New Pico2

[John Larkin]

On Sun, 11 Aug 2024 14:04:36 -0700, John Larkin

Surprised nobody has mentioned the Pico2 boards (based on RP2350A or
RP2350B chips, instead of RP2040).
2x ARM cores plus 2x RISC-V cores (perm any 2 from 4)
150 MHz with FPU instead of 133MHz without
lower power consumption
more I/O pins (B model only?)
I really ought to buy a couple for tinkering ...
official boards not available yet, but 3rd party boards are, e.g.
<https://shop.pimoroni.com/products/tiny-2350?variant=42092638699603>
<https://shop.pimoroni.com/products/pga2350?variant=42092629229651>

As of now, Digikey shows no stock on the Pico2 and doesn't recognize
the RP2350 chip as a product. Ditto Mouser.
The fast floats look great. I wonder how fast they are.

The RP2350 data sheet is 1347 pages!

[Andy Burns]

Post by John Larkin

As of now, Digikey shows no stock on the Pico2

Just the "pirate brand" boards for now, official boards on pre-order,
but i prefer USB-C to the official ones with micro-B

Post by John Larkin

and doesn't recognize
the RP2350 chip as a product. Ditto Mouser.
The fast floats look great. I wonder how fast they are.

The RP2350 data sheet is 1347 pages!

RP2040 is 644 pages, is the extra all accounted for by RISC-V info?
Regardless, SumatraPDF viewer whizzes through the pages

[Ahem A Rivet's Shot]

On Sun, 11 Aug 2024 14:07:59 -0700

Post by John Larkin
The RP2350 data sheet is 1347 pages!

The days of fitting an instruction set on a page or two are long
gone.

[Kerr-Mudd, John]

On Mon, 12 Aug 2024 12:04:43 +0100

Post by Ahem A Rivet's Shot
On Sun, 11 Aug 2024 14:07:59 -0700

Post by John Larkin
The RP2350 data sheet is 1347 pages!

The days of fitting an instruction set on a page or two are long
gone.

I liked the irony implied (or is it just me?) by wrapping that word. WYSi
WYG!

[Single Stage to Orbit]

Post by Ahem A Rivet's Shot
The days of fitting an instruction set on a page or two are long
gone.

The RISCV instruction set for the cores in the Pico is just 137. It's
all in a table on one page in the rp2350 datasheet.

[Don Y]

Post by Single Stage to Orbit

Post by Ahem A Rivet's Shot
The days of fitting an instruction set on a page or two are long
gone.

The RISCV instruction set for the cores in the Pico is just 137. It's
all in a table on one page in the rp2350 datasheet.

SoC/MCU datasheets have been "tomes" for more than a decade. It's
not the details of the instruction set that require the most
paper/pages but, rather, the details of all of the I/Os and
particulars of the programming model.

E.g., the datasheet for the A5 core that I'm using is almost 2000
pages (~1900). And, that doesn't count the Architecture Reference
Manual (2700 pp), details of the FPU, MMU, secure boot, debug
interface (300 pp), etc. Or, any of the tools available to
support hardware/software development!

They aren't "trivial components" that can be characterized in
a dozen pages of text, tables and graphs like MPUs of days gone by.

[At least we've moved past the days of needing hundreds of pages to
explain the *notion* of a microprocessor to the Reader!]

[Theo]

Post by Ahem A Rivet's Shot
On Sun, 11 Aug 2024 14:07:59 -0700

Post by John Larkin
The RP2350 data sheet is 1347 pages!

The days of fitting an instruction set on a page or two are long
gone.

These days, when you've got a 1K+ page manual you know it's the actual
manual. If it's 10 pages it's just a 'product brief' that shows some basic
information about the chip but not nearly enough to program it (contact the
OEM and they'll make you sign an NDA for the actual details, and maybe only
if you're going to buy a million units).

Most of the time you can ignore huge chunks of the manual - if you never use
the CAN bus transceiver, skip that section. But better to have the
information there if you need it.

Theo

[Phil Hobbs]

Post by Theo

Post by Ahem A Rivet's Shot
On Sun, 11 Aug 2024 14:07:59 -0700

Post by John Larkin
The RP2350 data sheet is 1347 pages!

The days of fitting an instruction set on a page or two are long
gone.

