Anyone experienced this? (Power Throttling A515-43)

Summan6145 said:

pssyche said:

I have the same issue with exact same model. Bought it for my niece as Xmas present, to use primarily for school/homework, but also some light gaming since I saw on YouTube reviews Ryzen 3500U running Fortnite comfortably over 60 fps, but mine barely does over 30 on lowest settings at 720p. I have latest BIOS installed and downloaded latest chipset drivers directly from AMD. I did manage to get better performance by forcing it to run at higher TDP using https://gitlab.com/ryzen-controller-team/ryzen-controller, but that's not something I'm comfortable running all the time since kids are using laptop primarily. It's ridiculous really, paying for hardware and then not be able to use it at full potential...

What settings are you using in Ryzen Controller? I just bought this laptop and im getting the same issue

It's not really an issue to begin with. I was wrong in my assessment.
It turned out that the "high performance" phase was just a turbo window, the APU uses more than 15 watts (~30) at first, that's why the performance is great, but then it goes back down to its rated TDP (15w), ~6 of which is what I can see reported in AIDA64 and HWiNFO64 (CPU cores) and the rest goes to the iGPU. A better monitoring program would be AMD μProf, it shows you the actual power consumption of the APU and PPT slow and fast limits, STAMP limits...etc.

If you want it to stay at that higher performance level using more power, you can always tweak the settings with Ryzen Controller but I wouldn't recommend it.
You see, this laptop is cheap, and Acer probably opted for the lowest grade chokes and ICs (VRM components) that can run this thing for a reasonable amount of time to make the biggest profit, so if you plan on gaming on it while running at 30+ watts, it will probably die (VRMs) much sooner than you'd think. I know I sound like some dead beat support guy by saying this, but it's the truth. That's why in the desktop world, the Z motherboards (overclocking ready) on the intel side for example, are usually equipped with more VRM phases and higher quality components + giant heatsinks to cool them. The VRMs on this laptop (and many other models too) not only lack heatsinks, but have no airflow of any kind.

So conclusion is that there's nothing wrong with the laptop. The only thing that's still puzzling me is how Acer Support didn't have a clue even though I asked for a higher level technician in the email, they just told me to send the laptop in for a repair service. If anyone from Acer Support is reading this, feel free to take notes.

PS: I selected my own response as the answer since there wasn't a problem to begin with. Thanks for everyone who tried to help.

All comments

>>>I want to know how many of you guys have experienced this? Is this behavior normal? Acer Support wouldn't tell me so I want to find out myself.If anyone has the same model even with an Intel CPU, please take 5 minutes of your time and test your CPU under load and report back. If this is programmed in the BIOS, like I believe, it would mean Acer is robbing us of compute power that we've paid for. >>>

Good luck with your survey. I'm staying tuned for all the responses.

Meanwhile, try changing ControlPanel's power plan from balanced to performance mode or performance to balanced mode to see if you find any difference between the two.

Jack E/NJ

JackE said:

>>>I want to know how many of you guys have experienced this? Is this behavior normal? Acer Support wouldn't tell me so I want to find out myself.If anyone has the same model even with an Intel CPU, please take 5 minutes of your time and test your CPU under load and report back. If this is programmed in the BIOS, like I believe, it would mean Acer is robbing us of compute power that we've paid for. >>>

Good luck with your survey. I'm staying tuned for all the responses.

Meanwhile, try changing ControlPanel's power plan from balanced to performance mode or performance to balanced mode to see if you find any difference between the two.

Jack E/NJ

Thanks but I already tweaked the power plan and even added all the hidden options with command lines like cpu performance increase and decrease thresholds...etc to no avail. The Processor state is at 100% at all times when plugged in, cooling policy active, AMD Power Slider and Graphics Power Settings all on maximum performance for both battery and plugged in modes.

Do you have an Aspire laptop? If you do, I'd highly appreciate it if you could test it out with AIDA64's stress test for 3 minutes or any kind of moderately demanding game. Doesn't matter if the CPU is from Intel or AMD.

>>>The Processor state is at 100% at all times when plugged in,>>>

Do you mean in the advanced power plan setting? Or its state according to AIDA64? Are you actually noticing a gaming performance issue? Jack E/NJ

JackE said:

>>>The Processor state is at 100% at all times when plugged in,>>>

Do you mean in the advanced power plan setting? Or its state according to AIDA64? Are you actually noticing a gaming performance issue? Jack E/NJ

No I mean in the power plan settings. Min and Max states are both at 100% plugged in. It makes no difference.

