For my real code, big applications only slow down 10-20%. I've measured it. Now, this is on Solaris/SPARC, which I suspect is a much more efficient platform than x86. How much slowdown do others see?
One of the many great things about purify is that (IME) it only slows down your code by 10-20%, which is small enough that you can always leave it in your code. Leaving it in for unit testing, integration testing, system testing, beta testing, etc., can make your life much easier.
Valgrind, however, runs your code 20-50 times slower, which means you can't have it on all the time. This is unfortunately, for it looks like a great tool, otherwise.
I get a ton of junk mail. Who doesn't? It usually gets tossed, unopened. Every now and then, I've tossed non-junk mail, as it looked like junk mail. It would be interesting to measure this "cost" of junk-mail.
It isn't just gprof that's broken by pthreads, other Linux tools fall victim as well. Core dump? Almost useless with pthreads running. Gdb? Getting better, but still a little wonky. Certain aspects of signal handling don't work as expected with pthreads.
BSD unix was a non-cleanroom reimplimentation of AT&T unix. BSD won when it went to court.
But only an organization like BSD, backed by the University of California and their lawyers, had the resources to stand up to AT&T in court. I wouldn't suggest being cavalier about clean-room issues to any random Open Source project.
This page has lots of paragraphs about distributed computing this, and xmlrpc that, and pretty animated gifs showing binary ones and zeros zipping from one computer to the other, but NOTHING about the actual algorithms or chess going on.
What is it actually doing? A complete tree-search for all the legal chess moves? That's a pretty big tree! Searching for conclusions to well-known games? Trying to crack into a Norwegian librarian's database?
After they get passed in the EU it is much easier to get them passed in the USA.
Huh? How about some evidence for this? Do you think that the Kyoto protocols are going to be mandated in the US because they've been accepted in Europe? Ditto for socialized medicine? Mandated shorter work weeks? The Euro?
In Blind Man's Bluff, the book covers tapping a coax cable in the 70s, by placing a device which detects RF below it, without physically modifying the cable at all. This is a lot different that tapping an optical fiber, but still pretty tricky to do.
There was a wired article about how the advertising company Chiat-Day started the "hot desk" concept.
Not coincidentally, the company tanked soon after this started, and had to be sold in order to survive. In their new offices, traditional offices are the rule of the day.
A couple of weeks ago, I went to Networld + Interop. At the vendor pavillion, I chatted with lots of engineers and designers of networking equipment of all kinds. 80% of the companies I talked to used Linux or *BSD. None of them were using any of the commercial embedded linux distributions, they all rolled their own, or started by hacking up a desktop distro.
Relatively few people are buying commercial embedded Linux distros, so these companies are going under (e.g Lineo). There's a lot of press, picking up on this trend, implying that embedded Linux is dead.
However, the fact that engineers don't need to buy expensive support to get their embedded open source OSes to run is really a positive sign for embedded Linux, despite the cries that it may be dead.
Wouldn't it be funny to form some BSA-like group, which would goose-step into corporate offices, auditing software. When the group found unlicensed software (or, more likely), software which probably is licensed, but for which no license can be found, it would either mandate that their victims pony up zillions of dollars -- OR, switch to OSS software.
I wonder how fast the BSA would shut down such a group?
He missed the one thing I want -- cheap clocks that set themselves. I've got cheap digital clocks in my VCR, TV, Microwave, coffee maker, etc. etc. Keeping them all set to the right time, especially when power is lost, is a real pain. They are never synchronized to each other, much less the right time. (Yes, my computers do run NTP, but that's another story).
I've seen clock radios which know the time via WWV, but that's a bit expensive to put into all these appliances -- there's several different ways you could do this, but I want one that just works -- maybe a time signal could be broadcast over the power cables? It needn't add but a few pennies to the cost of the item, and would make my life tremendously easier.
My cell phone sets its clock from the basestation automatically, and doesn't even have a way to manually set it. This is my favorite feature of my phone -- the time is always right.
Nice idea, but keep in mind that static compilers are extremely difficult to create for Itanium. Performance results I've seen show that while the theoretical maximum for IA-64 is pretty impressive, the actual results static compilers are generating are not so hot.
Now, try to write a dynamic, JIT compiler for Itanium, which is even hardware than a static compiler. I haven't seen any java or CLR performance numbers for IA-64, and suspect I know the reason why.:-)
Isn't it interesting how all the TiVo people crow about how hackable it is because it runs a linux kernel, and yet all the same kinds of hacks are possible on the Replay, which runs a closed-source OS.
Some huge percentage of my friends with (unhacked) TiVO's have had to send them back because of hardware failure. I think our peerless CmdrTaco falls into the same boat. I gotta think that a reputation for shabby quality has to have an effect on sales.
Of course, 300k units doesn't sound like a complete failure to me.
USENET -- works in practice, but not in theory
on
How Google Saved USENET
·
· Score: 4, Interesting
I find it very interesting that in the last 10 years of USENET, it's traffic (and presumably use) have grown dramatically. However, the number of servers has, I believe, dropped equally dramatically. USENET was one of the most distributed systems I remember using, with it's shared-nothing, "flood-fill" algorithm.
Yet, as it scales up to more and more messages, it actually is becoming less distributed. A good lesson for all the futurists forcasting the rise of distributed systems...
Eclipse is the best looking Java app I've ever seen. Congrats to IBM for taking the bold step of not using AWT/Swing, and replacing it with something decent (SWT/JFace).
Slashdot Mirror
For my real code, big applications only slow down 10-20%. I've measured it. Now, this is on Solaris/SPARC, which I suspect is a much more efficient platform than x86. How much slowdown do others see?