These days, when you've got a 1K+ page manual you know it's the actual
manual. If it's 10 pages it's just a 'product brief' that shows some basic
information about the chip but not nearly enough to program it (contact the
OEM and they'll make you sign an NDA for the actual details, and maybe only
if you're going to buy a million units).
Most of the time you can ignore huge chunks of the manual - if you never use
the CAN bus transceiver, skip that section. But better to have the
information there if you need it.
Theo

As long as the silicon errata sheet isn’t 1000 pages!

Cheers

Phil Hobbs

[druck]

Post by Phil Hobbs
As long as the silicon errata sheet isn’t 1000 pages!

That reminds me of only Intel's attempt at an ARM design; they were
tasked with updating the Strong ARM, instead they made the XScale which
was a complete dog of a core, and the errata alone was almost the size
of the

[David Higton]

Post by druck

Post by Phil Hobbs
As long as the silicon errata sheet isn’t 1000 pages!

That reminds me of only Intel's attempt at an ARM design; they were
tasked with updating the Strong ARM, instead they made the XScale which
was a complete dog of a core, and the errata alone was almost the size
ofthe complete SA110 manual.

When it comes to bugs in chips, Intel certainly have form.

David

[Charlie Gibbs]

Post by David Higton

Post by druck

Post by Phil Hobbs
As long as the silicon errata sheet isn’t 1000 pages!

That reminds me of only Intel's attempt at an ARM design; they were
tasked with updating the Strong ARM, instead they made the XScale which
was a complete dog of a core, and the errata alone was almost the size
ofthe complete SA110 manual.

When it comes to bugs in chips, Intel certainly have form.

I am Pentium of Borg.
Division is futile.
You will be approximated.

[Jan Panteltje]

On a sunny day (Mon, 12 Aug 2024 16:42:54 -0000 (UTC)) it happened Phil Hobbs

Post by Theo

Post by Ahem A Rivet's Shot
On Sun, 11 Aug 2024 14:07:59 -0700

Post by John Larkin
The RP2350 data sheet is 1347 pages!

The days of fitting an instruction set on a page or two are long
gone.

These days, when you've got a 1K+ page manual you know it's the actual
manual. If it's 10 pages it's just a 'product brief' that shows some basic
information about the chip but not nearly enough to program it (contact the
OEM and they'll make you sign an NDA for the actual details, and maybe only
if you're going to buy a million units).
Most of the time you can ignore huge chunks of the manual - if you never use
the CAN bus transceiver, skip that section. But better to have the
information there if you need it.
Theo

As long as the silicon errata sheet isn’t 1000 pages!

Yes, and then going through all the code you wrote...
Did I use that ? Ah, worked around it!

[Don Y]

Post by Theo
These days, when you've got a 1K+ page manual you know it's the actual
manual. If it's 10 pages it's just a 'product brief' that shows some basic

It's often 10 pages just to list (TABULATE!) the pins on the device.

Post by Theo
information about the chip but not nearly enough to program it (contact the
OEM and they'll make you sign an NDA for the actual details, and maybe only
if you're going to buy a million units).

Usually, NDA features are associated with higher volume buys (e.g., secure
boot). But, an average joe should be able to buy/design-in most of these
devices and have a functional product from just the publicly available
literature.

What's annoying is absence of a "click here to get EVERYTHING you need"...

Post by Theo
Most of the time you can ignore huge chunks of the manual - if you never use
the CAN bus transceiver, skip that section. But better to have the
information there if you need it.

It's usually worth familiarizing yourself with the content, at least.
You don't want to discover some detail that affects some other aspect
of your design buried in it. Or, belatedly realize that some part
of the device could have satisfied one of your needs.

Unfortunately, moving to a different device (series, manufacturer)
often means much of the information from the initial device is
just noise that MIGHT help you understand the new device... or,
might HINDER your understanding of it! ("But, I thought that..."
"No, that's the OTHER device!")

Sadly, it seems (esp with ARM) that these documents are just
pasted together with little attention to the OTHER "sections"
that may -- or may NOT! -- be present in a particular manufacturer's
device (as most of the source material comes from ARM and you have
to rely on your vendor to have added appropriate caveats IN THE
PLACES YOU ARE LIKELY TO CHECK.

