Microsoft Surface Pro 6 and Surface Book 2 Are Throttle Locking To 'Pentium 2 Speeds' of 400 MHz, Users Say (zdnet.com) 84
intensivevocoder writes: Owners of Microsoft's Surface Pro 6 and Surface Book 2 systems are finding themselves stuck at Pentium 2 speeds, as numerous user complaints indicate that the ultra-portables are throttling the processor down to 400 MHz, a state that -- in some instances -- persists across reboots. While similar issues with Surface devices have occurred in the past, reports of issues have increased in frequency following a firmware update for the Surface Pro 6.
The throttle-lock appears to be caused by an Intel CPU flag called BD PROCHOT (bi-directional processor hot), which can be set by any peripheral, telling the processor to throttle down in order to decrease system temperature -- a useful flag in cases where the CPU is operating within thermal limits, but other components tied to the CPU are running too hot, because of the demands placed on other components by processes on the CPU.
The throttle-lock appears to be caused by an Intel CPU flag called BD PROCHOT (bi-directional processor hot), which can be set by any peripheral, telling the processor to throttle down in order to decrease system temperature -- a useful flag in cases where the CPU is operating within thermal limits, but other components tied to the CPU are running too hot, because of the demands placed on other components by processes on the CPU.
Cooling Issue? (Score:5, Informative)
It is my guess that these devices are throttling to prevent overheating. I experienced similar in a desktop (several years back), but was able to add a supplemental fan and the situation improved.
That the issue persists across reboots makes perfect sense: Just because you rebooted doesn't mean you cooled the device... if anything, it only made it warmer.
Re: (Score:3)
Ya, that's exactly what it sounds like.
You'd think that someone along the way would have said "Hey, these run too hot, lets improve the cooling before these ship".
Re: (Score:3)
It's a problem for phones, too, but often it's related to being put into insulating cases.
Re: (Score:1)
Say what a person will about Apple and their prices, you don't find iOS devices that overheat just sitting there, or laptops that blast your legs with heat or need to run excessively noisy cooling fans.
They make sure their hardware has these issues sorted out - without having to throttle down the hardware they're advertising to make sales.
Re: (Score:2, Offtopic)
I'm not sure if you intended sarcasm there on not... but iPhones definitely suffer from thermal issues during normal* operations, and I've seen a fair number of MacBook Pros with uncomfortably-warm aluminum.
I'd rather not focus on particular vendors having this problem or not. Instead, I'd like to see it treated as a user experience problem. When a device is throttled, show an icon or message about it. Sacrifice a little bit of the "embedded device magic" in favor of keeping users informed. Go ahead and mak
Re: (Score:2)
Skrillex would like to have a word with you.
Re: (Score:2)
you don't find iOS devices that overheat just sitting there
Is that why the Surface Pro outperforms the Air despite having identical processors?
Not sure if you're trolling or ignorant but either way your comment couldn't be any sillier. The vast majority of portable devices on the market run in some kind of thermally limited fashion, Apple is not only not excluded from this, they were a fucking pioneer.
Re: (Score:2)
Surely you are joking? The "noisy" fans is what keeps the computer cool so it doesn't have to throttle. Apple's laptops don't have fans like that, and they most definitely overheat and throttle because of it. They also get uncomfortably warm because they use their aluminum cases as a heatsink.
I've also experienced iPhones getting pretty warm when there's a game running, whereas my Android phone stays pretty cool. With that said, I didn't see any evidence the iPhone was throttling back, though battery li
Re: (Score:3)
You'd think that someone along the way would have said "Hey, these run too hot, lets improve the cooling before these ship".
Would that perhaps entail making the devices bulkier, and spoiling their beautiful sleek form factor?
As the beancounters and PHBs have designated design and manufacturing "cost centres", whereas sales and marketing are "profit centres", it's easy to see whose arguments win out.
Re: (Score:2)
I'd also wager its a pretty complex bit of thermodynamics estimating to know if your final design will have sufficient cooling. There's probably a razor margin they're working with between the prototype designs and the ones that come off the assembly lines.
This could just be the case where they had so little working margin that slight variations in manufacturing tolerances change the heat dissipation enough to cause some to get hot.
And it's not just the case/cooling aspect, it could also be that some of th
Re: (Score:2)
Re: (Score:2)
My guess is the design margins of error are just kept too low to meet an external design criteria that's been elevated to some kind immutable status due to politics.
