Any reason to keep going? (for me, not in general)

[Confuzzled]

One of the many nice things about the Psion 5 series was the 'instant on' capability. Open up, press the 'Esc / On/ Off' key, and bam!, ready to use. I often wondered why turning on wasn't possible by just opening the device up. That was achieved by a deep sleep mode ( 3 microAmp power draw in standby) of the Cirrus Logic CL-PS7110 (datasheets availble from web-searches) ARM-710A (ARMv3 architecture) based processor running at 18.432 Mhz

Psion Series 5 CPU Processor

We could do with an updated Cirrus Logic cpu.

That said, as ED has said, the Pyra was not conceived as a Psion replacement, and while many people want to use it in the same way as 'always on' mobile phones, that wasn't in the requirements specification. It's basically a handheld gaming device, which, because of some very intelligent design choices, is capable of doing many other things, but is not optimised for them.

Now, I want a viable Psion 5 series replacement, and the Pyra can probably do that, although I would be wasting its gaming capabilities. A Pinephone with Keyboard might also be a possibility (although reports of significant problems with the keyboard put me off - the Psion 5 series keyboard is a hard act to equal, let alone beat). Similarly a Planet Gemini might be an option. An issue for me is that I'd prefer my pocket computer to not have a build in mobile chipset. If I want networking, then an RJ-45 Ethernet or USB Ethernet would be my choice, with some form of wireless networking (Bluetooth/WiFi) second. Ethernet 'just works', whereas all the wireless technologies have gone obsolete and have severe security issues in surprisingly short lengths of time. For a device that needs to last, that's poor design. At the very least, removable/replaceable (wireless) network devices are needed.

Anyway, I'm not going to complain that a Pyra is not like a Psion. It wasn't designed to be. But it is a lot closer than repurposed mobile phones.

Edit: replace it's with its. Sigh. Typos/brainos

 

[oskda]

By that definition they don't make those any more. All modern OSes protect to some extent against processes poking into others memory and certainly stop your process reading or writing kernel memory. I'm not sure when we switched from 8- and 16-bit machines where the only thing protecting you from a horrible crash was trust, maybe it was when we switched from real to protected mode on 386s and on or when we switched from win ME (final in the 9x line) to the Win NT line.

You are wrong in what you say, but right on other.

The OS you load/install on your computer, for example Window or Linux, is only software and you can install those OS or not, or you can install other without memory protection for process (but in general I don't see reason to do so), for example FreeDOS (I sed MS-DOS on old times but no FreeDOS for myself), or make your own OS without it or run hard directly without OS (for example for Raspberry Pi¹ there are bare metal programming). Even more, you can get Linux and access to all process if you wish or modify it to do so. And those protection are not limitations but a functionality, as you want a process can't mess with other space. You are not constrained by that protection (and you can "disable" or overtake it if you want).

¹: Raspberry Pi is not a good example of a computer where you have real control, as its SOC is a GPU with "auxiliar" ARM cores, and always there is a underlaying OS for GPU running with higher priority (and I suspect it is not casual that model design was selected for such a popular SBC).

You would be right if you look much deeper inside modern computer, there are occult OS and occult hardware on most modern computers. It is invisible to us (even most users don't know they exist). Those hardware+OS has access to all, runs more deep and with higher priority than any OS you install (like Windows or Linux; user OS don't see it). So in most modern hard user don't have real control of his computer although (that control is owned by other beings) he paid those computers with his money. And not only PC-compatible, but smartphones, etc, etc.

Even with that, although user don't have total control over his computer, he can do what he want (except access those occult parts; in that sense only beings with access to that part can do all they want, without owner permission and without he knows about). For example you can run what you want, access all CPU power/instructions/etc (except occult CPU of course).

In a programmable calculator not being a real personal computer you can't escape from its limited environment. For example if that calculator has a BASIC interpreter you can't escape form it: you are limited to what you can do in that BASIC and you can't make or run much more powerful machine language programs. On those BASIC language calculators there are no POKE command (to write to physical memory) or CALL/SYS command to run machine language code at desire. On some BASIC language calculators you have those commands so you CAN ESCAPE from BASIC and access all memory/hardware and run all machine languages you want, so you are not limited by a constrained calculator environment: they are real personal computers (of pocket size).

