Posted by
ryuzaki0
on from the we-wants-it-we-do dept.
Jonathan C. Patschke writes "SGI unveiled two new graphics workhorses today, the Tezro
(an Octane2 replacement) and the much-anticipated Onyx 4. The presence of the old "bug" logo warms the cockles of my heart, even if the desktop Tezro looks much like a subwoofer."
Yet more machines for geeks to dribble over.. I know I wouldn't mind one of those on my desk, even if all I used it for was browsing the net and checking my email..
Though its worth bearing in mind that you can still pick up some half decent SGI workstations on eBay.. seen some SGI Octane / 20" Monitor / 768MB RAM bundles on UK eBay for around £350 which is a superb deal.. these things might be getting on a bit, but they certainly do shift.
I used to own both an old Indy and an Indigo2, both of which would be the equivilant of an 8086 in PeeCee computing terms.. but they still cruised along even on the latest version of Irix, and were surprisingly usable:)
Really must get another SGI some day..
-- "Hey! Unless this is a nude love-in, get the hell off my property!!"
I used to own both an old Indy and an Indigo2, both of which would be the equivilant of an 8086 in PeeCee computing terms..
Actually, no, they aren't. A more accurate comparison would be a P5 series processor at a similar clock rate.
You forget the several previous generations of machines such as the Indigo or the Personal Iris and they were drastically faster than an 8086... To find the first machines produced you have to go waaaaay back to 1983 and the Iris1X00.
Re:Nice...
by
sql*kitten
·
· Score: 4, Interesting
Though its worth bearing in mind that you can still pick up some half decent SGI workstations on eBay.. seen some SGI Octane / 20" Monitor / 768MB RAM bundles on UK eBay for around £350 which is a superb deal.. these things might be getting on a bit, but they certainly do shift.
I used to own both an old Indy and an Indigo2, both of which would be the equivilant of an 8086 in PeeCee computing terms.. but they still cruised along even on the latest version of Irix, and were surprisingly usable:)
A PC is a general purpose device that is designed not to suck too badly at anything in particular. A workstation is a specialist device that is designed to retain some general purpose capability. Back in its day, the Indigo2 IMPACT was an impressive machine... you couldn't buy a PC that could do what it could do at any price. Even now, they can hold their own in solid modelling and CAD.
I have an Octane SE here, 1997 vintage, and my 2002-issue Dell beats it for small CPU bound jobs... but for anything involving a lot of memory accesses, or disk I/O the Octane wins hands down every time. And if I'm not using textures, SE graphics can easily beat a GeForce2.
How relevant are these boxes?
by
Surak
·
· Score: 1, Interesting
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?
A very GOOD THING [TM]
by
green+pizza
·
· Score: 4, Interesting
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.
Re:A very GOOD THING [TM]
by
fgodfrey
·
· Score: 2, Interesting
See, this is what I get for not using "preview". Right after the "I digress", insert the following:
The link to local memory is even faster. When you are doing scientific computing, ie. what these machines are sold for, odds are your problem isn't going to come close to fitting in cache in which case your poor P4 is going to spend 50% or more of its time waiting for the results of loads from memory.
-- Go Badgers! -- #include "std/disclaimer.h"
Onyx and LOTR
by
GillBates0
·
· Score: 5, Interesting
They have an interesting page about the success stories of SGI graphics workstations.
A particularly interestingone about their role in the making of the LOTR:
The Wellington, New Zealand, company is using a full complement of IRIX OS-based Silicon Graphics® Octane® and Silicon Graphics® Onyx2® visual workstations, SGI® Origin® family servers, and SGI Linux OS-based visual workstations and servers to create and manage up to 100TB of data. Cool pictures too.
-- An Indian-American Hindu committed to non-violent thought/speech/action alarmed by the global explosion of radical Islam
Re:cool
by
Anonymous Coward
·
· Score: 1, Interesting
Just for shits-n-grins, I'd like to see some sgi vs. apple rendering/modeling benchmarks.
Seems to me that sgi's only real computational advantages show up in the data modeling arenae; weather, molecules, etc...
They've both got their plusses and minuses, the most impressive of which differ greatly between machines. Where's the overlap?! That's what I wanna know. How close is Apple *really* to taking on sgi's last vestiges of profitability?
Abyss Nostalgia
by
nacturation
·
· Score: 4, Interesting
I remember at university when SGI came around with their trailer full of cool boxes. This was around 1990 - 1991. The one thing I remember about that event was the real-time demonstration of the water tentacle effect from The Abyss.