My guess with a lot of consumer electronics the visual/external design of the product is elevated to super high status, possibly even being "designed" by artists/non-technical people and driven by marketing/sales interests.
Combine this with brutal product delivery deadlines and so on, it's probably extremely difficult to engineer it when some ch
Re: (Score:2)
Re: (Score:2)
If the past is any indication, good design might cost money but the return on investment is not apparent to executives.
Exactly my point about profit centres and loss centres.
Re: (Score:1)
I'd also wager its a pretty complex bit of thermodynamics estimating to know if your final design will have sufficient cooling. There's probably a razor margin they're working with between the prototype designs and the ones that come off the assembly lines.
This could just be the case where they had so little working margin that slight variations in manufacturing tolerances change the heat dissipation enough to cause some to get hot.
And it's not just the case/cooling aspect, it could also be that some of the chips they use have variations in thermal footprint they didn't or couldn't easily account for.
I guess the bottom line is it's not an easy task, but then again maybe nobody was willing to challenge the people pushing for the "thinnest, sleekest ever" aesthetic up front and demand some extra margin for cooling.
Wow! I am amazed at the thoughtful responses that are coming out, now that ACs ARE BANNED!!! THANK YOU SLASHDOT for removing the scourge of the mindless ACs!!!
Re: Cooling Issue? (Score:2)
Re: (Score:2)
Re: (Score:2)
Would that perhaps entail making the devices bulkier, and spoiling their beautiful sleek form factor?
I thought we were talking about the Surface devices...
Re: (Score:2)
Ya, that's exactly what it sounds like.
It would be if you didn't look into the problem. The Surface Pro 6 is not new. The Surface Pro 6 throttling to 400MHz however is new.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
It seems like a bug.
The idea is to allow the system to run with less cooling capability than is required to cool all components at full load, because most of the time you aren't using all components at full load. Say you are only using the CPU, it can run at full speed, but if you are using both the CPU and GPU then the CPU might have to throttle back a bit to allow the GPU to be adequately cooled.
Problem is, if one of the peripherals has a bug in the driver or the firmware that gets it's "I'm on fire, plea
Re:Cooling Issue? (Score:4, Interesting)
It is my guess that these devices are throttling to prevent overheating.
My guess is it's a bug due to it becoming a major issue only after the most recent firmware update. All portable Surface devices are thermally limited but none of them in the past would throttle down to this level.
Re: (Score:2)
Pretty sure the surface book uses passive cooling. The entire perimeter of the "tablet" piece is vented.
Re: (Score:2)
Systems are getting to thin! (Score:2)
Systems are getting to thin!
Re: (Score:2)
Re: (Score:2)
From my experience this is not due to overheating, its a firmware bug. I experienced this a few times with a 7th gen intel cpu on a Dell laptop, a reboot would solve the issue. It was not overheating because it was 100% of the time on 0.39GHz, you could sit idle for hours and it would not move a single MHz. Under normal operation the laptop would NEVER reach speeds that low.
hmm,... new ransom ware angle? (Score:2, Interesting)
would be kind of 'funny' if this was a new angle off ransomware,...
Re: (Score:2)
This would be ineffective ransomware, as most user would just grudgingly accept it as the new normal.
Re: (Score:2)
A new recipe for a DoS attack? (Score:2)
The throttle-lock appears to be caused by an Intel CPU flag called BD PROCHOT (bi-directional processor hot), which can be set by any peripheral
It's probably only a matter of time (and it will be likely happen sooner than later) until hackers (black or white hat) figure out how to exploit this.
Re: (Score:2)
The outside AC units at some DC's are in the open and easy to DoS.
Re: (Score:2)
Not that I am aware of.... they're usually under surveillance and multiple layers of security due to the risk of copper thieves wanting to destroy them for scrap.
Same deal as backup generators.... tend to be in fenced off areas monitored by security cameras and often 24x7 guard patrols.
Pentium 2... ugh... (Score:2)
Re: (Score:2)
We're going to party like it's 2005!
Re: (Score:2)
Re: (Score:1)
And the fun thing is, you were a virgin during that whole time!
You still are!
Re: (Score:2)
My 2005 computer was Intel Core Duo 2.16 GHz.
