Isn't he confusing FreeBSD and NetBSD? Or is he referring to the FreeBSD userland in OS X/Darwin?
As he metioned that most of the free unices these days are alike, I'm guessing that he's referring to the FreeBSD userland.
I love FreeBSD though, I'd love to run it on my iMac instead of OS/X.
Mac OS X runs pretty well on the G4 iMac that I use. Apple's flavor of X11 runs well enough for me, and there's plenty of OSS goodies, links, and documentation to keep most folks happy.
It is kinda interesting that he mentions using SGI IRIX at Bell Labs... they don't seem to be the sort of shop that needs gobs of graphics pipes, hundreds of CPUs, or mountains of IO. Though I guess there are plenty of people that like IRIX simply for all of the performance tools that are part of the development environment. (An environment that's too anal to easily compile most open source software... but that's another rant for another time......)
Dispite using mostly X11 based desktops (SGIs, Suns, x86 Linux) I do really like the look and feel of Microsoft Entourage. It does way more than Outlook Express, but doesn't need to rely on an Exchange server.
Microsoft has recently dropped the price for standalone Entourage to $99, and I think they have an Entourage/Word bundle for $190. (It's still probably a better deal to make use of the Office OEM bundle when buying a new Mac from a reseller). http://www.microsoft.com/mac/products/entouragex/e ntouragex.aspx?pid=entouragex
Does anyone have a pointer to a binary for Silicon Graphics IRIX?
BTW, I see that SGI has posted an IRIX version of Mozilla 1.4 in both tarball and inst/swmgr native tardist installable format: ftp://oss.sgi.com/projects/mozilla/downlo ad/irix/1.4/
We spent so much time, money and effort fixing it, why not spend some more and upgrade it for another decade of use?
This is a bad economic arguement. If the costs of an extra 10 years of service outweigh the benefits, then it's not worth it.... regardless of how much money has been pumped into the project up until now.
I'm pretty much out of the DVD-RAM/DVD-RW/DVD+RW loop, but I do know that there are many existing firms that use DVD-RAM as part of their backup strategy. Because they don't distribute the discs they burn (it's a private backup) the compatibility issue isn't much a problem.
Plus there are still a fair number of "early adopter" folks and firms that have an existing investment in DVD-RAM drives and media.
Onyx4, for the most part, is just another Origin 3xxx class brick. In this case, it's the new Graphics Brick. Plug as many as you want into your existing Origin.
As most Onyx4s will probably be using Origin 350s as their host, then my best guess is R16K/700 CPUs.
The CPU performance doesn't matter quite as much in an SGI as it would in a Mac or PC.
Most folks that use SGIs for number cruching have picked that platform based on its trememdous amount of memory and I/O. If their task was simply CPU bound or didn't need more than a few hundred MB/sec of IO, they'd just use a PC cluster.
Most folks that use SGIs for graphics do so because they either need tight integration with video (HD or SD, see Discreet Inferno or IFX Piranha using SGI's DM3 HD video I/O subsystem).... or because they need multiple displays running of the same system. (http://www.sgi.com/newsroom/press_releases/2003/j une/planetarium.html) Either each pipe running one or two displays or multiple pipes running in parallel.
Folks that use SGIs for both reasons typically require gobs of number crunching combined with some sort of display system that is able to plot the trillions of data points without bringing the machine to its knees. SGI has a lot of such cloak and dagger government / defense users.
There's also the growing Altix series of machines, which use Origin-class architecture with the Itanium processor family. There are rumors of a totally new MIPS processor coming soon as well.
The main point is that the new Onyx4 graphics are delievered in brick form, they're modular, and they will probably be eventually used on multiple SGI systems. And because SGI is leaving most of the 3D work to the ATI/NVIDIA pixel war, they can save some money and focus on other engineering aspects.
The point is that if you want to render 3-D graphics on a wall of 36 LCD displays in a 6x6 grid
Or if you want to have, say, 16 GPUs working in parallel on one 1600x1200 display channel for an ungodly amount of detail... The Onyx4 (and previously, Onyx InfinitePerformance) can do that as well.
Other than driving multiple gfx displays, what is the main use for this workstation? I havnt used a SGI workstation, so share the info...
