Products for Machine Learning
Arm is the world's leading technology provider of silicon IP for the intelligent system-on-chips at the heart of billions of devices. Our portfolio of products enable partners to innovate and get-to-market faster on a secure architecture built for performance and power efficiency. Find the right processor IP for your application.
Product Filter
Showing: 57 Items
X
Selected filters:
Machine Learning
Product Families
Product Families
Clear All
|
X
Supreme performance at optimal power
Powering the most energy-efficient embedded devices
Robust real-time performance
Custom-designed for ultimate performance and specific market needs
Highest performance for machine learning inference
Highest graphics performance for flagship gaming experiences
Higher graphics performance combined with power efficiency
Advanced ISPs for human display and computer vision applications
Scalable and flexible for cloud to edge infrastructure
Fully verified, PPA-optimized, customizable Neoverse compute subsystems
Powerful solutions for physical security applications
Licensing Plan
Licensing Plan
X
Use Cases
Use Cases
X
Technologies
Technologies
X
Clear
Clear Filters
View 57 Items
Product Filter
Showing: 57 Items
Selected filters:
Machine Learning
Cortex-A
Cortex-A725
Second-generation Armv9.2 premium-efficiency, out-of-order CPU for sustained performance in a constrained power envelope.
- Designed to enhance consumer experience with AAA gaming and web browsing.
- Delivers increased sustained performance within fixed power envelope.
- Continues the area-optimized configurability option for cost-constrained markets.
Cortex-A720
First Armv9.2 premium-efficiency CPU based on DynamIQ technology. Offers sustained performance in a constrained power envelope.
- Built for next-generation consumer devices.
- Delivers consistently high performance for wearables, smartphones, and laptops.
- Offers new area configurability for an expanded range of use cases.
Cortex-A715
Second-generation Armv9 “big” CPU for best-in-class efficient performance.
- The CPU cluster workhorse across "big.LITTLE" configurations.
- Targeted microarchitecture optimizations for 20% power efficiency improvements.
- Consistent performance gains to match Cortex-X1, Arm’s first-generation Cortex-X CPU.
Cortex-A710
First-generation Armv9 “big” CPU that offers a balance of performance and efficiency.
- Addition of Armv9 architecture features for enhanced performance and security.
- Optimal for mobile compute use cases such as smartphones and smart TVs.
- 30% increase in energy efficiency compared to Cortex-A78.
Cortex-A520
First Armv9.2 high-efficiency “LITTLE” CPU.
- Most performant, high-efficiency CPU with improved power efficiency (up to 22% vs Cortex-A510) for DoU/real-world use cases.
- New QARMA3 PAC algorithm lowers the performance cost, strengthening PAC deployment in the consumer technology market.
- AArch64-only CPU for building big.LITTLE clusters across the consumer technology market.
Cortex-A510
First-generation Armv9 high-efficiency “LITTLE” CPU.
- Large performance increases for a highly efficient CPU.
- Innovative microarchitecture upgrades.
- Over 3x uplift in ML performance compared to Cortex-A55.
Cortex-A78
The fourth generation high-performance CPU based on DynamIQ technology. The most efficient premium Cortex-A CPU.
- Built for next generation consumer devices.
- Enabling immersive experiences on new form factors and foldables.
- Improving ML device responsiveness and capabilities such as face and speech recognition.
Cortex-A78C
Providing market-specific solutions with advanced security features and large big-core configurations.
- Performance for laptop class productivity and gaming on-the-go.
- Advanced data and device security with Pointer Authentication.
- Improved scalability with up to 8 big core only configuration and up to 8MB L3 cache.
Cortex-A78AE
A High-performance Cortex-A CPU based on Arm DynamIQ and designed for safety-critical applications.
- Designed for the software-defined vehicle.
- Split-lock capability with hybrid mode for efficient functional safety and post-silicon flexibility.
- ISO 26262 ASIL D certifications, and enhanced features targeted at ASIL B and quality management use cases.
Cortex-A77
Third-generation high-performance CPU based on DynamIQ technology.
- Leadership performance and efficiency for 5G mobile solutions.
- Improved responsiveness for on device machine learning.
- Built for next-gen smartphones and laptops.
Cortex-A76
Second-generation high-performance CPU based on DynamIQ technology.
- Designed for devices undertaking complex compute tasks.
- Greater single threaded performance and improved energy efficiency.
- Enables faster responsiveness and at-the-edge support for machine learning applications.
Cortex-A76AE
The world’s first autonomous-class processor with integrated safety features.
