linux-kernel - Re: [PATCH v2 00/15] misc: nnpi: New PCIe driver for Intel's NNP-I pcie device

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20210615145755.GA25911@aipg-stp-03.iil.intel.com>
Date:   Tue, 15 Jun 2021 17:58:09 +0300
From:   Guy Zadicario <guy.zadicario@...el.com>
To:     Greg KH <gregkh@...uxfoundation.org>
Cc:     "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "olof@...om.net" <olof@...om.net>,
        "alexander.shishkin@...ux.intel.com" 
        <alexander.shishkin@...ux.intel.com>,
        "Shevchenko, Andriy" <andriy.shevchenko@...el.com>,
        "Shefi-simchon, Yochai" <yochai.shefi-simchon@...el.com>
Subject: Re: [PATCH v2 00/15] misc: nnpi: New PCIe driver for Intel's NNP-I
 pcie device

On Fri, May 14, 2021 at 11:33:31AM +0300, Greg KH wrote:
> On Thu, May 13, 2021 at 11:57:10AM +0300, Guy Zadicario wrote:
> > Hi,
> > 
> > The following series is a driver for a new PCIe device from Intel named NNP-I
> > (Nirvana Neural Processor for Inference). NNP-I is a PCIe connected compute
> > device used for acceleration of AI deep learning inference applications in the
> > data-center.
> 
> How does this differ from the "intel_gna" driver being proposed here:
> 	https://lore.kernel.org/linux-doc/20210513110040.2268-1-maciej.kwapulinski@linux.intel.com/
> 
> Please work together to come up with a unified userspace api for these
> devices, and hopefully, shared kernel code as well.
> 
Ok, after consulting, we will try to share api and some kernel code
between the two drivers.

The plan is to suggest a common framework for both devices
and let each device driver register with that framework,
some functionality will be common and some will be device driver
specific (same model as drm for gpus). At first I think the common part
will be small and will deal only with host memory management, later,
it might grow, if/when some other drivers will adopt it as well.

We will first change the NNP-I driver to use this framework.
The GNA driver will wait for this API to be accepted and only then will
adopt it and request to upstream.

Thanks,
Guy.