Depends on which machine you're talking about... the Tezero workstation or the Onyx4 visual supercomputer... two totally different products.
The Tezro replaces the Octane/Octane2. These days SGI workstations are usually used for software development for the big iron, HD video work, and for smaller-scale data crunching. Octane2 and now Tezro both have pretty amazing HD abilities, due mostly to the wicked fast architecture. These usually run specialized apps such as Discreet Flame or IFX Piranha. Other big data companies like these workstations for similar reasons. GE Medical uses Fuel (Tezro's little brother) workstations to collect and display 3D data from MRI scanners.
99.999% of the population has no need for an SGI... but there are folks out there that have tasks that utilize such huge amounts of data and need as little latency as possible... for these folks there is the mighty expensive SGI kit.
Infinite Reality 4 has 1 GB of texture ram and 10 GB of frame buffer memory... so it doe have its advantages for a few specific users. But for the most part, using ATI gfx GPUs (working either independently or in parallel) makes far more sense than having SGI use the last of their resources to fight the ATI/NVIDIA 3D war.
SGI's strengths are with architecture and I/O. ATI's strenghts are in pixel and polygon pumps. Looks like a perfect union to me.
It's great to see SGI hanging in there, even though the industries in which they used to dominate have largely become the territory of cheap Linux PCs. While SGIs can no longer boast superior hardware of software, their brand still holds enough cachet for them to stick around a few more years a la Apple.
How many other PCs and Macs can handle hudreds of CPUs and 32 ATI gfx GPUs per system?
Up to 4 700 MHZ MIPS R4000 processors in the rackmount, or up to 2 in the tower. 12-bit alpha channel, 24-bit Z buffer. 128MB graphics memory. p to 8 GB main RAM in the tower, up to 16GB in the rackmount. Nice. SGI's were once the pinnacle of graphics performance, but one has to wonder with the predeominance of cheaper Wintel or Lintel boxes that have practically comparable performance, how relevant are these boxes still?
If you have gobs of IRIX code you need to run today, or if you need gobs of I/O on a desktop machine today, there isn't much other choice.
You're quoting specs from the Tezro workstation, which BTW, uses R16000 processors, not R4000. The Tezro uses Origin 350 architecture and has 3 PCI-X buses and two XIO buses (for gfx and HD/SD video I/O) as well as two builtin channels of SCSI. The thing is a full fledged data pump that I certainly don't need, but some folks do.
The new Onyx4 also uses Origin 350 and Origin 3000 host architecture, but can use all of that to feed 32+ ATI gfx cores per system. Can have each core drive one or two displays or can have multiple cores working in parallel. Two major uses -- doing crazy high end 3D or for visualizing big supercomputing data.
By using ATI GPUs, SGI can focus on their architecture, I/O, and SD/HD video options, rather than try to fight the ATI/NVIDIA 3D battle.
The new Onyx4 systems are able to drive multiple GPUs independently or in parallel for even more performance. All of this is backed by gobs of CPUs an many GB of RAM to feed the gfx.
Tezro comes in both desktop and rackmount form factors. 1 - 4 MIPS R16000 processors, up to 16 GB RAM, 7 PCI-X slots from 3 busses. Based on Origin 350 architecture.
Onyx4 "supports" up to 32 graphics GPUs, but more can be added. Each pipe can drive one or two displays or up to 16 GPUs can be used together in parallel for increased performance. Onyx4 is essentially a new graphics brick to be used on Origin 300 or 3000 class host systems.
There are gobs of new SD and HD video card available for both new systems, as well as new audio card offerings. Both machines will seem to require at least IRIX 6.5.21 (the August 2003 quarterly release) to run.
Up until now, 3D Labs has maintained a link to Xi Graphics for Linux drivers for the Wildcat cards. (www.xig.com)
So my question is this... are these drivers just a free re-release of the drivers originally built by Xi? Are they totally new drivers written by folks at 3D Labs? Are they from a combination of the two?
I'm curious (for both political and practical/technical reasons).
You can find a list here. For most computations and most hardware, you are probably still better off with MPI or PVM rather than shared memory.
