Hacker News Reader

Comment by anthonj | Hacker News Reader

I really don't understand how in 2025 a competent company can't implement a trival A/B flash update. Do they depend on some obscure contractor for the firmware and have no power here?

View on HN

As with many things, "It's complicated" but effectively, yes

Such system firmware is generally provided by AMI / Insyde / Phoenix and similar.

There are a few vendors who decide to go their own way (notably google using a fork of Coreboot for their chromebooks), but it's generally list of vendors that you can count on a single hand.

If it means anything I'm pretty sure AMI provides A/B updates on most products i've used with their stuff.

But that's always been the case. Phoenix and AMI have been the dominant BIOS providers for over 30 years. Both have always provided modular tookits for OEMs to customize their own ROM images, write their own ACPI tables, etc. UEFI gives even more control to the OEMs. If Framework isn't doing that properly, that's on them.

It appears that Framework contracted out the development of their (Insyde-based) firmware to somebody who, to put it mildly, does not treat them as a priority. Either way, Framework’s firmware is really bad, and that’s entirely on Framework; you can look around on the forums for the horror stories in the (open) EC code around power delivery for instance.

As another example of a vendor low-key screwing Framework in a way that I think it’s entirely appropriate to blame Framework themselves for, the first couple 13-inch models could charge off 5V chargers. The newer ones still have the necessary hardware, but whoever they get their power circuitry from cut that bit from the firmware on a newer chip version, and Framework obediently went “welp, guess we don’t have that feature anymore”.

I heard a lot of talk about porting to coreboot (they even sent some motherboard to some coreboot developers I think). Not sure what is stalling there, but sounds like a priority.

For the EC, I feel like a team of a single developer could arrange something there?

In addition to no longer charging off 5V chargers (their first laptop could literally charge off a USB type A with no USB-PD), on the AMD 7040 I've got ~10% chance every time I plug it in to my dock that it will simply not start charging. Everything else will connect (display, USB) but power won't flow. Resolving that is as simple as unplugging and replugging in my laptop but if I don't notice until I'm in a video conference then it is a problem because my webcam, microphone, and headset are also going over that cable. Oh how I miss the OG 11th gen framework charging.

Interestingly, I once saw them posting job for UEFI developer, and FW developer.

They definitely have FW team inhouse, but idk if that team handle the EC too.

Probably they depend on some "obscure" contactor which is one of the ODMs here: https://en.wikipedia.org/wiki/List_of_laptop_brands_and_manu... .

Intel / AMD and BIOS vendors simply do not expose their designs to everyone. So you have to go through an ODM to have an Intel chip-based motherboard design. The silicon industry is completely rotten at this point. Framework could simply be too small to get access to secret sauce.

This is not completely true, there is open source implementation of UEFI Spec(EDK 2), Microsoft also has Project MU for uefi.

If you are willing to pay, you can design from scratch everything.

The UEFI spec is open. The way to load Intel ME payload isn't. You need to do it otherwise the Intel CPU is as good as a brick. It will shut down automatically in 30 mins, none of the TPM functions will work, no power management will work.

It is not just payment. Without being "big enough" and guaranteeing some big volumes of CPU sales, Intel won't give you the specs. AFAIK only ODMs and BIOS makers like AMI has the full low-level CPU bootstrap spec.

This is just the CPU. Now add the other peripherals like HDMI controllers. You'd see the problem now.

I own the "original" gen11 Intel Framework 13. This is a company that wasn't able to ship a working RTC battery, a touchpad that would register a click when a physical click happened, a screen hinge that would reliably hold the screen in an upright position, or a USB-A port that wouldn't consume a watt of power when unused. You expect way too much engineering from them.

I also had one of this 11th Gen Intel FW13s, and the RTC battery on mine worked for a short time because they for some reason used some garbage rechargeable one with a 2 yr lifespan instead of the more common not-rechargeable battery that every other PC vendor uses with seemingly 10x the lifespan

Yep. Cheapest one from AliExpress works with no issues for four years now, so I shudder to think where they got the ones they shipped.

