[HN Gopher] Resizeable Bar Support on the Raspberry Pi
___________________________________________________________________
Resizeable Bar Support on the Raspberry Pi
Author : speckx
Score : 98 points
Date : 2025-10-09 15:12 UTC (8 days ago)
(HTM) web link (www.jeffgeerling.com)
(TXT) w3m dump (www.jeffgeerling.com)
| monocasa wrote:
| > That also led me to research why a card with only 12GB of VRAM
| is requesting a 16GB BAR, and as far as I can tell, it has
| something to do with potential uses like SR-IOV where multiple
| VMs could access a single device, thus requiring more BAR space
| for the physical device being shared... BAR is still a little bit
| of a mystery to me.
|
| Lengths of BARs have to be powers of two.
| tetris11 wrote:
| I'm always fascinated by hobbyist's storage compartments. Some of
| the labels of the boxes in the background: -
| "Broken and Delidded Pis" - "Non-PI ARM SBCs" -
| "-ers" - "Pico and Arduino Microcontrollers" -
| "Breadboards" / "Electric Doodads" / "Pepp(?) boards"
| - "M.2 / NVMe" - "Blades" / "Home Lab / Soquartz"
| - "Small Windows PCS"
| jasonjayr wrote:
| Jeff is a bit more than a hobbyist, between him & his father,
| they are rather accomplished professional radio + electric
| engineers.
| noelwelsh wrote:
| I think of this as a Midwestern thing, where land is cheap and
| one can afford to devote a lot of space to hobbies. It seems to
| be the same thing in fabric arts, where people will have sewing
| rooms larger than the footprint of many UK houses. I could be
| totally be off base here. Maybe some actual Midwesterners can
| weigh in.
| justin66 wrote:
| It's all true, when I moved to the west coast from the
| midwest I was struck by the way everyone just piled their
| possessions into a heap in the middle of their workspace
| rather than placing them in labeled containers. At times the
| heaps reach the ceiling.
| ssl-3 wrote:
| Small-town midwesterner here.
|
| There's some truth to what you're saying. Even the smallest
| apartment I've ever had was big enough to dedicate space to
| some serious hobbies, and it was (at that time) very
| inexpensive to rent compared a much smaller place in a more-
| dense area.
|
| My lifestyle isn't very compact because it has never needed
| to be compact. Compactness isn't a common expectation 'round
| these parts. I have space to keep collections of stuff that I
| find interesting, and to make use of it.
|
| There's tradeoffs to this lifestyle, though. The corner store
| is only a short walk away, but it mostly just sells beer,
| soda, and smokes. There's no walking to get something like
| groceries, or a new shirt (or a used shirt, for that matter).
| There isn't much for local entertainment. It's 15 miles to
| the next-largest city, and there's zero public transportation
| aside from the buses that get kids to/from school.
|
| (It's hard to imagine that there aren't areas of the UK that
| are of similar form, though, with roomy housing, space for
| things, and with very limited services and/or options for
| commerce nearby.)
| noelwelsh wrote:
| There are certainly places with very limited services! As
| for roomy houses, though, that is much harder to find. I
| did a little searching and found a UK govt report on
| housing sizes [1]. For a point of comparison, average house
| size broken out by house type is smaller for all types than
| the average _across_ all house types for Missouri [2; taken
| as a random exemplar of the Midwest]. Only when we break
| out to detached houses built in particular time periods
| (p.18) does the average become roughly equal to Missouri,
| and remember this is across all housing in Missouri (IIUC).
|
| The UK settlement pattern is quite weird. About 1/6 of the
| population lives within the South-East, which is roughly
| the area commutable to London by train. Then another 1/6 of
| the population lives in London. So land is very expensive
| there. The rest of the UK is on average a lot poorer. So
| the UK in general is not conducive to large housing. You
| can certainly find old farm houses out in the sticks, and
| small villages that are mostly expensive large houses.
| These are a tiny fraction of the housing, though. I believe
| UK houses are on average the smallest in Europe. (I think
| the UK would benefit a lot from some decentralisation, but
| that's a different topic. Side eye in cancelled HS2 lines.)
|
| [1]: https://assets.publishing.service.gov.uk/government/up
| loads/...
|
| [2]: https://fred.stlouisfed.org/series/MEDSQUFEEMO
| rcarmo wrote:
| As an EU resident with a modestly large flat who's dedicated
| 1/3 of it to a large living room because we prioritized room
| to relax and spend time together, I pine for the chance to
| have more than a pair of IKEA KALLAX to store (and pile upon)
| all my electronics stuff and 3D printer(s).
| tracker1 wrote:
| Not just the Midwest, but most of the US outside the East
| Coast cities tend to have pretty large homes and apartments
| compared to most of the rest of the world in terms of home
| sizes. The US is pretty huge, with much of it livable and
| relatively less total density than other nations.
|
| Of course, the US has also done a lot more than many places
| to preserve natural wilderness with national and ngo
| parks/reserves.
