Registers in modern processors get renamed. Intel gets away with having such few logical registers in their ISA (instruction-set architecture) because they have dozens of physical registers.
All hyperthreading will do is just maintain a different program counter and re-order buffer for each thread. There are probably other minor details as well, but don't get caught up in registers from a programmer's point of view. There is magic under the hood that the programmer will never ever be aware of. At some point in your program, their may be 8 or so "EAX" registers. Later on, this same register may be renamed to a "ESP" register.
Simultaneous Multithreading (SMT) is not a new idea, although no one to my knowledge has implemented it yet. Intel just calls it "Hyperthreading"...it is essentially SMT.
And yes, this is a very good idea. A modern superscaler out-of-order processor, like the Athlon and Pentium Pro (and later), can issue and retire multiple instructions per clock cycle. However, it can *only* do this if there is enough instruction-level parallelism (ILP). Turns out, there is not enough ILP in current programs to take full advantage of the chips processing capabilities. Issue slots and function units go unused due to dependencies in the program and cache misses that stall the processing. A typical processor can only look at about 32 instructions at a time. This is not a large enough window to execute future instructions out-of-order when such a stall occurs.
However, 2 threads of execution will likely fill all of the issue slots. They are also independent threads of execution, so dependencies don't exist between them. This means that when the pipeline stalls due to a cache miss, the other thread can keep on retiring instructions.
To all those saying that this is dumb, I suggest you study some modern architecture (I'm not talking about your undergrad architecture course either). A paper I read recently studied the affects of SMT on a simulated Alpha processor. The results were astounding with very little changes to the processor core. I heard that the next Alpha was slated to include SMT before Intel killed it.
Just because Win9x gets GPFs doesn't mean they run in user mode. The kernel can use virtual-memory capabilities. GPFs are a result of a page fault. They happen in kernel mode as well.
I know this is somewhat offtopic, however to make Linux more responsive, we need to improve X somehow. I am not saying that X sucks...I think it is a fantastic system
Anybody who uses X and Windows regularly knows the difference in responsiveness. X Windows does what it was for designed extremely well-- a client/server display system. However, due to the marshalling and de-marshalling of X calls, even if completely local, it will always be less responsive than other methods (winblows).
But I have an idea. Develop a system that implements the exact same interface as X but does no marshalling/demarshalling. Pixels can be written directly to the framebuffer. So you are thinking, "Yeah, but I want to use X apps without recompiling". Ok, use library interposition. This also allows you to use a "local" and "global" X library to maintain client/server capabilities. For those who aren't familar which library interpositioning, it essentially takes advantage of dynamic linking (set LD_LIBRARY_PATH on Unix). If you want to run a X program that directly writes to the framebuffer, then switch your LD_LIBRARY_PATH to a different directory before the program is executed. This could get annoying, but a Window Manager like Gnome could take care of this automatically.
Granted that our existing X server would have to be retrofitted to allow 2 different types of X libraries to update the same display to that we can run standard client/server X apps with the new "directXfree86" (no pun intended) apps.
However library interpositioning can be used to make X programs more responsive without sacrificing client/server capabilities and compatibility with existing applications (except those statically linked of course).
compiler technology crucial
on
Inside the Itanium
·
· Score: 2, Interesting
Itanium is making the compiler do alot of work! This presents a gigantic challenge for compiler writers. My concern is with GCC. We all know that GCC does not produce the tightest, fastest code. For IA-32, this is not a big deal because there is only so much optimization that can be done with that ISA (instruction set architecture). However with Itanium, the compiler will probably affect runtime execution speed by 100% or more. Will the GCC people have the resources and expertise to handle IA-64 (Itanium)??
interpretation is the only way to guarantee safety
on
Bill Joy's Takes on C#
·
· Score: 3, Insightful
Java is safe because it is interpreted. Sure, it is compiled, but the compiled code doesn't run on hardware.
If.NET code is interpreted, then they can make it safe. If they have a silly marker saying "This code is safe because it doesn't operate on memory directly", then that is just silly because some hacker can easily remove this marker.
Running code downloaded from the network, directly on your hardware, will always be somewhat dangerous. Of course that is what operating systems are for. However, there is always some way to figure out how to run malicious code in a privileged fashion.
I don't blame Linus for trying out Bitkeeper. After using Rational (formerly Atria) Clearcase for several years, and then working on a project using CVS, I can't believe that CVS is hyped the way it is. Sure, it works for small projects.
