application result more efficiently.
Treating a family headline as a SKU specification
Clock rates, AI throughput, I/O, memory, wireless options and power vary across IQ9, IQ8 and IQ6 parts. Use the exact product brief and ordering code for a design review.
Rank #4
- Hailo-10H AI accelerator delivering 40 TOPS (INT4) inferencing performance.
- Performance for computer vision models comparable to the Raspbery Pi AI HAT+ (26 TOPS).
- Runs generative AI models efficiently using 8GB on-board RAM.
- Fully integrated into Raspbery Pi’s camera software stack.
- Conforms to Raspbery Pi HAT+ specification.
Confusing industrial features with certification
ECC, a safety controller and SIL-3 support are useful building blocks, not proof that a finished machine meets a regulatory standard.
Assuming an industrial temperature label makes the product deployment-ready
The board and enclosure still need appropriate power delivery, cooling, EMC, vibration and environmental qualification.
Overlooking supply and software risk
A stated longevity program does not remove the need for written lifecycle terms, second-source planning and confirmation that the target framework and operators are supported efficiently.
What is and is not publicly established
The cited Qualcomm material provides family and selected SKU specifications, but it does not provide public system-level pricing or independent application benchmarks. Exact availability varies by part, module and design partner. Buyers should expect an evaluation, design-in and sourcing process rather than an ordinary retail checkout.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchPC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Competing industrial edge-AI choices include NVIDIA Jetson, NXP i.MX and Layerscape, AMD/Xilinx adaptive and embedded platforms, Intel edge processors, MediaTek solutions and lower-complexity MCU/MPU families from Renesas and Texas Instruments. They differ in accelerator architecture, software, camera support, power, certifications, lifecycle terms and module availability; a meaningful comparison requires the same workload and current SKU documentation.
Bottom line
Qualcomm’s Dragonwing IQ proposition is broader than “more AI.” IQ9, IQ8 and IQ6 combine different levels of local inference with camera processing, industrial networking, real-time resources, ECC options, temperature support and a software-and-partner framework. IQ9 is the candidate for the most demanding multi-camera and robotic workloads, IQ8 is the balanced high-performance choice, and IQ6 targets lower-power embedded systems. The right decision comes from benchmarking the complete model and sensor pipeline on the exact SKU—not from choosing the largest TOPS number.
Frequently Asked Questions
Are Qualcomm IQ processors available as ordinary retail products?
No. They are industrial OEM platforms typically accessed through Qualcomm, module manufacturers, distributors and design partners. Public system-level pricing was not provided in the cited material.
Does every IQ9 processor provide 100 TOPS?
No. Qualcomm’s IQ9 brief lists 50- and 100-dense-TOPS configurations for the IQ-9075 material. The maximum applies to an applicable configuration, not automatically to every IQ9 device.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Does Qualcomm’s SIL-3 support certify a finished machine?
No. It is a processor-level safety capability. Certification depends on the complete hardware, software, diagnostics, safety process and application.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
