# Ampere

Published articles for Ampere.

This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles.

## Security Week 2636: атака GPUThor обходит защиту ECC

DevFeed: [Security Week 2636: атака GPUThor обходит защиту ECC](<https://devfeed.tech/articles/security-week-2636-gputhor-ecc-23093.md>)

Original publisher: [Read original article](<https://habr.com/ru/companies/kaspersky/articles/1076720/>)

Author: Kaspersky\_Lab ("Лаборатория Касперского")

Published: 2026-08-31T14:24:08Z

Content type: news

Language: ru

Sources: ["Лаборатория Касперского" RU](<https://devfeed.tech/sources/ru-2.md>)

Topics: [rowhammer](<https://devfeed.tech/topics/rowhammer.md>), [Security](<https://devfeed.tech/topics/security.md>), [Nvidia](<https://devfeed.tech/topics/nvidia.md>)

Tags: [2026](<https://devfeed.tech/tags/2026.md>), [ampere](<https://devfeed.tech/tags/ampere.md>), [gpu](<https://devfeed.tech/tags/gpu.md>), [nvidia](<https://devfeed.tech/tags/nvidia.md>), [rowhammer](<https://devfeed.tech/tags/rowhammer.md>), [security](<https://devfeed.tech/tags/security.md>), [tag-9fe8963de219](<https://devfeed.tech/tags/tag-9fe8963de219.md>), [tag-cc2a6346d835](<https://devfeed.tech/tags/tag-cc2a6346d835.md>), [thor](<https://devfeed.tech/tags/thor.md>)

### AI overview

Researchers from the University of Toronto demonstrated GPUThor, a Rowhammer attack targeting NVIDIA Ampere A4000, A4500, A5000, and A6000 professional GPUs. The attack can produce double and triple bit flips that ECC does not correct, enabling denial-of-service and theoretically creating conditions for privilege escalation.

### Source excerpt

Исследователи из канадского Университета Торонто на прошлой неделе опубликовали научную работу, в которой продемонстрировали новую атаку на видеоускорители NVIDIA. Атака, получившая название GPUThor, относится к классу Rowhammer, то есть использует многократные обращения к ячейкам оперативной памяти с целью повлиять на соседние ячейки. Таким образом можно изменить данные в областях памяти, изначально недоступных потенциальному злоумышленнику. Наиболее актуален такой сценарий атаки в случае совместного доступа к профессиональному видеоускорителю. Именно поэтому в подобных работах традиционно исследуются устройства NVIDIA, в данном случае модели поколения Ampere A4000, A4500, A5000 и A6000. По сравнению с предыдущими атаками на подобные устройства, продемонстрированными в начале 2026 года, GPUThor обеспечивает изменение данных в целевых ячейках в сотни и даже тысячи раз чаще. Но самое главное -- новая атака в некоторых случаях приводит к двойным и тройным бит-флипам, которые не корректируются системой ECC. Это точно позволяет провести атаку типа "отказ в обслуживании" и теоретически создает условия для атаки с повышением привилегий, даже если коррекция ошибок включена. Читать далее

## Proxmox officially supports Arm, with some caveats

DevFeed: [Proxmox officially supports Arm, with some caveats](<https://devfeed.tech/articles/proxmox-officially-supports-arm-with-some-caveats-10479.md>)

Original publisher: [Read original article](<https://www.jeffgeerling.com/blog/2026/proxmox-ve-arm-official/>)

Author: jeff@jeffgeerling.com (Jeff Geerling)

Published: 2026-08-05T16:50:00Z

Content type: article

Language: en

Sources: [Jeff Geerling](<https://devfeed.tech/sources/jeff-geerling.md>)

Topics: [Hardware](<https://devfeed.tech/topics/hardware.md>), [systems](<https://devfeed.tech/topics/systems.md>), [Ubuntu](<https://devfeed.tech/topics/ubuntu.md>), [Linux](<https://devfeed.tech/topics/linux.md>)

Tags: [ampere](<https://devfeed.tech/tags/ampere.md>), [arm](<https://devfeed.tech/tags/arm.md>), [homelab](<https://devfeed.tech/tags/homelab.md>), [linux](<https://devfeed.tech/tags/linux.md>), [nvidia](<https://devfeed.tech/tags/nvidia.md>), [platform](<https://devfeed.tech/tags/platform.md>), [proxmox](<https://devfeed.tech/tags/proxmox.md>), [raspberry-pi](<https://devfeed.tech/tags/raspberry-pi.md>), [raspberry-pi-5](<https://devfeed.tech/tags/raspberry-pi-5.md>), [ubuntu](<https://devfeed.tech/tags/ubuntu.md>), [usb](<https://devfeed.tech/tags/usb.md>)

### AI overview

Proxmox Virtual Environment is now officially available for 64-bit ARM, with installation tested on an Ampere Altra platform. Official support is limited to NVIDIA Grace Hopper and Vera systems, while other UEFI- and ACPI-based ARM hardware receives best-effort support. Device-tree-only single-board computers are not officially supported, though community UEFI projects have enabled installations on Raspberry Pi systems.