Why nothing can beat Clearcase:
1. Clearcase is integrated into the OS and the repositories are actual filesystems. I create a "config_spec" that lists all of the branches and labels that I want to see. I don't have to checkout stuff, and copy a large project into a private directory. If I want to see another developer's change to a small set of files, I just add his/her branch to my config_spec. No recopying or anything like that.
2. Clearmake. If someone else in my organization builds the same object file (from the exact same version of the source file), I don't have to rebuild it. I love typing "clearmake" of something huge that I've never built before, and it is done in no time!! If something does have to be rebuilt, builds are distributed across a dozen or so machines.
3. Real branches. When I am developing something, I make a temporary branch on all the files. If I want to be able to let someone else use my "development" branches, I just let them know what the branch name is. When it is production-ready, we just merge the development branch into the main branch. No such thing as patch files in Clearcase world.
There is a reason that Clearcase is expensive and proprietary.
The Linux people should have realized that if AMD truly is "Pentium"-compatible, which is what AMD advertises, then there is no such thing as a Windows 2000 bug. Think about it.
To be classified as Pentium-compatible, the only difference between the behavior of an Athlon and a PIII should be the return value of the instruction that gives you the CPU ID.
I've always been an Intel-supporter. At least their chips don't burn up if your heat sink falls off. And their power consumption has generally been half of AMD's chips.
I avoided the early AMD chips because of compatibility issues. However I was planning on trying an AMD Duron chip because I thought that AMD had a decent chip out now. Guess I'm wrong...
Buying AMD will eventually bite you in the ass somehow someway.
No...UltraSPARCs are the choice for enterprise applications that are I/O intensive. However our simulation's are CPU-bound are require very little I/O. PC's have a better price/performance ratio for pure CPU performance...much better! Plus we use Condor to run several simulations at the same time on Linux.
Thats right, Microsoft is the one that calls the shots, not Intel. Back during the design of the Pentium Pro, which formed the basis of the PII and PIII, Intel wanted to implement a FPU that featured far more than 4 floating point registers. However, Microsoft refused to modify Windows to save those registers on a context switch. The result? Intel used only 4 FP registers which severely crippled its performance.
I recently tried installing FreeBSD on my old P166 machine. This box has seen about a dozen Linux distributions dating back to Slackware 3.0 with no problems.
The FreeBSD installation program freezes nearly immediately after booting the CD. There is nothing I could do (BIOS, options, etc) to get the installation to go.
So I installed NetBSD (with no problems) and am extremely impressed. I highly recommend NetBSD to anyone wanting a non-bloated pure UNIX installation. It just works and is very nice.
The package system blows me away. Adding a new package is as simple as 'pkg_add mozilla'. It downloads the package and all dependencies and installs flawlessly. I have not tried Debian's package system, but BSD's pkg and pkgsrc system is in my mind perfect.
Microsoft doesn't do code inspections. I believe that this explains alot of their "bugs". Any software development team, where quality is critical, inspects *every* line of code. And that doesn't mean "make the rookies do code inspections". I'm talking seasoned programmers spending hours and hours walking through code.
"Monsters, Inc." does not use Linux at all. They use Sun UltraEnterprise servers....about 3500 CPUs total. They are currently looking into Linux as a rendering machine, however that would require major rewrites of Pixar's proprietary software called RenderMan. Pixar does not use Maya which runs on Linux.
You are ripping on a man for leaving Microsoft. Do you really think he was responsible for all of Microsoft's security flaws? Do you really think that one man, no matter what level he is at, is capable of turning a corporate culture around? Do you think that one man, whom isn't the CEO, is capable of changing a company's philosophy?
If you do, you've never worked for a large company.
Whether you like them or not, Microsoft has a lot of smart people working for them. Grow up everybody...I'm sure he is very qualified.
Like I'm going to try this kernel! I used to grab kernel sources nearly right away, but with the latest releases, I'm going to wait until any serious bugs are found.
Slashdot Mirror
Registers in modern processors get renamed. Intel gets away with having such few logical registers in their ISA (instruction-set architecture) because they have dozens of physical registers.
All hyperthreading will do is just maintain a different program counter and re-order buffer for each thread. There are probably other minor details as well, but don't get caught up in registers from a programmer's point of view. There is magic under the hood that the programmer will never ever be aware of. At some point in your program, their may be 8 or so "EAX" registers. Later on, this same register may be renamed to a "ESP" register.
