[morlock_man]

Looked up QNX patents and found and interested on that was approved in 2006.

New Patent Recognizes QNX Innovation in Multi-Core Software Technology

�QNX pioneered the use of SMP for embedded systems and this patent confirms our undisputed leadership in multi-core software technology,� said Dan Dodge, CEO, QNX Software Systems. �Our patented approach to SMP not only gives QNX customers a distinct performance advantage, but also underscores the unique capabilities of the QNX Neutrino RTOS. As the industry�s only true microkernel, QNX Neutrino allowed us to implement an innovative and highly efficient SMP implementation that overcomes the drawbacks of traditional approaches and realizes the true potential of multi-core hardware.�

Traditionally, SMP has been implemented in large, complex, monolithic operating systems. These OSs require numerous modifications to support SMP, resulting in bloated code and suboptimal performance. In comparison, the QNX Neutrino microkernel is small and preemptible, allowing the patented SMP technology to be implemented in just a few kilobytes of code. The result is remarkably fast, lean, and reliable SMP.

Can't help but wonder if this plays a major part in the general hating on of the PlayBook. QNX seems to be, hands down the best OS for stable and efficent multi-core support and RIM is the only company with a tablet making use of this advantage.

For companies like NetFlix to insist that QNX on the PlayBook is a dying platform seems to ignore the facts. QNX has major advantages that are undeniable and are completely unique to the tablet world right now, especially with the growth of multi-processor tablets.

[CrackedBarry]

Looked up QNX patents and found and interested on that was approved in 2006.

New Patent Recognizes QNX Innovation in Multi-Core Software Technology



Can't help but wonder if this plays a major part in the general hating on of the PlayBook. QNX seems to be, hands down the best OS for stable and efficent multi-core support and RIM is the only company with a tablet making use of this advantage.

For companies like NetFlix to insist that QNX on the PlayBook is a dying platform seems to ignore the facts. QNX has major advantages that are undeniable and are completely unique to the tablet world right now, especially with the growth of multi-processor tablets.

Very doubtful. You forget that ANY company could have bought QNX if their technology and micro kernel was really that desirable. They didnt.

Google bought Android, and Microsoft and Apple both chose to develop something based on their existing technologies and assets, even though both companies have never known to be shy when it comes to buying their way to an advanced or tempting technology.
Nokia, Palm, Samsung etc. etc. etc. also all passed on buying QNX.

That would indicate that nobody in the industry believes that QNX is particularily competitive or advanced compared to their own offerings.

If they didnt want to acquire QNX it seems unlikely that they regard it highly enough to try to combat or kill it...

[Dapper37]

^Ya, we believe you
That's the thing about value, it's not there until someone unlocks it!
OP don't worry about crackedbarry He's here to kill anyone's enthusiasm that comes to the forum with positive news. His job I suspect

[morlock_man]

Very doubtful. You forget that ANY company could have bought QNX if their technology and micro kernel was really that desirable. They didnt.

Google bought Android, and Microsoft and Apple both chose to develop something based on their existing technologies and assets, even though both companies have never known to be shy when it comes to buying their way to an advanced or tempting technology.
Nokia, Palm, Samsung etc. etc. etc. also all passed on buying QNX.

That would indicate that nobody in the industry believes that QNX is particularily competitive or advanced compared to their own offerings.

If they didnt want to acquire QNX it seems unlikely that they regard it highly enough to try to combat or kill it...

That sounds plausible, except for two problems:

1. Unless I'm mistaken, multicore tablets didn't appear until 2010, the year RIM purchased QNX.

2. Microsoft, Apple and Google already had their own monolithic operating systems developed by 2010, so switching their own platforms to a QNX-based OS wouldn't be an easy task. (Google bought Android in 2005, the year before QNX recieved their patent for microkernel SMP.)

[Vindicators]

Looked up QNX patents and found and interested on that was approved in 2006.

New Patent Recognizes QNX Innovation in Multi-Core Software Technology



Can't help but wonder if this plays a major part in the general hating on of the PlayBook. QNX seems to be, hands down the best OS for stable and efficent multi-core support and RIM is the only company with a tablet making use of this advantage.