No other machine could even come close to rendering this kind of thing real-time. These days, we're spoiled by high-end graphics cards costing only hundreds of dollars which eclipse what SGI could do back then by a factor of 10.
-- Want to improve your Karma? Instead of "Post Anonymously", try the "Post Humously" option.
Hmm, let's see a 8086 do realtime capturing and displaying of an ntsc video source on a 24bit 1280x1024 display. Now to be honest, 486 to low end Pentium would be a better comparison. Of course assuming these machines had some type of video capture board installed and a pretty kick butt scsi setup. Not the best things in the world for day to day tasks, but if you're doing the right thing, then they are quite nice (Indy less so since it's not as expandable, but one can create a pretty beefy I2. Not to mention the O2.
Not so good reason imho. The new Onyx graphics have less texture memory than InfiniteReality, no 48 bit color and lacks all the extensions of IR. Sure, it's faster, but couldn't they have tried to speed up IR instead of going with Ati ?
This is Silicon Graphics we're talking about. They used to be the only option.
And the high price tag is worth it when you have very little down time doing 3D work compared to what we have with PCs.
And for real-time graphics we have yet to find a PC with ANY hardware that can output NTSC as nicely as the Onyx2 we use for live, on-air graphics. We were contemplating a PC, but even with the best video cards it couldn't run the scenes we had already created and ran on the SGI.
My department has switched mostly over to PCs, and because of pressure from accounting they are looking for ways to replace the SGIs with cheaper equipment, but there are areas where it's simply still not possible. I don't expect we'll be downgrading to consumer level graphics though, and we'll continue using the existing systems.
-- Stupid sexy Flanders.
A Very Odd Datasheet. Where's the processor?
by
Nova+Express
·
· Score: 4, Interesting
I actually downloaded the datasheet for the "Silicon Graphics Onyx4 Ultimate Vision Family," and found it a very curious document indeed. It has some interesting hard facts about the system (OpenGL 1.4, 8-32 graphic processor pipes on the "Extreme," up to 8 GB of graphic memory (sweet!), etc.), but what I was looking was the type and speed of the processors used. So I kept looking.
And looking.
And looking.
It's not there.
SGI's own datasheet for the Onyx4 Family doesn't tell you what processor it runs! Others in the thread have said it uses MIPS chips, but the word "MIPS" never appears in the datasheet (nor "RISC," for that matter). It tells you how many processors the system uses, but not what they are or how fast they are.
This is not just odd; for a datasheet, it's nearly unprecedented. Only three explanations for this abscence occur to me:
They have the world's most incompetent technical writers. (Very unlikely.)
They're actually ashamed of their CPU, and don't want to tell you what it is or how fast in runs. (Most likely.)
They're desperately working behind the scenes to port their software to commodity hardware (mostly likely x86, but the 970/G5 might be a smarter choice). (Unlikely, but not impossible.)
I have no idea how fast the current generation of MIPS chips are (I think the last time I saw a benchmark, they were slower than Alphas, which tells you it was back when they were still benchmarking Alphas rather than letting them die a quiet and undeserved death), but the fact that SGI isn't even willing to mention them in their datasheet doesn't give me confidence.
-- Lawrence Person (lawrencepersonh@gmailh.com (remove all "h"s to mail)
Hmm, let's see a 8086 do realtime capturing and displaying of an ntsc video source on a 24bit 1280x1024 display.
no problem... give me a scsi card and a hardware capture card... I know a 286 can do it, so get a 8 bit scsi card and I'll show you.
the computer is nothing more than a simple way of making the good powerful hardware talk to each other. Hell most high end capture cards are NOT PCI/ISA/or whatever but they are SCSI. same with the high end Video output cards. and using them takes almost no processing power.
This is how I stream 48 DIFFERENT dvd quality mpeg2 files out to NTSC video at the same time on a pentium 133 motherboard with 32 meg of ram on it running NT 3.51... simple scsi commands to high end hardware.
the computer has nothing to do with what is getting done except for making it easier to install the newest command software.
-- Do not look at laser with remaining good eye.
Its the Software that's expensive...
by
cutecub
·
· Score: 5, Interesting
I used to work for SGI and also did freelance video animation ( a very long time ago ) on an SGI Indy.
As an individual, the biggest problem I encountered wasn't the cost of the SGI system (a one-time cost), it was the cost of the system software and drivers.
OS upgrades were expensive.
Print drivers were expensive.