Pentium 2 is more like 1997. A few years later we were at Pentium III and 1.9 GHz; that was in 1999, so no.... Even in 2005 the P2 was considered ancient and dreadfully slow; that was not even enough MHz to run Windows ME, let-alone XP.
Re: (Score:1)
Surface Pros are very temperamental (Score:5, Interesting)
Just yesterday I was setting up 6 of them, all directly out of the box. I laid them all out on a table top, and set them side-by-side to go from one to the other at each operation along the way. Booted them up and started the processes on them all at the same time. It took them all different times to boot completely up. I'm not talking about only a few seconds (though some were only a few seconds behind each other), I mean some were over a minute behind others. During our setup process, I have to plug in a USB drive to copy some data to the Surface Pros. Each of the USB drives are identical, but the completion times varied by over 5 minutes. I even swapped out the USB drives with each Surface Pro in order to test if the problem was with the USB drives, but it didn't really seem to make a difference. I would have liked to have spent more time testing it, but life moves fast.
Re: (Score:3)
Re: (Score:1)
But hey, howabout that packaging, eh!!?? It's the bee's knees.
Re: (Score:2)
Unexpected thermal problems are often caused by individual components not operating within spec, or hot spots within the device, such that the local temperature is higher than spec. Without adequate ventilation, heating becomes highly localized.
Re: (Score:1)
oh yeah, every one of them performs differently. we have far too many at our org and they all do wildly different shit. some people have no issues at all, some people have constant issues. sometimes its docking situations, sometimes its just general use and poor performance.. makes me life a living hell trying to support all of them and their various differences
But, but.. "Mac Book" says they're ALL fast!
Re: (Score:2)
Re: (Score:2)
The Surface Pros are, at least partially, a gimmick. They have 1 USB port, and 1 mini-display port, no other ports. If you want to have additional USB ports, mini-display ports, network or headphone-jack, you'll need to purchase a Surface Dock [microsoft.com] for $200. Also, I hear a lot of folks complain about the $100 keyboard [microsoft.com] that has a lot of issues, as well as the $50 mouse [microsoft.com] that's a challenge too. Besides, $350 is a lot for a keyboard, mouse and a thing to plug them into.
Re: (Score:2)
Wrong, they also have the audio headphone/mic jack, the keyboard connector and the power connector (which doubles as docking station port too). Yes, I'm counting the power connector separately because remember the MacBook with ONE port, YES including the charging port? The I/O on the Surface Pros is pretty decent, if outdated. Whatever you want is there and ALL the adapters you might need are bog-standard somewhere in the single-digit (dollar) pr
Re: (Score:2)
Jesus Christ Bill, is THIS what you're doing with your time these days?
Re: (Score:2)
Re: (Score:1)
That sounds like the thermal solution isn't seated well in the manufacturing process, so you're gambling with how well it worked in any given unit.
I assume you mean that the thermal paste thickness isn't well-controlled during the mfg. process.
Re: (Score:2)
Examine teardown photos. It's not an impressive design and the heat pipes would be easy to slightly misalign.
So...1 star? (Score:1)
And I thought the Megahertz myth was long dead (Score:4, Insightful)
Re: (Score:1)
Doing what? Certainly not in getting work done. Give me 22 years ago any day over a modern PC with a 22 year old clock frequency :-P
Re: (Score:2)
Re: (Score:2)
Not when you remove the heatsink and bolt them into an 8mm thin case.
Re: (Score:2)
With the exception of the legendary Celeron 300A, those CPUs were pretty hard to overclock. The multiplier was locked, so your only option was to increase the bus speed. Since the PCI and AGP bus were also tied to the bus speed, you were overclocking your graphics card, and accessory cards, and your memory. Which is great so long as worked, but you almost always ran into an issue pretty quickly. Also, the Pentium II's separate L2 cache didn't care much for it either, which is why the Celerons actually o
Bad sensors? (Score:4, Informative)
I had a client with a similar problem in an older laptop. It would always run slow and when I looked into it saw that the CPU was always at 100%. After some more digging I discovered that the CPU was throttling down and that BD PROCHOT was the problem. There is a program that lets you turn this off (ThrottleStop) and one I did that the system ran fine (nothing was actually overheating).
