If you really don't want to run Linux, you can of course run Solaris 10 on the v40z.
I know it's heresy to say this here, but Solaris 10 is a far better server OS than Linux, and even more so on Opteron. Solaris 10 is also a better workstation OS on Opteron than Linux.
Linux will catch up to Solaris 10 in the next 2 years as 64-bit PeeCees become mainstream, but just now you can't beat S10 on Opteron.
Yes, I'm biased, I worked on S10.
Linux got me to where I am today, along with Solaris. Linux is still very immature in many respects. SGI and IBM have been paying it lip-service over the years, and tossing it the odd bone, but Linux has a long way to go.
I don't doubt that it will, and it will kill Solaris, like it killed Irix, AIX, HP/UX and the others. It's only a matter of time. However, right now, Solaris 10 is far superior technically.
Yes, the whole point of itanic was to kill the 64-bit RISC market and to make intel the sole provider of high-end 64-bit processors. Lots of companies unfortunately drank the cool aid back in the '90s, and SGI was one. MIPS was an excellent architecture. The only ones to survive were UltraSPARC (Sun) and POWER (IBM).
The other big mistake SGI made was when they took on Rocket Rick Belluzo (sp?) and he gave them a "Windows NT" strategy. In otherwords, a 10-year step backwards, and an attempt to sell over-priced 32-bit Pentium machines running Windows, where previously SGI had been selling 64-bit MIPS machines running UNIX.
When the pointy-hairs get in charge of a company, it spells death.
Sun will not be long for this world either. It is barely breaking even. Yes, Solaris 10 is superb and so are their Opteron servers and workstations, but the pointy-hairs are grinding the company down internally. The engineers are not longer free to innovate and work on the important stuff. They are given a constant diet of wild goose-chase projects which are ill-conceived and often cancelled upon completion, only to be give more with impossible deadlines and little, if any, thanks let alone financial reward.
Sun will only hover around break-even by continually making more and more staff redundant to "cut costs."
Sun can't market itself or its products to save itself (just look at it). The pointy-hairs keep changing company direction every three months. The engineers are over-worked, under-appreciated, under-rewarded and their opinions are not valued.
Sun PHBs make ill-judged and groundless attacks on Free Software and Open Source almost monthly, they did a deal with Microsoft, they continue to deride Linux where it could have been a great benefit to them and their customers, they can't develop processors for toffee (look at how slow UltraSPARC is, and how expensive).
Luckily Sun didn't do itanic, that's why it's not dead yet. Luckily they decided to go Opteron. Unluckily they left it a bit too late.
Sun should Open Source Java (purely for the good publicity), make 24-, 48- and 72-way Opteron servers and write a software UltraSPARC emulator to run legacy SPARC code. Scott should fire Schwartz and Weinberg. Oh, and they should cease and desist all further UltraSPARC development. It's a complete waste of money. Just use Opterons. They're cheap and very fast and software emulator technology is good nowadays (and I thought that "everything is written in Java" too).
This was the company that set most of the UNIX standards over the last 20 years and has given away more Open Source code than any other (including IBM, SGI, Red Hat,...)....
You EVIL PINKO COMMIES! You UNIONISED! Don't you realise that you are TERRORISTS? You are not WITH US you must be AGAINST us! You are against FREEDOM and the pursuit of WEALTH that is a God-given right of all AMERICANS.
You UNIONISED COMMIE SCUMBAGS. It's people like YOU who are causing POORNESS and shortages of OIL.
You are probably all ATHIESTS and worship the DEVIL and quite possibly Allah.
Next thing we know, you'll all be strapping dynamite to your bellies and blowing up INNOCENT AMERICAN CHRISTIANS who are not UNIONISED.
As a European, I'm more worried at the effect this will have on the world oil market and global warming. Running a car here in the UK is expensive enough due to high oil prices without the added demand from bugs with SUVs.
JDS on Linux was a tactical move to get some people off of Windows and on to an Open platform to run Star Office on the desktop and the Java Enterprise System on the server (preferably SPARC, but now that's so far behind the curve, they'd rather you bought an Opteron V20z or V40z than a box from Dell).
Solaris x86 got ressurected to reverse declining sales of low-end servers, and when Opteron came along, S10 went 64-bit on x86, and got support for commodity hardware (NICs, graphics cards, laptop chipsets).