And this has nothing to do with SIZE or aspect of pocket computer or complexity. I will show two examples:

Casio PB-770 from 1983 is labeled as "PERSONAL COMPUTER" (but it isn't, it is only a programmable calculator). It can be programmed in BASIC language but you can' t escape form it (no SYS/CALL command to escape from BASIC and run machine language program). You can't do all what hardware could do because you are restrained to BASIC software environment and you can't escape from it.
As you can see this calculator has a graphic display with 4 lines of text (continuos 32x160 píxels):

Now let's look at older Sharp PC-1251 (link to Wikipedia) from 1982:

Sharp is much smaller (Size/Weight: Casio PB-700 200 x 88 x 23 mm / 315 g VS Sharp PC-1251 135x70x9.5 mm / 115 g (with batteries), has a smaller display showing only 1 line of text and it is not graphic (each characters is graphic but they aren't continuous, there is a blank space between each character, so you can't draw a continuous graph around characters), BUT it is a REAL PERSONAL COMPUTER (not only a programmable calculator). Its BASIC language has POKE and CALL commands to access memory and create/run machine language programs. So it is not constrained by its BASIC LANGUAGE and you can do all what the hardware lets you to do.

They seem the same, but they aren't: one is a calculator (programmable in BASIC) the other is a PERSONAL COMPUTER.

PD: I hate BASIC, but it was near standard in eighties. Even our beloved Commodore 64/128 and ZX Spectrum run BASIC on start-up (most 8-bit computers did) and it makes like a command line interface to OS on those computers (Forth would be much better, much more powerful and consumes much less memory, but it seems it is "difficult" for normal people).

 

[oskda]

By that definition they don't make those any more. All modern OSes protect to some extent against processes poking into others memory and certainly stop your process reading or writing kernel memory. I'm not sure when we switched from 8- and 16-bit machines where the only thing protecting you from a horrible crash was trust, maybe it was when we switched from real to protected mode on 386s and on or when we switched from win ME (final in the 9x line) to the Win NT line.

Now I see you misunderstood my words or I wasn't clear enough. With «access all memory» on that context I wanted to refer that in BASIC without POKE command (or similar) you can't access/write to memory, you are limited to creating BASIC programs and variables. When you can write to memory as you want, you can do what you want, you can enter machine code and even run machine language code with SYS/CALL command from BASIC, escaping from calculator environment. That is the key difference between a programmable calculator and a real personal computer (being or not of pocket size).

For example in a modern OS with memory protection, even in your process so can access all its memory (memory from that process). It is not a limitation, but a protection not being capable of access memory from other process/program. It doesn't constraint you but it protects.

 

[oskda]

Yeah timed it from the time it hits the boot loader on the serial console to the log in screen is around 18 seconds, just felt like 7-10 seconds in my mind. That said, timed my modern smart phone it took 3.5 minutes to boot to a usable state, so what does boot up time really tell us?

My "modern" smartphone from 2016 (yes, my actual Android smartphone was launched in 2016, and today, in 2022, I don't plan to buy other) takes like 40 seconds to boot so you can use it (cell signal availability takes like 10 more seconds). So yours is really "slow" or it has too much bloatware (it is normal in Android, even more with some brands).

Of course boot time is not very important with actual values if you rely on good-working sleep mode because most time device is sleeping and you awake it instantly from seep mode.

 

[oskda]

One of the many nice things about the Psion 5 series was the 'instant on' capability. Open up, press the 'Esc / On/ Off' key, and bam!, ready to use. I often wondered why turning on wasn't possible by just opening the device up. That was achieved by a deep sleep mode ( 3 microAmp power draw in standby) of the Cirrus Logic CL-PS7110 (datasheets availble from web-searches) ARM-710A (ARMv3 architecture) based processor running at 18.432 Mhz

We could do with an updated Cirrus Logic cpu.

That said, as ED has said, the Pyra was not conceived as a Psion replacement, and while many people want to use it in the same way as 'always on' mobile phones, that wasn't in the requirements specification. It's basically a handheld gaming device, which, because of some very intelligent design choices, is capable of doing many other things, but is not optimised for them.

