Framework 13 AMD Ryzen 7040 BIOS 3.03b

Framework team - you should have replied to all tickets mentioning BSODs asking for customers to update. This would have saved me a lot of time.

@Kieran_Levin A suggestion: put these firmware posts in a specific forum or category, so people can subscribe to get emails when updates are released.

10 Likes

I like this.

2 Likes

Or unpin the welcome threads and perma pin the firmware one instead

1 Like

QuickCharge 4 is USB PD as far as I know.

QuickCharge 3 twiddles voltages on two of the 4 wires inside a “standard” USB cables (I think???)

Quick Charge 3.0 circuit – EmbedBlog seems more complicated than I thought it was.

AliExpress had some QC to PD boards, but I’d want to put them through pretty heavy testing before connecting them to an expensive Framework laptop.

1 Like

And if Discourse allows this, make them announce-only? That way they can be watched without chatter spamming one’s email.

This is still a Beta. Framework should not be telling everyone with a problem to install a beta BIOS until they know it is safe to do so.

1 Like

I was about to RMA the laptop - I think both parties would be willing to roll the dice here

Turns out rebooting resolved my issues.

Well, seems like supporting Qualcomm’s QuickCharge 2.0 might be possible? The twiddling of voltages is all withing the safe <3.3V of USB data lines, so given the frameworks USB controller supports this control, QC 2.0 support/compatibility should be possible?

As for QC 3.0, it seems to be a proper protocol instead of pulling some voltages on certain lines, so dunno how that entirely works… Though maybe thats not even required, if QC 3.0 is backward compatible with 2.0? Is it?

On the other hand, QC 4.0 seems to be a cross compatible variant with USB PD, so not entirely the same, but compatible just fine. So yeah, they should be able to charge a Framework alright, tho I don’t have a QC 4.0 charger at hand. Can anyone confirm this?

Framework laptops do not support QC 2.0. Or any charging algorithm that requires charge negotiation over the usb 2.0 data lines.

This requires special hardware on the usb data lines which we do not use in our design. Except for the framework 12th gen Intel Chromebook, which can support QC.

For laptops that support dumb chargers, we only support 5V@900mA charging.

3 Likes

@Alex_Shpilkin - Any chance you can elaborate on this more?

I took this to mean amd-ucode-20240312.3b128b60-1 at https://gitlab.com/kernel-firmware/linux-firmware/-/tree/main/amd-ucode?ref_type=heads did not include 7840U microcode, but that link dumps you into the 19h AMD Zen 4 microcode repository.

Similarly, amd-zen-ucode-platomav on Arch @ GitHub - platomav/CPUMicrocodes: Intel, AMD, VIA & Freescale CPU Microcode Repositories also contains 19h firmware.

Are you saying these are both server CPUs only, even though they indicate 19H / Zen4?

I hope openSIL is gonna take care of that in the future.

The password expiring on update or the password expiring at all(and having stupid complexity rules) one?

The protocol is still on the usb data lines.

Do you actually have a qc4 charger that doesn’t also have regular pd (or do you know of one)?

Pretty much all the qc3 chargers/powerbanks I have also support PD except for like 1 ancient usb-a only one.

Can we get propper usb 5v charging without kickstarting though?

1 Like

I’m having this problem with on my NixOS install with this BIOS installed:

Is this BIOS related?

New firmware installed fine, and I can confirm that after the update now my AMD FW13 charges with a 47W Anker charger, that didn’t work with last BIOS.

3 Likes

Circling back @Alex_Shpilkin

The gitlab I linked previously for the ucode links to a python script on github which will tell you if any of the patches are for your cpu, and which ones. I’ve listed them below for the 7840U.

Running the python script on my machine for the already loaded firmware:

[user@system]# python amd_ucode_info.py /lib/firmware/amd-ucode/microcode_amd_fam19h.bin -v
Microcode patches in /lib/firmware/amd-ucode/microcode_amd_fam19h.bin:
  Family=0x19 Model=0x11 Stepping=0x02: Patch=0x0a101244 Length=5568 bytes Start=148 bytes Date=2023-09-11 Equiv_id=0xa112
  Family=0x19 Model=0x01 Stepping=0x02: Patch=0x0a001236 Length=5568 bytes Start=5724 bytes Date=2023-08-31 Equiv_id=0xa012
  Family=0x19 Model=0x01 Stepping=0x00: Patch=0x0a001079 Length=5568 bytes Start=11300 bytes Date=2023-06-09 Equiv_id=0xa010
  Family=0x19 Model=0x01 Stepping=0x01: Patch=0x0a0011d3 Length=5568 bytes Start=16876 bytes Date=2023-08-23 Equiv_id=0xa011
  Family=0x19 Model=0xa0 Stepping=0x02: Patch=0x0aa00213 Length=5568 bytes Start=22452 bytes Date=2023-09-11 Equiv_id=0xaa02
  Family=0x19 Model=0xa0 Stepping=0x01: Patch=0x0aa00116 Length=5568 bytes Start=28028 bytes Date=2023-06-19 Equiv_id=0xaa01
  Family=0x19 Model=0x11 Stepping=0x01: Patch=0x0a101144 Length=5568 bytes Start=33604 bytes Date=2023-09-06 Equiv_id=0xa111

So, the microcode updates definitely do cover our CPU, and I count 7 microcode patches.

I just wanted to add, last time I looked at this the model doesn’t match most consumer CPUs (like @Alex_Shpilkin said), but I’m happy to be wrong.

Also when you do cat /proc/cpuinfo does the microcode line up with any of those versions? It doesn’t for me, but I may have broken it by removing my initramfs, which was why I looked into this before.

1 Like

Oh sheesh, I completely misunderstood earlier. I’ll have to dig into this another time and see where I get, or perhaps reach out to AMD / folks more knowledgeable. I see 19H covers Zen Zen 3, Zen3+, Zen4, including ryzen desktops, workstation cpus, and epyc cpus.

To your point, I haven’t found any proof that 7840/7640U have any microcode patches for them at all yet.

I appreciate you adding your input as I work to understand it better.

3 Likes

This BIOS update has been a huge W for me, I kept getting random crashes and then the laptop wouldn’t POST after (often when I had a charger plugged in) and that seems to be resolved. It occasionally won’t power on and I get a weird POST code on shutdown, but its much less common.

5 Likes