What with Solaris x86 running better than Linux on the desktop (and laptop), and in 64 bits, there ceased to be any need to sustain JDS on Linux.
A kernel is a kernel. It's cheaper to support and develop on one (Solaris) than on two (Solaris and Linux).
Sun's PHBs have been making some really bad decisions recently. Earlier this year, they RIF'd the Janus engineering team (in-kernel Linux emulation for Solaris) before it had been put into Solaris proper. They said it would be in Solaris by about now. Then they said maybe in a year's time. Now they are saying not at all and that Xen virtualisation would be a better solution... So much for Janus and Zones^H^H^H^H^HContainers.
And what about ZFS? The Zettabyte File System that was going to be lightyears ahead of the competition. Where is that? That was supposed to be in Solaris 10 as well.
And then they bought StorageTek... and then they RIF'd yet more people.
Off the West Coast of Scotland is Gruinard Island. It's uninhabited, and was used during the Second Word War for testing biological weapons. They put some sheep on it and dropped an anthrax bomb on it.
It was left contamintated for decades, until the 1990s IIRC. Allegedly it's all safe now, although they can't garantee that there isn't still some anthrax in the soil, but they put some more sheep on it and they all survived.
I was just going to post "boat-building" too. It goes back a long way in most cultures, and it's what got my dad (and hence me) into computers.
My dad was brought up a crofter in North West Scotland. He inherited an enthusiasm for the sea for his father and so on.
In the late '60s he went to university to study Marine Engineering (thanks to British state-funded Higher Education) and did an MSc after his BSc.
The MSc research involved developing some very early CAD and CAM systems for ship building (24-bit mainframes and CNC milling machines in the early '70s).
I was going on boats since I was born, and have never been sea sick. In fact, on my way to Holland on the SeaCat one day I didn't even spill my pint in the rough weather while all the scurvy land-lubbers were rolling around in puke in the bilges.
The computer stuff stuck. I got my first computer when I was eight, and here I sit this evening. having been frobbing with bootloader code all day on an embedded Linux system, writing my own language interpreter.
It is being aimed at a low to mid-end server environment (decidedly low-end, for Sun).
So is pentium and opteron. Can Sun really make a Niagara system cheap and fast enough to compete with them?
How much floating point do YOUR servers do?;)
Not much, but a few instructions deep in a function somewhere for calculating statistics could cause a major disruption on a Niagara processor as the thread containing the FP code gets migrated over to the one core out of 8 that contains the FP unit.
I suspect that Niagara machines will come with software floating-point emulation in Solaris to cope with such a scenario (so the thread migration doesn't need to happen).
Niagara 2 will have proper floating-point, so it goes.
My point is, that even a small piece of FP code will completely ruin the performance of a Niagara processor.
I don't doubt that for server processors, highly multi-threaded CPUs with low-latency context switches (e.g. 4 thread contexts with zero cycle switch latency per core) are the future.
My point is that Niagara is a bit too simplistic and despite it's radical new architecture, it will merely have comparable performance to the more traditional competition.
There will be no compelling reason to buy a Niagara 1 system over, say, a 4-core Opteron.
Niagara 2 allegedly will be much better according to Sun's marketing hype, but can they deliver it in time? Can they afford to prop up their dreadful CPU division financially until then?
Then I must be awfully clever, because most of the software I've written the past decade can easily be partitioned to run on multiple CPUs concurrently and benefit from it. In fact, most of it already does.
Compared to most people who call themselves programmers, you probably are.
Judging Sun on its processor track record of the last decade, the follow-on chips will be a year to two years late, under clockspeed and have over-all performance barely comparable with that of the more conventional competition.
Not that I'm cynical or anything.
I'm sure intel will be able to cobble together something with 4 pentium-m cores in it to compete, and AMD will have 4-core Opterons. And, as I said in my previous post, they'll be better suited to running general workloads, and will cope equally well with the multi-threaded ones.
Sun just doesn't seem to get it. Not everyone wants to run Solaris and Java.
Ask Sun if they will provide the IP necessary to get Linux ported to Niagara and Niagara 2.