For companies like NetFlix to insist that QNX on the PlayBook is a dying platform seems to ignore the facts. QNX has major advantages that are undeniable and are completely unique to the tablet world right now, especially with the growth of multi-processor tablets.

lol, "ignore the facts", facts from QNX official site, in a PR article? That is like saying iOS is the world's most advanced mobile OS, based on facts from Apple - iOS 5 - 200+ new features for iPad, iPhone, and iPod touch.

The fact is, that article from 2006, 4 years before RIM acquire QNX. Meanwhile, for a decade supercomputers still run some flavor of Unix/Linux monolithic kernel and Apple already implemented GCD to iOS4 before RIM acquire QNX.

Sorry but it seem like the world have very little intersting for QNX.

[morlock_man]

lol, "ignore the facts", facts from QNX official site, in a PR article? That is like saying iOS is the world's most advanced mobile OS, based on facts from Apple - iOS 5 - 200+ new features for iPad, iPhone, and iPod touch.

The fact is, that article from 2006, 4 years before RIM acquire QNX. Meanwhile, for a decade supercomputers still run some flavor of Unix/Linux monolithic kernel and Apple already implemented GCD to iOS4 before RIM acquire QNX.

Sorry but it seem like the world have very little intersting for QNX.

The PR article references the key patent, which is why I used it.

You can't compare a monolithic kernel used on a supercomputer to a microkernel used in a mobile device, thats comparing apples to oranges. Supercomputers have so much extra memory and processing power there's no concern about overall efficency, only brute force, which is not the paradigm that applies to mobile computing.

As the mobile computing industry scales up to quad core and beyond, there will be an increasing demand for properly implemented SMP. QNX happens to be the only company that got it right and patented it.

Damn smart Canucks.

QNX system runs 15 years without a single failure.

QNX: the first multicore-ready RTOS

(PlayBook != Avro Arrow)

[Vindicators]

Yes, its patented but I dont see any info prove that undeniable advantage, except from QNX's own words.

About your links, one said QNX is the first multicore-ready RTOS, it dont prove anything except that QNX kernel is the first RTOS ready for multicore system. But XNU/Darwin and Linux kernel already run on multicore system for a decade and already run on multicore mobile system for years.

The second link said QNX runs 15 years without a single failure. Yet its run on a embedded system, usually with only single program, no GUI, no flashy animation, no services, etc.

The thing is, QNX, Linux, Darwin in PlaybookOS, Android, iOS are merely just a kernel and everything about "undeniable advantage" you said just from a PR article on QNX official site.
iOS/Android already proved its ability to utilize multicore SoC, based on benchmark, browser javascript performance, OpenGL performance.

Remember tablet, smartphone on the market today are geared towards general purpose computing with the ability to handle multiprocess, load of services, multi-purpose app... And this is not the area where RTOS and microkernel really shine.

[morlock_man]

Yes, its patented but I dont see any info prove that undeniable advantage, except from QNX's own words.

I'll repeat myself for clarity.

Traditionally, SMP has been implemented in large, complex, monolithic operating systems. These OSs require numerous modifications to support SMP, resulting in bloated code and suboptimal performance. In comparison, the QNX Neutrino microkernel is small and preemptible, allowing the patented SMP technology to be implemented in just a few kilobytes of code. The result is remarkably fast, lean, and reliable SMP.

That would be the undeniable advantage in my eyes.

How many lines of code did GDC add to Apple's OS? Hundreds of kilobytes. Doesn't sound like a big difference, but in terms of coding complexity, its enormous.

[Vindicators]

again, its what QNX guys said, not the facts.

Theoretically, smaller code is easier maintain, it have nothing to do with software development for multicore.

The rest are PR words, where are the fast, lean, reliable, "hands down the best OS for stable and efficent multi-core support" you are talking about, any info, research, result to prove it?

How about some "real world" benchmarks from a creditable source?
AnandTech - The BlackBerry PlayBook Review

[Majestic Lion]

Very doubtful. You forget that ANY company could have bought QNX if their technology and micro kernel was really that desirable. They didnt.

Google bought Android, and Microsoft and Apple both chose to develop something based on their existing technologies and assets, even though both companies have never known to be shy when it comes to buying their way to an advanced or tempting technology.
Nokia, Palm, Samsung etc. etc. etc. also all passed on buying QNX.

That would indicate that nobody in the industry believes that QNX is particularily competitive or advanced compared to their own offerings.