[Dennis]

Post by Theo
These days, when you've got a 1K+ page manual you know it's the actual
manual. If it's 10 pages it's just a 'product brief' that shows some basic
information about the chip but not nearly enough to program it

Actually the "Product Brief" is 25 pages.

https://datasheets.raspberrypi.com/pico/pico-2-datasheet.pdf

[Lasse Langwadt]

Post by John Larkin
On Sun, 11 Aug 2024 14:04:36 -0700, John Larkin

Surprised nobody has mentioned the Pico2 boards (based on RP2350A or
RP2350B chips, instead of RP2040).
2x ARM cores plus 2x RISC-V cores (perm any 2 from 4)
150 MHz with FPU instead of 133MHz without
lower power consumption
more I/O pins (B model only?)
I really ought to buy a couple for tinkering ...
official boards not available yet, but 3rd party boards are, e.g.
<https://shop.pimoroni.com/products/tiny-2350?variant=42092638699603>
<https://shop.pimoroni.com/products/pga2350?variant=42092629229651>

As of now, Digikey shows no stock on the Pico2 and doesn't recognize
the RP2350 chip as a product. Ditto Mouser.
The fast floats look great. I wonder how fast they are.

The RP2350 data sheet is 1347 pages!

read the part on how the build in buck converter needs a custom inductor
with polarity marking to work, and tell there is something seriously
wrong with it

[John Larkin]

Post by Lasse Langwadt

Post by John Larkin
On Sun, 11 Aug 2024 14:04:36 -0700, John Larkin

Surprised nobody has mentioned the Pico2 boards (based on RP2350A or
RP2350B chips, instead of RP2040).
2x ARM cores plus 2x RISC-V cores (perm any 2 from 4)
150 MHz with FPU instead of 133MHz without
lower power consumption
more I/O pins (B model only?)
I really ought to buy a couple for tinkering ...
official boards not available yet, but 3rd party boards are, e.g.
<https://shop.pimoroni.com/products/tiny-2350?variant=42092638699603>
<https://shop.pimoroni.com/products/pga2350?variant=42092629229651>

As of now, Digikey shows no stock on the Pico2 and doesn't recognize
the RP2350 chip as a product. Ditto Mouser.
The fast floats look great. I wonder how fast they are.

The RP2350 data sheet is 1347 pages!

read the part on how the build in buck converter needs a custom inductor
with polarity marking to work, and tell there is something seriously
wrong with it

The polarized inducor is strange. I'd expect that a small shielded
inductor would work fine. It is interesting to have a switching
regulator on a CPU chip... near a 12-bit ADC!

The Pi designs struggle to save pennies and microwatts, which not all
of us care about.

[Edward Rawde]

Post by John Larkin

Post by Lasse Langwadt

Post by John Larkin
On Sun, 11 Aug 2024 14:04:36 -0700, John Larkin

Surprised nobody has mentioned the Pico2 boards (based on RP2350A or
RP2350B chips, instead of RP2040).
2x ARM cores plus 2x RISC-V cores (perm any 2 from 4)
150 MHz with FPU instead of 133MHz without
lower power consumption
more I/O pins (B model only?)
I really ought to buy a couple for tinkering ...
official boards not available yet, but 3rd party boards are, e.g.
<https://shop.pimoroni.com/products/tiny-2350?variant=42092638699603>
<https://shop.pimoroni.com/products/pga2350?variant=42092629229651>

As of now, Digikey shows no stock on the Pico2 and doesn't recognize
the RP2350 chip as a product. Ditto Mouser.
The fast floats look great. I wonder how fast they are.

The RP2350 data sheet is 1347 pages!

read the part on how the build in buck converter needs a custom inductor
with polarity marking to work, and tell there is something seriously
wrong with it

The polarized inducor is strange.

I've had buck converters sensitive to inductor orientation before.
Last time was with MIC4576
It was sensitive to the orientation of an inductor with the winding horizontal, similar to that shown in the RP2350 data.
But the inductor had no polarity markings.
It was changed to a stand up toroid which had no orientation issues.

Post by John Larkin
I'd expect that a small shielded
inductor would work fine. It is interesting to have a switching
regulator on a CPU chip... near a 12-bit ADC!
The Pi designs struggle to save pennies and microwatts, which not all
of us care about.

[Lawrence D'Oliveiro]

The Pi designs struggle to save pennies and microwatts, which not all of
us care about.

There are products other than the Raspberry Pi available for those with
more money to thr^H^H^Hspend.

[mm0fmf]

Post by Lawrence D'Oliveiro

The Pi designs struggle to save pennies and microwatts, which not all of
us care about.

There are products other than the Raspberry Pi available for those with
more money to thr^H^H^Hspend.

+1

[John Larkin]

On Tue, 13 Aug 2024 04:05:24 -0000 (UTC), Lawrence D'Oliveiro

Post by Lawrence D'Oliveiro

The Pi designs struggle to save pennies and microwatts, which not all of
us care about.

