On Wed, Feb 16, 2011 at 7:37 PM, Glen <speck...@specktech.com> wrote:
> rather than be peeved, I'd like to simply know where the problem is, make > it apparent to someone who cares, and have it get fixed, so that I can > deliver solutions based on Apple technology. > The problem is that the drivers don't support CL, but more importantly, have no GL Texture support. You can still deliver solutions based on Apple technology, just with less performant results. Nothing says that card has to support those features though, afaik. Maybe my wording of "being peeved" was off-task, but I in the face of being told that the hardware I bought (the computer en total) *does* support the tech (just not well), I would be peeved. I don't think anyone buys a tower with the expectation of a mini working better. > If I need to buy the Nvidia 4000,in order to apply OpenCL to my > problem/solution, ok, I need to know that too (whether I like it or not.. > Maybe Apple will swap cards) > Remember, OpenCL is pretty limited in use. I can't think of any major app that uses it (and I'm a bit of an OpenCL enthusiast, to say the least). For it not to be supported on the GPU with some cards is really no big deal to a co. like Apple (in my opinion), even though it basically sucks, for a very, very, verrrrry, small sect of users. > > So what is the best way to find a person who will take responsibility to > track this down? I can't believe Apple doesn't know that the ATI cards fail > their sample code. > There are *tons* of sample projects that don't work currently, or to varying degrees on different computers. There is some seriously sloppy stuff there in the archives. I don't think that sample code running or not can be used as a litmus (but can obviously point to issues). You can file a bug report, and take special care to point to the fact of what gpu and build you're running, and that whoever tests the bug needs to test on that gpu and particular build, so that they don't just try it out on whatever they have at hand. -gt > > On Feb 16, 2011, at 3:56 PM, George Toledo <gtole...@gmail.com> wrote: > > > > On Wed, Feb 16, 2011 at 6:28 PM, Patrick Sheffield <<psheffi...@earthlink.net> > psheffi...@earthlink.net> wrote: > >> I thought only Nvidia supported OpenCL without emulation? >> >> p >> >> > ATI has an actual OpenCL support area on their site, so I'm unsure about > that. > > I'm totally unsurprised about the result of the Grass and Terrain Example. > > I've had many conversations about this with a wide variety of ATI owners. > My understanding is that it's that the drivers are semi-inadequate, not > necessarily that the gpu's are strictly incapable of being performant. > > All I can say is that it's sort of .... special.... when you can go out and > get a mac mini and have it to perform better than a rocked-out tower. If I > had personally bought an ATI card with some idea of great graphics in SL, I > would be peeved. > > -gt > >> On Feb 16, 2011, at 2:53 PM, Glen Speckert wrote: >> >> There is an example called OpenCL Procedural Grass and Terrain Example. >> >> When trying this on my Westmere w/5870 it fails, while on a Mini with 9400 >> it works fine. >> >> The Mac Pro says: >> >> GNU gdb 6.3.50-20050815 (Apple version gdb-1518) (Thu Jan 27 08:34:47 UTC >> 2011) >> Copyright 2004 Free Software Foundation, Inc. >> GDB is free software, covered by the GNU General Public License, and you >> are >> welcome to change it and/or distribute copies of it under certain >> conditions. >> Type "show copying" to see the conditions. >> There is absolutely no warranty for GDB. Type "show warranty" for >> details. >> This GDB was configured as "x86_64-apple-darwin".tty /dev/ttys000 >> [Switching to process 17873 thread 0x0] >> * >> ---------------------------------------------------------------------------- >> * >> *Compute Engine: Using active OpenGL context...* >> * >> ---------------------------------------------------------------------------- >> * >> *Creating command queue for AMD ATI Radeon HD 5870...* >> * >> ---------------------------------------------------------------------------- >> * >> *Creating Buffer Object (6291456 bytes)...* >> *Creating Buffer Object (6291456 bytes)...* >> *Creating Buffer Object (262144 bytes)...* >> *Creating Buffer Object (262144 bytes)...* >> *Creating Buffer Object (262144 bytes)...* >> *Grass Simulator: BladeCount[16384] RowCount[1000] ColumnCount[1000]* >> *Compute Engine: Loading program 'gs' from file 'grass_kernels.cl'...* >> * >> ---------------------------------------------------------------------------- >> * >> *Building compute program 'gs'...* >> *Error: Failed to build program executable!* >> *OpenCL Error[-11]: Unknown error!* >> *Build log for device '0':* >> *Compute Engine: Empty Build Log! Unkown Error!* >> * >> ---------------------------------------------------------------------------- >> * >> *Grass Simulator: Failed to create device kernel!* >> Program ended with exit code: 1 >> >> >> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ >> >> However, when I run this on a Mac Mini, I get: >> >> GNU gdb 6.3.50-20050815 (Apple version gdb-1518) (Thu Jan 27 08:34:47 UTC >> 2011) >> Copyright 2004 Free Software Foundation, Inc. >> GDB is free software, covered by the GNU General Public License, and you >> are >> welcome to change it and/or distribute copies of it under certain >> conditions. >> Type "show copying" to see the conditions. >> There is absolutely no warranty for GDB. Type "show warranty" for >> details. >> This GDB was configured as "x86_64-apple-darwin".tty /dev/ttys001 >> [Switching to process 45120 thread 0x0] >> * >> ---------------------------------------------------------------------------- >> * >> *Compute Engine: Using active OpenGL context...