If they didnt want to acquire QNX it seems unlikely that they regard it highly enough to try to combat or kill it...

This assumes that the value was recognized for what it was and evaluated to not be worth buying. That's a big assumption here. Getting there first matters a lot, and if RIM did exactly that then there's only two real countermoves: one, you innovate something better, or two...you discredit it, cheapen it, make it anathema to anyone who would even potentially find any value in it. Then you wait until you can buy it cheaply and maybe quietly, and you've won.

True innovation takes longer to develop, doesn't work according to deadlines and can be expensive. I'll let folks guess which method is being employed now.

[Canuck671]

Ok, what about the thermal core dynamics of the intercooled bicellular hexaquad difibulating exctrogramite that is used to process the gigaquad memory overlay in the central processing hyperthreads?

Hey - ever think of that?

Didnt think so. I believe i have made my point.

[houshinto#IM]

again, its what QNX guys said, not the facts.

Theoretically, smaller code is easier maintain, it have nothing to do with software development for multicore.

The rest are PR words, where are the fast, lean, reliable, "hands down the best OS for stable and efficent multi-core support" you are talking about, any info, research, result to prove it?

How about some "real world" benchmarks from a creditable source?
AnandTech - The BlackBerry PlayBook Review

Theoretically? No. It's a proven fact, and a programmers golden rule that bigger code, unless it is to add more functionality, is always frowned upon. Frankly, big unnecessary or convoluted code is bad programming for any software development.

A prime example is how adobe just kept stacking more and more code onto their reader and it turned into a slow, bloated, and unreliable piece of turd. It's why a freeware pdf-viewer a fraction of the size in code can outperform.

Efficiency matters, and good code that can outperform previous ones are not easy to come by. Android, IOS, Windows OS's, etc. are good OS for different reasons, in their own right, but If QNX does have any validity in it's claims of bein the MOST efficient then I'm sure we'll see the results of that soon enough with later high multi-core devices.

And for the record that benchmark is hardly a credible source to make a judgement on QNX. It does, however, give insight on how RIM had to improve the integration of their hardware+software, via QNX.

With the arrival of OS2 I think they've made improvements in features and code to make
it necessary for a new test/benchmark be done.

It may not matter to you, but then again I'm pretty sure you're comfortable with parroting "Playbook sucks no matter what!"

The rest of us rational folk are open to see beyond the "i-hype" and RIM-slagging.

[CrackedBarry]

That sounds plausible, except for two problems:

1. Unless I'm mistaken, multicore tablets didn't appear until 2010, the year RIM purchased QNX.

Aaand? What's your point? That everybody was surprised by that development? Of course they weren't. Dual core mobile class CPUs had been shipping to vendors as samples in 2009, and we're road mapped in 2008. Everybody in the industry knew that this was coming as far back as 2005. Google, Microsoft, Apple et al. all knew this, but weren't interested in buying QNX, so it's fair to assume that QNX doesn't in itself offer an advantage thats somehow exclusive to QNX. Either that, or they believed that the disadvantages of using a micro kernel OS like QNX outweighs the advantages.

We've come along way since the 80ies where some computer scientists believed that micro kernel OS offers an advantage over monolithic kernels. Today we know it isn't really the case, and computer technology has come a long way since QNX was developed 30 years ago, though this is kinda OT...

2. Microsoft, Apple and Google already had their own monolithic operating systems developed by 2010, so switching their own platforms to a QNX-based OS wouldn't be an easy task. (Google bought Android in 2005, the year before QNX recieved their patent for microkernel SMP.)

And this is also beside the point. Google knew that it was only a matter of time before multicore CPUs became feasible and cheap enough on mobile devices. They could have easily bought QNX over Android, but chose to go with Android. Either they didn't see QNX as holding any advantage over another OS, or they didn't believe that the trade off was worth it.

Same goes for Apple. They had a couple of candidates for a mobile OS, one was iOS, the other was a smaller OS more closely based on Linux. They could also have bought QNX, but chose not to. And Apple is never afraid of buying a competitive technology they think might offer them an advantage.

Palm was looking for an OS for a while, but chose to develop WebOS instead of adapting QNX. Microsoft threw out what they had, and started pretty much from scratch.

Bottom line is... Anyone could have bought QNX, none of them did, even though they knew it was only a matter of time before multi core CPUs found a place in the mobile space.

