Slashdot Mirror

Fred Brooks original silver bullet paper

As for the "Brewing WW3 in the middle east," that's nothing new. For one thing it's been brewing for five thousand years. Even Israel invading its neighbors in the name of protecting itself is nothing new. On that topic I would like to quote a passage, "A Sermon on Ethics and Love" from the Principia Discordia:

One day Mal-2 asked the messenger spirit Saint Gulik to approach the Goddess and request Her presence for some desperate advice. Shortly afterwards the radio came on by itself, and an ethereal female Voice said YES?

"O! Eris! Blessed Mother of Man! Queen of Chaos! Daughter of Discord! Concubine of Confusion! O! Exquisite Lady, I beseech You to lift a heavy burden from my heart!"

WHAT BOTHERS YOU, MAL? YOU DON'T SOUND WELL.

"I am filled with fear and tormented with terrible visions of pain. Everywhere people are hurting one another, the planet is rampant with injustices, whole societies plunder groups of their own people, mothers imprison sons, children perish while brothers war. O, woe."

WHAT IS THE MATTER WITH THAT, IF IT IS WHAT YOU WANT TO DO?

"But nobody Wants it! Everybody hates it."

OH. WELL, THEN STOP.

At which moment She turned herself into an aspirin commercial and left The Polyfather stranded alone with his species.

In my introductory engineering/programming classes at Purdue University, the professors/TAs use a program called MOSS to check for code plagiarism. Apparently it's free to use; just email them to get an account.

Considering it's an intro class, I am sure lot of students will unintentionally end up with similar programs. http://www.cs.berkeley.edu/~aiken/moss.html

You forgot to add, "Nation-, national security information.".

Link goes to one transcript. Here's a list of related material - the second set is more relevant but doesn't produce quotes as lovely as the above example.

In one of the very first papers mentioning the Semantic Web, some paragraph was devoted to something then lost in the hype around the semantic web: the Web of trust, which had to be something like a certification of metadata. This is perhaps to be again regarded as important for the semantic web and the web in general (although not easy to manage).
By the way, Norvig is not only a Google exec, but also a well known AI researcher, author of one of most important books on that subject.

depression is more likely for people who don't get out much:

and heart disease but other than that, no, you should be just fine without a real life, er, I mean without real life.

My project is a paradigm, too.

There's more information on Kolmogorov's scaling laws here, not that I understand most of it. As far as I can tell, in a turbulent system the difference between the values of a physical property at two points follows a power law with respect to the distance between the points; the power laws for different physical properties have different exponents, but they all seem to be multiples of a third (?).

I wonder where people get the idea that OSX is fast. Apple marketing?

Most benchmarks that i've seen seem to indicate the opposite.
http://sekhon.berkeley.edu/macosx/

According to the update here, Sekhon acknowledges the discovery of the performance issue with malloc. He does, however, still find issues with medium-sized datasets. There only logical explanation for this is darwin system call overhead (discussed in the article I linked, AND the comments of the blog post you linked).

Next time, read the page you link.

Mac OS/X will then compete directly with Windows, and though it's faster, more stable and more secure, Windows has that whole 90%+ market share thing going for it.

I wonder where people get the idea that OSX is fast. Apple marketing?
Most benchmarks that i've seen seem to indicate the opposite.
http://sekhon.berkeley.edu/macosx/

Even my X41 Thinkpad with it's Pentium M 1.6GHz running debian testing with stock kernel does time echo "scale=5000; 4*a(1)" | bc -l faster (1m9s) than MacBook Pro 2GHz running OSX (1m18s). The very same MacBook Pro does (0m52s) when running linux.

Not very good benchmarks I know, but i'd like to see some prove that OSX does anything faster than windows or linux.

There are intel mac versions available. The BOINC application is a universal binary which will request the correct version of SETI@home for the architecture used. See this page for the stock application list. (And yes, the intel mac version does use SSE for its FFTs).

AIDS _and_ climate prediction? Which client is that? And how do they determine which is more important, the AIDS or the climate prediction? or do they just split it 50-50?

The client is BOINC.