Note also that there are several high speed interconnects for Linux clusters available from many different vendors, including InfiniBand, Gigabit Ethernet, FireWire, and Myrinet.
SGI systems (Origin and Altix) have massive interconnects that hold together the single-system architecture. They're fast for shmem-type shared memory apps, but also for MPI. In fact, SGI keeps tweaking their MPI implementation with every release of IRIX and the Linux ProPack, even though MPI is not the "best" way to run apps on their systems.
The interconnects in most Origins and Altix systems are 3.2 gigaBYTE per second with extremely low latency. I don't know about Infiniband, but I do know that GigE is only 125 MB/sec with really high latency... FireWire 800 is 100 MB/sec with better latency.... and I think the bst version of Myrinet is 500 MB/sec (4 gigabit) with about 5x the latency of SGI's 'numalink'.
The smaller Altix systems (and supposedly, future Altix and Origin systems this fall) can be double cabled or can run at a higher speed... for 6.4 gbyte/sec per interconnect.
Also, the Altix can handle up to 64 processors per single machine / single node (or 128 with a very beta set of patches). The cluster in the article is actually four Altix systems, each with 64 processors. The Origin 3800/3900 can handle 512 processors per node (or 1024 with a special "XXL" IRIX kernel).
Great stuff for I/O intensive tasks, but massive overkill for 3d rendering or calculating pi.
Just to nitpick... most Origins are not clusters but rather one large single machine. It is possible to partition the machine in firmware and have each partition talk to others over the existing (and now unused) numalink interconnects... but it's much faster (even for plain MPI code) to just run the beast as one large single machine.
Anyone else notice...
on
Random Humor
·
· Score: 3, Interesting
Anyone else notice that Pixar is mentioned in the credits?
Nice gfx for a Macintosh LC
on
Random Humor
·
· Score: 3, Funny
That must be one impressive video codec to fit the rapper and the interviews on that 1.44MB floppy! Auto-run too!
And the whole thing runs on the Motorola 68020 of the school's Macintosh LC!
They're supposedly banning cellphones at swimming pools because of people taking pictures of kids.
Sadly that sounds pretty typical of so many modern regulations. Ban cameras at public pools and the parents will no longer be able to take photos of their kids at play. But the pedaphiles, who already have plenty of screws loose, will continue to snap pictures... either hidden camera or at a distance with a telephoto lens.
We need a cheap source of e-readers / tablets. I mean *super* cheap, like $10 each.
Wait about 2 years. I'm sure some asian company will figure out how to mass produce e-readers for that sort of price. Meanwhile American companies will still be charging $400.
I just bought a true rms digital multimeter for $7 at a local discount electronics store, "SuperPower" batteries included! Thing even measures inductance and capacitance. It's enclosed in the cheapest plastic I've ever seen (the typical oily imported "everything's $1" type) and the bundled leads are junk... but it works well. It's amazing how cheaply some things can be mass producted.
I'm not too current on cell phones (I still use an original StarTac) but do these things have a high enough resolution to take a good snapshot of a page of a book or magazine? From the demos I've seen, I'm guessing the resolution of the cameras found on most phones is 640x480 at most.
Is this really a problem or is this just some case where *one* crazy guy walked into a bookstore and started taking snapshots with his phone (or camera)?
Netscape 2 (I don't think there ever was a released Netscape 1, was there? Netscape 1 was the commercialization of Mosaic)
You're right about Mosaic... the "original" Netscape was technically NSCA Mosaic. However, the first public official Netscape release was actually 0.something. I belive it was 0.9. I switched from Mosaic to Netscape at version 0.97 on my Mac and SGI. Then came Netscape 1.0, 1.1, 1.12...
Netscape 2.0 was the beginning of the end. Livescript ("Javascript"), animated GIFs, plugins, frames... it was all downhill from there. =)
Slashdot Mirror
Isn't he confusing FreeBSD and NetBSD? Or is he referring to the FreeBSD userland in OS X/Darwin?
As he metioned that most of the free unices these days are alike, I'm guessing that he's referring to the FreeBSD userland.
I love FreeBSD though, I'd love to run it on my iMac instead of OS/X.
Mac OS X runs pretty well on the G4 iMac that I use. Apple's flavor of X11 runs well enough for me, and there's plenty of OSS goodies, links, and documentation to keep most folks happy.