- Purpose-built for functional safety applications such as ADAS and autonomous vehicles.
- First application processor with Split-Lock capability.
- Safety capable to industry standards, including ISO 26262 ASIL D.
Cortex-A75
First-generation high-performance CPU based on DynamIQ technology.
- Flexible architecture provides a broad ecosystem of support.
- Executes up to three instructions in parallel per clock cycle.
- Broad market use covers smartphones, servers, automotive applications and more.
Cortex-A73
Highly power-efficient CPU that maintains high-performance.
- Increased power efficiency of up to 30 percent over predecessors.
- Smallest Armv8-A processor.
- Designed for mobile and consumer applications.
Cortex-A72
High-performance CPU that has multiple uses including mobile and embedded technologies.
- Advanced branch predictor reduces wasted energy consumption.
- Gain significant advantages in reduced memory requirements.
- Suitable for implementation in an Arm big.LITTLE configuration.
Cortex-A65AE
Arm’s first multithreaded Cortex-A CPU with Split-Lock for functional safety.
- Best-in-class throughput efficiency for memory intensive workloads.
- Highest levels of safety with Dual Core Lock-Step for demanding safety-critical tasks.
- Supports Split-Lock for improved cost efficiency in mixed-criticality applications.
Cortex-A55
Highest efficiency mid-range processor that can be paired with a high-performance CPU in a DynamIQ configuration.
- Flexible design meets requirements to support broad market application.
- Ideal for smaller devices with constrained environments.
- Designed for compatibilty with DynamIQ configurations.
Cortex-A53
The most widely-used mid-range processor with balanced performance and efficiency.
- Available in Arm Flexible Access.
- The choice for high single thread and FPU/Neon performance.
- Supports a wide range of applications across automotive and networking and more.
- Most widely deployed 64-bit Armv8-A processor.
Cortex-A34
Smallest and most power-efficient 64-bit Armv8-A processor.
- Scalable from a single core or up to a cluster of 4 processors.
- Advanced power management with idle power management features.
- Supports 64-bit processing suitable for embedded applications.
Cortex-M
Cortex-M85
Highest-performing Cortex-M processor with Arm Helium technology.
- Unprecedented scalar, DSP, and ML performance for demanding use cases.
- Offers enhanced software security with TrustZone and PACBTI extension to accelerate the route to PSA Certified silicon.
- Optional support for Arm Custom Instructions, enabling product differentiation without fragmentation.
Cortex-M55
Mainstream Cortex-M processor with Arm Helium technology.
- Efficient scalar, ML, and DSP performance for general purpose applications.
- Arm TrustZone accelerates the route to PSA Certified silicon.
- Optional support for Arm Custom Instructions, enabling product differentiation without fragmentation.
Cortex-M52
The smallest implementation of Armv8.1-M with Arm Helium Technology.
- Efficient scalar, ML, and DSP performance for cost sensitive designs.
- Offers enhanced software security with TrustZone and PACBTI extension to accelerate the route to PSA Certified silicon.
- Optional support for Arm Custom Instructions, enabling product differentiation without fragmentation.
Cortex-M33
Ideal blend of real-time determinism, efficiency and security.
- Simplifies digital signal processing with security.
- Differentiate your product with TrustZone software isolation.
- Brings 32-bit performance to even the simplest and cost-sensitive devices.
Cortex-M7
High-performance Cortex-M processor.
- Simplifies signal processing to bring high-performance DSP to the masses.
- Built in floating point processing reduces power consumption.
- Supports innovative MCUs for more intensive automation tasks.
Cortex-X
Cortex-X
Custom-design beyond traditional Arm Cortex products for next-generation devices.
- Delivers ultimate performance and compatibility with Arm Cortex-A.
- Addresses market-specific goals and relevant use cases.
- Provides DynamIQ support for intelligent, scalable solutions and flexibility.
Ethos - NPUs
Ethos-U85
Enabling edge AI use cases with generative AI capabilities.
- Delivers up to 4 TOPs scalable ML performance.
- 20% improvements in energy efficiency than previous Ethos-U NPUs.
- Native support for transformer networks.
Ethos-U65
Powering innovation in a new world of AI devices at the edge and endpoint.
- Delivers 1.0 TOP/s ML performance in about 0.6 mm2.
- Partner configurable from 256 to 512 8-bit MACs.
- Unified toolchain supports Cortex-M and Cortex-A based systems.
Ethos-U55
Configurable and efficient embedded ML inference.
- Delivers up to 0.5 TOP/s, a 480x ML up lift and 90% energy reduction.