The problem has been the cost of software development. It's almost always cheaper to throw more hardware at a problem than invest in cleverer code. Highly parallel designs require very clever code. The Pentoum 4 debacle has finally shown that we're now at the stage where we're going to have to bite the bullet at develop that cleverer code. With ubiquitous high-level laguages running on virtual machines (e.g. Java) this is becoming more feasable since a lot of the gory details and dangers can be hidden from the average programmer.
Today I have diarhea in the guts as well as the mind. I should have previewed that before I posted it.
On a good day, with a following wind, Niagara might be able to do 8 integer instructions per second,I meant per clock cycle, of course, not per second.
The thing is, all of the other CPU vendors with have
I am totally not privy to clock-rate numbers, but I see that Paul Murphy is claiming over on ZDNet that it runs at 1.4GHz.
Whatever the clock rate, multiply it by eight and it's pretty obvious that this puppy is going to be able to pump through a whole lot of instructions in aggregate.
Ho hum.
On a good day, with a following wind, Niagara might be able to do 8 integer instructions per second, provided it has 8 independent threads not blocking on I/O to execute.
It only has one floating-point execution unit attached to one of those 8 cores, so if you have a thread that needs to do some FP, it has to make its way over to that core and then has to be scheduled to be executed, and then it can only do one floating-point instruction.
Superb.
The thing is, all of the other CPU vendors with have super-scalar, out-of-order 2- and 4- core 64- bit processors running at over twice to three times the clock frequency.
Sun tried to become a software company and if it weren't for Fujitsu making modern sparcs, sparc would be dead.
This wasn't due to Sun planning to go to itanium, in fact, Sun saw the itanium for the turkey it is. Sun and intel fell out because Sun refused to give up UltraSPARC for itanium, hence no official itanium port of Solaris.
No, the UltraSPARC problems are becasue Sun's processor division sucks. They consistently deliver late (sometimes by several years), slower than planned and often with out-of-date technology.
Project Millennium was supposed to be UltraSPARC V, and should have come out in y2k. They killed the development in 2004, when it was still a year away from testing.
Sun is quite right to be discontinuing UltraSPARC development and buying in much faster SPARC64 CPUs from Fujitsu. Sun has scaled down its CPU operation and is concentrating on even smaller niche markets with Niagara and ROCK.
I expect Niagara will be a damp squib when launched, and development of ROCK will be terminated, with Sun's remaining CPU engineers getting their papers. Sun's entire line will probably eventually transition to Opteron with software emulation for legacy SPARC binaries, as it becomes difficult to justify expensive niche SPARC64 processors from Fujitsu. After all, it's only a matter of time until they become too expensive too, as Sun's competitors come out with 16-, 32- and 64-way Opteron machines.
Slashdot Mirror
I know it's heresy to say this here, but Solaris 10 is a far better server OS than Linux, and even more so on Opteron. Solaris 10 is also a better workstation OS on Opteron than Linux.
Linux will catch up to Solaris 10 in the next 2 years as 64-bit PeeCees become mainstream, but just now you can't beat S10 on Opteron.
Yes, I'm biased, I worked on S10.
Linux got me to where I am today, along with Solaris. Linux is still very immature in many respects. SGI and IBM have been paying it lip-service over the years, and tossing it the odd bone, but Linux has a long way to go.
I don't doubt that it will, and it will kill Solaris, like it killed Irix, AIX, HP/UX and the others. It's only a matter of time. However, right now, Solaris 10 is far superior technically.
The other big mistake SGI made was when they took on Rocket Rick Belluzo (sp?) and he gave them a "Windows NT" strategy. In otherwords, a 10-year step backwards, and an attempt to sell over-priced 32-bit Pentium machines running Windows, where previously SGI had been selling 64-bit MIPS machines running UNIX.
When the pointy-hairs get in charge of a company, it spells death.
Sun will not be long for this world either. It is barely breaking even. Yes, Solaris 10 is superb and so are their Opteron servers and workstations, but the pointy-hairs are grinding the company down internally. The engineers are not longer free to innovate and work on the important stuff. They are given a constant diet of wild goose-chase projects which are ill-conceived and often cancelled upon completion, only to be give more with impossible deadlines and little, if any, thanks let alone financial reward.
Sun will only hover around break-even by continually making more and more staff redundant to "cut costs."