It is kinda interesting that he mentions using SGI IRIX at Bell Labs... they don't seem to be the sort of shop that needs gobs of graphics pipes, hundreds of CPUs, or mountains of IO. Though I guess there are plenty of people that like IRIX simply for all of the performance tools that are part of the development environment. (An environment that's too anal to easily compile most open source software... but that's another rant for another time......)
Dispite using mostly X11 based desktops (SGIs, Suns, x86 Linux) I do really like the look and feel of Microsoft Entourage. It does way more than Outlook Express, but doesn't need to rely on an Exchange server.
e ntouragex.aspx?pid=entouragex
e ntouragex.aspx?pid=whatsnew
Microsoft has recently dropped the price for standalone Entourage to $99, and I think they have an Entourage/Word bundle for $190. (It's still probably a better deal to make use of the Office OEM bundle when buying a new Mac from a reseller).
http://www.microsoft.com/mac/products/entouragex/
There are also a few good screenshots of the app too... doesn't look like the commitee-designed beast that Outlook XP is:
http://www.microsoft.com/mac/products/entouragex/
Does anyone have a pointer to a binary for Silicon Graphics IRIX?
o ad/irix/1 .4/
BTW, I see that SGI has posted an IRIX version of Mozilla 1.4 in both tarball and inst/swmgr native tardist installable format:
ftp://oss.sgi.com/projects/mozilla/downl
We spent so much time, money and effort fixing it, why not spend some more and upgrade it for another decade of use?
This is a bad economic arguement. If the costs of an extra 10 years of service outweigh the benefits, then it's not worth it.... regardless of how much money has been pumped into the project up until now.
I'm pretty much out of the DVD-RAM/DVD-RW/DVD+RW loop, but I do know that there are many existing firms that use DVD-RAM as part of their backup strategy. Because they don't distribute the discs they burn (it's a private backup) the compatibility issue isn't much a problem.
Plus there are still a fair number of "early adopter" folks and firms that have an existing investment in DVD-RAM drives and media.
MIPS R16000 @ 700 MHz
j une/planetarium.html) Either each pipe running one or two displays or multiple pipes running in parallel.
Onyx4, for the most part, is just another Origin 3xxx class brick. In this case, it's the new Graphics Brick. Plug as many as you want into your existing Origin.
As most Onyx4s will probably be using Origin 350s as their host, then my best guess is R16K/700 CPUs.
The CPU performance doesn't matter quite as much in an SGI as it would in a Mac or PC.
Most folks that use SGIs for number cruching have picked that platform based on its trememdous amount of memory and I/O. If their task was simply CPU bound or didn't need more than a few hundred MB/sec of IO, they'd just use a PC cluster.
Most folks that use SGIs for graphics do so because they either need tight integration with video (HD or SD, see Discreet Inferno or IFX Piranha using SGI's DM3 HD video I/O subsystem).... or because they need multiple displays running of the same system. (http://www.sgi.com/newsroom/press_releases/2003/
Folks that use SGIs for both reasons typically require gobs of number crunching combined with some sort of display system that is able to plot the trillions of data points without bringing the machine to its knees. SGI has a lot of such cloak and dagger government / defense users.
There's also the growing Altix series of machines, which use Origin-class architecture with the Itanium processor family. There are rumors of a totally new MIPS processor coming soon as well.
The main point is that the new Onyx4 graphics are delievered in brick form, they're modular, and they will probably be eventually used on multiple SGI systems. And because SGI is leaving most of the 3D work to the ATI/NVIDIA pixel war, they can save some money and focus on other engineering aspects.
The point is that if you want to render 3-D graphics on a wall of 36 LCD displays in a 6x6 grid
Or if you want to have, say, 16 GPUs working in parallel on one 1600x1200 display channel for an ungodly amount of detail... The Onyx4 (and previously, Onyx InfinitePerformance) can do that as well.
Other than driving multiple gfx displays, what is the main use for this workstation? I havnt used a SGI workstation, so share the info...
Depends on which machine you're talking about... the Tezero workstation or the Onyx4 visual supercomputer... two totally different products.