You can run a number of different projects concurrently, and choose what percentage of your computer's resources are allocated to each.

Ummmm, the most recent stock SETI@home enhanced uses Altivec for FFT calculations. Better optimized versions for G4 and G5 are also available here if you want them.

SETI@home is GPL. Has been for some time. BOINC is LGPL. Has been for some time.

Which are "those" clients that you are talking about?

SETI@home is GPL. Has been for some time. BOINC is LGPL. Has been for some time.

Which are "those" clients that you are talking about?

I don't think so either. I also used to be a SETI@Home supporter and supplied them with over 2 years of CPU time. Then I found THE Extraterrestrial Intelligent life form and did not even need a computer or telescope. I have since dedicated my CPU time to things that are more important to me.

My SETI@Home Stats - http://setiathome.berkeley.edu/show_user.php?useri d=133483
The source of my own personal SETI discovery - http://www.hoei.com/blog/tlb/

Individuals contributing their spare processor cycles via BOINC are currently producing over 380 TeraFLOPS putting them clearly in first place (if such distributed systems were counted).

SETI@Home is now operated exclusively through BOINC and it alone is doing over 167 TeraFLOPS right now, putting the SETI@Home network in second place, only behind BlueGene/L (if such distributed systems were counted).

You can contribute your spare processor cycles too by downloading the BOINC client and attaching to a cool project such as Rosetta@Home which folds proteins as part of an effort to cure human diseases. Join the biggest "supercomputer" today!

Individuals contributing their spare processor cycles via BOINC are currently producing over 380 TeraFLOPS putting them clearly in first place (if such distributed systems were counted).

SETI@Home is now operated exclusively through BOINC and it alone is doing over 167 TeraFLOPS right now, putting the SETI@Home network in second place, only behind BlueGene/L (if such distributed systems were counted).

You can contribute your spare processor cycles too by downloading the BOINC client and attaching to a cool project such as Rosetta@Home which folds proteins as part of an effort to cure human diseases. Join the biggest "supercomputer" today!

Individuals contributing their spare processor cycles via BOINC are currently producing over 380 TeraFLOPS putting them clearly in first place (if such distributed systems were counted).

SETI@Home is now operated exclusively through BOINC and it alone is doing over 167 TeraFLOPS right now, putting the SETI@Home network in second place, only behind BlueGene/L (if such distributed systems were counted).

You can contribute your spare processor cycles too by downloading the BOINC client and attaching to a cool project such as Rosetta@Home which folds proteins as part of an effort to cure human diseases. Join the biggest "supercomputer" today!

Did they not know that SETI runs on the BOIC platform, which is open-source? So if you want to do some modelling, just write a BOINC plugin, and maybe people will install & run it.

Chip H.

It's not free, but you might enjoy this anyway.
Understanding the Human Body: An Introduction to Anatomy and Physiology

I've listened to lectures from the teaching company. They're very good. The only question I'd have is at what level are these lectures being presented. Usually, these courses are introductory or of a survey nature at best.

Also, these courses are free as well:
an introductory anatomy class from Berkeley

video lectures from "The neuronal basis of conciousness" course at Caltech

I know this may not be quite what your looking for. I've found that it's pretty rare for professors to post advanced undergraduate or introductory graduate lectures online.

Well, actually, you can, or at least could. Sort of. There was a weakeness in the encyrption algorithms of the older SIM cards that let you recover exactly that - the Ki value. As I understood the attack, it involved putting the SIM in a smart card reader and subjecting it to about half an hour's worth of continiuous querying.

I think there was some way you could work out the Ki by sending challange after challange and analysing the responses it gave.

I'd recommend reading this and the subsequent updates linked to from there. Yes, it is from 1998, and I'm hoping that modern SIM cards are more difficult (impossible?) to clone.

It's also possible there are countermeasures too these days built into the SIMs - as you point out, they are active devices.

You can extract the KI out of some SIMs by brute force.
I have 2 SIMs with the same KI and IMSI (and 2 phones with the same IMEI).
The last one to register on the network receives all incoming calls, you cannot make 2 calls simultanously, when a call is made on one it drops the call on the other.
This is what is does when both are in the same cell anyway.... i've not tried it with 2 phones on different cells or in different countries yet :)

Try reading http://www.isaac.cs.berkeley.edu/isaac/gsm-faq.htm l

Been done.