Intel is just building $#!7 now... (Score:3)
I'm a huge fan of the Surface series, as I think they're the best designed and thought-out PCs on the planet, but my problems with them due to Intel's continued incompetence has me really wanting to never buy another Intel-powered machine again. Intel definitely does NOT get power management and the firmware required to make it work.
The old Surface Pro 4 (which used the execrable Skylake) was a great example - although I loved the SP4, it was *well over a year* before the firmware was updated so that it could wake again after going to sleep. (Prior to that, sleep was almost always unrecoverable, and required a forced power-off and clean reboot, kind of nullifying the entire idea of "sleep", and guaranteeing the loos of all your work-in-progress every time you closed the keyboard cover to go to your next meeting. I'm a consultant with lots of client and prospect meetings, so this was especially bad...)
Given that 1) ARM CPUs' innovation curve has them getting better much faster than Intel's x86/x64 offerings, 2) The ARM-based chips are far more power-efficient, with lower power int he first place, plus deeper sleep and faster wake times, and 3) ARM is at least somewhat more resistant to many of the heinous hardware security vulnerabilities that we've seen exposed recently, I expect Intel will become an anti-brand for mobile PCs RealSoonNow, if it's not already. (Yes, I would happily buy a next-gen ARM-based Surface Pro if it had the ability to run Win32 apps at a reasonable pace via emulation (say, roughly equivalent to how they ran on the old Surface Pro 3 or 4...)
Re: (Score:2)
as I think they're the best designed and thought-out PCs on the planet
it was *well over a year* before the firmware was updated so that it could wake again after going to sleep
PC users have different expectations than some others, I suppose.
Re: (Score:2)
Both statements are accurate - but the problem was really with Intel's crappy Skylake power management circuitry and firmware code, not Microsoft's design. The problem was made worse because Microsoft believed the lie Intel told them: that Skylake power management was so smart, so awesome, that they could design an "all-day" PC with an even smaller battery than the previous Surface Pro 3. So they did. But then the power management *didn't* work. It took a long time to come up with a fix, but the SP4's b
Re: (Score:2)
"I think they're the best designed and thought-out PCs on the planet,"
By which specific metrics? Obviously performance and serviceability aren't among them.
Re: Intel is just building $#!7 now... (Score:2)
What about AMD? (Score:2)
I don't know the answer, just curious what you think -
How does AMD fit into this? They have been getting good press lately for their x86 implementations and performance. Would AMD processors warrant the same interest and affection as ARM? Is it x86 in general, or just Intel's latest offerings?
Re: (Score:2)
Intel definitely does NOT get power management and the firmware required to make it work.
I hate to break it to you, but this is not an Intel exclusive. The CPU market is now identical to the software market. Ship it and we'll patch it in software. For every Intel bug that has been solved via microcode and firmware, there's an AMD bug with a new emergency shipment of AGESA package shipped to motherboard vendors. Have you seen the Zen 2 launch from AMD? It's still a shitshow with people bouncing between betas of AGESA packages (if their motherboard vendors provide them) trying to get precision bo
Re: (Score:2)
This. People seem to either not know or conveniently forget that for the same performance, an ARM CPU is equally efficient or sometimes LESS efficient than an equivalent Intel or AMD. We view ARM as an efficient chip because most of the workloads for which it's designed are around efficiency. But crank up the clocks to similar speed to a datacenter-class chip and you're in the same or worse performance envelope.
And even then, ARM does some things well but other things it does rather badly. For all the grief
Simple solution (Score:3)
Re: (Score:2)
Cool, sure. But unless you're in a low-humidity environment, it will probably die a very damp death very quickly.
Re: (Score:2)
And then these people won't have to worry about their machine running like a Pentium 2.
My 2018 Macbook Pro does this (Score:2)
This is unfortunately a common issue across shitty hardware manufacturers.
Due to inadequate airflow and cooling Mac's will throttle the CPU to reduce the temperature. Unfortunately with the recent heatwaves in Europe this means that even when doing sweet-fa my Mac becomes stupid slow.
Buying those laptop cooling pads also isn't an option as there are no vents on the bottom of Apple laptops to blow cooler air in to anyway.
I'm contemplating building a desktop so that I can actually get work done.
So? (Score:3)
Re: (Score:2)
A 2018 MacBook Pro you say. Then I suggest you watch this [youtube.com]. Not that you should have to.
Another reason Surface GO is the best Surface (Score:1)
Thermal nightmares (Score:1)