Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

RTX2010 – Radiation-hardened stack machine microprocessor

en.wikipedia.org

Post reply on HN
Print view View original

31–40 of 61 posts

Re: RTX2010 – Radiation-hardened stack machine microprocessor

#31

Post by wlievens »

gus_massa wrote: post #20
wlievens wrote: post #15

Earlier quoted context omitted.

Wouldn't work for all applications, tough, if I'm not mistaken, as you specifically need Silicon for imaging chips (CCD/CMOS). Or is that nonsense?

Theoretically you can make integrated circuit with any semiconductor, but I don't know if anyone has been brave enough to try it. The smallest integrated circuit is a diode, it has 1/2 transistors :). OK, a diode is not actually an integrated circuit, but if you can use a material to make a diode, you probably can use the same material to make an integrated circuit with enough money, time and ingenuity. Some of the d…

I know that, but I don't think every material has the interesting properties that Si does for image sensors, i.e. it needs to basically turn photons of the right wavelength range into electrons.

Re: RTX2010 – Radiation-hardened stack machine microprocessor

#32

Post by wlievens »

Sanddancer wrote: post #27
wlievens wrote: post #15

Earlier quoted context omitted.

Wouldn't work for all applications, tough, if I'm not mistaken, as you specifically need Silicon for imaging chips (CCD/CMOS). Or is that nonsense?

Other materials have been used for imagers. Gallium Arsenide in particular has been used for high speed, and high radiation, imaging circuits.

Cool, I didn't know that.

Re: RTX2010 – Radiation-hardened stack machine microprocessor

#33

Post by SixSigma »

Animats wrote: post #30
tinco wrote: post #4

I wonder how hard it is to create a radiation hardened CPU. Is the premise that radiation can flip bits in the on-die cache and registers? would simply making those redundant be a solution? Or could a single flash of radiation invalidate all memory on the device?

If it's rad-hard, it's export-controlled in the US. See the US Munitions List, category XV, section (d), "Radiation-hardened microelectronic circuits". https://www.pmddtc.state.gov/regulations_laws/documents/offi... So mass market products aren't made radiation-hard. Rad-hard CPUs more modern than that FORTH engine exist; Atmel makes a rad-hard SPARC. But they're produced in tiny quantities and are thus very expensiv…

I'm pretty sure the marketing of "Our CPU is used on the probe" is enough to get a donation or two.

Re: RTX2010 – Radiation-hardened stack machine microprocessor

#34

Post by SixSigma »

A very closely related processor, the RTX2000, is described in detail in Koopman's book "Stack Computers: the new wave"

http://users.ece.cmu.edu/~koopman/stack_computers/index.html

The RTX2010 has had its own wikipedia entry too since 2008 : http://en.wikipedia.org/wiki/RTX2010

Re: RTX2010 – Radiation-hardened stack machine microprocessor

#35

Post by CHY872 »

s800 wrote: post #24
wlievens wrote: post #14

Earlier quoted context omitted.

Optimal register allocation is an NP-complete problem. By doing away with it, you don't need to allocate registers. How that compares on a practical level, I have no idea. But it's one of the reasons the JVM has a stack-based model and no registers: it makes writing compilers for JVM byte code a lot easier.

Interesting. I agree with your statement regarding it being an NP-complete problem, as intuitively it seems that way, but do you have a reference by chance? Thanks!

Ok, so we can view register allocation as being a case of graph colouring; create a graph where the nodes are your 'virtual registers' and there's an edge between two nodes if those registers are both live at the same time. Then we can run on n registers iff we can colour the graph with n colours. This is one of the first identified NP-complete problems.

https://en.wikipedia.org/wiki/Graph_coloring

Re: RTX2010 – Radiation-hardened stack machine microprocessor

#36

Post by spiritplumber »

Someone wrote: post #16
wlievens wrote: post #15

Earlier quoted context omitted.

Wouldn't work for all applications, tough, if I'm not mistaken, as you specifically need Silicon for imaging chips (CCD/CMOS). Or is that nonsense?

Why would you need silicon for that? Even if it were the only process we use now (which it isn't. As a simple example, Germanium-based photo diodes do exist), materials research could conceivably produce other materials that work, too. There are even more exotic examples to think of. Eyes turn light into electricity without (AFAIK) Silicon. Maybe part of that physics/chemistry can be practically used elsewhere?

Germanium diodes are still made because they have a much smaller voltage drop than silicon diodes do (0.1V nominal versus 0.7V nominal). This has its uses.

Re: RTX2010 – Radiation-hardened stack machine microprocessor

#37

Post by CHY872 »

wlievens wrote: post #14
programmer_dude wrote: post #13

Can someone point to resources comparing stack machines and register based machines? The RTX2010 technical documentation talks about all the advantages of a stack machine but doesn't say anything about its shortcomings.

Optimal register allocation is an NP-complete problem. By doing away with it, you don't need to allocate registers. How that compares on a practical level, I have no idea. But it's one of the reasons the JVM has a stack-based model and no registers: it makes writing compilers for JVM byte code a lot easier.

This isn't quite right - we don't in general really care whether we're allocating optimally (for example, if we have 16 registers we don't care if we can potentially allocate to 10), and heuristic based approaches are practically very useful (and cheap).

In terms of compiler writing, if you wanted to write any optimisations you'd probably want to manipulate 3-address code internally, if only because the data flow based analyses would be made easier (and then you would emit stack code in the last step).

This might be a reason why the JVM designers decided to use a stack machine, but I doubt it's an important one - they'd easily be able to get ahold of a few decent compiler writers if they needed to.

Re: RTX2010 – Radiation-hardened stack machine microprocessor

#38

Post by ryanmk »

jesrui wrote: post #10

Another microprocessor able to execute Forth natively is the b16, from one of the original authors of Gforth. http://bernd-paysan.de/b16.html

There's also GreenArrays GA144, which has 144 forth machines on a single microprocessor.

http://www.greenarraychips.com/

Re: RTX2010 – Radiation-hardened stack machine microprocessor

#39

Post by wlievens »

CHY872 wrote: post #37
wlievens wrote: post #14

Earlier quoted context omitted.

Optimal register allocation is an NP-complete problem. By doing away with it, you don't need to allocate registers. How that compares on a practical level, I have no idea. But it's one of the reasons the JVM has a stack-based model and no registers: it makes writing compilers for JVM byte code a lot easier.

This isn't quite right - we don't in general really care whether we're allocating optimally (for example, if we have 16 registers we don't care if we can potentially allocate to 10), and heuristic based approaches are practically very useful (and cheap). In terms of compiler writing, if you wanted to write any optimisations you'd probably want to manipulate 3-address code internally, if only because the data flow bas…

I don't doubt the JVM designers have good compiler writers. I rather think that they wanted the bytecode to be accessible to write for other compiler writers, such as the team that implemented JSP servlets a century ago, but also perhaps with a future vision for the alternative JVM languages.

Re: RTX2010 – Radiation-hardened stack machine microprocessor

#40

Post by naland »

naland wrote: post #23
mud_dauber wrote: post #21

How cool - I joined Harris after grad school expressly to work on RTX architecture issues. (Previously I was a Forth developer.) Nice to see this neat-but-underappreciated architecture in the news.

>Nice to see this neat-but-underappreciated architecture in the news. It's already appreciated. Let me recall here the 'scientist' character (A&B Srugatsky) who explains the World as a huge inertial mass to the good or evil in the same. Tertium non datur it is a principle of our mind sometimes used in computers.

edit: Strugatsky. hi ±HN.
Post reply on HN