By: MoTheG (better.delete@this.not.tell), November 11, 2009 11:33 pm
Room: Moderated Discussions
? (0xe2.0x9a.0x9b@gmail.com) on 11/11/09 wrote:
>- how many percent GCC spends doing IO? What is "GCC's IO" anyway?
That certainly depends very much on the project.
>- If you parallelize the parser
I never suggested this. Read Potatoswatters reply, he understood, so I must have been lesser randomly and arbitrary than you read me.
Maybe Potatoswatter said it better.
quote:
It sounds more like you're talking about pipelining the compiler: one thread parses while another thread generates code. Unfortunately that only works when you know the relative speeds of the pipeline stages. One program might parse slowly while another program optimizes slowly due to -O3 being passed. It's too much work to modify a compiler for such an uncertain gain, although it has been done.
The way one achieves parallel compilation is by launching one compiler instance on each core. .c files are turned into .o files in parallel. Then .o's get merged together in a tree.
All this is irrelevant to GPUs as they lack the data caches necessary to run any kind of compiler. A GPU could only parse C++ at a tiny fraction of CPU speed, no matter what.
end quote
>For example I am unsure whether you actually know what a Linux process is.
I do know. I even know the difference between a programm, a process and a thread. ("wow!")
I'm getting the impression that you don't only want to save me from being a backseater/know-it-all with no realworld experience, but simply bully me.
>- how many percent GCC spends doing IO? What is "GCC's IO" anyway?
That certainly depends very much on the project.
>- If you parallelize the parser
I never suggested this. Read Potatoswatters reply, he understood, so I must have been lesser randomly and arbitrary than you read me.
Maybe Potatoswatter said it better.
quote:
It sounds more like you're talking about pipelining the compiler: one thread parses while another thread generates code. Unfortunately that only works when you know the relative speeds of the pipeline stages. One program might parse slowly while another program optimizes slowly due to -O3 being passed. It's too much work to modify a compiler for such an uncertain gain, although it has been done.
The way one achieves parallel compilation is by launching one compiler instance on each core. .c files are turned into .o files in parallel. Then .o's get merged together in a tree.
All this is irrelevant to GPUs as they lack the data caches necessary to run any kind of compiler. A GPU could only parse C++ at a tiny fraction of CPU speed, no matter what.
end quote
>For example I am unsure whether you actually know what a Linux process is.
I do know. I even know the difference between a programm, a process and a thread. ("wow!")
I'm getting the impression that you don't only want to save me from being a backseater/know-it-all with no realworld experience, but simply bully me.
Topic | Posted By | Date |
---|---|---|
Article: Computational Efficiency in Modern Processors by DK | MoTheG | 2009/11/08 06:02 AM |
Article: Computational Efficiency in Modern Processors by DK | none | 2009/11/08 06:15 AM |
Silverthorne and OoO vs. InOrd | MoTheG | 2009/11/08 06:22 AM |
Silverthorne and OoO vs. InOrd | David Kanter | 2009/11/08 03:11 PM |
Magical 100x speedups | AM | 2009/11/09 08:03 AM |
Magical 100x speedups | David Kanter | 2009/11/09 11:41 AM |
Magical 100x speedups | none | 2009/11/09 12:36 PM |
Magical speedups | David Kanter | 2009/11/09 02:24 PM |
Magical speedups | none | 2009/11/09 02:40 PM |
Hardware Specs | MS | 2009/11/09 04:49 PM |
44x faster than a single cpu core | Vincent Diepeveen | 2009/11/10 07:17 AM |
Magical speedups | Vincent Diepeveen | 2009/11/10 07:02 AM |
Xeon 130x speedup vs Xeon | Eric Bron | 2009/11/10 07:20 AM |
Magical 100x speedups | AM | 2009/11/10 09:42 AM |
Magical 100x speedups | Linus Torvalds | 2009/11/10 12:19 PM |
Mega speedups | AM | 2009/11/11 05:21 AM |
Bogus 100x speedups | David Kanter | 2009/11/10 12:26 AM |
No speedups for CPUs for the general programming populace | MoTheG | 2009/11/10 04:26 AM |
Bogus 100x speedups | ? | 2009/11/10 04:45 AM |
Bogus 100x speedups | hobold | 2009/11/10 06:31 AM |
Bogus 100x speedups | Vincent Diepeveen | 2009/11/10 07:26 AM |
Bogus 100x speedups | sylt | 2009/11/10 09:00 AM |
Bogus 100x speedups | AM | 2009/11/10 09:47 AM |
GPU vs. CPU | MoTheG | 2009/11/09 10:30 AM |
GPU vs. CPU | a reader | 2009/11/09 06:58 PM |
ease of programming | MoTheG | 2009/11/09 10:45 PM |
yes for GPU programming you need non-public info | Vincent Diepeveen | 2009/11/10 07:36 AM |
yes for GPU programming you need non-public info | Potatoswatter | 2009/11/11 07:06 AM |
yes for GPU programming you need non-public info | Vincent Diepeveen | 2009/11/11 10:23 AM |
yes for GPU programming you need non-public info | Potatoswatter | 2009/11/11 12:26 PM |
Real businesses use GPGPU. | Jouni Osmala | 2009/11/11 10:00 PM |
GPU vs. CPU | ? | 2009/11/10 05:01 AM |
2. try but most is said, just clarifying | MoTheG | 2009/11/10 09:24 AM |
2. try but most is said, just clarifying | ? | 2009/11/11 12:11 AM |
you missread me | MoTheG | 2009/11/11 11:33 PM |
you missread me | ? | 2009/11/12 12:18 AM |
2. try but most is said, just clarifying | Potatoswatter | 2009/11/11 07:22 AM |
2. try but most is said, just clarifying | ? | 2009/11/12 12:22 AM |
loose, not so orderly | MoTheG | 2009/11/12 11:47 AM |
loose, not so orderly | Potatoswatter | 2009/11/12 05:50 PM |
2. try but most is said, just clarifying | rwessel | 2009/11/12 12:01 PM |
2. try but most is said, just clarifying | Gabriele Svelto | 2009/11/12 11:39 PM |
2. try but most is said, just clarifying | ? | 2009/11/13 12:14 AM |
2. try but most is said, just clarifying | Gabriele Svelto | 2009/11/13 12:30 AM |
2. try but most is said, just clarifying | rwessel | 2009/11/13 12:24 PM |
2. try but most is said, just clarifying | Michael S | 2009/11/14 12:08 PM |
2. try but most is said, just clarifying | Gabriele Svelto | 2009/11/14 10:38 PM |
2. try but most is said, just clarifying | Andi Kleen | 2009/11/15 12:19 AM |
2. try but most is said, just clarifying | Michael S | 2009/11/15 12:58 AM |
2. try but most is said, just clarifying | Eric Bron | 2009/11/15 01:25 AM |
/MP option | Eric Bron | 2009/11/15 01:33 AM |
/MP option | Paul | 2009/11/15 08:42 AM |
/MP option | Eric Bron | 2009/11/15 12:22 PM |
2. try but most is said, just clarifying | ? | 2009/11/15 02:13 AM |
2. try but most is said, just clarifying | Michael S | 2009/11/15 04:14 AM |
2. try but most is said, just clarifying | Eugene Nalimov | 2009/11/14 08:24 PM |
Atom point | AM | 2009/11/09 08:00 AM |
Atom TDP | David Kanter | 2009/11/09 11:48 AM |
Atom TDP | hobold | 2009/11/10 06:41 AM |
Atom TDP | AM | 2009/11/10 09:49 AM |