The R16000 has many significant architectural and memory-related improvements over the R14000A. However, you are correct in that it's not the speed demon that some folks are making it out to be.
But... keep in mind that it consumes far less than 20 watts of energy (and thus gives off little heat) and will eventually find itself packed in with other CPUs into Origin servers/supercomputers. The CPU bricks for the O3900, for example, have 16 CPUs in just 4U of rack space.
SGI's ccNUMA MIPS/IRIX machines are typically used for tasks that are severely I/O bound, that is, their strong point is chugging thru massive amounts of data where raw per-node CPU power is important, but not the largest factor. Somewhat like a mainframe, but with less redundancy and more CPU power.
as mentioned in a previous comment, NeXT computer (founded by Steve Jobs, now owned by Apple) named their server room the Bat Cave as well. Mostly because it was filled to the brim with black NeXT cubes and NeXTstations.
I've heard many stories about the server farm that NeXT used to have in their Redwood City, California office. Their "Bat Cave" was a small room packed full of black NeXT cubes and slabs. It's been rumored that much of the NeXT server farm still exists at Apple, as does a collection of every NeXT model and OS version on display for testing and "research".
I used an Apple IIe to trace the route and that was nearly 20 years ago. There was this really cool game for that computer. Anyone play it?
There is a huge difference between the two... Lewis and Clark (and their many assistants) were explorers. The Oregon trail was an established wagon train route from St Louis to Oregon.
That said, the (early) Oregon Trail games were quite cool. The original Apple II version had pretty crappy graphics, but the (two disk!!) second edition was much better (though required 128 KB of memory!). The early Mac version was cool as well, it supported LocalTalk LAN play for interactive wagon trains (voting, hunting, etc). The more recent versions for Mac/Win are nothing but eyecandy.
Some ISPs, especially the smaller guys or the co-op's, don't allow NATs for the simple reason that a un-clued user could hose the whole network.
A broadband ISP with midrange equipment (a few managed switches, a generic DSLAM and wireless gear, etc) could easily be crippled if a user were to plug his NAT in backwards. Plug the "LAN" side of a NAT box into the ISP's pipe and now it's broadcasting itself as a DHCP server. Not good. Giving each user a port on the core ISP router or a smarter switch would prevent this from happening, but would cost much much more.
Almost every travel agency is either A) connected diretly to SABRE, or B) connected to a database that is connected to SABRE.
SABRE being the company and computer system that handles an insane number of daily travel-related reservations and other similar date/time/person related events for *thousands* of companies (including Expedia, Orbitz, Travelocity, the airlines, etc). And yep, you guessed it, SABRE is based on mainframes.
What I don't get is why they need the fan. Why don't they just slap on a bigger heatsink. I don't like fans in my computer, they're noisy. Loss the fan and they got a nice system.
True, that. Would be even cooler if they'd use the PPC 750FX. That would bring the on-die L2 cache up to 512 KB plus would use even less power due to the smaller process, smaller die size.
the 17" widescreen iMac is quite nice... but not for $2000. I wish it was available with a plain DVD-ROM/CD-RW "combo" drive rather than the DVD-RW/CD-RW "superdrive", though. Save a few bucks that way.
It's a temping little machine, in fact it's almost moreso an executive system than a home/school machine. But right now I can get a refurbished single cpu G4 tower and a 19" samsung DVI monitor for the same price.
This is the slightly older version of the PPC 750 "G3". This 750 CXe model has 256 KB of on-die L2 cache and is fabbed at a.18 micron process.
The newer 750 FX model (as used in the current Apple iBooks) has 512 KB of on-die L2 cache and is fabbed at.13 micron with all of the buzzwords (silicon-on-insulator, etc).
I'm not sure it's something I'll rush out to buy, but it does sound interesting, if for no other reason than to add another dev/compile server to the stack we have at work.
Cobalt RaQ 4, IBM B50, generic 1U PC, Sun Netra X1, Apple Xserve.
Now if only SGI would make a 2P, 1U server. How thick is SGI's 4P Origin 300? 2U?
Very cool. I'm not a big GNOME fan, but if Sun's release version will run nicely on an Ultra 1, then I'll be way more likely to give it a shot. Might even be kinda zippy (heh) on the Ultra 30 (300 MHz) I may soon be getting.
BTW, how's Solaris 9 working out for you on that box? I'm still using 8, which feels a tad slower than 7, but it may just be a placebo effect.
This message might not get any replies, being down at the bottom of the stack, but it's worth a shot...
What has CDE been like over the past year? I keep hearing folks talk about how mature/stable/etc it has become. I first used CDE under Solaris 2.6, and later on Solaris 7 and Solaris 8 (revision 07/01). I never had much of a problem with it... it did the job and had a clean look to it. BUT...