Therefore it's highly unlikely that they find QNX to be a danger now, nevermind that there isn't even a shred of proof concerning some sort of alliance of evil directed at the Playook/QNX...

[sf49ers]

Very doubtful. You forget that ANY company could have bought QNX if their technology and micro kernel was really that desirable. They didnt.

Google bought Android, and Microsoft and Apple both chose to develop something based on their existing technologies and assets, even though both companies have never known to be shy when it comes to buying their way to an advanced or tempting technology.
Nokia, Palm, Samsung etc. etc. etc. also all passed on buying QNX.

That would indicate that nobody in the industry believes that QNX is particularily competitive or advanced compared to their own offerings.

If they didnt want to acquire QNX it seems unlikely that they regard it highly enough to try to combat or kill it...

Google didn't buy Facebook or Yahoo didn't buy Google ..does that mean facebook and google are worthless? QNX is such a beast and you don't need Google or Microsoft to approve the fact. Two reasons why nobody thought about QNX...either it was not up for sale or people didn't realize it's true potential to make use of it in what they wanted. Google bought Android because it wanted a mobile OS and the people who has expertise in mobile communications and if not for that reason they would have bought QNX or something else. On the Contrary RIM is a mobile OS company and wanted a secure kernel with a small foot print and they have the expertise to transform it into a full fledged mobile OS.

[NFLPLAYBOOK]

Seeing that you people are far more educated in this area. Is the XNU/Darwin and Linux kernel written to directly or on top of it? I've heard that the advantage of QNX is the the code is written of top of the kernel making for a more stable and easier correctable OS.

[CrackedBarry]

yes, if that was case Google should have bought Facebook or Yahoo should have bought Google instead of developing their own..right?

What does Facebook and Yahoo have to do with Moores Law and multicore CPUs?!?

Two reasons why nobody thought about QNX...either it was not up for sale or people didn't realize it's true potential to make use of it in what they wanted.

Aha... So you're assuming they weren't for sale, you couldnt really be bothered to check, but you're sorta guessing it wasn't. (Protip: It's not like you put a company like QNX up for sale on Kijiji or EBay. But you can always buy it regardless, especially as it was owned by Harman for a while)

You're also assuming here that an entire industry of very smart people, just sorta whoops, missed out on the potential of a widely known, 30 year old operating system, until RIM just happened to stumble over it in 2010.

Both assumptions are very feeble.

Google bought Android because it wanted a mobile OS and the people who has expertise in mobile communications.

You should really at least TRY to look up the stuff you're writing about instead of just making assumptions. Android inc. was a small shop, building an OS with little experience in "mobile communications". And that's Google. What about the rest of the entire mobile industry who at some point needed an OS? How come they overlooked QNX if it has such awesome built in advantages?

The whole question is based on the assumption that an entire industry filled with very smart people just happened to completely miss out on the fact that QNX has some amazing, mythical advantage. That assumption isn't very credible.

(I'm not saying that QNX is somehow worse than other OS. It's an operating system like any other, with some advantages over other competing OS, and some disadvantages. Both of which mean little in the end. Just ask Palm...)

[Vindicators]

Theoretically? No. It's a proven fact, and a programmers golden rule that bigger code, unless it is to add more functionality, is always frowned upon. Frankly, big unnecessary or convoluted code is bad programming for any software development.

A prime example is how adobe just kept stacking more and more code onto their reader and it turned into a slow, bloated, and unreliable piece of turd. It's why a freeware pdf-viewer a fraction of the size in code can outperform.

Efficiency matters, and good code that can outperform previous ones are not easy to come by. Android, IOS, Windows OS's, etc. are good OS for different reasons, in their own right, but If QNX does have any validity in it's claims of bein the MOST efficient then I'm sure we'll see the results of that soon enough with later high multi-core devices.

And for the record that benchmark is hardly a credible source to make a judgement on QNX. It does, however, give insight on how RIM had to improve the integration of their hardware+software, via QNX.

With the arrival of OS2 I think they've made improvements in features and code to make
it necessary for a new test/benchmark be done.

It may not matter to you, but then again I'm pretty sure you're comfortable with parroting "Playbook sucks no matter what!"

The rest of us rational folk are open to see beyond the "i-hype" and RIM-slagging.