See H.P. Lovecraft's At the Mountains of Madness. http://sprg.ssl.berkeley.edu/~wcoburn/hpl/mountain s.html

I think that their site is actually http://www.opengraphicsproject.org/

Does anyone know if there are any plans/projects out there to build an actual free HDL synthesizer? Something that can go from the Verilog or VHDL to a netlist? It seems that's kind of key to all of the "open source hardware" projects; without one it's like the FSF in 1986, before gcc. You can write all the code you want but doing anything with it requires finding someone with the right commercial software.

The concept of 'hacking hardware' is an attractive one, but it's hampered by the very high cost of entry. Having a Free simulator is certainly a big step, but I think a lot of people are turned off by the fact that they can't produce a netlist of their design for use on an FPGA without very, very expensive tools. (Although I've seen references to some old [c. 1983] tools published by Berkley on tape for VAX that might still be around.) Unless I'm just confused and you can program an FPGA directly from an HDL program without synthesizing to a netlist first...?

I'd be curious to see someone who's gotten involved in hardware, particularly FPGA, programming give a breakdown of the minimum costs required to experiment and actually fabricate (not just simulate) some circuits. Maybe the perception of high cost on my part is false, in which case I'd be happy to be corrected.

Either that, or a broken benchmark. Each Cell processor (Synergistic Processing Element -- SPE) shares its instruction fetch port with its data memory port. The SPE can buffer up 80 instructions at a time (2.5 fetch words), plus an additional 32 from a branch target. Fetch will stall if the memory system gets saturated with loads and stores. Properly written memory-intensive code includes explicit fetches to keep these buffers full. Incorrectly written code will cause problems. Still, that doesn't explain a 3 orders of magnitude drop.

If you look at the slides on the page I linked to above, you'll see the SPEs are not connected into the global address space. They connect to a private single ported memory, and to each other through two unidirectional rings. (The ring structure is not apparent from that diagram, but trust me, it's there.) These rings then connect to a DMA engine.

If you wade through this paper, you'll see that the Cell compiler implements a software cache. (The same paper also explains the instruction fetch mechanism mentioned above, BTW.) That is, it emulates a cache in software, using the DMA to actually move memory around. Depending on the nature of the benchmark and how it was written, it could be that the read benchmark spends all its time allocating stuff into this cache and waiting for it to arrive. Writes would be faster because the cache can "write behind" without having to wait for the allocation to happen, if the compiler is smart enough to know that the previous data will be entirely overwritten. So, if the benchmark goofed, then the results are meaningless.

Fact of the matter is that the SPEs are capable of reading 128 bits a cycle each (128 bytes / cycle across the 8 SPEs). Other benchmarks, such as the article recently posted to Slashdot about using Cell for scientific computation confirm that this thing hauls--and these are bandwidth-intensive tasks. The quoted paper did run some numbers on real silicon and showed numbers similar to their simulation results.

With all this in mind, I find it hard to believe that Cell is broken.

-L

There are some comparisons between the Cell, a Cray X1E MSP 2005, AMD64 Opertron and IA64 with regards to processing grunt, and power consumption. The Cell looks to be very powerful with low power requirements. (See table 1 in the link below) http://www.cs.berkeley.edu/~samw/projects/cell/CF0 6.pdf To see a PS3 processsor outperforming a Cray from last year in some of those criteria is quite impressive. Even more impresssive, is that these results are unbaised and have been nowehere near Sony's marketting department..

Our tests on nVidia 5600 series AGP cards (this was several years ago) showed that the net SETI@home throughput using the GPU was at best 1/5 of what we could obtain with the CPU. This was primarily due to transfers out of graphics memory and into main memory.

PCI Express allows for symmetric bandwidth to graphics memory and graphics memories are now typically larger than the size of our working set. The difficulty will be in benchmarking to see which is faster for a specific GPU/CPU combination.

