On 26 November 2014 at 00:14, Yichao Yu <yyc1992 at gmail.com> wrote:>> We're using MCJIT on ARM to implement an OpenCL on CPU implementation. It works! > > Thanks. Great to hear that!!! > > Maybe could someone please update that page.... It's really confusing.Hi Yichao, So, MCJIT works well for x86 and PPC, and it works to a degree on ARM. Tom's team has been successfully using it for OpenCL, and this is a very good use case, but I wouldn't dare say it's "supported" on ARM without more people showing it working on ARM on other codebases/projects. We could update the documentation to say ARM is supported, but the danger in that is that people will come blazing when they find really basic bugs on ARM. I haven't had the time yet in a long while to investigate MCJIT, but last I checked it was working ok. We still have some XFAIL on the tests, and that's what keeps me from claiming support. Maybe they're bad tests, maybe it's just broken. I'll see if I can find some time to look at this more closely and update the docs. But the end result is the same, please use it, and if you break it, we should fix it. thanks, --renato
It's been working for ARM's OpenCL on CPU driver for several years without fault too. On Wed, 26 Nov 2014 at 20:17, Renato Golin <renato.golin at linaro.org> wrote:> On 26 November 2014 at 00:14, Yichao Yu <yyc1992 at gmail.com> wrote: > >> We're using MCJIT on ARM to implement an OpenCL on CPU implementation. > It works! > > > > Thanks. Great to hear that!!! > > > > Maybe could someone please update that page.... It's really confusing. > > Hi Yichao, > > So, MCJIT works well for x86 and PPC, and it works to a degree on ARM. > Tom's team has been successfully using it for OpenCL, and this is a > very good use case, but I wouldn't dare say it's "supported" on ARM > without more people showing it working on ARM on other > codebases/projects. > > We could update the documentation to say ARM is supported, but the > danger in that is that people will come blazing when they find really > basic bugs on ARM. I haven't had the time yet in a long while to > investigate MCJIT, but last I checked it was working ok. We still have > some XFAIL on the tests, and that's what keeps me from claiming > support. Maybe they're bad tests, maybe it's just broken. > > I'll see if I can find some time to look at this more closely and > update the docs. But the end result is the same, please use it, and if > you break it, we should fix it. > > thanks, > --renato > _______________________________________________ > LLVM Developers mailing list > LLVMdev at cs.uiuc.edu http://llvm.cs.uiuc.edu > http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev >-------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20141126/d05e0856/attachment.html>
Hi Renato, On Wed, Nov 26, 2014 at 2:52 PM, Renato Golin <renato.golin at linaro.org> wrote:> On 26 November 2014 at 00:14, Yichao Yu <yyc1992 at gmail.com> wrote: >>> We're using MCJIT on ARM to implement an OpenCL on CPU implementation. It works! >> >> Thanks. Great to hear that!!! >> >> Maybe could someone please update that page.... It's really confusing. > > Hi Yichao, > > So, MCJIT works well for x86 and PPC, and it works to a degree on ARM. > Tom's team has been successfully using it for OpenCL, and this is a > very good use case, but I wouldn't dare say it's "supported" on ARM > without more people showing it working on ARM on other > codebases/projects. > > We could update the documentation to say ARM is supported, but the > danger in that is that people will come blazing when they find really > basic bugs on ARM. I haven't had the time yet in a long while to > investigate MCJIT, but last I checked it was working ok. We still have > some XFAIL on the tests, and that's what keeps me from claiming > support. Maybe they're bad tests, maybe it's just broken. > > I'll see if I can find some time to look at this more closely and > update the docs. But the end result is the same, please use it, and if > you break it, we should fix it.Thank you very much for the explaination. What I need should be fairly simple (basic flow control) but I will ask here if I see an issue.> > thanks, > --renatoYichao Yu
On 26 November 2014 at 21:31, Yichao Yu <yyc1992 at gmail.com> wrote:> Thank you very much for the explaination. What I need should be fairly > simple (basic flow control) but I will ask here if I see an issue.Please do. Maybe it's time to mark it as supported and cope with the bugs that are left... cheers, --renato