Simultaneous Multithreading (SMT) is not a new idea, although no one to my knowledge has implemented it yet. Intel just calls it "Hyperthreading"...it is essentially SMT.
And yes, this is a very good idea. A modern superscaler out-of-order processor, like the Athlon and Pentium Pro (and later), can issue and retire multiple instructions per clock cycle. However, it can *only* do this if there is enough instruction-level parallelism (ILP). Turns out, there is not enough ILP in current programs to take full advantage of the chips processing capabilities. Issue slots and function units go unused due to dependencies in the program and cache misses that stall the processing. A typical processor can only look at about 32 instructions at a time. This is not a large enough window to execute future instructions out-of-order when such a stall occurs.
However, 2 threads of execution will likely fill all of the issue slots. They are also independent threads of execution, so dependencies don't exist between them. This means that when the pipeline stalls due to a cache miss, the other thread can keep on retiring instructions.
To all those saying that this is dumb, I suggest you study some modern architecture (I'm not talking about your undergrad architecture course either). A paper I read recently studied the affects of SMT on a simulated Alpha processor. The results were astounding with very little changes to the processor core. I heard that the next Alpha was slated to include SMT before Intel killed it.
Just because Win9x gets GPFs doesn't mean they run in user mode. The kernel can use virtual-memory capabilities. GPFs are a result of a page fault. They happen in kernel mode as well.
I know this is somewhat offtopic, however to make Linux more responsive, we need to improve X somehow. I am not saying that X sucks...I think it is a fantastic system
Anybody who uses X and Windows regularly knows the difference in responsiveness. X Windows does what it was for designed extremely well-- a client/server display system. However, due to the marshalling and de-marshalling of X calls, even if completely local, it will always be less responsive than other methods (winblows).
But I have an idea. Develop a system that implements the exact same interface as X but does no marshalling/demarshalling. Pixels can be written directly to the framebuffer. So you are thinking, "Yeah, but I want to use X apps without recompiling". Ok, use library interposition. This also allows you to use a "local" and "global" X library to maintain client/server capabilities. For those who aren't familar which library interpositioning, it essentially takes advantage of dynamic linking (set LD_LIBRARY_PATH on Unix). If you want to run a X program that directly writes to the framebuffer, then switch your LD_LIBRARY_PATH to a different directory before the program is executed. This could get annoying, but a Window Manager like Gnome could take care of this automatically.
Granted that our existing X server would have to be retrofitted to allow 2 different types of X libraries to update the same display to that we can run standard client/server X apps with the new "directXfree86" (no pun intended) apps.
However library interpositioning can be used to make X programs more responsive without sacrificing client/server capabilities and compatibility with existing applications (except those statically linked of course).
Itanium is making the compiler do alot of work! This presents a gigantic challenge for compiler writers. My concern is with GCC. We all know that GCC does not produce the tightest, fastest code. For IA-32, this is not a big deal because there is only so much optimization that can be done with that ISA (instruction set architecture). However with Itanium, the compiler will probably affect runtime execution speed by 100% or more. Will the GCC people have the resources and expertise to handle IA-64 (Itanium)??
Java is safe because it is interpreted. Sure, it is compiled, but the compiled code doesn't run on hardware.
.NET code is interpreted, then they can make it safe. If they have a silly marker saying "This code is safe because it doesn't operate on memory directly", then that is just silly because some hacker can easily remove this marker.
If
Running code downloaded from the network, directly on your hardware, will always be somewhat dangerous. Of course that is what operating systems are for. However, there is always some way to figure out how to run malicious code in a privileged fashion.
I don't blame Linus for trying out Bitkeeper. After using Rational (formerly Atria) Clearcase for several years, and then working on a project using CVS, I can't believe that CVS is hyped the way it is. Sure, it works for small projects.
Why nothing can beat Clearcase:
1. Clearcase is integrated into the OS and the repositories are actual filesystems. I create a "config_spec" that lists all of the branches and labels that I want to see. I don't have to checkout stuff, and copy a large project into a private directory. If I want to see another developer's change to a small set of files, I just add his/her branch to my config_spec. No recopying or anything like that.