At any rate it's a fairly simple job to swap FFT routines in SETI@home. The source is available. Someone may have done it by now...

FFT is a data compression and encryption standard used by a wide variety of extraterrestrial civilizations. Seti@home spends most of its time running FFT code to look for signals. If we managed to communicate with any of these aliens we could ask them what it stands for.

There's VerbNet, FrameNet, Arabic WordNet, and probably others I don't know about.

WordNet has become a standard for working with semantic relations computationally these days. It works by storing all known senses of every dictionary word, and each sense has links to other words based on how it's semantically related(synonym, antonym, hyper/hyponym, meronym, troponym, cause, is_a, morphological derivative, etc...)

There's not any model that can compete with it currently, and it's widely accessible and very easy to use. As this tool improves, so will the semantic web.

Frankly, I cant believe this tech couldnt have been done already, even twenty or thirty years ago. I have to imagine we've had the tech to do adhesiveness on demand based on an external stimuli ( such as electricity ) for many years. We have had the ability when the opposite material is metal since atleast the beginning of the space race, but even sticking to any surface on demand shouldnt be too difficult.

OCW@MIT: Java Preparation
http://ocw.mit.edu/OcwWeb/Electrical-Engineering-a nd-Computer-Science/6-092January--IAP--2006/Course Home/index.htm

OCW@MIT: Software Engineering for Web Applications
http://ocw.mit.edu/OcwWeb/Electrical-Engineering-a nd-Computer-Science/6-171Fall2003/CourseHome/index .htm

Webcast@Berkeley: With real video and/or MP3 : Data Structures
http://webcast.berkeley.edu/courses/archive.php?se riesid=1906978271

Web development using XML:
http://cscie153.dce.harvard.edu/

A few seconds basic research would have found you an example - here's one:

The much celebrated Virginia Tech G5 cluster (ranked in the top five supercomputers for a time) replaced the rather average OS X Memory Manager with a real one.

There you go. They won't be able to do that when Apple are only selling Intel PCs.

Bloat or no bloat, OSX on Core Duo seems to trail the pack in simple horsepower.

http://sekhon.berkeley.edu/macosx/intel.html

MySQL is just one example. This comparison gives you Linux, Windows XP, and OSX on the same Core Duo machine. These benchmarks do almost no I/O, so your filesystem complaints are moot. They also did one using the G5. Same story. Not to say that the filesystem in freebsd doesn't hurt it, or that threading in freebsd doesn't hurt it, but you can't blame it for everything.

MySQL is just one example. This comparison gives you Linux, Windows XP, and OSX on the same Core Duo machine. These benchmarks do almost no I/O, so your filesystem complaints are moot. They also did one using the G5. Same story. Not to say that the filesystem in freebsd doesn't hurt it, or that threading in freebsd doesn't hurt it, but you can't blame it for everything.

What other platform? Apple? Yes they support Macs. But I can't personally attest to anything on mac because, quite frankly, OSX is inferior to both Linux and Windows on a technical level.

And yes you can run AGP + fb and switch between X and console just fine with the current driver ... and you could a few years back (2002)... I'm sorry if you can't configure an xorg.conf/XFree86.conf file.

Depends on what you mean by Micro Kernel and Monolithic.

True, the kernel of MacOS/X - Darwin, aka XNU, for performance reasons run the Mach and BSD layer both in superuser space to minimize the lattency.

Maybe this is what you call a hybrid kernel: http://en.wikipedia.org/wiki/Hybrid_kernel

You may call XNU whatever you wish but the fact remains:
- it's not a monolithic kernel by design
- it has Mach in it and Mach is some sort of microkernel. Maybe it does not reach "today's" standards of being called a microkernel but it was a very popular microkernel before.

So maybe the things running on top of Mach ( http://developer.apple.com/documentation/Darwin/Co nceptual/KernelProgramming/index.html ) are conceptually "different" from what the services of microkernel should be, and they do share indeed the address space, but this is very very very different from the architecture of a traditional monolithic kernel such as Linux

This guy ( http://sekhon.berkeley.edu/macosx/intel.html ) recently tested some stats software on his Mac running OS X and Linux, and found out that indeed MacOS X had performance issues, very likely due to the architecture of the kernel.