There are products other than the Raspberry Pi available for those with
more money to thr^H^H^Hspend.

I like the Pi for more reasons than the cost. The chips have long-term
availability guarentees, the documentation is excellent, and there are
zillions of kids out there who are familiar with the Pi culture.

Even ignoring the 70 cent price, the RP2040 is an amazing part.

[The Natural Philosopher]

Post by John Larkin
On Tue, 13 Aug 2024 04:05:24 -0000 (UTC), Lawrence D'Oliveiro

Post by Lawrence D'Oliveiro

The Pi designs struggle to save pennies and microwatts, which not all of
us care about.

There are products other than the Raspberry Pi available for those with
more money to thr^H^H^Hspend.

I like the Pi for more reasons than the cost. The chips have long-term
availability guarentees, the documentation is excellent, and there are
zillions of kids out there who are familiar with the Pi culture.
Even ignoring the 70 cent price, the RP2040 is an amazing part.

I agree. Being able to build a wifi connected mcrocontroller project for
under £10 including PCB and case is a win for me.

[Gerhard Hoffmann]

Post by John Larkin
I like the Pi for more reasons than the cost. The chips have long-term
availability guarentees, the documentation is excellent, and there are
zillions of kids out there who are familiar with the Pi culture.

I could not find a CPU chip manual in an hour for my new 8GB RP5
that allowed me to use SPI, so I shelved it and went with the old
BBB and its PRU. So far about documentation excellence.

Gerhard

[Jan Panteltje]

On a sunny day (Thu, 22 Aug 2024 02:01:25 +0200) it happened Gerhard Hoffmann

Post by Gerhard Hoffmann

Post by John Larkin
I like the Pi for more reasons than the cost. The chips have long-term
availability guarentees, the documentation is excellent, and there are
zillions of kids out there who are familiar with the Pi culture.

I could not find a CPU chip manual in an hour for my new 8GB RP5
that allowed me to use SPI, so I shelved it and went with the old
BBB and its PRU. So far about documentation excellence.
Gerhard

I decided to stop with newer Raspberry Pis after this 8 GB Pi4 I am using to post this.

A 35 dollar HD TV satellite box is a faster better media player,
Pis take ever more power..

Is not Pi5 a downgraded Pi4?

[druck]

Post by Jan Panteltje
I decided to stop with newer Raspberry Pis after this 8 GB Pi4 I am using to post this.
A 35 dollar HD TV satellite box is a faster better media player,

Try a media optimised OS such as Kodi, rather than a general purpose
distro if you want to use a Pi as a media player.

Post by Jan Panteltje
Pis take ever more power..

Don't confuse the availability of a larger power supply with the Pi 5
taking more power - it's not much different to the Pi 4. However, as you
*could* attach 4 USB devices and a PCIe device, the PSU is rated to cope
with the maximum current all of those devices could use simultanously.
If you won't be doing that, you can use the Pi 4 PSU.

Post by Jan Panteltje
Is not Pi5 a downgraded Pi4?

No, it's a new SOC which is about 3x faster than a Pi 4, and can use
NVME SSDs and other PCIe devices.

---druck

[Jan Panteltje]

On a sunny day (Thu, 22 Aug 2024 22:08:03 +0100) it happened druck

Post by druck

Post by Jan Panteltje
I decided to stop with newer Raspberry Pis after this 8 GB Pi4 I am using to post this.
A 35 dollar HD TV satellite box is a faster better media player,

Try a media optimised OS such as Kodi, rather than a general purpose
distro if you want to use a Pi as a media player.

Its ultra HD these days on satellite :-)
Just got a new Samsung TV...

In the old days I had to buy a decoder key from Raspberry to even look at mpeg!

Post by druck

Post by Jan Panteltje
Pis take ever more power..

Don't confuse the availability of a larger power supply with the Pi 5
taking more power - it's not much different to the Pi 4. However, as you
*could* attach 4 USB devices and a PCIe device, the PSU is rated to cope
with the maximum current all of those devices could use simultanously.
If you won't be doing that, you can use the Pi 4 PSU.

Post by Jan Panteltje
Is not Pi5 a downgraded Pi4?

No, it's a new SOC which is about 3x faster than a Pi 4, and can use
NVME SSDs and other PCIe devices.

OK, thank you for thr info
I have 2 Pi4 now each with a sitecom USB hub and a 4 TB Toshiba harddisk
plus some cameras (infrared too) plus several RTL_SDR stick plus some audio USB sticks etc..
all on 24/7 on a UPS...
Plus various older Pis ...

