Hello,
We’ve realized that the link to the documentation in our Compute Library v25.02.1 release announcement isn’t working at the moment. We’re sorry for the trouble!
Our team is actively working on getting this resolved as soon as possible. We’ll send an update once the link is accessible again.
Thanks for your patience, and let us know if you have any questions.
Best regards,
Yevgen
IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.
Hello,
The 25.02.1 release of Compute Library is out and comes with a
collection of improvements and new features.
Source code and prebuilt binaries are available at:
[1]https://github.com/ARM-software/ComputeLibrary/releases/tag/v25.02.1
Highlights of the release:
* Add stateless support for GEMM kernels that need working_space
IMPORTANT NOTICE: The contents of this email and any attachments are
confidential and may also be privileged. If you are not the intended
recipient, please notify the sender immediately and do not disclose the
contents to any other person, use it for any purpose, or store or copy
the information in any medium. Thank you.
References
1. https://github.com/ARM-software/ComputeLibrary/releases/tag/v25.02.1
Hello,
The 25.02 release of Compute Library is out and comes with a collection of improvements and new features.
Source code and prebuilt binaries are available at: https://github.com/ARM-software/ComputeLibrary/releases/tag/v25.02
Highlights of the release:
* Detect number of CPU cores in OpenBSD
* Support tensors with dynamic shapes in NEGEMM
* Support FP16 dequantization in NEGEMMLowpMatrixMultiplyCore
* Add a public API for CpuMeanStdDevNormalization
* Enable BF16 inputs in CpuFullyConnected
IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.
IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.
Dear ACL Development Team,
I hope this email finds you well. I am following up on my previous email
regarding the implementation of real-time inference using Arm Compute
Library.
To recap, I have successfully implemented inference for single images with
the object detection model on ARM devices, but I am looking to extend this
to real-time inference by feeding frames directly from a camera. From the
documentation I have reviewed, it seems that ACL supports input files in
formats like npy, jpg, and ppm.Could you please let me know if there is a
recommended approach or any existing functionality in ACL to achieve this?
In case there is no built-in support, I am currently exploring the ACL
source code to understand how this might be implemented. If possible, I
would appreciate any advice or suggestions on how to approach this, or any
resources that could assist in my efforts.
Thank you for your time and support. I look forward to hearing from you.
Best regards,
Darshan B Y
Dear ACL Development Team,
I am currently working on performing inference on ARM devices using an
object detection model with the Arm Compute Library (ACL). I have
successfully implemented inference for single images, obtaining correct
detections.
From the ACL documentation and examples I have reviewed, it appears the
library only supports input files in formats like npy, jpg, and ppm. I am
looking to implement real-time inference by feeding frames directly from a
camera. Could you please let me know if there is a recommended approach or
any existing functionality in ACL to achieve this?
Additionally, I have been exploring the ACL source code and am very
interested in working further with it. Any guidance or resources you could
provide would be greatly appreciated.
Thank you for your time and support.
Best regards,
Darshan B Y
Hello,
The 24.11.1 release of Compute Library is out and comes with a
collection of improvements and new features.
Source code and prebuilt binaries are available at:
[1]https://github.com/ARM-software/ComputeLibrary/releases/tag/v24.11.1
Highlights of the release:
* Add stateless GEMM execution via ICPPKernel::run_op
* TensorShape class supports dynamic shapes
* Add skeletons for Dynamic GEMM operator
* Convert Double rounding to Single rounding quantization behaviour
in both Cpu/Gpu backend
IMPORTANT NOTICE: The contents of this email and any attachments are
confidential and may also be privileged. If you are not the intended
recipient, please notify the sender immediately and do not disclose the
contents to any other person, use it for any purpose, or store or copy
the information in any medium. Thank you.
References
1. https://github.com/ARM-software/ComputeLibrary/releases/tag/v24.11.1
Hello,
The 24.11 release of Compute Library is out and comes with a collection
of improvements and new features.
Source code and prebuilt binaries are available at:
[1]https://github.com/ARM-software/ComputeLibrary/releases/tag/v24.11
Highlights of the release:
* Add SVE SoftmaxLayer kernel for BF16
* Provide stateless API for CpuGemmLowpMatrixMultiplyCore,
CpuQuantize, and DequantizationLayer
* Extend static quantization interface for both matmul and
convolution operations
IMPORTANT NOTICE: The contents of this email and any attachments are
confidential and may also be privileged. If you are not the intended
recipient, please notify the sender immediately and do not disclose the
contents to any other person, use it for any purpose, or store or copy
the information in any medium. Thank you.
References
1. https://github.com/ARM-software/ComputeLibrary/releases/tag/v24.11
Hello,
The 24.09 release of Compute Library is out and comes with a collection
of improvements and new features.
Source code and prebuilt binaries are available at:
[1]https://github.com/ARM-software/ComputeLibrary/releases/tag/v24.09
Highlights of the release:
* Provide a wrapper class to expose cpu::CpuSoftmaxGeneric
* Detect number of cores in Windows®
* Add Optimized SME kernel for QASYMM8_SIGNED elementwise addition
operation
IMPORTANT NOTICE: The contents of this email and any attachments are
confidential and may also be privileged. If you are not the intended
recipient, please notify the sender immediately and do not disclose the
contents to any other person, use it for any purpose, or store or copy
the information in any medium. Thank you.
References
1. https://github.com/ARM-software/ComputeLibrary/releases/tag/v24.09
Hello,
The 24.08.1 release of Compute Library is out and comes with a
collection of improvements and new features.
Source code and prebuilt binaries are available at:
[1]https://github.com/ARM-software/ComputeLibrary/releases/tag/v24.08.1
Highlights of the release:
* Change inheritance qualifiers of experimental Cpu operator
interface classes to public for cpu-wrappers.
* Mismatches in static quantization updated after configure tests
* CpuSoftmax configure ignores is_log on validation
* Linker errors in armv8.2a Windows® builds
IMPORTANT NOTICE: The contents of this email and any attachments are
confidential and may also be privileged. If you are not the intended
recipient, please notify the sender immediately and do not disclose the
contents to any other person, use it for any purpose, or store or copy
the information in any medium. Thank you.
References
1. https://github.com/ARM-software/ComputeLibrary/releases/tag/v24.08.1
Hello,
The v24.05 release of Compute Library is out and comes with a collection of improvements and new features.
Source code and prebuilt binaries are available at: https://github.com/ARM-software/ComputeLibrary/releases/tag/v24.05
Highlights of the release:
- Add CLScatter operator for FP32/16, S32/16/8, U32/16/8 data types.
- Various fixes to enable FP16 kernels in armv8a multi_isa builds.
- Updated logic in the OpenMP scheduler to exclude LITTLE cores.
IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.
