The unrealistic buy it now prices are here too, but a typical S7 Edge with at least a cracked screen goes for around £40. Every now and then one inexplicably gets bid up to £50+ despite having a cracked screen, it's odd. Over half the used market is Samsung and that's mainly S6/S7/S8 so there's no shortage of supply which may help with the price. At peak times prices are higher.

[QUOTE=ewmayer;507376]...
Those seem rather pricy vs what you paid, so "patient cheapskate" is the correct m.o.
...[/QUOTE]
The usual ebay tactics:[LIST][*]Don't bid until the last second and hope that the previous bidder put in a low max bid. A bidding war only helps the seller[*]Add a few pennies to win with £20.03 if the previous bidder bid £20. Normally the next bid would be £21 and if the price is tied the first bidder gets it which you won't have time to counter if you do point 1[*]Pay extra attention to listings that end at weird times, if they haven't already been bid beyond your price range they may go relatively unnoticed[*]Sort by newly listed and make offers to impatient sellers[/LIST]

Those are good tips for anyone looking to score a suitable broke-o-phone on ebay, but for myself I'm going to try a different cheapskate-tack:

o I've updated the OP with a link to your post #51, the one describing your utility scripts, a copy of your latest suitable-hardware PDF, and a note encouraging users who want to help but lack time to do their own rooting to consider donating a suitable device;

o Mlucas v18 is coming soon, I intend to add similar links and verbiage to the updated README page in conjunction with the release.

[QUOTE=ewmayer;507484]

o Mlucas v18 is coming soon, I intend to add similar links and verbiage to the updated README page in conjunction with the release.[/QUOTE]

:bow::tu:

Luigi, you may be disappointed - I decided that adding residue-shift and various other goodies was enough for v18, the PRP stuff will have to wait v19. I have already added the basic low-level stuff needed for PRP, but the attendant Gerbicz check needs some significant enhancements to the FFT API.

[QUOTE=ewmayer;507495]Luigi, you may be disappointed - I decided that adding residue-shift and various other goodies was enough for v18, the PRP stuff will have to wait v19. I have already added the basic low-level stuff needed for PRP, but the attendant Gerbicz check needs some significant enhancements to the FFT API.[/QUOTE]

I'm never disappointed when you present a new version of your software! :smile:
I understand that working for old rPi is actually becoming a bit old-fashion, and we might experience new faster boards before the end of the year (or version 19, whichever comes first).

[QUOTE=ewmayer;506815]
o Is there a way to tell if the CPUs are throttling under load? [/QUOTE]
I tested the throttling of my Galaxy S6 (Exynos 7420, 2 clusters, 4x A57 and 4xA53) about a year ago. It started to throttle within 2 minutes (it was running TheSkyNet POGS, from the BOINC wrapper on the 4 A57 cores). The internal temp of the 7420 was getting to 75-80C. The A57 ('big' cluster) is clocked at 2100MHz but it is only thermally stable at 1200MHz.

[url]https://andrewminalto.com/best-ebay-snipers/[/url]

[QUOTE=ewmayer;507376]
2. [url]https://www.ebay.com/itm/Samsung-Galaxy-S7-edge-SM-G935X-32-GB-Gold-Platinum-Smartphone/163510998469?epid=220203053&hash=item261203d5c5:g:RRUAAOSw~slcThGI:rk:9:pf:0[/url] -- Current bid $6.50, but meaningless since 2 days left. Buy It Now price set at $99:

"Samsung Galaxy S7 edge SM-G935X - 32 GB - Gold Platinum Smartphone. Condition is For parts or not working. Shipped with USPS Priority Mail Padded Flat Rate Envelope.

Device was a demo unit and has heavy damage. Screen is broken and has screen burn from being on all day. As is, no returns."[/QUOTE]

That one ended up going for $47, including $8 shipping.

[QUOTE=ET_;507539]I'm never disappointed when you present a new version of your software! :smile:
I understand that working for old rPi is actually becoming a bit old-fashion, and we might experience new faster boards before the end of the year (or version 19, whichever comes first).[/QUOTE]

There was an interview with Eben Upton a few days ago at tomshardware.co.uk which suggests that Raspberry Pi 4 won't be out this year (and therefore that, given the leap-year timing for RPi3 and the original RPi, it's likely to appear on 29 February 2020).

It sounds as if they are having to rework the system IP (the 40nm process that the original chips were built on is starting to be phased out, they would really like a memory map allowing more than 1GB of DRAM, they would really like a memory controller that isn't LPDDR2 only). Probably moving to TSMC 28nm; I wouldn't be amazed if they stayed with a quad-A53, and a more-64-bit interconnect design with a bag on the side to present a 32-bit interface to a not very changed VideoCore.

From my understanding raspberry pi is a vehicle for Broadcom to unload it's older designs that would otherwise be obsoleted. 28nm sounds about right but I'd hope for 20nm if they're going to stick with it for years. Whatever the case it's likely to be a breaking change, quad-A53 is out of the question IMO.

[QUOTE=M344587487;507617]From my understanding raspberry pi is a vehicle for Broadcom to unload it's older designs that would otherwise be obsoleted. 28nm sounds about right but I'd hope for 20nm if they're going to stick with it for years. Whatever the case it's likely to be a breaking change, quad-A53 is out of the question IMO.[/QUOTE]

Raspberry Pi definitely would not exist without Broadcom, and RPi1 was the result of working out that the firmware-management-and-shader-compiling processor on a Broadcom GPU could be used to make the chip into a real computer. By RPi3 the Foundation (having hired a lot of the Broadcom Cambridge.uk team after Broadcom Cambridge closed down) was calling the shots and telling Broadcom what to put on their chip; I believe that the Foundation now have chip design capabilities and ARM licenses and can contrive a custom chip.

Maybe RPi4 would use Cortex-A55, but the major point in A55 was compatibility with the AMBA5 bus, and I don't think that's necessary for RPi-type designs; so I would not be at all surprised with sticking with A53. A75/A76 is much more expensive to license than A53.

28nm is I think one of the processes that TSMC are going to stay with for years; 20nm was an odd intermediate node (shrinking geometry without introducing FinFET produced hot processors), the next plausibly long-term node after 28nm is 16nm.

There's quite a lot of continuity, through the set-top-box market that Acorn moved into as the market for non-PC-compatible home desktops faded away, from the original Acorn of the BBC Micro through to the Raspberry Pi of today.