News AMD targets embedded market with RTOS deal

AMD has signed a deal with Green Hills to add G-Series support to the Integrity RTOS platform.

http://www.bit-tech.net/news/hardware/2012/03/28/amd-rtos-deal/1

 

I think the front page article blurb should read ARM and MIPS feel the squeeze. Not AMD and MIPS...

 

so i don't get it, this is going to be 1/3 of an x86 cpu, 1/3 of an ARM cpu, and 1/3 of a gpu? or, is this basically going to be an ARM cpu on steroids?

 

i would not be surpirsed if AMD do really well in this market considering there Apu technology.

 

When they can make self-checkout machines run quickly and not randomly hang, I'll be impressed...

 

Oooh ok, that makes a lot more sense now. However, I'm not exactly sure what the point of this is. Why do these low-end systems need DX11 or need to be 64 bit? But more importantly, how hard could it possibly be to customize any OS to be a RTOS on any CPU architecture? Maybe for Windows this might not be so easily accessible but most of these systems don't run windows, and if they do, the programs they are running are often java based (meaning they should be easy to port to another platform). To me this seems to be entirely a software issue, so unless there is some tremendous positive performance or power consumption difference by intel and AMD making processors work specifically for a RTOS, I don't see why they can't just customize an OS like linux to act as an RTOS on any standard x86 CPU. Unless I'm completely misinterpreting the point of this, it just seems like a market fad to me.

 

Um, running programs written in a language that requires a VM seems like a bad idea for an RTOS. I would think the way you would want to go would be to get as low-level as possible - C or even Assembly. And market fad? Not familiar with industrial control systems, are you? These are the reason Via still has so many boards with RS232 ports - sometimes multiple ports on one board. You think I need that much back-panel real-estate being eaten up by what you would think are long-dead legacy ports? I don't, but people running industrial machinery often do. (And that's Via's bread and butter - embedded/industrial machines...x86 boards [and some supporting 64-bit, iirc], at that.)

Relating to what ch424 says, yes, exactly. The largest chain in my area uses self-check stations running WinXP (they bluescreen often enough that I've seen it many times) So right there, yes, DX11 isn't strictly needed, nor 64-bit now, but if I can get it without the cost going up because of it? All the better. Once those systems have been ported to a more modern OS, they're ready to roll.

As to Linux, some (maybe even a lot) probably already do. But Linux is not Jesus, Miracle Worker of Galilee. It gets bit tiring, all the evangelists acting as if the solution to all computing's problems is Linux. I would guess quite a number of systems in these markets are running OS's most of us have never heard of, or wouldn't even recognize as an OS. Obscure (to the mass market) systems that nevertheless represent millions of dollars within the markets they serve.

Seriously, though, anyone here know? Could Linux do it?

 

I read Green Hill and thought about playing Sonic at the checkouts.

/thread contribution.

 

Hey, I'm just saying what I've seen. I haven't really seen any of these types of low-end computers for industrial purposes, so I don't really know what they require. I'm not saying that RTOSes are a fad, because I think that is an excellent idea. I think that processors designed specifically for an RTOS is a fad, because it seems to me to be more of a software thing than hardware. Even though I've never seen any RTOS computers myself, I can easily imagine dozens of purposes for them.

Yes, absolutely. But if it does actually affect price then that's where I begin to question it. 3D in general isn't the kind of thing you want on a system like these, except for maybe self-checkout machines and that's pretty much the 1 and only thing. It wouldn't take much for a 3D task to slow down the rest of the system to a possible instability. As for 64 bit, since these computers aren't very good to begin with, the extra performance you would gain from 64 bit is minuscule. If you're looking for more than 4GB of RAM, use PAE.
I'm not against these features at all, but if adding them affects the price then they might want to remove them.

Well, have you ever used linux yourself? I use it as my main OS on all of my computers, I'm making a robot based on it, and it is used in the most obscure purposes from routers to cars to TVs or even microwave ovens. If you can get a computer kernel to operate in devices like these then it would be relatively effortless to make the kernel change the way it processes tasks in order for it to become a RTOS.

The cool thing about intel or AMD joining in on this is if they do a dual-core CPU, they could have each core do a different thing. The first core could be used for OS tasks while the 2nd core could be used for RTOS. Obviously the only problem would be making sure both processes don't clash. Or, in a more likely and practical scenario, the first core could be used strictly for the kernel and any background processes while the 2nd core is for user processes only, which would possibly help improve stability. I have a feeling that many current RTOS program out there are not multi-threaded.

 

It's funny to see how ignorant self-appointed "geeks" can be. RTOS systems are used everywhere - from your car's ABS to NASA Mars landers. They run factories, cell phone BTS stations and ballistic nuclear warheads, yet I see words like "fad" used. When hardware is considered, it's all about reliability. Of course, you can run RTOS on any x86 CPU, but no home or server grade CPU would work reliably in space - that's why NASA still uses specially shielded 386/486 CPUs in probes worth milions. Basically, AMD say that they made a faster CPU that can reliably work in harsh enviroments. If they prove as reliable as they say, it will really make people in embedded systems world happy.

 

Did you not read my response? I said that the CPU itself seems like a fad, not the RTOS. I state this again - RTOSes are practical and a good idea. I asked for someone to prove me wrong about why you can't just use any CPU in an RTOS. So explain to me, how is requesting to be proven wrong being ignorant or a self-appointed geek? Ok, maybe your run-of-the-mill x86 based CPU won't work (well) in a low-gravity vacuum. Well guess what - most RTOS processors will exist on earth for their entire lifespan. Way to make yourself look like an ass.