Networking options were expensive.
The compilers were unbundled.
Most of the software Open Source geeks nowadays take for granted as being free, cheap, and readily available was expensive and exotic on the SGI.
I ultimately switched to a high-end Macintosh. Today, the Mac is an even more compelling alternative to a low-end SGI for media production.
I don't know about SGI's other niches, such as Scientific Visualization, but I would expect high-end PCs to have the edge over low-end SGIs in other areas.
-S
Re:A few notes...
by
pmz
·
· Score: 2, Interesting
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.
Trash! My new PC supports AGP 31.415x and has DDR 7000pHz RAM all for $8.65! Hyperthreading and RAM hacks to the max!
(I'm just joking, here; my most powerful PC is actually a old SunPCi card, for better or worse)
What would be interesting is a comparison of the recent high-end offerings. Sun released their V880z machine recently, and I'm sure IBM could throw something into the mix. However, I guess the manufacturers would hesitate to mail $150,000 samples to magazine reviewers:(
I run an Indigo 2 with a MIPS R4400 200mhz 1mb cache....which is a fairly low end model. By doing some rough cpu benchmarks (rendering scenes in POV-Ray), it seemed to place a little shy of my old PC, which is a 233mhz Pentium II.
Note that for the video capture above on an Indigo 2, you'd need the galileo video option, or some similar device, but that's nit-picking.
The MIPS R10k and R12k systems are far superior, though even with the Max Impact graphics set, you will be disappointed with textured 3d if you're used to current hardware (max is similar to a TnT2, I'd say).
Still, my indigo was out (with the "extreme" graphics set - 8 GL processors, 2 raster engines, 1 geometry pre-processor, no texture) when the best PC you could buy was a decked-out 486, or possibly a pentium 60.
I still want a Crimson Reality Engine, just for the coolness factor. Best looking case ever!:)
I guess the geeks don't hang out on/. as they once did. The original IBM Personal Computer (circa 1981 - 1983) used the Intel 8088 chip, not the 8086. Although related, the 8088 is a distinct chip that uses an 8-bit (as opposed to 16-bit) instruction/data bus and intergrates a few additional features that allow for 5 less glue logic chips, resulting in lower manufacturing costs in addition to the 8-bit expansion slots being cheaper.
Although IBM considered upgrading the design to the 80186 when it appeared that Intel could not deliver the 286 chip on schedule, they wisely skipped that step and the PC-AT first appeared with a 6MHz 80286 processor -- crippled addressing and all.
Now for extra points, what clock-rate did the original IBM PC operate at, and why?
-- "It's the height of ridiculousness to say for those 9 lines you get hundreds of millions."
Apple should take over SGI
by
afantee
·
· Score: 3, Interesting
>> Wouldnt the new Apple G5's with dual 2 ghz cpus crush it?
It most certainly will, in probably every single aspect. The dual 2 GHz G5 Power Mac has 2 independent 1 GHz FSBs, dual channel 128-bit 400 MHz DDR RAM, dual 800 MHz HyperTransport interconnects, dual SATA drives with 1.5 Gbps throughput per channels.
Not only the G5 is 3x faster than the MIPS R16000 in clock speed, it also has 2 FPUs and can handle 215 simultaneous in-flight instructions, so most likely will beat the MIPS per cycle as well, not to mention the Altivec vector unit.
Of course, there are much more native Mac software, and the G5 is probably much cheaper. The only place where SGI beats Apple is at the high end super computing market, but even there it's probably better to use G5 clusters.
Currently SGI is only valued for $260, about 6% of Apple's $4.5 B cash pile, so maybe Apple should acquire SGI in order to move into the scientific computing and visualization market.
Re:What?
by
Anonymous Coward
·
· Score: 1, Interesting
Actually, 4.77MHz, which allowed relatively easy division to the 3.58MHz color burst signal of NTSC.
The actual crystals were 14.32... MHz -
four times the NTSC subcarrier, and cheaply
available due to their use in home VCRs.
This was divided by three to give the 4.77 MHz
CPU clock.
It was further divided by four to give the 1.193 MHz clock for the main timer chip - a frequency
still in use today.
Slashdot Mirror
Yet more machines for geeks to dribble over.. I know I wouldn't mind one of those on my desk, even if all I used it for was browsing the net and checking my email..
:)
Though its worth bearing in mind that you can still pick up some half decent SGI workstations on eBay.. seen some SGI Octane / 20" Monitor / 768MB RAM bundles on UK eBay for around £350 which is a superb deal.. these things might be getting on a bit, but they certainly do shift.