- Partner configurable from 32 to 256 8-bit MACs in around 0.1mm2.
- Rapid development with a single tool chain for Cortex-M and Ethos-U.
Arm NN SDK
Bridges the gap between existing neural network frameworks and the underlying IP.
- Free of charge.
- Supports Arm Cortex CPUs, Arm Mali GPUs and the Arm Machine Learning processor.
- Arm NN for NNAPI accelerates neural networks on Android devices.
Ethos-N78
Scalable and efficient second-generation ML inference processor.
- 2x faster inference with 40% lower bandwidth, 25% increased efficiency.
- Multiple markets from 1 to 10 TOP/s and up to 90 unique configurations.
- Develop once, deploy anywhere with online and offline compilation.
Arm Immortalis Graphics Processors
Arm Immortalis-G925
Arm's most performance and efficient GPU till date, offering unparalled mobile gaming and ML performance.
- Fragment prepass.
- Doubled tiler throughput.
- Doubled shift-convert unit thoughput.
- Improved command stream frontend.
- Improved ray tracing performance.
Arm Immortalis-G720
Arm’s latest flagship GPU is based on the new 5th Gen GPU architecture, bringing the next generation of visual computing to mobile.
- Deferred vertex shading geometry pipeline.
- Doubled 64bpp texture throughput.
- Optimized 2xMSAA (multisampling anti-aliasing).
- Hardware-based ray tracing support on mobile.
Arm Immortalis-G715
Arm’s flagship GPU providing ultimate mobile gaming experiences.
- First Arm GPU to offer hardware-based ray tracing support on mobile.
- New variable-rate shading for gaming boost.
- 2x architectural ML improvements.
Mali Graphics Processors
Mali-G720
5th Gen based graphics processing unit (GPU) for premium mobile market.
- Deferred vertex shading geometry pipeline.
- Doubled 64bpp texture throughput.
- Optimized 2xMSAA (multisampling anti-aliasing).
Mali-G715
Fourth-generation Valhall-based graphics processing unit (GPU) for premium mobile market.
- New variable rate shading for gaming boost.
- Improved execution engine for more compute power.
- 2x architectural ML improvements.
Mali-G710
Third- generation Valhall-based graphics processing unit (GPU) for the premium market.
- Game changing features to improve on-device graphics.
- Inclusion of command stream frontend for a more realistic gaming experience.
- Larger cores to boost device efficiency to improve battery life.
Mali-G620
5th Gen based graphics processing unit (GPU).
- Adopts features from Mali-G720, including deferred vertex shading geometry pipeline.
- Design work from Mali-G720 can be used to address wider audience.
Mali-G615
Fourth-generation Valhall-based graphics processing unit (GPU) in Mali-G600 series.
- Adopts features from Mali-G715, including variable rate shading, for gaming boost.
- Enables high-quality battery life.
- Design work from Mali-G715 can be used to address wider audience.
Mali-G610
Third-generation Valhall-based graphics processing unit (GPU) in Mali-G600 series.
- Brings premium use cases to a wider audience of developers and consumers.
- Inclusion of command stream frontend for an improved gaming experience.
- Industry-leading device efficiency to boost battery life.
Mali-G510
Third-generation Valhall-based graphics processing unit (GPU) for the mainstream market.
- Bringing premium features to a range of different devices.
- Doubling on device graphics performance and increased visual fidelity on smart TVs.
- Doubling of machine learning capabilities on previous generation mainstream GPUs.
Mali-G78AE
High-performance GPU suitable for complex autonomous applications, safety capable to ASIL B.
- Flexible Partitioning enables hardware separation for mixed-criticality workloads.
- Hardware virtualisation support.
- Highly scalable, from 1-24 shader cores from a single IP.
Mali-G78
Second-generation premium GPU based on the Mali Valhall architecture.
- Enables superior digital immersion for consumers.
- Asynchronous Top Level for power savings and greater battery life for next-generation devices.
- On device improvements in ML enabling more realistic gaming experiences on mobile.
Mali-G68
Valhall-based graphics processing unit (GPU) for the sub premium market.
- Fused Multiply-Add (FMA) built from the ground up offers improved power consumption and prolonged battery life.
- Enables developers to target larger consumer markets with high-performing applications and mobile gaming.
Mali-G77
First-generation premium GPU based on the Mali Valhall architecture.
- 60 percent performance improvement for on-device machine learning.
- Improved high-fidelity gaming performance.
- Increased performance due to the innovative super scalar engine.
Mali-G76
Premium GPU based on the Mali Bifrost architecture, delivering high energy and area-efficiency.