### Source excerpt

Proxmox today announced their Proxmox Virtual Environment is now available for 64-bit ARM. I tested it on my Ampere Altra Dev Platform--the same machine on which I've booted Windows on Arm the first time, messed with multiple GPUs, and most recently tested Houdini's native arm64 support. Install was easy, as the Ampere Altra uses UEFI / ACPI for its hardware, meaning Proxmox didn't have to tailor its ISO to specific platforms, like you have to do with Raspberry Pis and most SBCs using a Device Tree setup.

## Mixed-input matrix multiplication performance optimizations

DevFeed: [Mixed-input matrix multiplication performance optimizations](<https://devfeed.tech/articles/mixed-input-matrix-multiplication-performance-optimizations-28543.md>)

Original publisher: [Read original article](<http://blog.research.google/2024/01/mixed-input-matrix-multiplication.html>)

Author: Google AI (noreply@blogger.com)

Published: 2024-01-26T19:56:00Z

Content type: article

Language: en

Sources: [Google Research](<https://devfeed.tech/sources/google-research.md>)

Topics: [Large language models (LLMs)](<https://devfeed.tech/topics/large-language-models-llms.md>), [quantization](<https://devfeed.tech/topics/quantization.md>), [Hardware](<https://devfeed.tech/topics/hardware.md>), [GPU](<https://devfeed.tech/topics/gpu.md>), [Nvidia](<https://devfeed.tech/topics/nvidia.md>), [Tensor Cores](<https://devfeed.tech/topics/tensor-cores.md>), [Open Source](<https://devfeed.tech/topics/open-source.md>), [Software](<https://devfeed.tech/topics/software.md>), [data type](<https://devfeed.tech/topics/data-type.md>)

Tags: [accelerators](<https://devfeed.tech/tags/accelerators.md>), [ai](<https://devfeed.tech/tags/ai.md>), [algorithms](<https://devfeed.tech/tags/algorithms.md>), [ampere](<https://devfeed.tech/tags/ampere.md>), [compute](<https://devfeed.tech/tags/compute.md>), [conversion](<https://devfeed.tech/tags/conversion.md>), [data](<https://devfeed.tech/tags/data.md>), [deep-learning](<https://devfeed.tech/tags/deep-learning.md>), [effective](<https://devfeed.tech/tags/effective.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [hardware](<https://devfeed.tech/tags/hardware.md>), [implementation](<https://devfeed.tech/tags/implementation.md>), [language-models](<https://devfeed.tech/tags/language-models.md>), [large-language-models-llms](<https://devfeed.tech/tags/large-language-models-llms.md>), [models](<https://devfeed.tech/tags/models.md>), [nvidia](<https://devfeed.tech/tags/nvidia.md>), [optimization](<https://devfeed.tech/tags/optimization.md>), [performance](<https://devfeed.tech/tags/performance.md>), [precision](<https://devfeed.tech/tags/precision.md>), [quantization](<https://devfeed.tech/tags/quantization.md>), [research](<https://devfeed.tech/tags/research.md>), [software](<https://devfeed.tech/tags/software.md>), [tensor-cores](<https://devfeed.tech/tags/tensor-cores.md>)

### AI overview

This Google Research article explains software techniques for mapping mixed-input matrix multiplication onto NVIDIA Ampere hardware. It describes using lower-precision weights with higher-precision inputs, data-type conversion, and layout transformations to support weight-only quantization. The authors report minimal software overhead and performance close to peak hardware capabilities, and state that the techniques were released in the open-source NVIDIA/CUTLASS repository.

### Source excerpt

Posted by Manish Gupta, Staff Software Engineer, Google Research AI-driven technologies are weaving themselves into the fabric of our daily routines, with the potential to enhance our access to knowledge and boost our overall productivity. The backbone of these applications lies in large language models (LLMs). LLMs are memory-intensive and typically require specialized hardware accelerators to efficiently deliver tens of exaflops of computing power. This blog post shows how we can start addressing the computational challenges by utilizing memory more effectively. The bulk of an LLM's memory and compute are consumed by weights in matrix multiplication operations. Using narrower data types reduces memory consumption. For example, storing weights in the 8-bit integer (i.e., U8 or S8) data type reduces the memory footprint by 4x relative to single-precision (F32) and 2x relative to half-precision (F16) or bfloat16 (BF16). Furthermore, previous work has shown that LLM models running matrix multiplications with weights in S8 and input in F16 (preserving higher precision of the user-input) is an effective method for increasing the efficiency with acceptable trade-offs in accuracy. This technique is known as weight-only quantization and requires efficient implementation of matrix multiplication with mixed-inputs, e.g., half-precision input multiplied with 8-bits integer. Hardware accelerators, including GPUs, support a fixed set of data types, and thus, mixed-input matrix multiplication requires software transformations to map to the hardware operations. To that end, in this blog we focus on mapping mixed-input matrix multiplication onto the NVIDIA Ampere architecture. We present software techniques addressing data type conversion and layout conformance to map mixed-input matrix multiplication efficiently onto hardware-supported data types and layouts. Our results show that the overhead of additional work in software is minimal and enables performance close to the peak har