Sun can't market itself or its products to save itself (just look at it). The pointy-hairs keep changing company direction every three months. The engineers are over-worked, under-appreciated, under-rewarded and their opinions are not valued.
Sun PHBs make ill-judged and groundless attacks on Free Software and Open Source almost monthly, they did a deal with Microsoft, they continue to deride Linux where it could have been a great benefit to them and their customers, they can't develop processors for toffee (look at how slow UltraSPARC is, and how expensive).
Luckily Sun didn't do itanic, that's why it's not dead yet. Luckily they decided to go Opteron. Unluckily they left it a bit too late.
Sun should Open Source Java (purely for the good publicity), make 24-, 48- and 72-way Opteron servers and write a software UltraSPARC emulator to run legacy SPARC code. Scott should fire Schwartz and Weinberg. Oh, and they should cease and desist all further UltraSPARC development. It's a complete waste of money. Just use Opterons. They're cheap and very fast and software emulator technology is good nowadays (and I thought that "everything is written in Java" too).
This was the company that set most of the UNIX standards over the last 20 years and has given away more Open Source code than any other (including IBM, SGI, Red Hat,...)....
You UNIONISED COMMIE SCUMBAGS. It's people like YOU who are causing POORNESS and shortages of OIL.
You are probably all ATHIESTS and worship the DEVIL and quite possibly Allah.
Next thing we know, you'll all be strapping dynamite to your bellies and blowing up INNOCENT AMERICAN CHRISTIANS who are not UNIONISED.
As a European, I'm more worried at the effect this will have on the world oil market and global warming. Running a car here in the UK is expensive enough due to high oil prices without the added demand from bugs with SUVs.
As is plainly evindent from the quality of its software.
/me ducks.
Solaris x86 got ressurected to reverse declining sales of low-end servers, and when Opteron came along, S10 went 64-bit on x86, and got support for commodity hardware (NICs, graphics cards, laptop chipsets).
What with Solaris x86 running better than Linux on the desktop (and laptop), and in 64 bits, there ceased to be any need to sustain JDS on Linux.
A kernel is a kernel. It's cheaper to support and develop on one (Solaris) than on two (Solaris and Linux).
Sun's PHBs have been making some really bad decisions recently. Earlier this year, they RIF'd the Janus engineering team (in-kernel Linux emulation for Solaris) before it had been put into Solaris proper. They said it would be in Solaris by about now. Then they said maybe in a year's time. Now they are saying not at all and that Xen virtualisation would be a better solution... So much for Janus and Zones^H^H^H^H^HContainers.
And what about ZFS? The Zettabyte File System that was going to be lightyears ahead of the competition. Where is that? That was supposed to be in Solaris 10 as well.
And then they bought StorageTek... and then they RIF'd yet more people.
Did Bill Gates pay Scott McNeally to kill Sun?
Arse! Maybe a terrorist will turn that there traffic island into an anthrax island and I'll have been right all along :-)
It was left contamintated for decades, until the 1990s IIRC. Allegedly it's all safe now, although they can't garantee that there isn't still some anthrax in the soil, but they put some more sheep on it and they all survived.
My dad was brought up a crofter in North West Scotland. He inherited an enthusiasm for the sea for his father and so on.
In the late '60s he went to university to study Marine Engineering (thanks to British state-funded Higher Education) and did an MSc after his BSc.
The MSc research involved developing some very early CAD and CAM systems for ship building (24-bit mainframes and CNC milling machines in the early '70s).
I was going on boats since I was born, and have never been sea sick. In fact, on my way to Holland on the SeaCat one day I didn't even spill my pint in the rough weather while all the scurvy land-lubbers were rolling around in puke in the bilges.
The computer stuff stuck. I got my first computer when I was eight, and here I sit this evening. having been frobbing with bootloader code all day on an embedded Linux system, writing my own language interpreter.
I hear George W. takes one with him wherever he goes. Handy for writing speeches.
Don't try to say that with false teeth.
I can't wait to get my legal Slackware images 20-30% faster :-)
And the English are all POOFS!
Oh ye of little faith.
So is pentium and opteron. Can Sun really make a Niagara system cheap and fast enough to compete with them?