- Execution engine width doubled with eight-thread pipelines.
- Uses three wider engines to double overall compute capability per shader core.
- 8-bit integer dot product accelerates neural network (NN) interface.
Mali-G57
First-generation mainstream GPU based on the Mali Valhall architecture.
- Improved performance and increased efficiency.
- Premium high-fidelity content delivered to mass markets.
- Re-engineered for Vulkan to be positioned for the future of gaming.
Mali-G72
Second-generation Bifrost-based high-performance GPU for high-fidelity gaming and complex machine learning (ML) workloads.
- Arithmetic optimizations for performance and efficiency improvements.
- Advanced technologies to increase tile buffer memory support.
Mali-G52
Mainstream Mali Bifrost architecture GPU with a scalable core.
- Large range of configurations - two or three 8-thread wide pipelines per shader core.
- Significant performance over Mali-G51 for complex graphics.
- 8-bit integer dot product accelerates neural network (NN) interface.
Mali Image Signal Processors
Mali-C52
Multi-camera image signal processor for display applications.
- High precision and high dynamic range.
- Image quality focused.
- Use cases range from security cameras to high-end drones, and more.
Mali-C32
Configuration version of Mali-C52, optimized for area.
- High precision and high dynamic range.
- Use cases include low-power, cost-sensitive embedded vision devices where the full feature set of Mali-C52 is not required.
Neoverse
Neoverse N3
Performance-per-watt optimized for hyperscale, 5G, enterprise networking, and infrastructure edge workloads.
- 20% greater performance-per-watt efficiency compared to Neoverse N2.
- 2MB L2 cache option offers nearly 3x performance gains on ML workloads.
Neoverse V3
The first Armv9.2 Neoverse CPU to support the Arm Confidential Compute Architecture.
- Highest single-thread performance for cloud, HPC, and AI/ML workloads.
- Supports up to 3MB private L2 cache per core.
- The perfect foundation for AI accelerator development.
Neoverse V2
A New Benchmark for Cloud Computing, HPC, and ML Performance.
- With an enhanced pipeline and up to 2MB L2 cache per core, Neoverse V2 can deliver 2x the performance of Neoverse V1 on cloud and ML applications.
- Arm's first V-series core with Armv9 features like memory tagging extension (MTE) and performance defined power (PDP).
Neoverse V1
A performance-first tier targeting HPC, HPC in the Cloud and AI/ML-accelerated applications.
- Arm’s first SVE implementation for the HPC market, with 2x floating and 4x ML uplift over Neoverse N1.
- Expands market-leading performance with 50% IPC uplift over Neoverse N1.
Neoverse Compute Subsystems
Neoverse CSS V3
The high-performance Arm Neoverse V3 platform is configured and verified by Arm for cloud, HPC, and AI/ML workloads.
- Up to 64 high-performance Neoverse V3 cores per die.
- Offers a highly customizable, high-performance memory subsystem.
- Supports high-speed, low-latency AI accelerator attachment options.
Neoverse CSS N3
The Arm Neoverse N3 platform, validated and optimized by Arm, is helping reduce time-to-market, cost, and risk.
- Highly configurable to target 5G, enterprise networking, and infrastructure edge use cases.
- Supports from 8 to 32 Neoverse N3 cores per die.
Neoverse CSS N2
The market leading performance-per-watt of the Arm Neoverse N2 platform, delivered as a fully verified, customizable compute subsystem.
- Up to 64 Neoverse N2 cores in a 5nm advanced process.
- Up to 1MB L2 private cache per core and up to 64MB shared system-level cache.
- Up to 8x DDR5 40b or LPDDR5 channels.
- Up to 4x x16 PCIe/CXL Gen5 lanes.
Show more - 51 items
Understanding AI Inference on Arm CPUs
Demand for running AI workloads on CPU is growing. This comprehensive guide provides a deep dive into CPU inference and the use cases for which this may be the practical choice. Explore the industries that are already benefiting from AI on CPU and learn about real-world examples.
Unlock the Power of Arm Technology
Arm’s cutting-edge solutions are easily accessible through our subscription-based licensing options. In just a few clicks, find out if your company has an active subscription to the technology that’s shaping the future of computing.
Once you confirm your subscription, explore a treasure trove of IP, powerful tools, and innovative models—all designed to help elevate your projects and get you started on building the future of computing on Arm.
Plus, our experts are here to guide you every step of the way. With seamless access and guided excellence, learn how to harness the full potential of Arm technology. From IP integration to advanced modeling, we’ve got you covered.