Re: rocblas and llama.cpp performance profiling

To: Debian ROCm Team <debian-ai@lists.debian.org>
Subject: Re: rocblas and llama.cpp performance profiling
From: Christian Kastner <ckk@debian.org>
Date: Fri, 7 Feb 2025 09:42:26 +0100
Message-id: <[🔎] 373acbd0-255c-4ef9-8063-d196723feda0@debian.org>
In-reply-to: <[🔎] 8fe91676-35fb-45db-8270-6f4749a64218@slerp.xyz>
References: <[🔎] 8fe91676-35fb-45db-8270-6f4749a64218@slerp.xyz>

On 2025-02-07 02:06, Cordell Bloor wrote:
> I wanted to share a bit of information on how I do performance profiling
> for AMD GPU BLAS operations on Debian [...]

Thanks Cory, this was not just interesting but very enlightening.

Best,
Christian

Reply to:

References:
- rocblas and llama.cpp performance profiling
  - From: Cordell Bloor <cgmb@slerp.xyz>

Prev by Date: Re: Architectures for packages
Next by Date: Re: ci.rocm.debian.net: gfx1100, gfx1101 broken with (possibly) 6.12.11
Previous by thread: rocblas and llama.cpp performance profiling
Next by thread: Re: rocblas and llama.cpp performance profiling
Index(es):
- Date
- Thread