How much floating point do YOUR servers do? ;)
Not much, but a few instructions deep in a function somewhere for calculating statistics could cause a major disruption on a Niagara processor as the thread containing the FP code gets migrated over to the one core out of 8 that contains the FP unit.
I suspect that Niagara machines will come with software floating-point emulation in Solaris to cope with such a scenario (so the thread migration doesn't need to happen).
Niagara 2 will have proper floating-point, so it goes.
My point is, that even a small piece of FP code will completely ruin the performance of a Niagara processor.
I don't doubt that for server processors, highly multi-threaded CPUs with low-latency context switches (e.g. 4 thread contexts with zero cycle switch latency per core) are the future.
My point is that Niagara is a bit too simplistic and despite it's radical new architecture, it will merely have comparable performance to the more traditional competition.
There will be no compelling reason to buy a Niagara 1 system over, say, a 4-core Opteron.
Niagara 2 allegedly will be much better according to Sun's marketing hype, but can they deliver it in time? Can they afford to prop up their dreadful CPU division financially until then?
Compared to most people who call themselves programmers, you probably are.
Judging Sun on its processor track record of the last decade, the follow-on chips will be a year to two years late, under clockspeed and have over-all performance barely comparable with that of the more conventional competition.
Not that I'm cynical or anything.
I'm sure intel will be able to cobble together something with 4 pentium-m cores in it to compete, and AMD will have 4-core Opterons. And, as I said in my previous post, they'll be better suited to running general workloads, and will cope equally well with the multi-threaded ones.
Sun just doesn't seem to get it. Not everyone wants to run Solaris and Java.
Ask Sun if they will provide the IP necessary to get Linux ported to Niagara and Niagara 2.
You're right. I'm full of shit.
Yes, and I corrected myself straight away in another post. In true slashdot style, the post where I corrected myself got modded down to Offtopic.
The problem has been the cost of software development. It's almost always cheaper to throw more hardware at a problem than invest in cleverer code. Highly parallel designs require very clever code. The Pentoum 4 debacle has finally shown that we're now at the stage where we're going to have to bite the bullet at develop that cleverer code. With ubiquitous high-level laguages running on virtual machines (e.g. Java) this is becoming more feasable since a lot of the gory details and dangers can be hidden from the average programmer.
On a good day, with a following wind, Niagara might be able to do 8 integer instructions per second, I meant per clock cycle, of course, not per second.
The thing is, all of the other CPU vendors with have
I meant "will have" not "with have".
Whatever the clock rate, multiply it by eight and it's pretty obvious that this puppy is going to be able to pump through a whole lot of instructions in aggregate.
Ho hum.
On a good day, with a following wind, Niagara might be able to do 8 integer instructions per second, provided it has 8 independent threads not blocking on I/O to execute.
It only has one floating-point execution unit attached to one of those 8 cores, so if you have a thread that needs to do some FP, it has to make its way over to that core and then has to be scheduled to be executed, and then it can only do one floating-point instruction.
Superb.
The thing is, all of the other CPU vendors with have super-scalar, out-of-order 2- and 4- core 64- bit processors running at over twice to three times the clock frequency.
You do the mathematics.
This wasn't due to Sun planning to go to itanium, in fact, Sun saw the itanium for the turkey it is. Sun and intel fell out because Sun refused to give up UltraSPARC for itanium, hence no official itanium port of Solaris.
No, the UltraSPARC problems are becasue Sun's processor division sucks. They consistently deliver late (sometimes by several years), slower than planned and often with out-of-date technology.
Project Millennium was supposed to be UltraSPARC V, and should have come out in y2k. They killed the development in 2004, when it was still a year away from testing.
Sun is quite right to be discontinuing UltraSPARC development and buying in much faster SPARC64 CPUs from Fujitsu. Sun has scaled down its CPU operation and is concentrating on even smaller niche markets with Niagara and ROCK.
I expect Niagara will be a damp squib when launched, and development of ROCK will be terminated, with Sun's remaining CPU engineers getting their papers. Sun's entire line will probably eventually transition to Opteron with software emulation for legacy SPARC binaries, as it becomes difficult to justify expensive niche SPARC64 processors from Fujitsu. After all, it's only a matter of time until they become too expensive too, as Sun's competitors come out with 16-, 32- and 64-way Opteron machines.
Well, Sun certainly doesn't.
/me ducks.