Time for a change ;-)

Best performance is from an older Pi4 4 GB in a metal housing with a build in fan..
been extremly reliable so far, but no WiFi that way.
Records security cams 24/7, radiation. weather, air traffic, shipping traffic..
more stuff... plays audio too.

[Lawrence D'Oliveiro]

Post by Jan Panteltje
In the old days I had to buy a decoder key from Raspberry to even look at mpeg!

Presumably that was only to activate the decoder hardware. Otherwise
FFmpeg, VLC etc could play it just fine, just with a bit more CPU usage.

[Jan Panteltje]

On a sunny day (Fri, 23 Aug 2024 07:09:03 -0000 (UTC)) it happened Lawrence

Post by Lawrence D'Oliveiro

Post by Jan Panteltje
In the old days I had to buy a decoder key from Raspberry to even look at mpeg!

Presumably that was only to activate the decoder hardware. Otherwise
FFmpeg, VLC etc could play it just fine, just with a bit more CPU usage.

I would not play smooth.

[Jan Panteltje]

On a sunny day (Fri, 23 Aug 2024 07:09:03 -0000 (UTC)) it happened Lawrence

Post by Lawrence D'Oliveiro

Post by Jan Panteltje
In the old days I had to buy a decoder key from Raspberry to even look at mpeg!

Presumably that was only to activate the decoder hardware. Otherwise
FFmpeg, VLC etc could play it just fine, just with a bit more CPU usage.

I would not play smooth.

PS:
Maybe good to look around a bit:
https://arstechnica.com/gadgets/2024/04/what-i-learned-when-i-replaced-my-cheap-pi-5-pc-with-a-no-name-amazon-mini-desktop/

My core I5 Samsung laptop, now > 10 years old, is soooo much faster
browsing with Ubuntu and firefox.

ARM is not a solution to everything.

Sure GPIO is cool,
but I have PCI cards with parport for I/O in my normal PCs..
So maybe for I/O applications just use a <10 dollar Pico2?
(at least it also has a RISC core?)
Have not ordered one yet... Not in the online shop here yet last monday.

Or if no extreme speed is needed I use Microchip PICs:
https://panteltje.nl/panteltje/pic/index.html
You do not need a filesystem if you have only one 'subject' that needs
storing data, just write sector by sector.
https://panteltje.nl/panteltje/quadcopter/index.html

Too much bloat these days and Linux with all the rathead shit is becoming a nuisance.

We went to the moon in the sixties of last century and came back
with computing power less than a Raspberry.
Now astronuts get stuck on the ISS with billions of dollars and sup[p]er computers to do the work.

Back to basics guys!

[Lawrence D'Oliveiro]

Post by Jan Panteltje
Too much bloat these days and Linux with all the rathead shit is becoming a nuisance.

Linux can be as small as you want. Look at Damn Small Linux, for example.
You can actually build a kernel to run on certain CPUs with no memory-
management hardware.

Post by Jan Panteltje
We went to the moon in the sixties of last century and came back with
computing power less than a Raspberry.

There was actually a whole lot of computing power on the ground, in the
form of those IBM mainframes, backing up the guys in space. Everything
they did had to be managed in coordination with ground control, all the
steps worked out in advance.

Post by Jan Panteltje
Now astronuts get stuck on the ISS with billions of dollars and
sup[p]er computers to do the work.

Bit more Government management of the process back then, bit more trust to
private corporations like Boeing, with its stellar reputation for
relentlessly pursuing reliability over profits, these days.

Coincidence? You be the judge.

[Jan Panteltje]

On a sunny day (Sat, 24 Aug 2024 07:26:54 -0000 (UTC)) it happened Lawrence

Post by Lawrence D'Oliveiro

Post by Jan Panteltje
Too much bloat these days and Linux with all the rathead shit is becoming a nuisance.

Linux can be as small as you want. Look at Damn Small Linux, for example.
You can actually build a kernel to run on certain CPUs with no memory-
management hardware.

