Re: [Haifux] Lecture proposal - high performance multi threaded programming
Great, I'm totally in. > Abstract: > With a bit of manual reading, anyone can learn how to create a program > that has more than one thread of execution. This breaks down, very > rapidly, however, as the inter-dependencies inside the program start to > burden you down, to the point where you get race bugs that are close to > impossible to find. Fixing those typically involve using the various > locking mechanisms. The result is, more often than not, a program that > both works much slower than the number of threads and processors would > suggest it should, AND at the same time still has race conditions. > > This lecture will try to give rules relating to how to construct your > program to begin with so that it will provide high performance, be > maintainable (for some definition of maintainable), and be bug > efficient. In other words, this lecture is about learning to think > "multi-threaded". In fact, efficient enough multi-threaded design do not > even need to have more than one thread of execution.. > > All in favor say "aye". > > Shachar > ___ > Haifux mailing list > Haifux@haifux.org > http://hamakor.org.il/cgi-bin/mailman/listinfo/haifux > > -- "Change is inevitable; progress is optional". ___ Haifux mailing list Haifux@haifux.org http://hamakor.org.il/cgi-bin/mailman/listinfo/haifux
Re: [Haifux] Lecture proposal - high performance multi threaded programming
Yeah, great topic! It will be great if this issue will be covered as well: http://www.airs.com/blog/archives/79 On Tue, Apr 29, 2008 at 8:56 AM, Leon Romanovsky <[EMAIL PROTECTED]> wrote: > Great, > I'm totally in. > > > > > > Abstract: > > With a bit of manual reading, anyone can learn how to create a program > > that has more than one thread of execution. This breaks down, very > > rapidly, however, as the inter-dependencies inside the program start to > > burden you down, to the point where you get race bugs that are close to > > impossible to find. Fixing those typically involve using the various > > locking mechanisms. The result is, more often than not, a program that > > both works much slower than the number of threads and processors would > > suggest it should, AND at the same time still has race conditions. > > > > This lecture will try to give rules relating to how to construct your > > program to begin with so that it will provide high performance, be > > maintainable (for some definition of maintainable), and be bug > > efficient. In other words, this lecture is about learning to think > > "multi-threaded". In fact, efficient enough multi-threaded design do not > > even need to have more than one thread of execution.. > > > > All in favor say "aye". > > > > Shachar > > ___ > > Haifux mailing list > > Haifux@haifux.org > > http://hamakor.org.il/cgi-bin/mailman/listinfo/haifux > > > > > > > -- > "Change is inevitable; progress is optional". > > > ___ > Haifux mailing list > Haifux@haifux.org > http://hamakor.org.il/cgi-bin/mailman/listinfo/haifux > -- Zaar ___ Haifux mailing list Haifux@haifux.org http://hamakor.org.il/cgi-bin/mailman/listinfo/haifux
Re: [Haifux] Lecture proposal - high performance multi threaded programming
Shachar Shemesh wrote: > Abstract: > With a bit of manual reading, anyone can learn how to create a program > that has more than one thread of execution. *cough* How about defining "anyone"? ;) > All in favor say "aye". > I say "Yay!" Eli -- Web: http://www.billauer.co.il ___ Haifux mailing list Haifux@haifux.org http://hamakor.org.il/cgi-bin/mailman/listinfo/haifux
Re: [Haifux] Lecture proposal - high performance multi threaded programming
Eli Billauer wrote: > Shachar Shemesh wrote: > > >> Abstract: >> With a bit of manual reading, anyone can learn how to create a program >> that has more than one thread of execution. >> > *cough* > > How about defining "anyone"? ;) > Fine, I'll cover the basics of multi-threaded programming, but I won't go into the API discussion. It's not really relevant. Shachar ___ Haifux mailing list Haifux@haifux.org http://hamakor.org.il/cgi-bin/mailman/listinfo/haifux
Re: [Haifux] Lecture proposal - high performance multi threaded programming
