AMD fixes HandBrake bottlenecks on Threadripper systems

HandBrake 1.11.0 includes AMD-backed threading and scheduling fixes that help high-core-count Threadripper CPUs stay busy during transcoding. Tested gains ranged from 16% to 215% across workloads.

AMD has addressed threading bottlenecks in HandBrake that limited transcoding performance on high-core-count systems, with the fixes now included in HandBrake 1.11.0. The issues involved inefficient handling of systems with more than 64 logical processors and workloads divided into jobs that were too small, which increased scheduling overhead instead of improving throughput.

The problems were especially visible on Threadripper systems, where available compute could sit idle as more cores were assigned. AMD found cases where performance dropped by up to 60% rather than scaling, with 720p workloads particularly affected. The updated thread management and job scheduling are designed to distribute transcoding work more effectively across many cores.

According to AMD, a Threadripper 7980X (64 cores, 128 threads) with 128 GB of DDR5-5600 and a Radeon RX 9070 XT delivered gains up to 215% in Perfume H.264 720p, 203% in LG_8K HEVC 8-bit 4320p and 105% in LG 8K 60 FPS HEVC 10-bit 4320p. Across tested workloads, improvements ranged from 16% to 215%. A Threadripper PRO 9995WX (96 cores, 192 threads) with the same RX 9070 XT reached gains of up to 181% in HEVC, depending on the workload.

53

Impact Score

France brings NVIDIA-backed AI infrastructure into production

France’s AI buildout is shifting from planning to deployment, with new compute capacity, open models and enterprise applications moving into active use. The ecosystem spans cloud providers, startups, public institutions and major companies across healthcare, telecom, manufacturing, energy and beauty.

FERC moves to speed grid connections for large power users

Federal regulators have advanced a large-load interconnection framework aimed at helping AI factories, semiconductor projects and advanced manufacturing sites connect to the grid more quickly while limiting costs for existing customers.

General Atlantic weighs lead role in Kling AI funding round

General Atlantic is in talks to lead a funding round of more than $2 billion for Kuaishou’s Kling AI, with a proposed post-investment valuation of $18 billion. The discussions come as Kuaishou prepares the video generation unit for outside capital and a future public listing.

Imec advances ferroelectric memory for AI workloads

Imec is testing ferroelectric capacitors and vertically stacked FeFETs as possible routes to denser, lower-voltage memory for data-intensive systems. The work targets scaling limits in DRAM and SRAM as AI workloads increase memory demands.

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.