new laptop's charger inadequate for GIMPS load?

[kriesel]

Has anyone else run into a situation where a laptop discharges the battery and shuts down while plugged in and charging and running GIMPS work?

In the midst of accessing a new laptop remotely, the session dropped. Investigating locally, laptop had shut down due to drained battery. This is my newest hardware, less than 2 months old. On resume, it was at 6% battery charge, discharging. Unplug, replug, 5% charging. While I watch, it drops to 4% and charging, and shuts down with low battery warning. It pops messages to plug it in. But it's already plugged in when that happens.
So I've deactivated the main power usages on the laptop, prime95 on the cpu cores and mfaktc instances on the gtx1050Ti in it, left mfakto on the igp running, and it's climbed back to 16% over several minutes with nearly 2 hours of charging to go. Then resume prime95 and an mfaktc instance; in a few minutes it has slipped back to 15% charge. Pause prime95, and it's SLOWLY creeping back up, 16% charge after minutes.

This system is a Dell G3 3579, Win 10 x64, i7-8750H, 16GB ram, internal GTX1050Ti gpu and uhd630 igp, power plan set to best performance, display operations set to use uhd630 so the GTX1050Ti is free for compute-only.

Apparently the power adapter doesn't always keep up with the load of full GIMPS utilization and a best performance power setting. I'm trying reduced screen power settings first to try to fit the power budget within its steady state capabilities, before paring back number and thereby aggregate performance of GIMPS applications on it. Power demands would increase when expanded from 16 GB current ram configuration to 32GB max memory.

Idle: package power 10W per HWMonitor;
prime95 resumed, 45W (didn't change noticeably when mfakto on uhd630 resumed)
GTX1050Ti laptop power 40-64W per https://www.nvidia.com/en-us/geforce...ries/laptops/# (other sources say around 70)
power adapter nominal rating on its case is 130W

With one mfaktc instance on the GTX1050Ti, I could see the battery charge slip away at 15mwh/24 seconds in HWMonitor. That's a discharge rate of 2.25w, 37mwh/minute, or 1493 minutes (24.9 hours) to discharge from full charge the 56WH battery. But it's variable. It seems for now to have stabilized at 59% battery charge, treading water at 33212 mwh charge, number not changing for several minutes, with full GIMPS workload resumed.

[Madpoo]

Originally Posted by kriesel

Has anyone else run into a situation where a laptop discharges the battery and shuts down while plugged in and charging and running GIMPS work?

In the midst of accessing a new laptop remotely, the session dropped. Investigating locally, laptop had shut down due to drained battery. This is my newest hardware, less than 2 months old. On resume, it was at 6% battery charge, discharging. Unplug, replug, 5% charging. While I watch, it drops to 4% and charging, and shuts down with low battery warning. It pops messages to plug it in. But it's already plugged in when that happens.
So I've deactivated the main power usages on the laptop, prime95 on the cpu cores and mfaktc instances on the gtx1050Ti in it, left mfakto on the igp running, and it's climbed back to 16% over several minutes with nearly 2 hours of charging to go. Then resume prime95 and an mfaktc instance; in a few minutes it has slipped back to 15% charge. Pause prime95, and it's SLOWLY creeping back up, 16% charge after minutes.

This system is a Dell G3 3579, Win 10 x64, i7-8750H, 16GB ram, internal GTX1050Ti gpu and uhd630 igp, power plan set to best performance, display operations set to use uhd630 so the GTX1050Ti is free for compute-only.

Apparently the power adapter doesn't always keep up with the load of full GIMPS utilization and a best performance power setting. I'm trying reduced screen power settings first to try to fit the power budget within its steady state capabilities, before paring back number and thereby aggregate performance of GIMPS applications on it. Power demands would increase when expanded from 16 GB current ram configuration to 32GB max memory.

Idle: package power 10W per HWMonitor;
prime95 resumed, 45W (didn't change noticeably when mfakto on uhd630 resumed)
GTX1050Ti laptop power 40-64W per https://www.nvidia.com/en-us/geforce...ries/laptops/# (other sources say around 70)
power adapter nominal rating on its case is 130W

With one mfaktc instance on the GTX1050Ti, I could see the battery charge slip away at 15mwh/24 seconds in HWMonitor. That's a discharge rate of 2.25w, 37mwh/minute, or 1493 minutes (24.9 hours) to discharge from full charge the 56WH battery. But it's variable. It seems for now to have stabilized at 59% battery charge, treading water at 33212 mwh charge, number not changing for several minutes, with full GIMPS workload resumed.

I've been dealing with a bunch of Dell Latitudes lately... I don't remember the details but their newer models do have a bunch of options related to battery charging in the BIOS setup.

