Pi Model Suggestions

[Ken Wiebe]

Hi,

I read on the PiDP8 pages that the Pi Zero W was a good choice for the 8.  Is that true for the 11 also, or should I go with the Pi 3?

And if Pi 3, would that be the 'B' or 'B+'?

Thx

[Garry Lockyer]

Based purely on feel, not measurements, the 3B+ is significantly more capable than Zero W. The GUI on a 3B+ is usable while it’s barely so on a Zero W. This could be because the 3B+ has a GPU. While the simulations will run on a Zero W, they feel much slower on a Zero W than a 3B+.

Regards,

Garry

[Tony]

There was a comment posted in this google group (link) :  Testing the PiDP 11/70  :

"After completing the assembly of the PiDP-11, I installed the Raspberry Pi (for now a Pi 3 B until software runs on the 3B+)."

that suggests the SimH software for the B+ was not able to run.   Has that changed?

Tony

[terry-...@glaver.org]

On Friday, August 10, 2018 at 9:22:56 PM UTC-4, Tony wrote:

that suggests the SimH software for the B+ was not able to run.   Has that changed?

I've always run mine on a 3 B+ with no apparent problems. 

[Randy Mongenel]

All of the Raspberry Pi models have the exact same GPU, they only differ in CPU and RAM. The Zeroes all have a single-core ARMv6 1GHz CPU with 512MB of RAM, while the Pi2 and Pi3 models have a quad-core ARMv8 1.4GHz CPU and up to 1GB of RAM. 

[oscarv]

Ken,

Right - for the PiDP-8 the Pi Zero was (IMO) the best choice.

For the PiDP-11, a Pi 2 or 3+ makes more sense. The Pi Zero is a bit underpowered, it will work fine but you'll appreciate the extra speed of the 3 when booting things like RSX.

Kind regards,

Oscar.

[Warren Young]

On Saturday, August 11, 2018 at 7:25:42 AM UTC-6, oscarv wrote:

for the PiDP-8 the Pi Zero was (IMO) the best choice.

Compiling the PiDP-8 software on a Pi B+ was painful, about 11x slower than on a Pi 3B, as I recall. A Zero is faster; I'd estimate the delta at about 8x.

If you're only going to be running pre-built binaries, then this difference won't matter.

The Pi Zero is a bit underpowered, it will work fine but you'll appreciate the extra speed of the 3 when booting things like RSX.

The benchmarking I did for the PiDP-8/I on a Pi B+ resulted in about 3x more instructions per second (IPS) than a real PDP-8/I could have done, running the same code. With a Pi Zero, that might be 4x the IPS.

I don't have a PiDP-11 yet, but based on memory speed differences and such, it looks like the speed of a PDP-11 is about that same rough speed difference relative to a PDP-8/I. If so, that means a Pi Zero running SIMH is going to be able to process roughly the same IPS rate as a PDP-11; maybe a bit higher, maybe significantly lower.

If someone reads that and thinks, "Great, that means a Zero will give the most accurate emulation," that's not true. First, you want some background power left over for the Pi to do other things, else the front panel LEDs might visibly stutter if a background process starts taking a lot of CPU. Second, SIMH has a powerful throttling feature, so if you want your PiDP-11 to run at the same speed as some particular -11 model, you can make it do so, leaving even more power for other background tasks.

Not only are the individual CPU cores in a 3B or 3B+ faster than the single core in a Zero, there are four of them, so you can dedicate one or two of them to the simulation. Background processes tend not to conflict with the simulator at all on a multi-core Pi.

When I get my PiDP-11, I'm going to be putting a 3B or 3B+ into it.

[Manfred Koethe]

The PI 3 and 3+ are ARM Cortex 53 processors (64 bit), while all other PIs are ARM Cortex 8 (32 bit).

The weak point is the SD Card, which is really not designed to be a "disk" (heavy CRUD-style operation instead of just storing stuff as, for example, a camera does), and the interface is slow.

I have an Pine64 on order, which is also a Cortex 53, but has one USB-3 port and an interface to eMMC cards, which are much more SSD-like. The USB-3 also allows to connect an external disk with reasonable performance, if needed. (the Pine64 is also a little cheaper than a PI 3+, btw.).

I will report back when I get the board, but I can only report "headless" experience until I get a PiDP kit.

Kind regards,

Manfred

[Warren Young]

On Saturday, August 11, 2018 at 6:34:11 PM UTC-6, Manfred Koethe wrote:

The PI 3 and 3+ are ARM Cortex 53 processors (64 bit), while all other PIs are ARM Cortex 8 (32 bit).

That's a relevant distinction only if you're using an OS that supports it. Raspbian remains 32-bit, even on a 3+, which is why there is only one download each for Raspbian Lite and Desktop. Since Raspbian continues to support the older Pi boards, you'd expect two downloads for each version of Raspbian if there were a 64-bit version available.

This article explains the "why" well: https://raspberrypi.stackexchange.com/q/43921/13589

 

I have an Pine64 on order

That price makes me nervous: more functionality for less money than a nonprofit is able to charge for the less capable offering. That makes me wonder whether they'll be able to honor their "LTS" guarantee.

I will report back when I get the board

Please do. I'd be particularly interested to know:

1. What's the output of "make clean ; time make -j6" for either the PiDP-8 or -11 software?
2. Once you've gotten your PiDP-11, is it physically compatible? I see the Pi 2 header on the Pine 64, but the board looks bigger, so does it crash into anything on the PiDP-11 board? I think it would be entirely incompatible with the PiDP-8/I, short of adding a ribbon cable to let you move the Pine64 out of the way of various obstructions.