"Are you actually noticing a gaming performance issue?"
Yes absolutely. The GPU starts at around 900-1100Mhz, fps depends on the game for example in Alien Isolation it starts at 55-60 fps. Then it power throttles itself without any apparent reason and the GPU core clocks drop to 400-500Mhz and so does the fps to 30-40. It's a huge difference, it goes from smooth and playable to a laggy mess and what angers me is there's no obvious reason for it to happen. Plugged in to the wall and temps are in check.
It's like buying a car that can do 120 km/h for a few seconds then caps the speed at 20 km/h in a highway. It's a $cam and I don't understand how people are okay with this.

Image: https://us.v-cdn.net/6029997/uploads/editor/7m/ljbiebx540p9.png

Image: https://us.v-cdn.net/6029997/uploads/editor/uk/js9fep7f5fwa.png

Image: https://us.v-cdn.net/6029997/uploads/editor/w8/yl2pxidr7rgl.png

Some screenshots showing what I mean. (A bit old I know I posted them somewhere else)
As you can see, the temperatures are absolutely in check (special setup for the experiment) and it still throttles down.
Would you be okay with something like this?

For those who might argue that this is not a "gaming" laptop, yes I know and agree. Never bought it for gaming to begin with, wasn't at all my intent BUT it includes a Ryzen 5 3500U and I expect the full performance of said CPU. I didn't buy a 3500U Laptop to get the constant performance of a 3200U after a couple seconds.
However poor the performance of the Vega 8 graphics is, I want to be able to utilize it to its full extent, or at least close to that.
If this was some sort of a thermal limit, whereby upon reaching 80c or something, it starts to clock itself down and decrease power little by little, I would've been completely okay with it as it is understandable but this thing was maxing out at 54c, 54 degrees! (see AIDA64's screenshot above) and was plugged in, NO AMOUNT OF THROTTLING SHOULD TAKE PLACE!

What is your BIOS version number? There was an update released last year that could influence this issue one way or another depending on if the issue existed or didn't exist in the factory delivered mainboard. Jack E/NJ

JackE said:

What is your BIOS version number? There was an update released last year that could influence this issue one way or another depending on if the issue existed or didn't exist in the factory delivered mainboard. Jack E/NJ

Laptop came with the 1.05 version, I updated it to the latest which is 1.06.
It's the Insyde H2O Rev5.0.

Did you try 1.06 due this issue or something else? Jack E/NJ

JackE said:

Did you try 1.06 due this issue or something else? Jack E/NJ

No I didn't even know about this issue. First thing I did when I got the laptop is to check that everything is up to date.

I have the same issue with exact same model. Bought it for my niece as Xmas present, to use primarily for school/homework, but also some light gaming since I saw on YouTube reviews Ryzen 3500U running Fortnite comfortably over 60 fps, but mine barely does over 30 on lowest settings at 720p. I have latest BIOS installed and downloaded latest chipset drivers directly from AMD. I did manage to get better performance by forcing it to run at higher TDP using https://gitlab.com/ryzen-controller-team/ryzen-controller, but that's not something I'm comfortable running all the time since kids are using laptop primarily. It's ridiculous really, paying for hardware and then not be able to use it at full potential...

Aye! To both of you! Almost all BIOS updates, unlike device drivers updates, shouldn't be applied unless they specifically address issues the mainboard is having. They carry risks of adverse side effects or worse (brick) Try this. Download and extract the 1.05 zip. Then follow directions directions here for reinstalling the 1.05 exe. https://community.acer.com/en/discussion/comment/611003/#Comment_611003 It also carries a brick risk.

Jack E/NJ

Mine was performing the same on 1.05, actually it was main reason I installed 1.06, hoping to improve performance.

pssychehen Then you should forget 1.05. You should return the machine to the vendor as defective for an exchange or refund ASAP before the vendor's RMA policy period expires. Acer warranty service will take longer.. The level and timing of throttling the OP originally decribed is not normal. Jack E/NJ

pssyche said:

I have the same issue with exact same model. Bought it for my niece as Xmas present, to use primarily for school/homework, but also some light gaming since I saw on YouTube reviews Ryzen 3500U running Fortnite comfortably over 60 fps, but mine barely does over 30 on lowest settings at 720p. I have latest BIOS installed and downloaded latest chipset drivers directly from AMD. I did manage to get better performance by forcing it to run at higher TDP using https://gitlab.com/ryzen-controller-team/ryzen-controller, but that's not something I'm comfortable running all the time since kids are using laptop primarily. It's ridiculous really, paying for hardware and then not be able to use it at full potential...

That's the point, we shouldn't need any external third party tools to solve an issue that shouldn't exist in the first place.
The funny thing is that Acer support pretends to not know what's going on. They asked me to send the laptop in for "repairs" instead of actually addressing the problem and giving an honest explanation.
"It's ridiculous really, paying for hardware and then not be able to use it at full potential."
Absolutely. How do people not regularly complain about this?