* >> * >> ---------------------------------------------------------------------------- >> * >> *Creating command queue for NVIDIA GeForce 9400...* >> * >> ---------------------------------------------------------------------------- >> * >> *Creating Buffer Object (6291456 bytes)...* >> *Creating Buffer Object (6291456 bytes)...* >> *Creating Buffer Object (262144 bytes)...* >> *Creating Buffer Object (262144 bytes)...* >> *Creating Buffer Object (262144 bytes)...* >> *Grass Simulator: BladeCount[16384] RowCount[1000] ColumnCount[1000]* >> *Compute Engine: Loading program 'gs' from file 'grass_kernels.cl'...* >> * >> ---------------------------------------------------------------------------- >> * >> *Building compute program 'gs'...* >> * >> ---------------------------------------------------------------------------- >> * >> *Compute Engine: Creating kernel 'ComputeGrassOnTerrainKernel' for >> program 'gs'...* >> *Compute Engine: Creating OpenGL buffer reference 'grass_vertices' for >> buffer id '1'...* >> *Compute Engine: Creating OpenGL buffer reference 'grass_colors' for >> buffer id '2'...* >> *Terrain Simulator: Vertices[16384] Count[128, 128]* >> *Compute Engine: Loading program 'tk' from file 'terrain_kernels.cl'...* >> * >> ---------------------------------------------------------------------------- >> * >> *Building compute program 'tk'...* >> * >> ---------------------------------------------------------------------------- >> * >> *Compute Engine: Creating kernel 'ComputeTerrainKernel' for program >> 'tk'...* >> *Compute Engine: Creating OpenGL buffer reference 'terrain_vertices' for >> buffer id '3'...* >> *Compute Engine: Creating OpenGL buffer reference 'terrain_normals' for >> buffer id '4'...* >> *Compute Engine: Creating OpenGL buffer reference 'terrain_texcoords' for >> buffer id '5'...* >> On Feb 6, 2011, at 9:32 AM, George Toledo wrote: >> >> Apologies; make that 4870. >> >> On Sun, Feb 6, 2011 at 12:26 PM, George Toledo < <gtole...@gmail.com> >> gtole...@gmail.com> wrote: >> >>> Is this card known to perform badly with QC? I had some issues with CL, >>> GLSL and feedback loops (separately) that gave me serious pause when using >>> this gpu. I was really surprised. It's not a stretch to say that my intel >>> 3100/macbook 13" works better than a mac pro loaded with the ati. I seem to >>> remember that it's not CL supported (I forgot to check w/ the cl info >>> patch), but I was surprised that it just performs badly, in general. Is this >>> consistent with the experience of others? >>> >>> -- >>> George Toledo >>> <gtole...@gmail.com>gtole...@gmail.com >>> <http://www.georgetoledo.com/>www.georgetoledo.com >>> >>> The information contained in this E-mail and any attachments may be >>> confidential. >>> If you have received this E-mail in error, please notify us immediately >>> by telephone or return E-mail. >>> You should not use or disclose the contents of this E-mail or any of the >>> attachments for any purpose or to any persons. >>> >>> >> >> >> -- >> George Toledo >> <gtole...@gmail.com>gtole...@gmail.com >> <http://www.georgetoledo.com/>www.georgetoledo.com >> >> The information contained in this E-mail and any attachments may be >> confidential. >> If you have received this E-mail in error, please notify us immediately by >> telephone or return E-mail. >> You should not use or disclose the contents of this E-mail or any of the >> attachments for any purpose or to any persons. >> >> _______________________________________________ >> Do not post admin requests to the list. They will be ignored. >> Quartzcomposer-dev mailing list (<Quartzcomposer-dev@lists.apple.com> >> Quartzcomposer-dev@lists.apple.com) >> Help/Unsubscribe/Update your Subscription: >> >> <http://lists.apple.com/mailman/options/quartzcomposer-dev/speckert%40specktech.com> >> http://lists.apple.com/mailman/options/quartzcomposer-dev/speckert%40specktech.com >> >> This email sent to <speck...@specktech.com>speck...@specktech.com >> >> >> _______________________________________________ >> Do not post admin requests to the list. They will be ignored. >> Quartzcomposer-dev mailing list (<Quartzcomposer-dev@lists.apple.com> >> Quartzcomposer-dev@lists.apple.com) >> Help/Unsubscribe/Update your Subscription: >> >> <http://lists.apple.com/mailman/options/quartzcomposer-dev/psheffield%40earthlink.net> >> http://lists.apple.com/mailman/options/quartzcomposer-dev/psheffield%40earthlink.net >> >> This email sent to <psheffi...@earthlink.net>psheffi...@earthlink.net >> >> >> > > > -- > George Toledo > <gtole...@gmail.com>gtole...@gmail.com > <http://www.georgetoledo.com>www.georgetoledo.com > > The information contained in this E-mail and any attachments may be > confidential. > If you have received this E-mail in error, please notify us immediately by > telephone or return E-mail. > You should not use or disclose the contents of this E-mail or any of the > attachments for any purpose or to any persons. > > -- George Toledo gtole...@gmail.com www.georgetoledo.com The information contained in this E-mail and any attachments may be confidential. If you have received this E-mail in error, please notify us immediately by telephone or return E-mail. You should not use or disclose the contents of this E-mail or any of the attachments for any purpose or to any persons.
_______________________________________________ Do not post admin requests to the list. They will be ignored. Quartzcomposer-dev mailing list (Quartzcomposer-dev@lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/quartzcomposer-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