Heh, I remember the “remove the USB A” debacle when I had my OG.

That thing really hacked me off generally. Coming from a MBP 2016 that sipped power, the battery anxiety was real (especially standby drain), and I spent more time than I’d care to admit tweaking tlp to be still a pretty poor drain.

Then the backlight would only work in certain positions.

Love the dream, but it got sold and replaced by a MBP of 2x the price. 4years later the MBP has been used and abused like a tool, dropped from my car boot while trying to carry it and a child, and is still fresh as a daisy. I worry that FW doesn’t fit into “buy once buy well”

Well, at least I haven't had the backlight issue.

I did eventually tweak it to a very particular state where it had decent idle power use. Of course, it had to be re-tweaked occasionally.

Apart from that laptop, I've only ever used good-gen Thinkpads (and I'm back on one; the Framework now lives on my son's desk). The whole "we ship a screwdriver" gimmick gets old fast if you have to use that screwdriver on a regular basis (_of course_ they couldn't just use Phillips screws).

I still don't understand we are comparing a brand new company's engineering to Apple, who have been making laptops for what, 40 years now? Apple's GPUs used to over heat a fry the laptop. They used to make keyboards that would fully break because of a piece of dust, but they were riveted down so they were irreplaceable. Titanium powerbooks had hinges that would break in a year.

And Apple are still unable to master cables that don't fray within a few months of use.

I still don't see how Framework's relative youth compensates my issues.

Because this isn’t a charity, I’m not paying to work out a new company’s engineering issues. I feel confident, based on the last few laptops, in spending £2k and having it last as my primary for 5 years or so, and then either still having some resale value, or being repurposed for simpler tasks.

I don’t have the same confidence in FW

I know some higher end gaming motherboards that were targeted towards the overclocking community had two bios chips for exactly this purpose and switching between them was trivial from what I remember. If you overclocked too far, you wouldn't brick your system. It seemed like overkill to me. I'm guessing this is less common than it's worth to fix in software, but then at least continue to give support for the issue.

If you overclocked too far, you could remove the CMOS battery or hold the CLR_CMOS button that these high-end boards probably had.

Two BIOS chips was supposed to protect you from bad BIOS updates or a bad BIOS mod, just like the scenario from TFA.

Original author here, and I agree. When overclocking, the most important feature is probably resetting the CMOS with one easy to reach button.

I was testing ECC previously in https://qt.ax/ecc and was trying to overclock the RAM to induce faults to confirm ECC was fixing them. Since I was using a cheaper board, I had to use keys to short the CLR_CMOS header and that got old really quick... I didn't ultimately succeed in creating a barely stable overclock that generated a ton of ECC errors, but that was a very awful experience.

Having some sort of BIOS recovery is definitely the baseline I'd expect these days, and having two BIOS chips so you can just keep using the device as if nothing had happened is the best solution.

If you need to do that again, maybe try turning the tREFI DRAM timing to the moon? That controls the refresh interval, which is very temperature sensitive. You could configure a marginally stable setting, let it boot, start running `memtester`, and then apply hair dryer directly to DRAM.

Possible they're thinking of certain OC editions of GPUs. Some of them have 2 BIOSes you can switch between with a little slider. Some even 3.

Isn’t that just a gimmick?

I was under the impression that GPU overclocking, overvoltage and fan curve settings were not persistent across reboots.

It's not a gimmick when the serious overclocking people do custom VBIOSes. Similarly for miners. That's when you need the recovery switch if things go wrong.

TIL about custom VBIOSes.

Why don’t the OEMs of GPUs aimed at OCing just ship with a BIOS that exposes more tweaks through software though…

Custom Video Card firmware has been around a long time. It was possible to convert PC video cards to Mac use by flashing them. When I was faffing around with PPC64 Macs (I mean hey I got it for free) I picked up an ATI card on ebay that had been flashed.

I had sleep issues. It was the lowest-end G5. I ended up recycling it.

Reply on news.ycombinator.com