JackE said:

Aye! To both of you! Almost all BIOS updates, unlike device drivers updates, shouldn't be applied unless they specifically address issues the mainboard is having. They carry risks of adverse side effects or worse (brick) Try this. Download and extract the 1.05 zip. Then follow directions directions here for reinstalling the 1.05 exe. https://community.acer.com/en/discussion/comment/611003/#Comment_611003 It also carries a brick risk.

Jack E/NJ

The other guy confirmed the behavior is present in the older version too.
"shouldn't be applied unless they specifically address issues the mainboard is having."
That's... what any kind of update is for? I'm not aware of any BIOS updates that don't specifically address issues with the motherboard or any other component. It's not like the engineers get bored and release an update that does nothing.
Even if what you're suggesting is true (which is not), it's not my fault to seek out the latest BIOS version and if that one is bugged then it's high time they fix it and release another update.

PM pssyche to confirm his performed exactly as you originally described with 1.05. If it is and you can't return it to the vendor as defective, then arrange for warranty repair by using the ACER service link at the bottom of this page. Jack E/NJ

I did a bit of stress testing using OCCT, under default settings with CPU running at 100%, it acts exactly as gwynbleidd reported, CPU usage starts at about 10 watts and then after couple minutes drops to 6-7. Clocks max at about 2.4Ghz, temperature 76C. Even when setting TDP to 25 and temp limit to 95C with ryzen controller, I couldn't get it to pull more than 10 watts, and the temperature went well above 90C. So I suspect it's set to throttle on purpose, because of the poor cooling properties.

pssyche said:

I did a bit of stress testing using OCCT, under default settings with CPU running at 100%, it acts exactly as gwynbleidd reported, CPU usage starts at about 10 watts and then after couple minutes drops to 6-7. Clocks max at about 2.4Ghz, temperature 76C. Even when setting TDP to 25 and temp limit to 95C with ryzen controller, I couldn't get it to pull more than 10 watts, and the temperature went well above 90C. So I suspect it's set to throttle on purpose, because of the poor cooling properties.

JackE said:

gwynbleidd
PM pssyche to confirm his performed exactly as you originally described with 1.05. If it is and you can't return it to the vendor as defective, then arrange for warranty repair by using the ACER service link at the bottom of this page. Jack E/NJ

I did some more digging and here's what I found out:
I found a tool named AMD μProf, there are two different values for power namely Socket0 Socket Power and Socket0 Package Power.
I believe the Package power is a measurement of how much power the CPU is pulling excluding the iGPU. Socket power is both combined plus the I/O and all that jazz so that's why it's always higher in value than Package power. I don't know that for sure but it seems like the most probable explanation, so take this info with a grain of salt.
I can also see that there are 3 power limits; STAPM (15 watts), PPT Fast (30 watts) and PPT Slow (25 watts).
It looks to me that the first "boost" period where the performance is great is actually the CPU being ran outside of its "spec" of 15 watts. It's the 25 watts PPT Slow that we were experiencing this whole time and the real (advertised) performance of the chip is when it settles down to 15 watts (aka 6-7 for the CPU and the rest for the GPU and other integrated components). So it looks to me that there's no foul play here, it's all good.
I was able to override those limits using RyzenController and making it apply the settings every 30 seconds but the thing reached 78c with a laptop cooler, so I don't know how viable it really is to run it like that for extended periods of time. (I didn't touch any of the amps settings for fear of blowing up the VRMs since this is a low end laptop).
I need to do more testing, for instance I want to see what happens when I only load the CPU and leave the iGPU idling and vice versa, will keep you posted if you're interested. Also, the GPU core frequency is still a bit odd, it is advertised at 1200Mhz after all and it drops significantly after the boost window so they should really only advertise 400-500Mhz on the Vega 8.

Though this is not the only problem I've encountered. There's another issue when you're on battery, the NVMe SSD's sequential speeds drop down to ~400MB/s instead of ~1500MB/s while plugged in. I posted a thread about this issue here but all I got was the usual "reinstall drivers" that obviously didn't help.
I noticed something else, when on battery, the memory controller also drops in frequency leaving you with a measly 8GB/s read and write on your RAM instead of 34+GB/s and double the latency (~220ns compared to 115ns for my dual channel kit / AIDA64 Cache and Memory Benchmark).
Neither problems can be fixed with RyzenController, so there's absolutely nothing I could do (already tried Power plan settings).

Sure I'm interested. But as I said earlier, the level and timing of throttling that you originally described in your first post doesn't seem normal. Don't let the warranty expire while you're fooling around with this thing. Jack E/NJ