I used to own both an old Indy and an Indigo2, both of which would be the equivilant of an 8086 in PeeCee computing terms.. but they still cruised along even on the latest version of Irix, and were surprisingly usable
Really must get another SGI some day..
"Hey! Unless this is a nude love-in, get the hell off my property!!"
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?
My journal has hot
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.
A particularly interestingone about their role in the making of the LOTR:
The Wellington, New Zealand, company is using a full complement of IRIX OS-based Silicon Graphics® Octane® and Silicon Graphics® Onyx2® visual workstations, SGI® Origin® family servers, and SGI Linux OS-based visual workstations and servers to create and manage up to 100TB of data. Cool pictures too.
An Indian-American Hindu committed to non-violent thought/speech/action alarmed by the global explosion of radical Islam
This guy is a troll, mod him down... check out his other posts specifically this one. (rather amusing)
Just for shits-n-grins, I'd like to see some sgi vs. apple rendering/modeling benchmarks.
Seems to me that sgi's only real computational advantages show up in the data modeling arenae; weather, molecules, etc...
They've both got their plusses and minuses, the most impressive of which differ greatly between machines. Where's the overlap?! That's what I wanna know. How close is Apple *really* to taking on sgi's last vestiges of profitability?
I remember at university when SGI came around with their trailer full of cool boxes. This was around 1990 - 1991. The one thing I remember about that event was the real-time demonstration of the water tentacle effect from The Abyss.
No other machine could even come close to rendering this kind of thing real-time. These days, we're spoiled by high-end graphics cards costing only hundreds of dollars which eclipse what SGI could do back then by a factor of 10.
Want to improve your Karma? Instead of "Post Anonymously", try the "Post Humously" option.
Hmm, let's see a 8086 do realtime capturing and displaying of an ntsc video source on a 24bit 1280x1024 display. Now to be honest, 486 to low end Pentium would be a better comparison. Of course assuming these machines had some type of video capture board installed and a pretty kick butt scsi setup. Not the best things in the world for day to day tasks, but if you're doing the right thing, then they are quite nice (Indy less so since it's not as expandable, but one can create a pretty beefy I2. Not to mention the O2.
Not so good reason imho.
The new Onyx graphics have less texture memory than InfiniteReality, no 48 bit color and lacks all the extensions of IR. Sure, it's faster, but couldn't they have tried to speed up IR instead of going with Ati ?
This is Silicon Graphics we're talking about. They used to be the only option.
I have to agree...
And the high price tag is worth it when you have very little down time doing 3D work compared to what we have with PCs.
And for real-time graphics we have yet to find a PC with ANY hardware that can output NTSC as nicely as the Onyx2 we use for live, on-air graphics. We were contemplating a PC, but even with the best video cards it couldn't run the scenes we had already created and ran on the SGI.
My department has switched mostly over to PCs, and because of pressure from accounting they are looking for ways to replace the SGIs with cheaper equipment, but there are areas where it's simply still not possible. I don't expect we'll be downgrading to consumer level graphics though, and we'll continue using the existing systems.
Stupid sexy Flanders.
And looking.
And looking.
It's not there.
SGI's own datasheet for the Onyx4 Family doesn't tell you what processor it runs! Others in the thread have said it uses MIPS chips, but the word "MIPS" never appears in the datasheet (nor "RISC," for that matter). It tells you how many processors the system uses, but not what they are or how fast they are.
This is not just odd; for a datasheet, it's nearly unprecedented. Only three explanations for this abscence occur to me:
I have no idea how fast the current generation of MIPS chips are (I think the last time I saw a benchmark, they were slower than Alphas, which tells you it was back when they were still benchmarking Alphas rather than letting them die a quiet and undeserved death), but the fact that SGI isn't even willing to mention them in their datasheet doesn't give me confidence.
Lawrence Person (lawrencepersonh@gmailh.com (remove all "h"s to mail)
http://www.lawrenceperson.com/
Hmm, let's see a 8086 do realtime capturing and displaying of an ntsc video source on a 24bit 1280x1024 display.
no problem... give me a scsi card and a hardware capture card... I know a 286 can do it, so get a 8 bit scsi card and I'll show you.
the computer is nothing more than a simple way of making the good powerful hardware talk to each other. Hell most high end capture cards are NOT PCI/ISA/or whatever but they are SCSI. same with the high end Video output cards. and using them takes almost no processing power.