Now, I want a viable Psion 5 series replacement, and the Pyra can probably do that, although I would be wasting its gaming capabilities. A Pinephone with Keyboard might also be a possibility (although reports of significant problems with the keyboard put me off - the Psion 5 series keyboard is a hard act to equal, let alone beat). Similarly a Planet Gemini might be an option. An issue for me is that I'd prefer my pocket computer to not have a build in mobile chipset. If I want networking, then an RJ-45 Ethernet or USB Ethernet would be my choice, with some form of wireless networking (Bluetooth/WiFi) second. Ethernet 'just works', whereas all the wireless technologies have gone obsolete and have severe security issues in surprisingly short lengths of time. For a device that needs to last, that's poor design. At the very least, removable/replaceable (wireless) network devices are needed.

Anyway, I'm not going to complain that a Pyra is not like a Psion. It wasn't designed to be. But it is a lot closer than repurposed mobile phones.

I think you are wrong, at least from my point of view you are wrong: Pyra and Psion 5 series both are pocket computers. The key difference is that Pyra is much more modern and powerful and it also include gaming controls on its case. In fact those gaming controls are not only good for gaming: having 4 shoulder buttons is a very nice point because you can configure them as modifiers, and it is a great thing because when you use this pocket computer holding them with hands and typing with both thumbs, you can use those 4 modifiers on shoulder buttons. On other side having analog control is nice for moving mouse pointer (on Psion yoy rely on pen with resistive touch input).

Pyra would have good working sleep mode and instant on (from sleep) if OMAP 5 would not have that bug. Not having that sleep mode is not planned but a consequence of OMAP 5 bug as this version used in Pyra was designed by TI for car infotainment and when they discovered bug it was not relevant for that market (that SOC on car is not in sleep mode). When Pyra have other CPU board surely it will get seep mode.

About Psion 5 keyboard I think it is very overrated, I commented it on #5. I own a Psion 5mx (36 Mhz CPU) and keyboard is horrible to use, only good when using when computer is relying on a table (and not perfect because of unstable moving base under keyboard). It is unusable when you hold computer with hands and type with thumbs.

A pocket computer is a type of computer you can carry with you near like a smartphone, and you want to use it in a lot of places and positions where a table is not an option. Screen/text can be so small in same modern machines, that using it like a notebook on a desktop is not optimal.

I haven't tested Pyra keyboard, but I have tested Psion 5mx keyboard (I own a Psion 5mx), and my instinct says to me it will be a much better keyboard than Psion 5.

PD: Planet Gemini is never an option for me: it has no real Linux support. Its SOC is a mobile SOC only supported by Android from manufacturer, so it uses sonetrhing like Linux with Android kernel/drivers, so you are limited in Linux availability and kernel updates. MNT Pocket Reform could be a interested option as it will have a NXP SOC with real Linux support (and near no blobs/etc).

 

[oskda]

One of the many nice things about the Psion 5 series was the 'instant on' capability. Open up, press the 'Esc / On/ Off' key, and bam!, ready to use. I often wondered why turning on wasn't possible by just opening the device up. That was achieved by a deep sleep mode ( 3 microAmp power draw in standby) of the Cirrus Logic CL-PS7110 (datasheets availble from web-searches) ARM-710A (ARMv3 architecture) based processor running at 18.432 Mhz

Psion Series 5 CPU Processor

We could do with an updated Cirrus Logic cpu.

About«an updated Cirrus Logic cpu»: I think this company is out of CPU market from time ago, but you can look for actual CPU/SOC from other manufacturers with lowest consumption. But don't forget CPU consumption is only one of others. For example when in sleep mode you must keep RAM power and I suspect modern 4GB DDR (Dual Channel PC DDR3L) memory could consume more than old 16/32MB low power SDRAM (and those consumes more than old small SRAM, static RAM, memory; On my HP 48SX consumption on sleep mode is 15 µA for complete machine and batteries last for years, even it has a "coma" mode where it halts real time clock and total consumption goes down to 7 µA for complete machine minus real time clock (when running code it consumes no more than 10 mA) but that machine has STATIC RAM, no DRAM or DDR which needs refreshing), even if modern memory chips is manufactured in much higher density. Of course today we can't have 4GB SRAM

I can say batteries on Psion 5, even with that 3 µA CPU in standaby, doesn't last so much like on HP 48, and I think RAM consumptio is one of the reasons. And there are more elements than CPU and memory even on sleep mode.

As a note we would need to talk on W (mW, µW) as amperes does not state energy but only current (it is not the same 1A at 3v than 1A at 30kV).

