Please note that there are some gaps (spaces) in the middle of the two URLs I pasted. They need to be removed in order for the URL to play nice with your web browser.
IRIX 6.5 can best be described by the neat poster SGI made about a year ago... it features a train chugging along past 5.3 all the way to 6.5 and through the 6.5.X updates. The front of the locomotive sports the designation "6.5.oo" (infinity). While 6.5 most likely won't be around forever, there are NO current plans to release a 6.6 or a 7.0.
Why?
Becasue 6.5 is the best thing SGI has ever done. There is an update released quarterly, on time, along two different streams: [M]aintenance and [F]eature. No patch dependency hell (though urgent patches are posted between 6.5.X updates). The Feature release gets new goodies every quarter, some of which are slowly rolled into Maintenance. Both streams are HEAVILY tested, tuned, reviewed, and compiled using the latest available MIPSpro compilers. In fact, SGI likes to stay a release or two ahead of their users, using it on most of their production systems to ensure a rock solid release.
The (large) IRIX 6.5 team has been plodding away ever since 6.5.0. When asked "where's 6.6" they will usually respond: "you didn't ask us to break application compatibility, so we aren't working on a 6.6".
I wouldn't want it any other way. Aside from a few small networking issues early on, 6.5 has been rock solid for me over the years. Each quarterly update has been surprise-free and without incident.
SGI MIPS/IRIX Roadmap:
http://www.sgi.com/developers/feature/2000/irix_ mi ps.html
The Mandate of Application Compatibility in SGI IRIX 6.5
(An excellent whitepaper on the goals and future of IRIX 6.5, written by an IRIX 6.5 engineer)
http://techpubs.sgi.com/library/tpl/cgi-bin/brow se.cgi?coll=0650&db=bks&cmd=toc&pth=/SGI_Developer/m andate_IRIX
DMA is direct memory access, that is, an operation that will bypass the CPU for better performance.
To enable DMA for a hard drive:
hdparm -d1/dev/hda
To disable DMA for a hard drive:
hdparm -d0/dev/hda
To enable Ultra-DMA/66:
hdparm -X66/dev/hda
To measure the transfer rate of a hard drive:
hdparm -Tt/dev/hda
To see what options are enabled for a hard drive:
hdparm/dev/hda
To see many details about your drive:
hdparm -i/dev/hda
I have an onboard ATA/33 IDE controller and a hard drive that supports ATA/66, so obviously I cannot use ATA/66. But by using stock ATA/33 and turning DMA on, I was able to get about a 30% boost in performance when tested by hdparm. I only know the basics and a lot of is still cloudy and/or seems to be voodoo magic. Could someone please explain this better for me as well? Also, should the above commands be entersted into an RC startup script?
This is what you can tell people when they tell you that linux is a toy. The best search engine in the world is *not* a toy
I think you pretty much said it all. Running 8000 personal computers to run a seach engine site, heh. You are correct, the best search engine in the world isn't a toy. It isn't Google either.
Seeing how SGI is strapped for cash, yet makes some awesoeme hardware and a rock solid OS, I bet they would have cut Google a good deal on a set of four maxed out O3Ks (512 CPU and 1TB RAM each).
First of all, there is no way that Google is sustaining a full load of 16 GB/sec from disk or fully streaming of their 16 TB of ram. The overhead is probably with the branching and routing of the reguests and really can't be overcome without a major software overhaul... adding more hardware is cheaper than developer time, plus adds redundancy and storage.
As far as no single memory system being that large, you are correct. The largest *single* system I can think of is the SGI Origin 3000 which maxes out at "just" 512 MIPS R14K CPUs with 1TB RAM. Storage wouldn't be from SCSI, but rather Fibrechanel... 4 XIO fc_al cards per CPU brick (or 1 fc_al card per cpu) would be about as dense as you'd want to go... that's be about 51.2 GB/sec.
I was thinking they were opensource. Knowing this I don't feel like even giving them the time of day. Anyone know of a good opensource google-like engine?
Why bother to put together 8,000 Linux boxes, when one could obtain high-powered 64-bit computers to accomplish the same task?
Redundancy is often a common answer, they could have *FOUR THOUSAND* failures and still keep on chugging. Still, I agree, I wouldn't want to be in charge of keeping 8000 personal computers happy. I would probably load balance the thing between, oh, four different 128-CPU (four racks each) SGI Origin 3800s. Maybe just 64-CPU models or a large Sun. Keeping 4 machines going is a hellofa lot easier than 8000, plus you get just about the same amount of bandwidth. The backplane-less Origin 3000 series uses *gobs* of 3.6 GB/sec "NUMAlink" interconnects. Not exactly your dad's gigE or myranet.