|
| My house in Phoenix, AZ, USA is relatively typical 3br/3ba
| and just over 2K sqft (around 192 M^2), and I know a lot of
| people with larger homes than this.
| bityard wrote:
| I mean, yes, housing and land is cheaper in the midwest US,
| but you mostly have to get at least an hour away from a major
| city before it becomes what I consider "cheap." (Otherwise
| you're still looking at more desirable suburban or bedroom
| community housing prices.)
|
| But to defend Jeff, the pictures in the article show like a
| 4x4 corner of a room, even if we extrapolate to say, 8x8x8,
| that amount of space can usually be found in even a 1-bedroom
| apartment. I have seen many pictures of amazingly complete
| electronics/hacking workbenches from around the world that
| easily fit into much smaller spaces.
| ptaffs wrote:
| ...and they have time and access to label maker, as opposed to
| sharpie. That's some dedication to having _order_.
| dmd wrote:
| A label maker is about $20. I'm not sure that speaks to
| intense dedication all by itself.
| munificent wrote:
| I wonder if they have a drawer labeled "Label Makers".
| tracker1 wrote:
| Label makers aren't all that expensive, even for relatively
| high volume and fancy options. It's been relatively useful
| IMO to have bins and labels available. That said, my own
| office space is a bit cluttered right now with a handful of
| half-completed projects that I've been too lazy to finish.
| abbeyj wrote:
| > "Pepp(?) boards"
|
| "Perfboards", perhaps?
| eqvinox wrote:
| > Resizeable Bar Support on the Raspberry Pi
|
| Why'd the capitalisation of BAR get lost between the article and
| here? It's a little triggering.
|
| Other than that, nice to see, a lot of ARM SoCs & SBCs have
| asininely small PCIe memory map sizes. Looking at you, RK3399,
| what was it, 64MB? Not even enough for some NICs.
| monocasa wrote:
| That's an hn title munging things I think.
| throw74845858 wrote:
| I am still waiting for Pi5 to correctly support USBC power
| delivery spec. 5.1 Volts at 5 amps is a dirty hack and fire
| hazard, not consumer electronics!
| klempner wrote:
| Huh? There are plenty of good reasons to complain about the
| 5V5A thing, but "fire hazard" is not one of them.
|
| It is even entirely within spec for a PD power supply to offer
| a 5V5A PDO, as long as it is only doing so with a 5A capable
| cable (i.e. 100W or 240W). 5V5A is no more a fire hazard than
| 20V5A.
|
| The spec violation isn't that it negotiates 5V5A when
| available, but that it isn't willing to buck from 9V or 15V to
| get those 25W which means that power supply compatibility is
| incredibly limited.
| prism56 wrote:
| Shame a 5v "fake" PPS voltage couldn't somehow be obtained or
| patched in. Loads of chargers would work then.
|
| My pocket PD can request 5v5a from quite a few chargers in
| PPS mode.
| tracker1 wrote:
| It's a shame that RPi didn't just adopt a proper PD
| interface for power. For that matter, if they had USB-C +
| TB/USB4 with display support, then I could just plug it
| into my display without any other cables like I do my
| laptop, with all the peripherals connected to the display.
| throw84944994 wrote:
| > negotiates 5V5A
|
| Even the voltage is not matching spec (Pi power supply has
| 5.1 volts, not 5.0 volts!). That is because historically Pi
| had shitty cables, with high resistance and voltage drops.
| 5V5A is not even in spec, limit for 5 volts is 3 amps!
|
| > fire hazard than 20V5A
|
| That would be 100 watts! Many people just grab any usbc
| cable, and solder it directly to GPIO power pins. But good
| luck with that!
|
| Initial batches of Pi4 did not even had a resistor, to
| request 3.0 amps!
| toast0 wrote:
| 5.1 volts is 2% off 5.0 volts. I don't have a copy of the
| USB specs, but a voltage 2% higher than nominal is almost
| certainly within specifications.
| bmurphy1976 wrote:
| Oh I'm with you. I've been trying to clean up my setup and the
| Pi5 is still a problem. Even my Intel N100 NUC is using a USB-C
| to barrel jack adapter and working great with a perfectly
| normal multi-port GAN charger, but the Pi5? "Undervoltage
| Detected"
|
| sigh
| burnte wrote:
| Keep waiting. I really hoped they'd not repeat this problem
| from the 4 to the 5, but they did.
| checker659 wrote:
| Isn't it the DMA engine that's responsible for the data movement?
| How does the size of the BAR even factor into that?
| namibj wrote:
| If you can map all VRAM into host linear/physical address
| space, you can just use write combining page table entries to
| it and get efficient buffered write back caching for the CPU
| writes that created the very bytes you think of DMA-ing in the
| first place.
| bityard wrote:
| Hold up, when did htop get the ability to graph things?
| foobar12321 wrote:
| pretty sure thats nvtop
| bityard wrote:
| Ah, that makes sense. Thank you.
___________________________________________________________________
(page generated 2025-10-17 23:01 UTC)