I have take a look at NXP i.MX 8M CPU/SOC consumption (here you have a 54 pages PDF on "i.MX 8M Mini Power Consumption Measurement") and it states like 9 mW total power in Suspend mode (This mode is called either Dormant mode or Deep sleep mode in the Linux BSP. This is the lowest possible power state where the external supplies are still on) on page 11, On idle default it states 617 mW. On a core mark it states 1786 mW. On 4-core Streamcpy + VPU + Taiji it states 2359 mW. But this SOC includes much more elements than Cirrus Logic on Psion 5 (and 4 main cores, not only one).

Finally you need to know total power consumption of your system on sleep (or other modes) because there is much more than CPU/SOC.

 

[levi]

You are wrong in what you say, but right on other.

The OS you load/install on your computer, for example Window or Linux, is only software and you can install those OS or not, or you can install other without memory protection for process (but in general I don't see reason to do so), for example FreeDOS (I sed MS-DOS on old times but no FreeDOS for myself), or make your own OS without it or run hard directly without OS (for example for Raspberry Pi¹ there are bare metal programming). Even more, you can get Linux and access to all process if you wish or modify it to do so. And those protection are not limitations but a functionality, as you want a process can't mess with other space. You are not constrained by that protection (and you can "disable" or overtake it if you want).

Yes, you're right, notably if you boot from a USB stick, when the BIOS loads there are no memory protections and you can access all of it. Actually that's true even when you normally boot, memory protections only come in finally when the kernel loads, but you need to boot from something other to install your own homegrown random OS that lets you control the computer but doesn't put any memory protections on you. But if I ignore the SBC category nobody mainstream makes a computer without an OS installed already, and unless you boot your own howebrew woowoo you won't have completely unfettered access.

PD: I hate BASIC, but it was near standard in eighties. Even our beloved Commodore 64/128 and ZX Spectrum run BASIC on start-up (most 8-bit computers did) and it makes like a command line interface to OS on those computers (Forth would be much better, much more powerful and consumes much less memory, but it seems it is "difficult" for normal people).

You should have bought a Jupiter Ace back in the day. Those ran forth instead of BASIC, and that's normally credited as the reason that machine died a death back in the day.

 

[Confuzzled]

You should have bought a Jupiter Ace back in the day. Those ran forth instead of BASIC, and that's normally credited as the reason that machine died a death back in the day.

I got a Forth interpreter for my CP/M machine, (was tempted by the Ace at the time) and while I could see the benefits, the pain of having to load it from tape instead of the ROM's built-in BASIC meant I didn't use it as much as I should have. I also had got so used to thinking procedurally (from BASIC) , that the Forth approach was cognitively difficult, much like my later introduction to Object-oriented programming. Thankfully, in my career, I used mostly VAX FORTRAN.

 

[Confuzzled]

I think you are wrong, at least from my point of view you are wrong: Pyra and Psion 5 series both are pocket computers. The key difference is that Pyra is much more modern and powerful and it also include gaming controls on its case. In fact those gaming controls are not only good for gaming: having 4 shoulder buttons is a very nice point because you can configure them as modifiers, and it is a great thing because when you use this pocket computer holding them with hands and typing with both thumbs, you can use those 4 modifiers on shoulder buttons. On other side having analog control is nice for moving mouse pointer (on Psion yoy rely on pen with resistive touch input).

Pyra would have good working sleep mode and instant on (from sleep) if OMAP 5 would not have that bug. Not having that sleep mode is not planned but a consequence of OMAP 5 bug as this version used in Pyra was designed by TI for car infotainment and when they discovered bug it was not relevant for that market (that SOC on car is not in sleep mode). When Pyra have other CPU board surely it will get seep mode.

About Psion 5 keyboard I think it is very overrated, I commented it on #5. I own a Psion 5mx (36 Mhz CPU) and keyboard is horrible to use, only good when using when computer is relying on a table (and not perfect because of unstable moving base under keyboard). It is unusable when you hold computer with hands and type with thumbs.

A pocket computer is a type of computer you can carry with you near like a smartphone, and you want to use it in a lot of places and positions where a table is not an option. Screen/text can be so small in same modern machines, that using it like a notebook on a desktop is not optimal.