I assume Google's search engine software is opensource? Where can I download a copy? I would like to use it on about 4 Linux PCs to index and search my company's web-based intranet.
If they know what they're doing it's probably a Clariion (I don't think EMC^2 disk arrays even come smaller than 20 TB). If they're a bunch of linux kiddies, it's probably some IDE RAID with a a pair of Promise cards.
... they're supposedly archiving and restoring historical documents for corporations with rich and detailed backgrounds. A tour of their complex reveals only two consumer scanners, storage facilities that consist of 3 "IDE RAIDs", and not a single paper/book restoration lab. Yet their employee rec center has a 140" projected TV and movie room and a retro arcade. Their "data center" has a 12-node beowulf cluster that seems to run benchmark scripts around the clock. When I asked what sort of scanning and storage facilities they had, the rep dodged the question and told me that they have achieve a sustained 25 GFLOPs with their cluster.
I wasn't impressed and took my business elsewhere.
...pretty much the same. Many of my machines are Pentium 233 MMX-based. A few sport the AMD K6-2. My most powerful is a 350 MHz Pentium II (with a TNT2). Ya know what, it doesn't feel all that slow at all. Apache takes more than a few seconds to compile, but that's about it. I can even play Quake 3 Arena in Windows at 640x480 with nice framerates. The best thing is, I'm not shelling out money every 8 months to keep up with the Joneses.
I'm happy to see that Intel, AMD, and others keep chugging along. My next upgrade will probably be this fall sometime and I'm looking forward to what will be offered then. In the mean time, my current x86 systems (and my SGI MIPS/IRIX and Sun SPARC/Solaris) machines from the days of yore are even more powerful than I'll probably need for the next 24 months. The announcement of the 1.7 GHz P4 is neat, but it's not something I'm drooling/lusting over or "need" in any meaning of the word. Neat stuff, but my hardware is doing fine, thank you.
AFAIK, the support cost of your average corporate personal computer is often higher than what the machine itself costs. Not to mention that cost is not always the sole concern, especially at firms that have huge profits -- many can afford to spend a little (or a lot) more. Do you really think they need $800 Aereon chairs at every desk?
Why would any server farm have graphics on the servers? Plug the servers into the lan via eithernet and connect the serial/console port of the server into a Portmaster or similar terminal server. I never understood why sooo many folks think that their servers need to have framebuffers/gfxcards/"video"cards.
Both NCD and Tektronix have terminals that handle Windows (Citrix/WinDD/WindowsTerminalServer) as well as X and even plain vt100 telnet sessions. Very cool setups. Walk up and login to whatever flavor machine you desire. Even xdmcp into your home linux/bsd/unix machine. Neat stuff.
Terminal server would work quite well. Set up a fairly powerful Sun server and install a bunch of SunRay terminals about the building. The hot session management is quite slick. Generic X terminals with an unix or unix-like host works well too. NCD makes some nice termials that work with X as well as Citrix/Windows.
It sounds a lot like an -interactive- demo that SGI has for some of their 3-projector panoramic "reality center" setups. Fly around earth and then "bungie jump" down all the way to one of about 8 cities, then pull back up. I forget how large they said their texture database was but 400 GB sounds about right. Their Ciprico fibrechannel disk arrays were beside the Onyx2's in the corner of the room and golly were their activity lights every going like crazy. If I recall correctly, we were able to from space to street level in about 15 seconds smooth as butter on that crazy high resolution screen, though we pulled back to space a lot slower but just as smooth.
Anyone else recall this demo?
Ever been to the Hayden Planetarium?? SGI+Zeiss!!
on
Solar System Simulator
·
· Score: 1
The (new) Hayden Planetarium has a *very cool* projection system on a hemispherical dome. Combines output from a 28-CPU, 7-GFX-pipe SGI Onyx2 (~1999 technology) with some computer controlled lasers and a new computer controlled Zeiss star projector. *Very* impressive stuff. This ain't the planetarium that your parents took you to as a kid, it'll blow you away.
Yeah, yeah, I know the jokes... Plane-Arium. And yeah, yeah, SGI doesn't include their compilers with thier OS... but it's still the coolest thing I have ever visited.
Slashdot Mirror
Please note that there are some gaps (spaces) in the middle of the two URLs I pasted. They need to be removed in order for the URL to play nice with your web browser.