There's even a rumor that says that since Avie Tevanian left Apple ( http://www.neoseeker.com/news/story/5553/ ), some guys are now working for removing the Mach microkernel and migrate to a full BSD kernel in the next release of the operating system.

And now my personal touch. I agree with Linus when he says that having small components doing simple parts on their sides and putting them together with pipes and so on, is somehow the UNIX way and is attracting (too lazy to find the quote). However as he demonstrates later, distributed computing is not easy, and there's also the boundray crossing issue. I guess he has a point when he says this is a problem for performance and the difficulty on designing the system... So if performance is what you indeed expect from a kernel, then you must stop dreaming of a clean-centralized good software architecture like those we have for our high oo-oriented software.

But the truth is that, although developing a monolithic kernel is an easier task to do from scratch than a microkernel, I guess the entry ticket (learning curve) for a monolithic kernel developes is more expensive. The main reason being, "things ARE NOT separated". Anyone, anywhere in the kernel could be modifying the state of that thing, for non obvious reason, even if there's a comment that says "please don't do that" or it shoulld not be the case etc.... Microkernel can obviouisly provide some kind of protection and introspections to these things, but have always hurt performances to do so.

Now it has everything to do on what you expect. Linux has many many many developpers and obviously can afford having a monolithic design that changes every now and then and you may prefer a kernel that goes fast than one whose code is clearn, well organized and easy to read. But the corrolary of that observation is that for the same reasons, grep, cat, cut, find, sort, or whatever unix tools you use with pipes and redirection are similarly a cleaner but YET INEFFICIANT design. However, it's been proven (with time) to be a good idea..

I think things that are "low level" will be bound to have a poor spagehtti software architecture because performance matters and the code is smaller.. but the higher level you go, the less performance matters, and the more code maintenance and evolutivity matters... Everything is a tradeof: good design practice depends on the type of problems your software tackles.

That said, it does not mean no progress can be made in kernel developments. Linux already uses a somewhat different C lang

http://kabuki.eecs.berkeley.edu/~luns/papers/241re p.html

-John

http://petroleum.berkeley.edu/papers/patzek/CRPS41 6-Patzek-Web.pdf

Not again:

http://people.fluidsignal.com/~luferbu/misc/Linus_ vs_Tanenbaum.html

We've got Andy Tanenbaum coming up with nothing practical in the fifteen or sixteen years he's been promoting microkernels, and then turning around and telling us he was right all along. Meanwhile, the performance of OS X sucks like a Hoover, as we all knew:

http://sekhon.berkeley.edu/macosx/intel.html

I'll just pretend I didn't see this article.

So you want a completely detailed model of the world, down to bricks and individual grains of sand?
You want it all be simulated with physics so that you can interact with everything in a plausible way?

Well, I can tell you that any one of these things currently is a struggle to get to work at all,
even assuming you are willing to wait hours per frame. You want a pile of thousands of bricks
falling into a pile, with correct collision detection? This is an area of active research.

http://www.cs.cmu.edu/~djames/

You want the water on the beach to swirl and splash?
How about a piece of paper that you can burn?
Again, a challenging set of problems that we are just beginning to solve in a way that looks good.
http://graphics.stanford.edu/~fedkiw/

How about the snot you pull out of your nose?
You want to pick your nose and have the snot squish in a gooey fashion?
We can do it, but just barely, if you want to wait all week for a few seconds of animation.

http://www.cs.berkeley.edu/b-cam/Papers/Goktekin-2 004-AMF/index.html

Now, what you want is to combine all these simulations, plus many more.
Also you want it to run in real time on a desktop PC.

I predict we will have this in 50 years, and that is being extremely optimistic.
If Moore's law is really ending, then maybe much longer.

Hardware physics cards may be just the thing we need to make it possible one day.

There are a couple of bundles of lecture notes (#41 and #42 at http://www-inst.eecs.berkeley.edu/~cs164/sp05/lect ures/index.html) about proof checking in programs. Simply put, solvers can't handle programs of normal size in industry. Proof CHECKING, on the other hand, is quite fast, but that means the programmer has to provide a proof, which the slides show can get quite tedious for even small examples.