I find
https://www.damnsmalllinux.org/
interesting

My Raspis and PCs all use fvwm as window manager and xfm as file manager
and have 9 virtual screens with terminals.

xfm -appmgr -geometry 1910x1054+0+0 &

# xterms
/usr/bin/X11/rxvt -font 7x14 -ls -sl 10000 -geometry 270x75+1920+0 &
/usr/bin/X11/rxvt -font 7x14 -ls -sl 10000 -geometry 270x75+3840+0 &
/usr/bin/X11/rxvt -font 7x14 -ls -sl 10000 -geometry 270x75+0+1080 &
/usr/bin/X11/rxvt -font 7x14 -ls -sl 10000 -rv -geometry 270x75+1920+1080 &
/usr/bin/X11/rxvt -font 7x14 -ls -sl 10000 -geometry 270x75+3840+1080 &
/usr/bin/X11/rxvt -font 7x14 -ls -sl 10000 -geometry 270x75+0+2160 &
/usr/bin/X11/rxvt -font 7x14 -ls -sl 10000 -geometry 270x75+1920+2160 &
/usr/bin/X11/rxvt -font 7x14 -ls -sl 10000 -geometry 270x75+3840+2160 &

I use zsh as shell, much less typing required

Post by Lawrence D'Oliveiro

Post by Jan Panteltje
We went to the moon in the sixties of last century and came back with
computing power less than a Raspberry.

There was actually a whole lot of computing power on the ground, in the
form of those IBM mainframes, backing up the guys in space. Everything
they did had to be managed in coordination with ground control, all the
steps worked out in advance.

Indeed, but a 'whole lot' in those days was not even a small PC today.

Post by Lawrence D'Oliveiro

Post by Jan Panteltje
Now astronuts get stuck on the ISS with billions of dollars and
sup[p]er computers to do the work.

Bit more Government management of the process back then, bit more trust to
private corporations like Boeing, with its stellar reputation for
relentlessly pursuing reliability over profits, these days.
Coincidence? You be the judge.

Motivation, indeed now the motivation is profit only it seems.
Back then it was to beat Russia to the moon...
And US had Von Braun, who knew about rockets.
Without him nothing would have flown, the US first rocket tests were a failure.

I have heard Elon wants to go to Mars, I am waiting.
That super heavy thing is interesting, he has some very good engineers.
But chances are by the time they arrive at Mars they need Chinese Visa and money, pay landing rights.
China just did a sample return mission from the backside of the moon.
They invented the gunpowder too, long ago.

As to going back to 'simple'
https://panteltje.nl/panteltje/quadcopter/index.html
https://panteltje.nl/panteltje/quadcopter/drop.html
that was 2016-2017
now they use it to drop explosives etc...

All just done with a simple Microchip PIC
https://panteltje.nl/panteltje/quadcopter/hsign.html

It seems to me we only need ever more complex Linux just for the browser
the plot is to make the OS ever more complex to force people to buy ever more powerful hardware
The media content you see has not much more 'depth' but advertising increases.
Microsoft used to buy shares in hardware companies before releasing the next bigger bloat
the bloat would force the people to buy ever more powerful hardware, they use
automatic updates to force the buying, how evil can you get.

Evils like rathead would try to make their Linux versions non-portable non-compatible to lock in customers to their systems.
Unfortunately some of their silly crap made it into the kernel.

When Linus dies (is he still alive) and greed only rules where will it go.

[Ahem A Rivet's Shot]

On Sat, 24 Aug 2024 12:32:59 GMT

Post by Jan Panteltje
But chances are by the time they arrive at Mars they need Chinese Visa
and money, pay landing rights.

<splort!> I always thought that that was one of the things Jos
Whedon got right in Firefly - two languages in space Chines and English
with people speaking a pidgin of both.

[Lawrence D'Oliveiro]

Post by Ahem A Rivet's Shot

Post by Jan Panteltje
But chances are by the time they arrive at Mars they need Chinese Visa
and money, pay landing rights.

<splort!> I always thought that that was one of the things Jos
Whedon got right in Firefly - two languages in space Chines and English
with people speaking a pidgin of both.

But there are multiple Chinese languages. I know they are commonly called
“dialects”, but it is also pointed out that, say, Mandarin and Cantonese
are more different from each other than, say, English and Norwegian.

There is the saying “a language is a dialect with an army”. I suppose
there is a corollary to that: China ceased being a bunch of independent
states around 220 BC.

[Ahem A Rivet's Shot]

On Fri, 23 Aug 2024 10:33:21 GMT

Post by Jan Panteltje
We went to the moon in the sixties of last century and came back
with computing power less than a Raspberry.

Hohmann worked out the details of travel around the solar system
using the computing power in his head and the storage afforded by a pencil
and paper. The computing requirements for space travel are not that large.

Post by Jan Panteltje
Now astronuts get stuck on the ISS with billions of dollars and sup[p]er
computers to do the work.

