Date: Sun, 7 Feb 2021 02:17:25 -0500 From: Paul Procacci <pprocacci@gmail.com> To: Aryeh Friedman <aryeh.friedman@gmail.com> Cc: Patrick Mahan <plmahan@gmail.com>, User Questions <freebsd-questions@freebsd.org> Subject: Re: Suggested upgrade for a GeForce GT 450 Message-ID: <CAFbbPujKK9_2vttv%2BL7Nj8owarYyf7XiRjBY02JhzgaL-ugYQg@mail.gmail.com> In-Reply-To: <CAFbbPuj8iswz73pZFtyxSjWmasCMgvozbVjNRN-NvE=gn_uT3A@mail.gmail.com> References: <CAFDHx1LRvcJHNnYFJB=JQzk_p%2B17i9e2qEHZOMGRCojiyaV5aw@mail.gmail.com> <CAFbbPugv13CZVVyZAtshPtn-kKM=7oHon-iQoy8dSHH2N9z3xQ@mail.gmail.com> <CAGBxaXmdUnQANVBwottmu8j_LuB9vJubK=sQxBUGRfQEFMxskA@mail.gmail.com> <CAFbbPuhOQYwJpX57cciXgH9EKv2OmtJBANkFc99UHib-Gj-Q0g@mail.gmail.com> <CAGBxaXmEjdkxWLRugWsJf6_1JqLYuNQF=xaaMANhiQ5CD=Whmg@mail.gmail.com> <CAFbbPuj8iswz73pZFtyxSjWmasCMgvozbVjNRN-NvE=gn_uT3A@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
I should also mention, it seems we're talking about two different things. There is limited CUDA support w/in the linux emulation layer for nvidia/cuda, but the native nvidia blob for FreeBSD hasn't had this support for well over the ~20 years (or more) I've been using FreeBSD. That's an important distinction. Also, it's maybe true that they've added support within the past year to the native blob, but when I was building my machine not than a year ago, it wasn't there. ~Paul On Sun, Feb 7, 2021 at 2:13 AM Paul Procacci <pprocacci@gmail.com> wrote: > It's not incorrect. The necessary bits to run anything related to CUDA > _are missing from_ FreeBSD's native nvidia blob. > > ~Paul > > On Sun, Feb 7, 2021 at 2:07 AM Aryeh Friedman <aryeh.friedman@gmail.com> > wrote: > >> Again DO NOT TOP POST! >> >> On Sun, Feb 7, 2021 at 1:56 AM Paul Procacci <pprocacci@gmail.com> wrote: >> >>> I think you're mistaken. I like Nvidia and always have liked Nvidia. >>> There are limitations to the card however when it comes to gpu >>> offloading of tasks that utilize either cuda or similar. >>> It's simply not possible. Full stop. >>> >> >> Again incorrect the documentation for the nVidia driver even discusses >> how to set up CUDA here are two of the relevant quotes from it: >> >> "Disabling acceleration is useful when another component, such as CUDA, >> requires exclusive use of the GPU's processing cores. Performance of >> the X >> server will be reduced when acceleration is disabled, and some >> features >> may not be available." >> >> "The NVIDIA driver will not perform any >> mode validation or mode setting for this X screen. This is intended >> for >> use in conjunction with CUDA or in remote graphics solutions such as >> VNC >> or Hewlett Packard's Remote Graphics Software (RGS)." >> >> From an article on the release of 12.0-BETA4 ( >> https://www.phoronix.com/scan.php?page=news_item&px=FreeBSD-12.0-Beta-4): >> >> "This is necessary for FreeBSD 64-bit CUDA support with NVIDIA's driver. >> Previously the FreeBSD CUDA support played nicely with 32-bit, but that was >> dropped in CUDA 9.0. This should help too for other 64-bit Linux emulation >> code for working with NVIDIA's binary graphics driver." >> >> -- >> Aryeh M. Friedman, Lead Developer, http://www.PetiteCloud.org >> > > > -- > __________________ > > :(){ :|:& };: > -- __________________ :(){ :|:& };:
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAFbbPujKK9_2vttv%2BL7Nj8owarYyf7XiRjBY02JhzgaL-ugYQg>