SGI MIPS/IRIX Roadmap
The Mandate of Application Compatibility in SGI IRIX 6.5 (An excellent whitepaper on the goals and future of IRIX 6.5, written by an IRIX 6.5 engineer
IRIX 6.5 can best be described by the neat poster SGI made about a year ago... it features a train chugging along past 5.3 all the way to 6.5 and through the 6.5.X updates. The front of the locomotive sports the designation "6.5.oo" (infinity). While 6.5 most likely won't be around forever, there are NO current plans to release a 6.6 or a 7.0.
_ mi ps.html
w se .cgi?coll=0650&db=bks&cmd=toc&pth=/SGI_Developer/m andate_IRIX
Why?
Becasue 6.5 is the best thing SGI has ever done. There is an update released quarterly, on time, along two different streams: [M]aintenance and [F]eature. No patch dependency hell (though urgent patches are posted between 6.5.X updates). The Feature release gets new goodies every quarter, some of which are slowly rolled into Maintenance. Both streams are HEAVILY tested, tuned, reviewed, and compiled using the latest available MIPSpro compilers. In fact, SGI likes to stay a release or two ahead of their users, using it on most of their production systems to ensure a rock solid release.
The (large) IRIX 6.5 team has been plodding away ever since 6.5.0. When asked "where's 6.6" they will usually respond: "you didn't ask us to break application compatibility, so we aren't working on a 6.6".
I wouldn't want it any other way. Aside from a few small networking issues early on, 6.5 has been rock solid for me over the years. Each quarterly update has been surprise-free and without incident.
SGI MIPS/IRIX Roadmap:
http://www.sgi.com/developers/feature/2000/irix
The Mandate of Application Compatibility in SGI IRIX 6.5
(An excellent whitepaper on the goals and future of IRIX 6.5, written by an IRIX 6.5 engineer)
http://techpubs.sgi.com/library/tpl/cgi-bin/bro
DMA is direct memory access, that is, an operation that will bypass the CPU for better performance.
/dev/hda
/dev/hda
/dev/hda
/dev/hda
/dev/hda
/dev/hda
To enable DMA for a hard drive:
hdparm -d1
To disable DMA for a hard drive:
hdparm -d0
To enable Ultra-DMA/66:
hdparm -X66
To measure the transfer rate of a hard drive:
hdparm -Tt
To see what options are enabled for a hard drive:
hdparm
To see many details about your drive:
hdparm -i
I have an onboard ATA/33 IDE controller and a hard drive that supports ATA/66, so obviously I cannot use ATA/66. But by using stock ATA/33 and turning DMA on, I was able to get about a 30% boost in performance when tested by hdparm. I only know the basics and a lot of is still cloudy and/or seems to be voodoo magic. Could someone please explain this better for me as well? Also, should the above commands be entersted into an RC startup script?
So, do tell us ... what is the best search engine in the world?
Veronica
This is what you can tell people when they tell you that linux is a toy. The best search engine in the world is *not* a toy
I think you pretty much said it all. Running 8000 personal computers to run a seach engine site, heh. You are correct, the best search engine in the world isn't a toy. It isn't Google either.
why is this marked flamebait?
Seeing how SGI is strapped for cash, yet makes some awesoeme hardware and a rock solid OS, I bet they would have cut Google a good deal on a set of four maxed out O3Ks (512 CPU and 1TB RAM each).
Amen. Good point.
First of all, there is no way that Google is sustaining a full load of 16 GB/sec from disk or fully streaming of their 16 TB of ram. The overhead is probably with the branching and routing of the reguests and really can't be overcome without a major software overhaul... adding more hardware is cheaper than developer time, plus adds redundancy and storage.
As far as no single memory system being that large, you are correct. The largest *single* system I can think of is the SGI Origin 3000 which maxes out at "just" 512 MIPS R14K CPUs with 1TB RAM. Storage wouldn't be from SCSI, but rather Fibrechanel... 4 XIO fc_al cards per CPU brick (or 1 fc_al card per cpu) would be about as dense as you'd want to go... that's be about 51.2 GB/sec.
I was thinking they were opensource. Knowing this I don't feel like even giving them the time of day. Anyone know of a good opensource google-like engine?
Why bother to put together 8,000 Linux boxes, when one could obtain high-powered 64-bit computers to accomplish the same task?
Redundancy is often a common answer, they could have *FOUR THOUSAND* failures and still keep on chugging. Still, I agree, I wouldn't want to be in charge of keeping 8000 personal computers happy. I would probably load balance the thing between, oh, four different 128-CPU (four racks each) SGI Origin 3800s. Maybe just 64-CPU models or a large Sun. Keeping 4 machines going is a hellofa lot easier than 8000, plus you get just about the same amount of bandwidth. The backplane-less Origin 3000 series uses *gobs* of 3.6 GB/sec "NUMAlink" interconnects. Not exactly your dad's gigE or myranet.
