Ryzen 7 7840HS is not running at full speed

[Neurrone]

> But now I have 1.x watts when idle but up to 5-7 watts when I am doing some little things.

Is that total battery draw? Or CPU Package power?

I believe the problem under Linux is that you can reduce the CPU package power, but the discharge rate is still high.

[Jy_D]

Cpu package draw. It is impossible that total laptop has only 1 watts.

[Jy_D]

I think that either you have config problem or your laptop has hardware problem. I know that elitebook 845 g10 is great at battery time.

[Neurrone]

> Cpu package draw. It is impossible that total laptop has only 1 watts.

Good point. What about actual discharge rate then?

> I think that either you have config problem or your laptop has hardware problem. I know that elitebook 845 g10 is great at battery time.

I guess it depends on what the definition of great is. I'm not satisfied with getting only 6-7 hours though (light activity like email and web browsing, low power screen, 30% brightness). It could easily be 2 hours longer , based on CPU package power of the same processor from other non HP laptops. This is also an expensive laptop, which raises my expectations accordingly.

[TzortzisG]

As already pointed out on Reddit, and as promised by Inthemirror on 08-29-2023, a new BIOS version has replaced the previous buggy firmware updates:

HP BIOS and System Firmware (V82) VERSION: 01.03.02 REV: A (info) 

BIOS VERSION: 01.03.02 (softpaq link) 

[Neurrone]

The latest bios posted doesn't fix the absurd power draw at idle unfortunately.

@Inthemirror could you paste the business number that you used?

I tried talking to the consumer support in the US, but they refused to work with me and redirected me to the consumer line for where I am.

[inthemirror]

I can now confirm the issue has been solved. It's been almost four months.

Maybe I'm lucky but my CPU now boosts up to 5137MHz. Single threaded performance has increased (or should I say "restored") by more than 10%. Multi-threaded performance strangely has slightly increased as well.

Check this GeekBench 5 result under Linux. Your Windows results will be lower because GB loves Linux for some reasons.

GB6 sucks but here's my results for it with the new BIOS. Overall this tiny CPU now runs faster ... than the desktop Ryzen 7 5800X with around 142W TDP (AMD has been openly lying about its desktop CPUs power consumption since Zen 1).

Idle power consumption has not been addressed but at least under Linux it's possible to fix it.

[TzortzisG]

Yes, it was mentioned on Reddit -unfortunately. However many users are interested on the max frequency bug issue which has finally been fixed.

Idle power draw is basically affected by Cstates. Meaning CPU time on the high Cstates equals cores shutting down various parts and power draw plummeting. On Intel CPUs modern standby lets the cores spend time on C10, and the power draw can be as low as 0.1W. AMD CPUs have slightly different Cstates, but the principle is the same. HP's badly written UEFI firmware code often causes problems with full Cstate use. Even worse is the fact that HP is one of those laptop manufacturers that started selling laptops or retrofitting older laptops with BIOS versions that removed S3. This was around the time Intel TigerLake CPUs came out (which lack S3 on a design level) and as a result no modern AMD HP laptops support S3 either, only the "better" Modern Standby......

Have you tried Project-sbc 's Handheld-Control-Panel to change EPP and other CPU settings? Maybe you can reduce power draw by forcing the AMD cores to prefer lower frequencies when idle. Although the true fix will of course have to be a BIOS update.

[Neurrone]

HP has been pushing newer bios than what they offer on their website via Windows Update. I got my unit in Aug and was updated to the 24th Aug bios which fixed the max frequency bug for me.

> Maybe you can reduce power draw by forcing the AMD cores to prefer lower frequencies when idle. Although the true fix will of course have to be a BIOS update

Which settings would be able to do that? My impression was that I could only change TDP, which doesn't do anything for idle. I've been obsessively trying to reduce this because most of the time between keystrokes or mouse movements, most things should be powered off to save battery life.

If I can't easily solve this or if HP doesn't fix it, then I'll be returning the laptop and asking for a full refund. I hate to do so though because its disappointing to have this great machine ruined by awful firmware. I'm not willing to tolerate 2 hours less battery life though due to sloppy coding on a laptop at this price point.

I'm also not sure whether this ASPM issue is universal across all of HP's phoenix notebooks. If so, perhaps some units are hit more than others. For example, I don't think that the review units on Notebookcheck are affected, since it is mathematically impossible for my unit to reach anywhere close to those battery life results even at idle.

[TzortzisG]

I don't trust the Notebookcheck reviews 100% anymore. I've commented about false and misleading information many times this last year on their pages. I mostly use Laptopmedia and some Asia-based review sites nowadays.

I was talking about the EPP, which you can change through hidden windows settings as well. I'm pretty sure inthemirror 's Linux patch doesn't really fix the BIOS problem, but activates some hidden CPU power settings to force the package to lower levels. There are many hidden CPU settings in any OS, like fine-tuning core parking and many many others.....

RyzenAdj "power-saving" explained 

Returning a laptop that you find unsatisfactory makes complete sense. HP will eventually fix this as well, especially if users keep nagging about it on the HP support lines. The real question is.......how soon.