...my big beef with it was the blasted memory leaks. All three versions I tried would gobble up insane amounts of ram over about a month's time. Logging out at night solved the problem, but was a bit of a pain on non-networked, always-logged-in boxes. I was used to Openwin as well as SGI's "IndigoMagic" desktop, both of which could run logged-in for months without sucking up more than an extra mb or so of ram. I guess maybe CDE's developers felt the average user would logout after a day or two... or reboot every day like the Wintel PC crowd. I dunno. *shrug*
My long term solution was to ditch CDE on my own box and just use mwm as my wm and have a nicely configured "right-click" root menu. xterm and xwsh are my program launchers, damnit!:)
How's the performance? Have you tried Sun's packaged GNOME 1.4 on the same box? Why do I ask?... because I tried GNOME 1.4 on an Ultra 1 as well, and it was miserably slow. I really can't imagine that 2.0 would be that much faster, maybe even a tad bit slower.
For the record, I was using an Ultra 1/200E, 512 MB RAM, Creator3D gfx, Solaris 8 7/01 with latest patchset.
CDE/dtwm on the same box was about as zippy as it could possibly be. Vanilla plain, but fast.
I've continued to use Pine for years, as have many of my friends and coworkers. It's fast, it's (somewhat) lean, and it just works. I'll probably keep using it for many more years.
I have a choice at work: 1.dtmail 2.Netscape 4.9's mail 3.Pine Which would you use?
LOL, good point. Actually, I've been using Pine by choice for about 6 years. I might use it for another 6. I try a new/different GUI email app every few months, but none has impressed me enough to switch.
dtmail is pretty weak, but it works. Doesn't crash nearly as often as Netscape's mail... but like most of the CDE apps, it has memory leaks up the wazoo.
I can totally relate. I started with Elm and was forced to switch to pine many years ago when our sysadmin decided to overhaul the system to his likings. After a week or so, I got used to Pine and have been using it ever since. I haven't even tried Mutt as Pine works fine for me.
Maybe I'll give Mutt & Nano a try this weekend. Or maybe sometime around 2005. Who knows.
Slashdot Mirror
The R16000 has Out-of-order instruction execution?.
So did the R14000. And the R12000. And the R10000...
The R16000 has many significant architectural and memory-related improvements over the R14000A. However, you are correct in that it's not the speed demon that some folks are making it out to be.
But... keep in mind that it consumes far less than 20 watts of energy (and thus gives off little heat) and will eventually find itself packed in with other CPUs into Origin servers/supercomputers. The CPU bricks for the O3900, for example, have 16 CPUs in just 4U of rack space.
SGI's ccNUMA MIPS/IRIX machines are typically used for tasks that are severely I/O bound, that is, their strong point is chugging thru massive amounts of data where raw per-node CPU power is important, but not the largest factor. Somewhat like a mainframe, but with less redundancy and more CPU power.
as mentioned in a previous comment, NeXT computer (founded by Steve Jobs, now owned by Apple) named their server room the Bat Cave as well. Mostly because it was filled to the brim with black NeXT cubes and NeXTstations.
I've heard many stories about the server farm that NeXT used to have in their Redwood City, California office. Their "Bat Cave" was a small room packed full of black NeXT cubes and slabs. It's been rumored that much of the NeXT server farm still exists at Apple, as does a collection of every NeXT model and OS version on display for testing and "research".
I used an Apple IIe to trace the route and that was nearly 20 years ago. There was this really cool game for that computer. Anyone play it?
There is a huge difference between the two... Lewis and Clark (and their many assistants) were explorers. The Oregon trail was an established wagon train route from St Louis to Oregon.
That said, the (early) Oregon Trail games were quite cool. The original Apple II version had pretty crappy graphics, but the (two disk!!) second edition was much better (though required 128 KB of memory!). The early Mac version was cool as well, it supported LocalTalk LAN play for interactive wagon trains (voting, hunting, etc). The more recent versions for Mac/Win are nothing but eyecandy.
The Lewis & Clark IMAX movie is even showing at the local Omnimax/IMAXdome in Lubbock Texas.
Some ISPs, especially the smaller guys or the co-op's, don't allow NATs for the simple reason that a un-clued user could hose the whole network.
A broadband ISP with midrange equipment (a few managed switches, a generic DSLAM and wireless gear, etc) could easily be crippled if a user were to plug his NAT in backwards. Plug the "LAN" side of a NAT box into the ISP's pipe and now it's broadcasting itself as a DHCP server. Not good. Giving each user a port on the core ISP router or a smarter switch would prevent this from happening, but would cost much much more.
That's only one reason, there are many.
Notice that floppy disk quality/reliability began to sink about the same time the FDD-less iMac came out?
Hmmm... =)
Search for "sgi" and you'll find even more silicon graphics machines. Most auctions don't have "workstation" in the title.
Almost every travel agency is either A) connected diretly to SABRE, or B) connected to a database that is connected to SABRE.
SABRE being the company and computer system that handles an insane number of daily travel-related reservations and other similar date/time/person related events for *thousands* of companies (including Expedia, Orbitz, Travelocity, the airlines, etc). And yep, you guessed it, SABRE is based on mainframes.
(dont ask me how to modify BIOS settings or whatever on a Mac :))
:)
OpenFirmware, baby! Hope you brushed up on FORTH!