I had one laptop that was doing something similar even though it was just sitting there not doing anything except showing a logon screen. No idea what was tweaked in the settings but I reset to default and it started charging normally again.

Anyway, I know I'm light on any details but that might be a good place to check - all of that "smart charging" stuff if your laptop has them.

[kriesel]

Originally Posted by Madpoo

I've been dealing with a bunch of Dell Latitudes lately... I don't remember the details but their newer models do have a bunch of options related to battery charging in the BIOS setup.

I had one laptop that was doing something similar even though it was just sitting there not doing anything except showing a logon screen. No idea what was tweaked in the settings but I reset to default and it started charging normally again.

Anyway, I know I'm light on any details but that might be a good place to check - all of that "smart charging" stuff if your laptop has them.

I had quite a battle getting the same laptop in post 1 to not shut down overnight initially, when it was new. It seemed like it was not taking "never" seriously on some power settings, like when to shut down, sleep, hibernate. There are more than enough switches, sliders, etc in Windows power settings to keep a user entertained/distracted/unproductive. Haven't looked in the BIOS for more, lately. In retrospect, maybe it was power adapter inadequacy from early on. I've been through the usual and advanced power settings on this quirky machine many times. I now have the screen brightness dialed down below 30% when in use, 20% fallback.It's regained <1% in 20 minutes on battery charge. I started puzzling over the emptied battery about 90 minutes before the first post in the thread was submitted. Now 33455mwh 59% in HWMonitor, while the little system tray applet calls it 60% and projects 51 minutes to complete charge (that would be if it was idle, and it's not); 20 minutes ago it said 52 minutes.
I avoid restarting systems as much as practical, but will keep a BIOS battery charge related deep dive in mind for next time. Thanks.
(Now where did my KillaWatts get to, I want to look at the power adapter AC plug input wattage draw, PF, and line voltage.)

[M344587487]

I've had a laptop bluescreen because the transformer couldn't keep up, it kept fully discharging every second or so causing the laptop to constantly cycle between "on battery" and "on mains". I fixed it by disabling the nvidia GPU entirely, you might be able to get away with a slight underclock and/or undervolt of the GPU.

[kriesel]

In the past hour, battery charge has crept up by 3%; it's now indicating "47 minutes to fully charged" (instead of 51 minutes an hour ago). Whoever programs those battery charge projections must be a failed meteorologist. (The forecast for later, possibly overnight, is dark...)


Next possible project is to revive or pronounce terminal an old laptop with imminent HD failure.

[aurashift]

Grab a kill-a-watt and see what the laptop is drawing vs what the power supply supplies. Also check drivers. My system76 bonobo is weird about power in windows until drivers get installed.

[kriesel]

I futzed with lots of power settings, and as I recall dimming the screen and other tweaks, seem to have gotten the total power down below what the charger can dish. it is the biggest laptop charger I own, nominally 130W, runs hots, and is feeding 6 cpu cores running prime95, a UHD630 running mfakto, and a gtx1050ti running mfaktc. It's stood up for weeks now. Thanks for the suggestions.

An insertable line power meter is on order. (FYI those things have gotten very inexpensive.)

[dcheuk]

Originally Posted by kriesel

Has anyone else run into a situation where a laptop discharges the battery and shuts down while plugged in and charging and running GIMPS work?

In the midst of accessing a new laptop remotely, the session dropped. Investigating locally, laptop had shut down due to drained battery. This is my newest hardware, less than 2 months old. On resume, it was at 6% battery charge, discharging. Unplug, replug, 5% charging. While I watch, it drops to 4% and charging, and shuts down with low battery warning. It pops messages to plug it in. But it's already plugged in when that happens.
So I've deactivated the main power usages on the laptop, prime95 on the cpu cores and mfaktc instances on the gtx1050Ti in it, left mfakto on the igp running, and it's climbed back to 16% over several minutes with nearly 2 hours of charging to go. Then resume prime95 and an mfaktc instance; in a few minutes it has slipped back to 15% charge. Pause prime95, and it's SLOWLY creeping back up, 16% charge after minutes.

This system is a Dell G3 3579, Win 10 x64, i7-8750H, 16GB ram, internal GTX1050Ti gpu and uhd630 igp, power plan set to best performance, display operations set to use uhd630 so the GTX1050Ti is free for compute-only.

Apparently the power adapter doesn't always keep up with the load of full GIMPS utilization and a best performance power setting. I'm trying reduced screen power settings first to try to fit the power budget within its steady state capabilities, before paring back number and thereby aggregate performance of GIMPS applications on it. Power demands would increase when expanded from 16 GB current ram configuration to 32GB max memory.