High quality engineering is a major requirement of space travel.
NASA in the sixties had an effectively infinite budget and a culture of
careful engineering going back to Edward Murphy of Murphy's law.

Boeing are busy acquiring a reputation of careless engineering.

[Lawrence D'Oliveiro]

Post by Ahem A Rivet's Shot
Hohmann worked out the details of travel around the solar system
using the computing power in his head and the storage afforded by a
pencil and paper. The computing requirements for space travel are not
that large.

He was only able to come up with “minimum energy” transfer orbits that
went fairly directly from body A to body B. Since then, we have worked out
more complicated low-energy “slingshot” orbits that involve bouncing
around the Solar System to match velocities with difficult-to-reach target
bodies like comets. That required real computing power: there’s no way
anybody could have worked that out with pencil and paper, even with a
slide rule.

And what about “halo” orbits, like at Earth-Moon L2, where the Chinese put
their relay station for maintaining contact with their far-side rover, and
where the James Webb telescope is located? You think you could work out
the right numbers for those without a modern digital electronic computer?
They’re not even properly stable, for a start.

Post by Ahem A Rivet's Shot
Boeing are busy acquiring a reputation of careless engineering.

True, that. But are they too big to be allowed to fail?

[Theo]

Post by Lawrence D'Oliveiro

Post by Jan Panteltje
In the old days I had to buy a decoder key from Raspberry to even look at mpeg!

Presumably that was only to activate the decoder hardware. Otherwise
FFmpeg, VLC etc could play it just fine, just with a bit more CPU usage.

It was to pay the patent licence fees for using the hardware decoder. FOSS
software decode got around them by not having anyone the patent holders
could meaningfully sue.

Since that was 12 years ago, perhaps the patents have now expired?

Theo

[The Natural Philosopher]

Post by druck

Post by Jan Panteltje
Pis take ever more power..

Don't confuse the availability of a larger power supply with the Pi 5
taking more power - it's not much different to the Pi 4. However, as you
*could* attach 4 USB devices and a PCIe device, the PSU is rated to cope
with the maximum current all of those devices could use simultanously.
If you won't be doing that, you can use the Pi 4 PSU.

No, that is true, and my Pi 4 can just handle two USB SSD drives. JUST.
Not three though.

BUT even doing sod all it still draws enough to get its temperature up
into the 50's°C. As do the drives into the 40s°C

Which the Pi zero also does

I think that as a desktop machine Pi 4s and 5s power consumption is
approaching that of a good intel chipset *for equivalent performance*.

(these temps may be of interest)

PI 4B (headless) equipped with 2 SSD
=======================
CPU temp=52.1'C
Disk 1= 40°C
Disk 2 =44°C

PI zero 1 (headless)
===========
CPU temp=43.9'C

PI zero 2 (headless)
===========
CPU temp=39.5'C

[Ahem A Rivet's Shot]

On Fri, 23 Aug 2024 10:36:20 +0100

Post by The Natural Philosopher
I think that as a desktop machine Pi 4s and 5s power consumption is
approaching that of a good intel chipset *for equivalent performance*.

The rise of ARM caused Intel and AMD to pay a *lot* more attention
to power consumption which has indeed resulted in the gap closing
especially at desktop levels of performance.

[The Natural Philosopher]

Post by Ahem A Rivet's Shot
On Fri, 23 Aug 2024 10:36:20 +0100

Post by The Natural Philosopher
I think that as a desktop machine Pi 4s and 5s power consumption is
approaching that of a good intel chipset *for equivalent performance*.

The rise of ARM caused Intel and AMD to pay a *lot* more attention
to power consumption which has indeed resulted in the gap closing
especially at desktop levels of performance.

Well MIPs per mW is where its at these days as parallelism can sorta
increase MIPS to whatever level you want.

Funnily enuff, the race to more MIPS is likely to drive adoption of
nuclear power more than any other factor...

I don't see data centres 'throttling back' servers after sunset, or on
calm days...

[Andy Burns]

Post by The Natural Philosopher
the race to more MIPS is likely to drive adoption of
nuclear power

In the USA, Big Data is either buying-up the full output of nuclear
plants, or buying the nuclear plants themselves.

Post by The Natural Philosopher
I don't see data centres 'throttling back' servers after sunset, or on
calm days...

Throttling in tune with renewable generation, maybe not, but "herding"
multiple virtual servers onto a reduced number of physical hosts in tune
with demand and shutting down some of the physical hosts is done.