The Tezro replaces the Octane/Octane2. These days SGI workstations are usually used for software development for the big iron, HD video work, and for smaller-scale data crunching. Octane2 and now Tezro both have pretty amazing HD abilities, due mostly to the wicked fast architecture. These usually run specialized apps such as Discreet Flame or IFX Piranha. Other big data companies like these workstations for similar reasons. GE Medical uses Fuel (Tezro's little brother) workstations to collect and display 3D data from MRI scanners.
99.999% of the population has no need for an SGI... but there are folks out there that have tasks that utilize such huge amounts of data and need as little latency as possible... for these folks there is the mighty expensive SGI kit.
Check out the press release... 34 gfx pipes (ATI GPUs) in that machine!
Infinite Reality 4 has 1 GB of texture ram and 10 GB of frame buffer memory... so it doe have its advantages for a few specific users. But for the most part, using ATI gfx GPUs (working either independently or in parallel) makes far more sense than having SGI use the last of their resources to fight the ATI/NVIDIA 3D war.
SGI's strengths are with architecture and I/O. ATI's strenghts are in pixel and polygon pumps. Looks like a perfect union to me.
Tezro uses Origin 350 architecture and has 3 PCI-X buses and two XIO buses
I should clarify... Tezro does have 7 PCI-X slots, but they are fed by 3 buses (mainly so one or two cards don't swamp the entire PCI subsystem).
It's great to see SGI hanging in there, even though the industries in which they used to dominate have largely become the territory of cheap Linux PCs. While SGIs can no longer boast superior hardware of software, their brand still holds enough cachet for them to stick around a few more years a la Apple.
How many other PCs and Macs can handle hudreds of CPUs and 32 ATI gfx GPUs per system?
Up to 4 700 MHZ MIPS R4000 processors in the rackmount, or up to 2 in the tower. 12-bit alpha channel, 24-bit Z buffer. 128MB graphics memory. p to 8 GB main RAM in the tower, up to 16GB in the rackmount. Nice. SGI's were once the pinnacle of graphics performance, but one has to wonder with the predeominance of cheaper Wintel or Lintel boxes that have practically comparable performance, how relevant are these boxes still?
If you have gobs of IRIX code you need to run today, or if you need gobs of I/O on a desktop machine today, there isn't much other choice.
You're quoting specs from the Tezro workstation, which BTW, uses R16000 processors, not R4000. The Tezro uses Origin 350 architecture and has 3 PCI-X buses and two XIO buses (for gfx and HD/SD video I/O) as well as two builtin channels of SCSI. The thing is a full fledged data pump that I certainly don't need, but some folks do.
The new Onyx4 also uses Origin 350 and Origin 3000 host architecture, but can use all of that to feed 32+ ATI gfx cores per system. Can have each core drive one or two displays or can have multiple cores working in parallel. Two major uses -- doing crazy high end 3D or for visualizing big supercomputing data.
By using ATI GPUs, SGI can focus on their architecture, I/O, and SD/HD video options, rather than try to fight the ATI/NVIDIA 3D battle.
The new Onyx4 systems are able to drive multiple GPUs independently or in parallel for even more performance. All of this is backed by gobs of CPUs an many GB of RAM to feed the gfx.
Tezro comes in both desktop and rackmount form factors. 1 - 4 MIPS R16000 processors, up to 16 GB RAM, 7 PCI-X slots from 3 busses. Based on Origin 350 architecture.
u ly/lanl.html
Onyx4 "supports" up to 32 graphics GPUs, but more can be added. Each pipe can drive one or two displays or up to 16 GPUs can be used together in parallel for increased performance. Onyx4 is essentially a new graphics brick to be used on Origin 300 or 3000 class host systems.
SGI has issued a press release discussing a monster Onyx4 they've already sold:
http://www.sgi.com/newsroom/press_releases/2003/j
There are gobs of new SD and HD video card available for both new systems, as well as new audio card offerings. Both machines will seem to require at least IRIX 6.5.21 (the August 2003 quarterly release) to run.
Up until now, 3D Labs has maintained a link to Xi Graphics for Linux drivers for the Wildcat cards. (www.xig.com)
So my question is this... are these drivers just a free re-release of the drivers originally built by Xi? Are they totally new drivers written by folks at 3D Labs? Are they from a combination of the two?