Idle: package power 10W per HWMonitor;
prime95 resumed, 45W (didn't change noticeably when mfakto on uhd630 resumed)
GTX1050Ti laptop power 40-64W per https://www.nvidia.com/en-us/geforce...ries/laptops/# (other sources say around 70)
power adapter nominal rating on its case is 130W

With one mfaktc instance on the GTX1050Ti, I could see the battery charge slip away at 15mwh/24 seconds in HWMonitor. That's a discharge rate of 2.25w, 37mwh/minute, or 1493 minutes (24.9 hours) to discharge from full charge the 56WH battery. But it's variable. It seems for now to have stabilized at 59% battery charge, treading water at 33212 mwh charge, number not changing for several minutes, with full GIMPS workload resumed.

Hello!

I have a laptop from MSI, running i7-8750h too, ,1060 6gb graphics, 32gb ddr4 2666 memory, m.2 ssd.

I have the laptop running on 6/6 cores for LL test and mfaktc at the same time. My battery has always been at 94% all the time, and Windows would always say "fully charged." I tried unplug the power and plug it back in, restart it, it remained the same. So I just ignore it now, since the laptop just always sits there and I never really uses it.

The temperature hovers around 75-80C, sometimes hit as high as 95C (maximum in HWmonitor, yeah it's not good lol I don't know what to do).

Software Version	Windows64,Prime95,v29.4,build 8
Model	Intel(R) Core(TM) i7-8750H CPU @ 2.20GHz
Features	6 core, hyperthreaded, Prefetch,SSE,SSE2,SSE4,AVX,AVX2,FMA,
Speed	3.881 GHz (17.295 GHz P4 effective equivalent)
L1/L2 Cache	32 / 1536 KB
Computer Memory	32615 MB   configured usage 24576 MB day / 24576 MB night
Reliability, Confidence	0.98, 10.0  Reset, I fixed the hardware
Status	 Trusted software version
Hours per day	24 hours    (day memory use starts 07:30 and ends 23:30)
Work Preference	CPU/Work thread 1: 
DC instead of
LL percentage*	
0
 (in 1% increments up to 99%)
Workload Cache	
3
 days
Email options	 Send an email if computer is 
1
 hours late contacting the server
 Send an email if computer returns a suspicious LL result
GHz-days credited	4099.210 GHz-days (in the last 365 days)
Work units completed	22 (in the last 365 days)
Work units assigned	3
CPU rolling average	1285 / 1000 (128%)   Execution priority 1 (default)
Benchmarks	1 recorded, most recent 2019-01-03 04:45

[kriesel]

Originally Posted by dcheuk

Hello!

I have a laptop from MSI, running i7-8750h too, ,1060 6gb graphics, 32gb ddr4 2666 memory, m.2 ssd.
...
The temperature hovers around 75-80C, sometimes hit as high as 95C (maximum in HWmonitor, yeah it's not good lol I don't know what to do).

Nice system. The gtx1060 6gb would outrun the gtx1050ti and uhd630 combined. Lots of system ram is good too. If you want to cool it down a bit you could use prime95 throttle setting in prime.txt. Which would give the charger a bit of a reprieve too.

[aurashift]

Originally Posted by dcheuk

I have the laptop running on 6/6 cores for LL test and mfaktc at the same time. My battery has always been at 94% all the time, and Windows would always say "fully charged."[/CODE]

My best guess is that since the battery is probably in-line acting somewhat as a UPS, that the voltage from the battery drops just enough under full load to give you a readout of 94%. I bet it goes back to 100 as soon as it isn't under load. It probably isn't a problem as long as it doesn't drain your battery.


Also, my laptop has 2x300w power supplies for gtx1080 SLI.... they're both the size of a novel. each. bleh.

[retina]

Originally Posted by aurashift

My best guess is that since the battery is probably in-line acting somewhat as a UPS, that the voltage from the battery drops just enough under full load to give you a readout of 94%. I bet it goes back to 100 as soon as it isn't under load. It probably isn't a problem as long as it doesn't drain your battery.

It's usually more complicated than that. The batteries have "fuel gauge" chips in them that count the electrons going in and coming out and give an accumulated state of charge value to the system.

Those fuel gauge chips require regular recalibrating to reset the the offset and max values. It is common for them to show bad values IME.

The voltage level is only used for charge termination and supply cut-off. That is why you can sometimes find the system still runs for some time after it has reached 0.0%, or prematurely cuts-off at >0%.

So the 94% reading just means it needs to recalibrate the fuel gauge to account for the battery decline.