2. Clearmake. If someone else in my organization builds the same object file (from the exact same version of the source file), I don't have to rebuild it. I love typing "clearmake" of something huge that I've never built before, and it is done in no time!! If something does have to be rebuilt, builds are distributed across a dozen or so machines.
3. Real branches. When I am developing something, I make a temporary branch on all the files. If I want to be able to let someone else use my "development" branches, I just let them know what the branch name is. When it is production-ready, we just merge the development branch into the main branch. No such thing as patch files in Clearcase world.
There is a reason that Clearcase is expensive and proprietary.
Clearcase is a VFS (virtual file system). I set a "config_spec", do a 'ls', and I see the files I want. That is why CVS is a child's toy to me.
The way he was portrayed in the movie, there is absolutely no way that such a man could score a woman as beautiful as Jennifer Connelly!
The Linux people should have realized that if AMD truly is "Pentium"-compatible, which is what AMD advertises, then there is no such thing as a Windows 2000 bug. Think about it.
To be classified as Pentium-compatible, the only difference between the behavior of an Athlon and a PIII should be the return value of the instruction that gives you the CPU ID.
I just purchased a 900 MHz T-Bird Athlon for another Linux box today. Hopefully it can turn my opinion of AMD...
I've always been an Intel-supporter. At least their chips don't burn up if your heat sink falls off. And their power consumption has generally been half of AMD's chips.
I avoided the early AMD chips because of compatibility issues. However I was planning on trying an AMD Duron chip because I thought that AMD had a decent chip out now. Guess I'm wrong...
Buying AMD will eventually bite you in the ass somehow someway.
Go ahead...flame me.
No...UltraSPARCs are the choice for enterprise applications that are I/O intensive. However our simulation's are CPU-bound are require very little I/O. PC's have a better price/performance ratio for pure CPU performance...much better! Plus we use Condor to run several simulations at the same time on Linux.
Whats pathetic is that Windows NT (2000 and XP) can only run on little-endian machines.
Try running microprocessor simulations. We need all the power we can get. I'm talking simulations that run for over a month...
Thats right, Microsoft is the one that calls the shots, not Intel. Back during the design of the Pentium Pro, which formed the basis of the PII and PIII, Intel wanted to implement a FPU that featured far more than 4 floating point registers. However, Microsoft refused to modify Windows to save those registers on a context switch. The result? Intel used only 4 FP registers which severely crippled its performance.
I recently tried installing FreeBSD on my old P166 machine. This box has seen about a dozen Linux distributions dating back to Slackware 3.0 with no problems.
The FreeBSD installation program freezes nearly immediately after booting the CD. There is nothing I could do (BIOS, options, etc) to get the installation to go.
So I installed NetBSD (with no problems) and am extremely impressed. I highly recommend NetBSD to anyone wanting a non-bloated pure UNIX installation. It just works and is very nice.
The package system blows me away. Adding a new package is as simple as 'pkg_add mozilla'. It downloads the package and all dependencies and installs flawlessly. I have not tried Debian's package system, but BSD's pkg and pkgsrc system is in my mind perfect.
At my university, it crashes every other day.
Microsoft doesn't do code inspections. I believe that this explains alot of their "bugs". Any software development team, where quality is critical, inspects *every* line of code. And that doesn't mean "make the rookies do code inspections". I'm talking seasoned programmers spending hours and hours walking through code.
Intel has always won the money battle. They have about $25 billion in cash. Meanwhile, AMD is still in the red.
"Monsters, Inc." does not use Linux at all. They use Sun UltraEnterprise servers....about 3500 CPUs total. They are currently looking into Linux as a rendering machine, however that would require major rewrites of Pixar's proprietary software called RenderMan. Pixar does not use Maya which runs on Linux.
You are ripping on a man for leaving Microsoft. Do you really think he was responsible for all of Microsoft's security flaws? Do you really think that one man, no matter what level he is at, is capable of turning a corporate culture around? Do you think that one man, whom isn't the CEO, is capable of changing a company's philosophy?
If you do, you've never worked for a large company.
Whether you like them or not, Microsoft has a lot of smart people working for them. Grow up everybody...I'm sure he is very qualified.
Anyone can have sex, but the question is with whom?!!!
CDMA IS-95a is a standard that defines everything.
GSM also has a similar standard, but the word "GSM" does not define everything.
Like I'm going to try this kernel! I used to grab kernel sources nearly right away, but with the latest releases, I'm going to wait until any serious bugs are found.