I haven't tested Pyra keyboard, but I have tested Psion 5mx keyboard (I own a Psion 5mx), and my instinct says to me it will be a much better keyboard than Psion 5.

PD: Planet Gemini is never an option for me: it has no real Linux support. Its SOC is a mobile SOC only supported by Android from manufacturer, so it uses sonetrhing like Linux with Android kernel/drivers, so you are limited in Linux availability and kernel updates. MNT Pocket Reform could be a interested option as it will have a NXP SOC with real Linux support (and near no blobs/etc).

Well, your use-case and mine are likely different, which is fine. I'm very happy with the Psion Series 5 keyboard for my use. I take your point about the use of the stylus - I simply used my thumb or fingernail, which was good enough for my purposes. I've never got on with trackpoint/nipple pointing devices, while I know some people love them. De gustibus non est disputandum. And yes, the MNT Pocket Reform looks interesting.

To be honest, the virtual keyboard on my phone is pretty good, allowing me to flip between input languages easily, although its implementation of how to access the full available character set could be improved. Having the word suggestions is nice, which means I can compose SMSs and short emails quickly, but I do prefer a physical keyboard for more extended use. It is a preference, and I wouldn't want to impose it on other people, but I would like the choice to have it.

Thank you for sharing your thoughts.

Postscript: I just looked at the renderings of the MNT Pocket Reform here: https://tuxphones.com/mnt-pocketreform-open-hardware-linux-pda-keyboard-arm/

The Control-key is where God intended, to the immediate left of the 'A'-key! Superlative.

 

[Confuzzled]

About«an updated Cirrus Logic cpu»: I think this company is out of CPU market from time ago, but you can look for actual CPU/SOC from other manufacturers with lowest consumption. But don't forget CPU consumption is only one of others. For example when in sleep mode you must keep RAM power and I suspect modern 4GB DDR (Dual Channel PC DDR3L) memory could consume more than old 16/32MB low power SDRAM (and those consumes more than old small SRAM, static RAM, memory; On my HP 48SX consumption on sleep mode is 15 µA for complete machine and batteries last for years, even it has a "coma" mode where it halts real time clock and total consumption goes down to 7 µA for complete machine minus real time clock (when running code it consumes no more than 10 mA) but that machine has STATIC RAM, no DRAM or DDR which needs refreshing), even if modern memory chips is manufactured in much higher density. Of course today we can't have 4GB SRAM

I can say batteries on Psion 5, even with that 3 µA CPU in standaby, doesn't last so much like on HP 48, and I think RAM consumptio is one of the reasons. And there are more elements than CPU and memory even on sleep mode.

As a note we would need to talk on W (mW, µW) as amperes does not state energy but only current (it is not the same 1A at 3v than 1A at 30kV).

I have take a look at NXP i.MX 8M CPU/SOC consumption (here you have a 54 pages PDF on "i.MX 8M Mini Power Consumption Measurement") and it states like 9 mW total power in Suspend mode (This mode is called either Dormant mode or Deep sleep mode in the Linux BSP. This is the lowest possible power state where the external supplies are still on) on page 11, On idle default it states 617 mW. On a core mark it states 1786 mW. On 4-core Streamcpy + VPU + Taiji it states 2359 mW. But this SOC includes much more elements than Cirrus Logic on Psion 5 (and 4 main cores, not only one).

Finally you need to know total power consumption of your system on sleep (or other modes) because there is much more than CPU/SOC.

You are quite right to point out that total system power consumption is what should be measured, not just the processor power draw in sleep mode; and for pointing out that DRAM is a 'power-hog'. Let me visit fantasy-land for a moment and proposed that a pocket computer should use 3D Xpoint memory...

 

[netcat]

Ah.. now i understand what Confuzzled means.

 

[Confuzzled]

Ah.. now i understand what Confuzzled means.

Glad you understand my point of view!

It's like psed, but longer.

(3D Xpoint memory is non-volatile, so no power draw when in 'sleep' (power -off). Of course, it is a power-hog in operation (if you look at Intel Optane SSDs), and orders of magnitude slower than DRAM, even though it is significantly faster than NAND flash - so, no. Essentially looking for RAM that is fast, high density and non-volatile, which doesn't exist. Possibly spintronics or memristors might offer a solution, but there's nothing on the horizon.)