Hai Zaar wrote: > Yeah, great topic! > It will be great if this issue will be covered as well: > http://www.airs.com/blog/archives/79 > First, I've only read this through twice, and so have not, yet, understood what the compiler is doing to this code. As such, I don't feel qualified to comment. The other aspect is that the lecture is more algorithmic than implementation. I'm trying to answer the "how do you structure your program so that algorithmic complexity doesn't kill you of an heart attack before you are 30". Just as the article points out, a simple memory barrier here would solve the problem, so this is obviously an implementation question. I'll try to point out some of these problems, but that won't be the focus. Shachar ___ Haifux mailing list Haifux@haifux.org http://hamakor.org.il/cgi-bin/mailman/listinfo/haifux
Re: [Haifux] Lecture proposal - high performance multi threaded programming
Shachar Shemesh wrote: > Fine, I'll cover the basics of multi-threaded programming, but I won't > go into the API discussion. It's not really relevant. > Actually, my remark about your use of the word "anyone" was more like "isn't grandma somebody?". I'm fine with going right to business about threads. But having that subject brought up, I just realized out that Guy Keren's legendary POSIX Thread lecture set (see http://www.haifux.org/lectures/51/ ) was six years ago. Pheew! Maybe a rerun is in place? Eli -- Web: http://www.billauer.co.il ___ Haifux mailing list Haifux@haifux.org http://hamakor.org.il/cgi-bin/mailman/listinfo/haifux
Re: [Haifux] Lecture proposal - high performance multi threaded programming
On Tue, Apr 29, 2008 at 5:21 PM, Shachar Shemesh <[EMAIL PROTECTED]> wrote: > Hai Zaar wrote: > > > Yeah, great topic! > > It will be great if this issue will be covered as well: > > http://www.airs.com/blog/archives/79 > > > > > First, I've only read this through twice, and so have not, yet, understood > what the compiler is doing to this code. As such, I don't feel qualified to > comment. > > The other aspect is that the lecture is more algorithmic than > implementation. I'm trying to answer the "how do you structure your program > so that algorithmic complexity doesn't kill you of an heart attack before > you are 30". Just as the article points out, a simple memory barrier here > would solve the problem, so this is obviously an implementation question. Any remarks about the what memory barrier is would be appreciated during the lecture. > > > I'll try to point out some of these problems, but that won't be the focus. > > Shachar > -- Zaar ___ Haifux mailing list Haifux@haifux.org http://hamakor.org.il/cgi-bin/mailman/listinfo/haifux
Re: [Haifux] Lecture proposal - high performance multi threaded programming
Great! aye. DS On Tue, Apr 29, 2008 at 8:30 AM, Shachar Shemesh <[EMAIL PROTECTED]> wrote: > Abstract: > With a bit of manual reading, anyone can learn how to create a program > that has more than one thread of execution. This breaks down, very > rapidly, however, as the inter-dependencies inside the program start to > burden you down, to the point where you get race bugs that are close to > impossible to find. Fixing those typically involve using the various > locking mechanisms. The result is, more often than not, a program that > both works much slower than the number of threads and processors would > suggest it should, AND at the same time still has race conditions. > > This lecture will try to give rules relating to how to construct your > program to begin with so that it will provide high performance, be > maintainable (for some definition of maintainable), and be bug > efficient. In other words, this lecture is about learning to think > "multi-threaded". In fact, efficient enough multi-threaded design do not > even need to have more than one thread of execution.. > > All in favor say "aye". > > Shachar > ___ > Haifux mailing list > Haifux@haifux.org > http://hamakor.org.il/cgi-bin/mailman/listinfo/haifux > ___ Haifux mailing list Haifux@haifux.org http://hamakor.org.il/cgi-bin/mailman/listinfo/haifux
Re: [Haifux] Lecture proposal - high performance multi threaded programming