Hit google, lots of stuff on OF out there, it's sort of a standard.
What I don't get is why they need the fan. Why don't they just slap on a bigger heatsink.
I don't like fans in my computer, they're noisy. Loss the fan and they got a nice system.
True, that. Would be even cooler if they'd use the PPC 750FX. That would bring the on-die L2 cache up to 512 KB plus would use even less power due to the smaller process, smaller die size.
You are the reason people hate geeks, going around acting like something this insignificant is worthy of any sort of comment.
You say that like it's a bad thing...
the 17" widescreen iMac is quite nice... but not for $2000. I wish it was available with a plain DVD-ROM/CD-RW "combo" drive rather than the DVD-RW/CD-RW "superdrive", though. Save a few bucks that way.
It's a temping little machine, in fact it's almost moreso an executive system than a home/school machine. But right now I can get a refurbished single cpu G4 tower and a 19" samsung DVI monitor for the same price.
oops, I guess it's a RaQ 2, the model that uses the MIPS cpu.
It's the IBM PowerPC 750CXe.
.18 micron process.
.13 micron with all of the buzzwords (silicon-on-insulator, etc).
This is the slightly older version of the PPC 750 "G3". This 750 CXe model has 256 KB of on-die L2 cache and is fabbed at a
The newer 750 FX model (as used in the current Apple iBooks) has 512 KB of on-die L2 cache and is fabbed at
I belive this board uses PC133 RAM. 133 MHz x 4.5
I hate to nitpick... but it's "Mac", not "MAC".
Mac is short for Macintosh, a series of computers sold by Apple Computer Inc.
Anyone else notice how small the CPU heatsink and fan is on that PPC mobo? I guess it is only 600 MHz and it's the newer G3.
But still... it's smaller than the little heatsink/fan on my PC's motherboard chipset!
I'm not sure it's something I'll rush out to buy, but it does sound interesting, if for no other reason than to add another dev/compile server to the stack we have at work.
Cobalt RaQ 4, IBM B50, generic 1U PC, Sun Netra X1, Apple Xserve.
Now if only SGI would make a 2P, 1U server. How thick is SGI's 4P Origin 300? 2U?
Very cool. I'm not a big GNOME fan, but if Sun's release version will run nicely on an Ultra 1, then I'll be way more likely to give it a shot. Might even be kinda zippy (heh) on the Ultra 30 (300 MHz) I may soon be getting.
BTW, how's Solaris 9 working out for you on that box? I'm still using 8, which feels a tad slower than 7, but it may just be a placebo effect.
This message might not get any replies, being down at the bottom of the stack, but it's worth a shot...
:)
What has CDE been like over the past year? I keep hearing folks talk about how mature/stable/etc it has become. I first used CDE under Solaris 2.6, and later on Solaris 7 and Solaris 8 (revision 07/01). I never had much of a problem with it... it did the job and had a clean look to it. BUT...
...my big beef with it was the blasted memory leaks. All three versions I tried would gobble up insane amounts of ram over about a month's time. Logging out at night solved the problem, but was a bit of a pain on non-networked, always-logged-in boxes. I was used to Openwin as well as SGI's "IndigoMagic" desktop, both of which could run logged-in for months without sucking up more than an extra mb or so of ram. I guess maybe CDE's developers felt the average user would logout after a day or two... or reboot every day like the Wintel PC crowd. I dunno. *shrug*
My long term solution was to ditch CDE on my own box and just use mwm as my wm and have a nicely configured "right-click" root menu. xterm and xwsh are my program launchers, damnit!
How's the performance? Have you tried Sun's packaged GNOME 1.4 on the same box? Why do I ask?... because I tried GNOME 1.4 on an Ultra 1 as well, and it was miserably slow. I really can't imagine that 2.0 would be that much faster, maybe even a tad bit slower.
For the record, I was using an Ultra 1/200E, 512 MB RAM, Creator3D gfx, Solaris 8 7/01 with latest patchset.
CDE/dtwm on the same box was about as zippy as it could possibly be. Vanilla plain, but fast.
I've continued to use Pine for years, as have many of my friends and coworkers. It's fast, it's (somewhat) lean, and it just works. I'll probably keep using it for many more years.
I have a choice at work: 1.dtmail 2.Netscape 4.9's mail 3.Pine Which would you use?
LOL, good point. Actually, I've been using Pine by choice for about 6 years. I might use it for another 6. I try a new/different GUI email app every few months, but none has impressed me enough to switch.
dtmail is pretty weak, but it works. Doesn't crash nearly as often as Netscape's mail... but like most of the CDE apps, it has memory leaks up the wazoo.
I can totally relate. I started with Elm and was forced to switch to pine many years ago when our sysadmin decided to overhaul the system to his likings. After a week or so, I got used to Pine and have been using it ever since. I haven't even tried Mutt as Pine works fine for me.
Maybe I'll give Mutt & Nano a try this weekend. Or maybe sometime around 2005. Who knows.