Their full and final product is not a toy, that's for sure. Using *EIGHT THOUSAND* personal computers to run it sure is... it's quite goofy.
I assume Google's search engine software is opensource? Where can I download a copy? I would like to use it on about 4 Linux PCs to index and search my company's web-based intranet.
If they know what they're doing it's probably a Clariion (I don't think EMC^2 disk arrays even come smaller than 20 TB). If they're a bunch of linux kiddies, it's probably some IDE RAID with a a pair of Promise cards.
Have you looked at or used his code?
... they're supposedly archiving and restoring historical documents for corporations with rich and detailed backgrounds. A tour of their complex reveals only two consumer scanners, storage facilities that consist of 3 "IDE RAIDs", and not a single paper/book restoration lab. Yet their employee rec center has a 140" projected TV and movie room and a retro arcade. Their "data center" has a 12-node beowulf cluster that seems to run benchmark scripts around the clock. When I asked what sort of scanning and storage facilities they had, the rep dodged the question and told me that they have achieve a sustained 25 GFLOPs with their cluster.
I wasn't impressed and took my business elsewhere.
...pretty much the same. Many of my machines are Pentium 233 MMX-based. A few sport the AMD K6-2. My most powerful is a 350 MHz Pentium II (with a TNT2). Ya know what, it doesn't feel all that slow at all. Apache takes more than a few seconds to compile, but that's about it. I can even play Quake 3 Arena in Windows at 640x480 with nice framerates. The best thing is, I'm not shelling out money every 8 months to keep up with the Joneses.
I'm happy to see that Intel, AMD, and others keep chugging along. My next upgrade will probably be this fall sometime and I'm looking forward to what will be offered then. In the mean time, my current x86 systems (and my SGI MIPS/IRIX and Sun SPARC/Solaris) machines from the days of yore are even more powerful than I'll probably need for the next 24 months. The announcement of the 1.7 GHz P4 is neat, but it's not something I'm drooling/lusting over or "need" in any meaning of the word. Neat stuff, but my hardware is doing fine, thank you.
AFAIK, the support cost of your average corporate personal computer is often higher than what the machine itself costs. Not to mention that cost is not always the sole concern, especially at firms that have huge profits -- many can afford to spend a little (or a lot) more. Do you really think they need $800 Aereon chairs at every desk?
Why would any server farm have graphics on the servers? Plug the servers into the lan via eithernet and connect the serial/console port of the server into a Portmaster or similar terminal server. I never understood why sooo many folks think that their servers need to have framebuffers/gfxcards/"video"cards.
Both NCD and Tektronix have terminals that handle Windows (Citrix/WinDD/WindowsTerminalServer) as well as X and even plain vt100 telnet sessions. Very cool setups. Walk up and login to whatever flavor machine you desire. Even xdmcp into your home linux/bsd/unix machine. Neat stuff.
Terminal server would work quite well. Set up a fairly powerful Sun server and install a bunch of SunRay terminals about the building. The hot session management is quite slick. Generic X terminals with an unix or unix-like host works well too. NCD makes some nice termials that work with X as well as Citrix/Windows.
It uses lossless compression and a differential digital signal. Read the whitepapers on any of the long-distance KVM gizmos. Interesting stuff.
Why would a failed compile require you to walk over to your computer? Debug your code and cc again.
It sounds a lot like an -interactive- demo that SGI has for some of their 3-projector panoramic "reality center" setups. Fly around earth and then "bungie jump" down all the way to one of about 8 cities, then pull back up. I forget how large they said their texture database was but 400 GB sounds about right. Their Ciprico fibrechannel disk arrays were beside the Onyx2's in the corner of the room and golly were their activity lights every going like crazy. If I recall correctly, we were able to from space to street level in about 15 seconds smooth as butter on that crazy high resolution screen, though we pulled back to space a lot slower but just as smooth.
Anyone else recall this demo?
The (new) Hayden Planetarium has a *very cool* projection system on a hemispherical dome. Combines output from a 28-CPU, 7-GFX-pipe SGI Onyx2 (~1999 technology) with some computer controlled lasers and a new computer controlled Zeiss star projector. *Very* impressive stuff. This ain't the planetarium that your parents took you to as a kid, it'll blow you away.
. html
http://www.sgi.com/features/2000/feb/hayden/index
http://www.amnh.org/rose/digitaldome.html
Yeah, yeah, I know the jokes... Plane-Arium. And yeah, yeah, SGI doesn't include their compilers with thier OS... but it's still the coolest thing I have ever visited.