Thanks for the offer, I am in. Please not in the period 11-22/may/2008 as I'll be in mil. (If some one wants to hear more about Wireshark, I can do that after the 22-may-2008) Nir. On Tue, 29 Apr 2008, Shachar Shemesh wrote: > Hai Zaar wrote: > Yeah, great topic! > It will be great if this issue will be covered as well: > http://www.airs.com/blog/archives/79 > First, I've only read this through twice, and so have not, yet, understood what the compiler is doing to this code. As such, I don't feel qualified to comment. The other aspect is that the lecture is more algorithmic than implementation. I'm trying to answer the "how do you structure your program so that algorithmic complexity doesn't kill you of an heart attackbefore you are 30". Just as the article points out, a simple memory barrier here would solve the problem, so this is obviously an implementation question. I'll try to point out some of these problems, but that won't be the focus. Shachar ___ Haifux mailing list Haifux@haifux.org http://hamakor.org.il/cgi-bin/mailman/listinfo/haifux ___ Haifux mailing list Haifux@haifux.org http://hamakor.org.il/cgi-bin/mailman/listinfo/haifux
Re: [Haifux] Lecture proposal - high performance multi threaded programming
Hi, >If some one wants to hear more about Wireshark, I can do that after >the 22-may-2008) Thanks ,this will be great ! I am all for it ! DS On Wed, Apr 30, 2008 at 12:06 PM, Nir Abulaffio <[EMAIL PROTECTED]> wrote: > Thanks for the offer, I am in. > Please not in the period 11-22/may/2008 as I'll be in mil. > > (If some one wants to hear more about Wireshark, I can do that after the > 22-may-2008) > Nir. > > > On Tue, 29 Apr 2008, Shachar Shemesh wrote: > > Hai Zaar wrote: > > Yeah, great topic! > > It will be great if this issue will be covered as well: > > http://www.airs.com/blog/archives/79 > > > First, I've only read this through twice, and so have not, yet, > understood what the compiler is doing to this code. As such, I don't > feel qualified to comment. > > The other aspect is that the lecture is more algorithmic than > implementation. I'm trying to answer the "how do you structure your > program so that algorithmic complexity doesn't kill you of an heart > attackbefore you are 30". Just as the article points out, a simple > > > memory barrier here would solve the problem, so this is obviously an > implementation question. > > > I'll try to point out some of these problems, but that won't be the focus. > > Shachar > ___ > Haifux mailing list > Haifux@haifux.org > http://hamakor.org.il/cgi-bin/mailman/listinfo/haifux > ___ > Haifux mailing list > Haifux@haifux.org > http://hamakor.org.il/cgi-bin/mailman/listinfo/haifux > ___ Haifux mailing list Haifux@haifux.org http://hamakor.org.il/cgi-bin/mailman/listinfo/haifux
Re: [Haifux] Lecture proposal - high performance multi threaded programming
Another aye from here. David. On Tue, Apr 29, 2008 at 8:30 AM, Shachar Shemesh <[EMAIL PROTECTED]> wrote: > Abstract: > With a bit of manual reading, anyone can learn how to create a program > that has more than one thread of execution. This breaks down, very > rapidly, however, as the inter-dependencies inside the program start to > burden you down, to the point where you get race bugs that are close to > impossible to find. Fixing those typically involve using the various > locking mechanisms. The result is, more often than not, a program that > both works much slower than the number of threads and processors would > suggest it should, AND at the same time still has race conditions. > > This lecture will try to give rules relating to how to construct your > program to begin with so that it will provide high performance, be > maintainable (for some definition of maintainable), and be bug > efficient. In other words, this lecture is about learning to think > "multi-threaded". In fact, efficient enough multi-threaded design do not > even need to have more than one thread of execution.. > > All in favor say "aye". > > Shachar > ___ > Haifux mailing list > Haifux@haifux.org > http://hamakor.org.il/cgi-bin/mailman/listinfo/haifux > ___ Haifux mailing list Haifux@haifux.org http://hamakor.org.il/cgi-bin/mailman/listinfo/haifux