I'm curious (for both political and practical/technical reasons).
You can find a list here. For most computations and most hardware, you are probably still better off with MPI or PVM rather than shared memory.
Note also that there are several high speed interconnects for Linux clusters available from many different vendors, including InfiniBand, Gigabit Ethernet, FireWire, and Myrinet.
SGI systems (Origin and Altix) have massive interconnects that hold together the single-system architecture. They're fast for shmem-type shared memory apps, but also for MPI. In fact, SGI keeps tweaking their MPI implementation with every release of IRIX and the Linux ProPack, even though MPI is not the "best" way to run apps on their systems.
The interconnects in most Origins and Altix systems are 3.2 gigaBYTE per second with extremely low latency. I don't know about Infiniband, but I do know that GigE is only 125 MB/sec with really high latency... FireWire 800 is 100 MB/sec with better latency.... and I think the bst version of Myrinet is 500 MB/sec (4 gigabit) with about 5x the latency of SGI's 'numalink'.
The smaller Altix systems (and supposedly, future Altix and Origin systems this fall) can be double cabled or can run at a higher speed... for 6.4 gbyte/sec per interconnect.
Also, the Altix can handle up to 64 processors per single machine / single node (or 128 with a very beta set of patches). The cluster in the article is actually four Altix systems, each with 64 processors. The Origin 3800/3900 can handle 512 processors per node (or 1024 with a special "XXL" IRIX kernel).
Great stuff for I/O intensive tasks, but massive overkill for 3d rendering or calculating pi.
SGI Origin 3800 cluster
Just to nitpick... most Origins are not clusters but rather one large single machine. It is possible to partition the machine in firmware and have each partition talk to others over the existing (and now unused) numalink interconnects... but it's much faster (even for plain MPI code) to just run the beast as one large single machine.
Anyone else notice that Pixar is mentioned in the credits?
That must be one impressive video codec to fit the rapper and the interviews on that 1.44MB floppy! Auto-run too!
And the whole thing runs on the Motorola 68020 of the school's Macintosh LC!
They're supposedly banning cellphones at swimming pools because of people taking pictures of kids.
Sadly that sounds pretty typical of so many modern regulations. Ban cameras at public pools and the parents will no longer be able to take photos of their kids at play. But the pedaphiles, who already have plenty of screws loose, will continue to snap pictures... either hidden camera or at a distance with a telephoto lens.
Does anyone else notice the pattern with bans?
We need a cheap source of e-readers / tablets. I mean *super* cheap, like $10 each.
Wait about 2 years. I'm sure some asian company will figure out how to mass produce e-readers for that sort of price. Meanwhile American companies will still be charging $400.
I just bought a true rms digital multimeter for $7 at a local discount electronics store, "SuperPower" batteries included! Thing even measures inductance and capacitance. It's enclosed in the cheapest plastic I've ever seen (the typical oily imported "everything's $1" type) and the bundled leads are junk... but it works well. It's amazing how cheaply some things can be mass producted.
There is clearly a requirement here for some sort of DRM for printed materials...
This is a joke
*Sigh* It's a joke for now, but give the knee-jerk reactionary "think of the children!" pointy haired bosses some time...
I'm not too current on cell phones (I still use an original StarTac) but do these things have a high enough resolution to take a good snapshot of a page of a book or magazine? From the demos I've seen, I'm guessing the resolution of the cameras found on most phones is 640x480 at most.
Is this really a problem or is this just some case where *one* crazy guy walked into a bookstore and started taking snapshots with his phone (or camera)?
Netscape 2 (I don't think there ever was a released Netscape 1, was there? Netscape 1 was the commercialization of Mosaic)
You're right about Mosaic... the "original" Netscape was technically NSCA Mosaic. However, the first public official Netscape release was actually 0.something. I belive it was 0.9. I switched from Mosaic to Netscape at version 0.97 on my Mac and SGI. Then came Netscape 1.0, 1.1, 1.12...
Netscape 2.0 was the beginning of the end. Livescript ("Javascript"), animated GIFs, plugins, frames... it was all downhill from there. =)