 

[oskda]

You should have bought a Jupiter Ace back in the day. Those ran forth instead of BASIC, and that's normally credited as the reason that machine died a death back in the day.

I know about Jupiter ACE. The problem was not only having a much better integrated language which normal people can´t use because it is difficult for them.

It was released in 1982 with only 1KB of RAM and with monochrome graphics.. Well, it didn't had graphic mode, only text (monochrome text). In those parameters it is nearer to Sinclair ZX81 than to Sinclair ZX Spectrum which was launched in same year: 1982.

How could a 1KB/monochrome/no-graphics Jupiter Ace for £90 compete with a 16KB/color/graphics ZX Spectrum for £125? It could not (on other side both had horrible bad keyboards).

Forth was not its weak point, but its stronger and only point, although it is true for a lot of people it is not a point. But let's remember than most ZX Spectrum users didn't programed in BASIC (nor in other language). A lot of them only learned LOAD "" to load games.

If I would go back to eighties, in 8-bits I would get a Forth cartridge for my Commodore 64/128 (now I have that cartridge as retrocomputing).

PD: Text mode is not bad by itself. In fact it is good: a display in text mode consumes less memory and can be updated much more quickly (for example 1 vs 8 speed and memory). The bad is not having bitmap mode too. For example Commodore 64 VIC chip has a lot of video modes, including text modes and graphics modes.

 

[oskda]

I got a Forth interpreter for my CP/M machine, (was tempted by the Ace at the time) and while I could see the benefits, the pain of having to load it from tape instead of the ROM's built-in BASIC meant I didn't use it as much as I should have. I also had got so used to thinking procedurally (from BASIC) , that the Forth approach was cognitively difficult, much like my later introduction to Object-oriented programming. Thankfully, in my career, I used mostly VAX FORTRAN.

Some machines could use ROM cartridges. For example for my Commodore 64/128 I have cartridges with Forth. On other side Forth interpreter/environment consumes uses very little space in ROM (and in RAM), so making a ROM chip/cartridge should not be expensive.

Of course loading for tape is tedious. With disk that was different. But disk was not common on some countries, even some computers didn't have a common disk drive from manufacturer.

I liked to program and "play" with assembler/machine language, and you had to load an assembler/machine language monitor (Commodore 128 has that in ROM, but it was a simple monitor/assemble/etc, not a assembler with labels/etc, so if you wanted that you had to load from tape or disk).

All BASIC languages I have see in 8-bit computers are totally garbage (no procedures/etc), except only one: BBC BASIC which includes procedures (with parameters/etc) and even assembly language from that BASIC. Of course, BBC Micro was designed by best team in Acorn (Acorn also designed original ARM CPU and architecture) with education focus (BBC computer was oriented that way).

 

[oskda]

I take your point about the use of the stylus - I simply used my thumb or fingernail, which was good enough for my purposes. I've never got on with trackpoint/nipple pointing devices, while I know some people love them. De gustibus non est disputandum. And yes, the MNT Pocket Reform looks interesting.

I used my fingernail too with those resistive touch displays (my first one was Casio PB-1000 from 1987) but the worst part and real problem of resistive touch system is that it decrease screen clarity and adds a lot of reflections and even some rippling (if you look your face reflection on a capacitive screen it looks like in a mirror, but in a resistive screen it is deformed). So I hate resistive touch system, for those reason. And Psion 5 LCD display is horrible: low contrast, a lot of reflections by resistive layer...

To be honest, the virtual keyboard on my phone is pretty good.... but I do prefer a physical keyboard for more extended use.

So do I, but ONLY with a good physical keyboard, because having a bad one is worse than no having keyboard. A lot of small keyboards I have tried were from really bad to horrible.

I just looked at the renderings of the MNT Pocket Reform here: https://tuxphones.com/mnt-pocketreform-open-hardware-linux-pda-keyboard-arm/
The Control-key is where God intended, to the immediate left of the 'A'-key! Superlative.

Like in original 83/84 keys Model F keyboard on IBM PC Later they ruined that key position with Model M keyboard.

If finally MNT Pocket Reform has a ortholinear keyboard (like we see on renders and prototype), you can put each key where you want: all them have same size

 

[oskda]

Let me visit fantasy-land for a moment and proposed that a pocket computer should use 3D Xpoint memory...

I liked MRAM: It is as good as SRAM (Static RAM) but it is non-volatile.