Back when the wish list threads were going for the -11 on the -8's mailing list, one of the suggestions was to support boards like the ODROID-C2. That required skipping the 2 pins that aren't electrically compatible with the Pi. Did that happen?

[Ken Wiebe]

Thanks for all the tips and info.  I am up and running PiDP11 on a 3B+.  Anxious for the kit!

Ken

[Ken Wiebe]

This post confused me.  Pi Zero and 3B outperform the 3B+?  I thought the 3B+ was the most powerful Pi at the moment.

On Saturday, August 11, 2018 at 1:24:05 PM UTC-7, Warren Young wrote:

[Tony]

I believe Warren is referring to the original Pi B+ - not the 3B - when he makes those comments about relative speed.

[Whit Turner]

Correct - the 3B+ is the most powerful. The Zero is the least powerful (in comparison to the 3B or the 3B+).

BTW - if you are in the US and near a Micro Center store, they are currently selling the 3B+ for only $29.99 (in-store pickup only). Seems like a pretty good price. I recently paid that much for a plain 3B.

Whit

[Warren Young]

On Sunday, August 12, 2018 at 7:51:53 PM UTC-6, Tony wrote:

I believe Warren is referring to the original Pi B+ - not the 3B - when he makes those comments about relative speed.

Yes. Sometimes people say "Pi 1B+" or similar to make it clearer, but that's a retronym. The original product was called the "Raspberry Pi Model A", no "1" at all. Then there was the model B, then the model B that I referred to repeatedly in that post. That in turn is about 3/4 the speed of the Zero we're focusing on so much in this thread.

So yeah, don't put a Pi B+ into your PiDP-11, unless you like underclocked retrocomputers.

[Warren Young]

On Sunday, August 12, 2018 at 9:29:02 PM UTC-6, Warren Young wrote:

Then there was the model B, then the model B

I meant "then the model B+".

Apparently you can't edit posts in this forum, as you could in the -8 forum. 

[oscarv]

Warren,

On Sunday, August 12, 2018 at 3:00:47 AM UTC+2, Warren Young wrote:

Back when the wish list threads were going for the -11 on the -8's mailing list, one of the suggestions was to support boards like the ODROID-C2. That required skipping the 2 pins that aren't electrically compatible with the Pi. Did that happen?

No, actually... I never played with the Raspberry Pi competitors. But a small trace cut should solve that problem with the Odroid!

Kind regards,

Oscar

 

[Jörg Hoppe]

Guys,

maybe this post is obsolete now but here is a nice RPi model comparision chart:
http://socialcompare.com/en/comparison/raspberrypi-models-comparison

Joerg

--
You received this message because you are subscribed to the Google Groups "[PiDP-11]" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pidp-11+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pidp-11/d473eeaa-7bac-4b5f-bd0f-d67b6f2b7a64%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Warren Young]

On Sunday, August 12, 2018 at 10:23:52 PM UTC-6, oscarv wrote:

I never played with the Raspberry Pi competitors. But a small trace cut should solve that problem with the Odroid!

I'm taking that as confirmation that you did manage to avoid using the two problematic pins. If so, thank you! I have no immediate plans to use that board, but it looks interesting, so I hope someone does try it. 

[Manfred Koethe]

Warren, Oscar,

I have experience in other tasks with the ODROID C2, it works very well. Also I will get a ROCK64 (Pine) soon. Both have 4-core ARM Cortex A53 processors and eMMC connectors. While the eMMC does not have the speed of a real SSD, its controller is designed for disk-like operations, while the SD cards are primarily designed for storage. The difference is wear-leveling strategy of the flash controller. The appeal of the ROCK64 is the USB-3 port (with good speed), that allows the use of an external disk. The weak USB-2 performance is the biggest drawback on the Raspberry PI line, in particular before the PI-3, where even the ethernet was routed through the USB controller. The ODROID also has only USB-2, but has the eMMC with decent performance.

I will build software images for both processors as soon as I can free-up (probably next weekend) and as soon as I get the ROCK64. But so far I can only test headless.

Regarding the case-depth discussion, it would be good to get the computer board further away from the PiDP board. I don't know at this moment if higher PI connectors exist. When I ran (headless) RSX-11M Plus Sysgen on my PI 3 B+, it got extremely hot. So at minimum a heatsink is recommended. But the way the computer is mounted, the airflow is restricted, and the heatsink height may become a problem. I consider the use of a small (and very silent) fan. [ on the other hand, a noisy fan would improve the PDP11/70 experience :-) ]. I would also recommend openings for air circulation in the back panel.

Regards,

Manfred

[Oscar Vermeulen]

Warren,

On Mon, 13 Aug 2018 at 07:41, Warren Young <tange...@gmail.com> wrote:

On Sunday, August 12, 2018 at 10:23:52 PM UTC-6, oscarv wrote:

I never played with the Raspberry Pi competitors. But a small trace cut should solve that problem with the Odroid!

I'm taking that as confirmation that you did manage to avoid using the two problematic pins.

I have to dig deep into my notes - which two pins were that? I might - but I am not sure!

Kind regards,

Oscar.

[Warren Young]

I was told 2, but it looks like 3 to me: there are two ADCs down on pins 37 and 39 instead of GPIO, but there's also a 1.8V reference for these ADCs on pin 40. I don't know if these ADCs are inside the microprocessor, but if they are, they're apparently not reassignable as GPIO.

[Warren Young]

On Monday, August 13, 2018 at 9:32:22 PM UTC-6, Warren Young wrote:

there are two ADCs down on pins 37 and 39...1.8V reference for these ADCs on pin 40

Not that it matters for our purposes here, but my spectrum disorder is making me report that the pin numbers are swapped between ADC0 and the reference: the ADC inputs are on 37 and 40, and the reference is on 38.