According to most evolutionary biologists, meat was the catalyst for the evolution of human intelligence. Frankly, you owe /. to meat eating, as well as your computer, your room, your excellent house, and the country you live in. http://news.nationalgeographic.com/news/2005/02/02 18_050218_human_diet.html http://www.berkeley.edu/news/media/releases/99lega cy/6-14-1999a.html Oh yes; that 4 days thing is bull. Your stomach maintains a sterilized environ, and food goes through it. Hence, sterile food, unless there's tapeworms. Then that's your fault for not cooking it properly.

For more indepth information that any human should really posess about the topic, see Setuid Demystified.

Year old statistics of crashes indeed shows that in 2005 (date of the article, crash collection started as early in July 2004) indeed showed that Firefox crashed more often than IE (see Figure 1, need to devide bottom % by top %).

The data cited is definitely outdated since there were releases of analyzed soft after that.

Firefox 1.5.0.2 crashes very rarely on me and I am happy with it.

You're running Windows in a non-admin account ? Whoa ! That's rare, but very good !

Solution that is recommanded almost everywhere :
- Use the "Run as..." feature of WinXP to run only "explorefs2" in an admin account and the rest in normal accounts. Therefor, only exploits directly aimed at explorefs2 will have admin privilege, if other exploits are encountered (you got an MS-Office-only document. you reboot under windows, you import this MS-Office document using explorefs2 and MS-Word gets exploited) they won't pwn the whole machine.

Solution I use here :
- Get some of the old hardware you have lying around (some Pentium-II/III era mother board and CPUs), a nice netword card (1GBps if you get one. Even if PCI bus won't max it out (33Mhz * 32bits), they're cheap) a lot of memory, nice new shiny disks (the only realy new stuff you buy), and maybe some controller to put them on (if you're unlucky with some pre-LBA48 chipset like 440BX)
- Install a headless linux on it. You may use LVM2 and RAID, even software RAID5 (you don't give a damn if software raid slows your machine : the CPU on this machine is used only by the server. It doesn't slow the CPU on which you're gaming/working). It's like turning your old hardware into a glorified hardware RAID controller. And the good part is, if the CPU or motherboard dies, you'll be able to re-plug the harddrive into any other linux box with instant raid & lvm2 support. (Unlike trying to find a RAID controller of the same exact model). (Besides, as almost nothing runs on this machine, most of the memory will be used as cache.)
- Install a file server on it, using Samba and requiring log-in (no guest accounts).
- Voilà ! You can mount your share from whatever OS you want, underwhich ever user access level you want, the files remain on the server and are only accessed with the right of the user loged in samba. And on top of that, you get a nice journaled file system you choose, with support for >4GB files, even if the clients you connect with don't support it. (like FreeDOS. Reiser & Ext2 DOS tools don't support the journaling. But you can still SFTP or SCP files to/from your server)
Other ideas :
- run clamav periodically on it : virus scanning may slow computers, but it won't slow the computer you're working on.
- use Smart : the disk are the only precious thing that must be monitored. The rest is old recycled crap.
- run some P-2-P software that has distinct core (running on file server) and GUI (on your box). Sancho and mlDonkey are a nice combination. Your torrent keep being downloaded at night, but your girlfriend doesn't complain because you have to keep you "OMG!!LOL!!11!!!"-overclocked-with-ponnies cooled-with-20-12cm-fans-running-at-10'000rpms Pentium Ultra Extreem edition running at night.
- If you need to access other machines at home remotely, no need to keep them on. Only keep your server on, log-in with SSH, and use Wake-On-Lan feature to turn on the other machines.
- You can give the spare CPU cycles that aren't used in software RAID or clamav-scanning to some distributed project at BOINC

- I heard that part of the StarForce protection scheme detects when data is simultaneously streamed from both the harddrive and the cd-rom (id est : DaemonTools reads image from hard-drive and then feeds it as virtual Cd-ROM). Using a server to store disc images supposed to circumvent this part of the protection scheme.