Im just talking about the effect of kernel and SMP stack size to whole OS performance and app development. I dont even talk about the software framework implement on top of kernel which is I believe have more effect to device performance than the kernel itself.

The benchmark link isn't to make judgement on QNX, it's to point out device performance depend on many things. A pr article about the kernel patent in 2006 hardly is a fact, let alone to prove it is a undeniable major advantage for the Playbook. Even more ridiculous to based on that to come with a conclusion other tech companies scared of QNX.

[DaedalusIcarusHelios]

@CrackedBarry: So what you are really saying is that all of RIM's acquisitions must be worthless because it was RIM, not someone else, that bought them? Come on. Mike Lazaridis is a smart guy, and he understood the technical merits of QNX and connected the dots of how it could give RIM a technical advantage on a new platform they needed. It doesn't matter what reasons other potential suitors had for not acquiring them. It is entirely possible QNX wasn't on their radar, for whatever reason, or they weren't looking to buy. RIM had a need and wanted to go big. Looking at their acquisitions, it clearly shows a lot of ambition but also shows they didn't believe they could do it alone.

[morlock_man]

Im just talking about the effect of kernel and SMP stack size to whole OS performance and app development.

I've rerun those benchmarks myself and the scores show much improvement. It's beating the iPad2 now in the Sunspider 0.9.1 tests.

QNX has SMP within the microkernel, they have since 1997. The patent was only awarded in 2006. Apple's implementation of SMP is patched on and the code for it is about the same size as QNX's entire microkernel.

[Tre Lawrence]

So... is the theory that RIM is hated because of the patents it holds?

[Vindicators]

I've rerun those benchmarks myself and the scores show much improvement. It's beating the iPad2 now in the Sunspider 0.9.1 tests.

QNX has SMP within the microkernel, they have since 1997. The patent was only awarded in 2006. Apple's implementation of SMP is patched on and the code for it is about the same size as QNX's entire microkernel.

You just proved my point, software performance depend on many things, not only the kernel.
And monolithic kernel (usually have a bigger size than microkernel) can have a much better performance than microkernel. Kernel size do not necessarily translate to performance.

[morlock_man]

You just proved my point, software performance depend on many things, not only the kernel.
And monolithic kernel (usually have a bigger size than microkernel) can have a much better performance than microkernel. Kernel size do not necessarily translate to performance.

Unless, you know, you're talking about scalability and a robust end-user experience.

[app_Developer]

Apple's implementation of SMP is patched on and the code for it is about the same size as QNX's entire microkernel.

I worked on Hurd and then Mach back then. NeXT had Mach running SMP in production by 1988. What part do you say is "patched on"?

Also, what score are you getting on sunspider now?

[morlock_man]

I worked on Hurd and then Mach back then. NeXT had Mach running SMP in production by 1988. What part do you say is "patched on"?

Also, what score are you getting on sunspider now?

You worked on the early versions of Mach? So what would you give as a reason why their early implementations of SMP sucked so much? 1988 should be around the right time.

Just ran the test again three times. Got 2166ms, 2182ms and 2176ms. Average of 2174ms. Thats about 150ms faster than their launch score and about 60ms faster than the iPad 2's results for the same time.

[app_Developer]

You worked on the early versions of Mach? So what would you give as a reason why their early implementations of SMP sucked so much? 1988 should be around the right time.

"sucked" compared to what? Obviously the state of the art has progressed quite a bit in 25 years.

But again, the question was what part of this is "patched on". SMP has been part of Mach since very early on. I believe they were doing this at CMU even before we got it at NeXT. That was part of the reason why it was chosen. It was also somewhat of a microkernel at the time, which the team thought was a good idea at the time. But as you know at NeXT and at Apple, it was made more "macro", if you will, to avoid so many context switches.

Don't you think context switches are relatively more expensive now (especially on ARM) than they were back then? I would say they are.

Just ran the test again three times. Got 2166ms, 2182ms and 2176ms. Average of 2174ms. Thats about 150ms faster than their launch score and about 60ms faster than the iPad 2's results for the same time.

That's odd, i just got 2179, 2194ms on my PB, and 1722, 1724ms on my iPad2.

EDIT: My iPhone gets 2233, and 2234. Perhaps you were comparing the PB to the iPhone?