One of the many great things about purify is that (IME) it only slows down your code by 10-20%, which is small enough that you can always leave it in your code. Leaving it in for unit testing, integration testing, system testing, beta testing, etc., can make your life much easier.
Valgrind, however, runs your code 20-50 times slower, which means you can't have it on all the time. This is unfortunately, for it looks like a great tool, otherwise.
Two words: sendmail
I get a ton of junk mail. Who doesn't? It usually gets tossed, unopened. Every now and then, I've tossed non-junk mail, as it looked like junk mail. It would be interesting to measure this "cost" of junk-mail.
It isn't just gprof that's broken by pthreads, other Linux tools fall victim as well. Core dump? Almost useless with pthreads running. Gdb? Getting better, but still a little wonky. Certain aspects of signal handling don't work as expected with pthreads.
BSD unix was a non-cleanroom reimplimentation of AT&T unix. BSD won when it went to court.
But only an organization like BSD, backed by the University of California and their lawyers, had the resources to stand up to AT&T in court. I wouldn't suggest being cavalier about clean-room issues to any random Open Source project.
What is it actually doing? A complete tree-search for all the legal chess moves? That's a pretty big tree! Searching for conclusions to well-known games? Trying to crack into a Norwegian librarian's database?
Huh? How about some evidence for this? Do you think that the Kyoto protocols are going to be mandated in the US because they've been accepted in Europe? Ditto for socialized medicine? Mandated shorter work weeks? The Euro?
Check out this. Slashdot hasn't won a web award since April of 2000. That must mean Slashdot is going downhill, right?
is to put the jokes in the source.
"You are not expected to understand this".
In Blind Man's Bluff, the book covers tapping a coax cable in the 70s, by placing a device which detects RF below it, without physically modifying the cable at all. This is a lot different that tapping an optical fiber, but still pretty tricky to do.
"'You come in early, you get a good a parking space, you get a good office.' Chief Executive Scott McNealy has summed up the iWork program. "
How early do I have to come in to get Scott's office?
Not coincidentally, the company tanked soon after this started, and had to be sold in order to survive. In their new offices, traditional offices are the rule of the day.
Relatively few people are buying commercial embedded Linux distros, so these companies are going under (e.g Lineo). There's a lot of press, picking up on this trend, implying that embedded Linux is dead.
However, the fact that engineers don't need to buy expensive support to get their embedded open source OSes to run is really a positive sign for embedded Linux, despite the cries that it may be dead.
Wouldn't it be funny to form some BSA-like group, which would goose-step into corporate offices, auditing software. When the group found unlicensed software (or, more likely), software which probably is licensed, but for which no license can be found, it would either mandate that their victims pony up zillions of dollars -- OR, switch to OSS software.
I wonder how fast the BSA would shut down such a group?
He missed the one thing I want -- cheap clocks that set themselves. I've got cheap digital clocks in my VCR, TV, Microwave, coffee maker, etc. etc. Keeping them all set to the right time, especially when power is lost, is a real pain. They are never synchronized to each other, much less the right time. (Yes, my computers do run NTP, but that's another story).
I've seen clock radios which know the time via WWV, but that's a bit expensive to put into all these appliances -- there's several different ways you could do this, but I want one that just works -- maybe a time signal could be broadcast over the power cables? It needn't add but a few pennies to the cost of the item, and would make my life tremendously easier.
My cell phone sets its clock from the basestation automatically, and doesn't even have a way to manually set it. This is my favorite feature of my phone -- the time is always right.
Can't we have this for appliances?
That's why I used an alias for mine, I can get the "normal" prices and the gov doesn't have to know about my M&M addiction
Do you always pay in cash? Otherwise with a credit/debit card or check, its easy to associate your real name with your card.
Do you mean StrToInt?
If so, I'm always surprised at how many people miss the common case of correctly parsing the most negative integer.
Praise grid for libraries, in this case.
I suppose that is reasonable. How many security problems have you found in your audit?
What about the compilers? Are you sure they aren't inserting any funny business into the binary code?
While we're at it, what about the CPU, and other support chips? Have you inspected the VHDL?
Nice idea, but keep in mind that static compilers are extremely difficult to create for Itanium. Performance results I've seen show that while the theoretical maximum for IA-64 is pretty impressive, the actual results static compilers are generating are not so hot.
:-)
Now, try to write a dynamic, JIT compiler for Itanium, which is even hardware than a static compiler. I haven't seen any java or CLR performance numbers for IA-64, and suspect I know the reason why.
Isn't it interesting how all the TiVo people crow about how hackable it is because it runs a linux kernel, and yet all the same kinds of hacks are possible on the Replay, which runs a closed-source OS.
Some huge percentage of my friends with (unhacked) TiVO's have had to send them back because of hardware failure. I think our peerless CmdrTaco falls into the same boat. I gotta think that a reputation for shabby quality has to have an effect on sales.
Of course, 300k units doesn't sound like a complete failure to me.
I find it very interesting that in the last 10 years of USENET, it's traffic (and presumably use) have grown dramatically. However, the number of servers has, I believe, dropped equally dramatically. USENET was one of the most distributed systems I remember using, with it's shared-nothing, "flood-fill" algorithm.
Yet, as it scales up to more and more messages, it actually is becoming less distributed. A good lesson for all the futurists forcasting the rise of distributed systems...
Eclipse is the best looking Java app I've ever seen. Congrats to IBM for taking the bold step of not using AWT/Swing, and replacing it with something decent (SWT/JFace).