- I think that my windows Object Pascal Processor Groups unit is optimal on NUMA systems - 8 Updates
- cmsg cancel <nkeqpt$eqo$2@dont-email.me> - 5 Updates
- escape sequence '\(' - 2 Updates
| Ramine <ramine@1.1>: Jun 22 04:12AM -0700 Hello.... I think that my windows Object Pascal Processor Groups unit is optimal on NUMA systems, because as you will notice in the source code , i am assigning the mask and the group in a round-robin fashion so that it distributes NUMA nodes efficiently on threads, so it makes the threads run optimally on NUMA systems, please read the following from Microsoft as a confirmation of what i am saying about NUMA nodes: "Each node must be fully contained within a group. If the capacities of the nodes are relatively small, the system assigns more than one node to the same group, choosing nodes that are physically close to one another for better performance. If a node's capacity exceeds the maximum number of processors in a group, the system splits the node into multiple smaller nodes, each small enough to fit in a group." Read here: https://msdn.microsoft.com/en-us/library/windows/desktop/aa363804(v=vs.85).aspx So as you have noticed Microsoft is using algorithms that optimize efficiently. So if you read the following about gabr: http://www.thedelphigeek.com/ As you have noticed he is using a function called RunInNUMANode(), look here: https://docs.google.com/document/d/1dHiJyHj80TJ2oFAOtFxNtXLMrAhzm7F5SoczBl5ssxM/edit# But i don't think i need this function because my Object Pascal unit is optimal i think, as i have just explained to you. You will find my Object Pascal unit called ProcessorGroups.pas inside the following zip file that you can download from here: https://sites.google.com/site/aminer68/an-efficient-threadpool-engine-that-scales-well Look at the test2.pas example to know how to use it. Thank you, Amine Moulay Ramdane. |
| Real Troll <real.troll@trolls.com>: Jun 22 09:19PM +0100 I thought you were told not to spam these newsgroups with your Object pascal tools. What is so difficult for you to understand this? Personally, I don't have any problems with your work because Object Pascal Code can be used in VS2015 as CLI but people here are obsessed with everything Microsoft is bad!! I don't use anything other than MS VS products. Delphi is just too expensive these days. On 22/06/2016 12:12, Ramine wrote: |
| Ramine <ramine@1.1>: Jun 22 04:39AM -0700 On 6/22/2016 1:19 PM, Real Troll wrote: I have ported other of my works in Object Pascal to C++.. Please look here to my C++ synchronization objects library that is powerfull: https://sites.google.com/site/aminer68/c-synchronization-objects-library And look here. i have just ported from Object Pascal my scalable Parallel C++ Conjugate Gradient Linear System Solver Library to C++: https://sites.google.com/site/aminer68/scalable-parallel-c-conjugate-gradient-linear-system-solver-library I will port soon more of my works to C++... Thank you, Amine Moulay Ramdane. |
| Ramine <ramine@1.1>: Jun 22 04:42AM -0700 On 6/22/2016 4:39 AM, Ramine wrote: > Please look here to my C++ synchronization objects library that is > powerfull: > https://sites.google.com/site/aminer68/c-synchronization-objects-library Those are all my algorithms that i have invented. |
| "Chris M. Thomasson" <invalid@invalid.invalid>: Jun 22 01:36PM -0700 On 6/22/2016 4:42 AM, Ramine wrote: [...] >> I will port soon more of my works to C++... [...] FWIW, make sure to test the shi% out of them, at least on Relacy: http://www.viva64.com/en/a/0041 (read all) Or, ThreadSanitzer: http://googletesting.blogspot.com/2014/06/threadsanitizer-slaughtering-data-races.html http://static.googleusercontent.com/media/research.google.com/en//pubs/archive/35604.pdf :^) |
| Real Troll <real.troll@trolls.com>: Jun 22 09:53PM +0100 thank you. You now need to change your timezone so that we can find your posts at the top of our messages. At present they get hidden down the list because the date/time is incorrect or not calculating correctly. You are using UTC timezone when in fact you should be using your own. Are you based in India? Your timezone should be UTC+9.5 or even UTC+8.5. We in UK are 1 hour ahead so our time is UTC+1 On 22/06/2016 12:42, Ramine wrote: |
| Ramine <ramine@1.1>: Jun 22 05:08AM -0700 On 6/22/2016 1:53 PM, Real Troll wrote: >>> Thank you, >>> Amine Moulay Ramdane. Thank you. I will correct this soon. Thank you, Amine Moulay Ramdane. |
| Ramine <ramine@1.1>: Jun 22 05:13AM -0700 On 6/22/2016 1:36 PM, Chris M. Thomasson wrote: > http://static.googleusercontent.com/media/research.google.com/en//pubs/archive/35604.pdf > :^) Delphi and FreePascal compilers with wich i am compiling my Dynamic Link Libraries source codes, don't reorder loads and stores, so that make the reasonning much easy, other than that my algorithms are not so difficult, for example my LW_Asym_RWLockX algorithm uses a technic that looks like seqlock without looping as seqlock, my other algorithm are more easy also to reason about, so even if i have not used tools such us relacy, i think that you can be more confident with my C++ library. Thank you, Amine Moulay Ramdane. |
| bleachbot <bleachbot@httrack.com>: Jun 22 10:01PM +0200 |
| bleachbot <bleachbot@httrack.com>: Jun 22 10:29PM +0200 |
| bleachbot <bleachbot@httrack.com>: Jun 22 10:32PM +0200 |
| bleachbot <bleachbot@httrack.com>: Jun 22 10:56PM +0200 |
| bleachbot <bleachbot@httrack.com>: Jun 22 11:01PM +0200 |
| Ben Bacarisse <ben.usenet@bsb.me.uk>: Jun 22 08:07PM +0100 >> What is the following warning given by g++ supposed to mean? >> es_test.cc:6:32: warning: unknown escape sequence: '\)' >> std::string legal="\(",illegal="\)"; <snip> > recognises '\%', '\(', '\[' and '\{', which all translate to their > unescaped versions. Additionally, it recognises '\E' and '\e', > translating both to ASCII <ESC>. My g++ version (5.2.1) rejects them all if asked to be conforming (-std=c++11 -pedantic). -- Ben. |
| scott@slp53.sl.home (Scott Lurndal): Jun 22 07:40PM >Good god, no. >Just use raw string literals. >C++11, ya know. It's already 5 years old. But not an option for everyone, unfortunately. |
| You received this digest because you're subscribed to updates for this group. You can change your settings on the group membership page. To unsubscribe from this group and stop receiving emails from it send an email to comp.lang.c+++unsubscribe@googlegroups.com. |
No comments:
Post a Comment