This is how I stream 48 DIFFERENT dvd quality mpeg2 files out to NTSC video at the same time on a pentium 133 motherboard with 32 meg of ram on it running NT 3.51... simple scsi commands to high end hardware.
the computer has nothing to do with what is getting done except for making it easier to install the newest command software.
Do not look at laser with remaining good eye.
I used to work for SGI and also did freelance video animation ( a very long time ago ) on an SGI Indy.
As an individual, the biggest problem I encountered wasn't the cost of the SGI system (a one-time cost), it was the cost of the system software and drivers.
OS upgrades were expensive.
Print drivers were expensive.
Networking options were expensive.
The compilers were unbundled.
Most of the software Open Source geeks nowadays take for granted as being free, cheap, and readily available was expensive and exotic on the SGI.
I ultimately switched to a high-end Macintosh. Today, the Mac is an even more compelling alternative to a low-end SGI for media production.
I don't know about SGI's other niches, such as Scientific Visualization, but I would expect high-end PCs to have the edge over low-end SGIs in other areas.
-S
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.
:(
Trash! My new PC supports AGP 31.415x and has DDR 7000pHz RAM all for $8.65! Hyperthreading and RAM hacks to the max!
(I'm just joking, here; my most powerful PC is actually a old SunPCi card, for better or worse)
What would be interesting is a comparison of the recent high-end offerings. Sun released their V880z machine recently, and I'm sure IBM could throw something into the mix. However, I guess the manufacturers would hesitate to mail $150,000 samples to magazine reviewers
Healthcare article at Kuro5hin
I run an Indigo 2 with a MIPS R4400 200mhz 1mb cache....which is a fairly low end model. By doing some rough cpu benchmarks (rendering scenes in POV-Ray), it seemed to place a little shy of my old PC, which is a 233mhz Pentium II.
:)
Note that for the video capture above on an Indigo 2, you'd need the galileo video option, or some similar device, but that's nit-picking.
The MIPS R10k and R12k systems are far superior, though even with the Max Impact graphics set, you will be disappointed with textured 3d if you're used to current hardware (max is similar to a TnT2, I'd say).
Still, my indigo was out (with the "extreme" graphics set - 8 GL processors, 2 raster engines, 1 geometry pre-processor, no texture) when the best PC you could buy was a decked-out 486, or possibly a pentium 60.
I still want a Crimson Reality Engine, just for the coolness factor. Best looking case ever!
man tunefs | grep fish
I guess the geeks don't hang out on /. as they once did. The original IBM Personal Computer (circa 1981 - 1983) used the Intel 8088 chip, not the 8086. Although related, the 8088 is a distinct chip that uses an 8-bit (as opposed to 16-bit) instruction/data bus and intergrates a few additional features that allow for 5 less glue logic chips, resulting in lower manufacturing costs in addition to the 8-bit expansion slots being cheaper.
Although IBM considered upgrading the design to the 80186 when it appeared that Intel could not deliver the 286 chip on schedule, they wisely skipped that step and the PC-AT first appeared with a 6MHz 80286 processor -- crippled addressing and all.
Now for extra points, what clock-rate did the original IBM PC operate at, and why?
"It's the height of ridiculousness to say for those 9 lines you get hundreds of millions."
>> Wouldnt the new Apple G5's with dual 2 ghz cpus crush it?
It most certainly will, in probably every single aspect. The dual 2 GHz G5 Power Mac has 2 independent 1 GHz FSBs, dual channel 128-bit 400 MHz DDR RAM, dual 800 MHz HyperTransport interconnects, dual SATA drives with 1.5 Gbps throughput per channels.
Not only the G5 is 3x faster than the MIPS R16000 in clock speed, it also has 2 FPUs and can handle 215 simultaneous in-flight instructions, so most likely will beat the MIPS per cycle as well, not to mention the Altivec vector unit.
Of course, there are much more native Mac software, and the G5 is probably much cheaper. The only place where SGI beats Apple is at the high end super computing market, but even there it's probably better to use G5 clusters.
Currently SGI is only valued for $260, about 6% of Apple's $4.5 B cash pile, so maybe Apple should acquire SGI in order to move into the scientific computing and visualization market.
The actual crystals were 14.32... MHz - four times the NTSC subcarrier, and cheaply available due to their use in home VCRs. This was divided by three to give the 4.77 MHz CPU clock. It was further divided by four to give the 1.193 MHz clock for the main timer chip - a frequency still in use today.
Simon