[Chris Jones]

Post by John Larkin

Post by Lasse Langwadt

Post by John Larkin
On Sun, 11 Aug 2024 14:04:36 -0700, John Larkin

Surprised nobody has mentioned the Pico2 boards (based on RP2350A or
RP2350B chips, instead of RP2040).
2x ARM cores plus 2x RISC-V cores (perm any 2 from 4)
150 MHz with FPU instead of 133MHz without
lower power consumption
more I/O pins (B model only?)
I really ought to buy a couple for tinkering ...
official boards not available yet, but 3rd party boards are, e.g.
<https://shop.pimoroni.com/products/tiny-2350?variant=42092638699603>
<https://shop.pimoroni.com/products/pga2350?variant=42092629229651>

As of now, Digikey shows no stock on the Pico2 and doesn't recognize
the RP2350 chip as a product. Ditto Mouser.
The fast floats look great. I wonder how fast they are.

The RP2350 data sheet is 1347 pages!

read the part on how the build in buck converter needs a custom inductor
with polarity marking to work, and tell there is something seriously
wrong with it

The polarized inducor is strange. I'd expect that a small shielded
inductor would work fine. It is interesting to have a switching
regulator on a CPU chip... near a 12-bit ADC!

I don't think they know much about how to do ADCs yet.

Have a look at the performance of the RP2040 ADC - it is awful!

http://pico-adc.markomo.me/

I hope the new one is better.

[john larkin]

On Mon, 19 Aug 2024 22:04:19 +1000, Chris Jones

Post by Chris Jones

Post by John Larkin

Post by Lasse Langwadt

Post by John Larkin
On Sun, 11 Aug 2024 14:04:36 -0700, John Larkin

Surprised nobody has mentioned the Pico2 boards (based on RP2350A or
RP2350B chips, instead of RP2040).
2x ARM cores plus 2x RISC-V cores (perm any 2 from 4)
150 MHz with FPU instead of 133MHz without
lower power consumption
more I/O pins (B model only?)
I really ought to buy a couple for tinkering ...
official boards not available yet, but 3rd party boards are, e.g.
<https://shop.pimoroni.com/products/tiny-2350?variant=42092638699603>
<https://shop.pimoroni.com/products/pga2350?variant=42092629229651>

As of now, Digikey shows no stock on the Pico2 and doesn't recognize
the RP2350 chip as a product. Ditto Mouser.
The fast floats look great. I wonder how fast they are.

The RP2350 data sheet is 1347 pages!

read the part on how the build in buck converter needs a custom inductor
with polarity marking to work, and tell there is something seriously
wrong with it

The polarized inducor is strange. I'd expect that a small shielded
inductor would work fine. It is interesting to have a switching
regulator on a CPU chip... near a 12-bit ADC!

I don't think they know much about how to do ADCs yet.
Have a look at the performance of the RP2040 ADC - it is awful!
http://pico-adc.markomo.me/
I hope the new one is better.

It's tough to put a good 12-bit ADC on a 70 cent dual-core CPU chip.
The silicon process, the thermals, the ground loop and noise
environment, are all wrong.

The 2040 ADC has chunks of missing codes. It's probably usable as a
7-bit, 1% ADC. Some lowpass filtering, with some dithering, would
improve it but if you want precision, buy a separate ADC.

We use the ADCs in FPGAs and some other ARM processors, for crude
things like checking power supplies. Not for sellable instrumentation.

[wmartin]

Hi All,
Anyone doing stuff with both cores on the Pico? I just grabbed the Pico
extension for Visual Code to poke it & see what it does, had an
interesting first adventure. It does not seem to know anything about
setting up a muticore project, so I had it generate a simple project
that it did understand, just "example" level stuff, replaced the
generated .c file with my own little test case...#include
"pico/multicore.h" is not findable. So, looking at the generated
compiler commands file, I added in a -I/home/.../include explicitly to
see what happens. No change, still not found. hmm. Is there some magic
caching going on here? Next, I added a -D for the multicore library,
still no change. Noticed that I had forgotten to add the pico_multicore
lib to the CMakeLists.txt file and behold it builds now, no missing .h
files. So my "solution" isn't really good for anything more than a
casual kicking the tires thing, and I'd really appreciate it if sommeone
could guide me a little on how to do this "right". Do I just have to
wait for the VSCode extension to get smarter, or is there a better way
to get those .h & .lib paths known to the compiler?
Thanks,
Bill