Can a 7 days full 100% CPU load "burn-in" / "stress test" damage a modern notebook?

Question

When I buy new computers, I usually do a "burn-in" with 100% CPU load for, say, seven days, in a well ventilated place.

This is to

• find out if the computer has problems before I invest time into setting it up for me

• let it "steam out" some of the "plastic/chemical gasses" before I use it in my home

But, now I'm wondering, am I creating hardware problems by doing this? Are modern notebooks built to withstand this?

I guess a server surely is, but is a notebook? I'm talking one from a "well" known brand as Apple, Lenovo, HP, ASUS, Dell, etc.

So I did a burn-in for a new notebook recently. It went fine and didn't crash the OS even once. The notebook is of a model that doesn't get that very hot.

But after some weeks the hardware seems to "give up", recently the SSD has started to act up.

So, can a seven days full 100% CPU load "burn-in" / "stress test" damage a modern notebook??

How would you do a "burn-in" of a notebook? I guess one could put a fan next to it, but then the temperature would not get that high, and huge part of the reason for burn-in would disappear.

---

Thank you all for answers!

But reading them, I really wonder where "computing" is today. Many say that this seven days 100% CPU load will be damaging (/wear out early) to the computer.

From my computing background (1990- ) this sounds really strange - though I'm not saying it is wrong.

What if you have some encoding work to do? Or some 3D-rendering work? Those can easily run for three days, 100% CPU load. Shouldn't notebooks be used for those things?

I can accept that some people are skeptical about my worries about "toxic fumes"; you want to see proof. OK. But really, if one can't use a new notebook at 100% CPU for some days, I really would like to see some disclaimers from computer manufacturers, or some kind of admissions from them, that this is the case.

"This notebook is not suitable for 3D rendering." LOL!!!

---

Added 22 Oct 2014: I took a look at the SSD using some diagnostics software. It had information on the temperatures:

• Highest temperature recorded ever was around 42 °C

• Highest temperature recommended was circa 70 °C

• Shutdown temperature was circa 75 °C

Conclusion: My seven days 100% CPU burn-in did not create a high temperature in the SSD, and it kept very good margin to the given max. Remember that I had the notebook in a room that never had an air temperature over, say, 26 °C.

Solution 1

Can damage? Yes. Should damage? Usually not. It will reduce the lifetime though, and I do not see any benefit of doing such an extreme burn-in for workstations or even laptops - you'd get nearly all of the benefits by running the same tests for 10-30 minutes, instead of 150 hours.

When you use a device, there's wear and tear. Lifetime of components is not infinite - many components die after a few years of normal use (some even after a few months) even though they should last forever.

There are some things to consider in your case:

• Some Notebooks are quite simply not designed to run continuously at maximum load for ~150 hours straight. For most laptops, the cooling can not keep up with running the system at max load, so when the temperature increases the system reduces max speed to protect itself from overheating. That means the system will run at full speed, get hot, throttle, get cooler, release the throttle, get hot, throttle, get cooler, et cetera. The frequency and the severity of these temperature changes depend on the laptop. This will shorten the lifetime of components.
• Anything with moving parts, such as hard disks, will degrade by being used.
• SSDs, despite having no moving parts, have limited lifetime based on how many writes are made on the disk.

Engineering always involves compromises. When designing a PC, one goal is to keep costs for production down while also keeping costs for warranty repairs/replacements down. This involves designing for the most likely usage scenarios. Unlike laptops, servers can be expected to run under load 24/7, which is why they are designed to last longer than consumer desktops and laptops. That increases their cost (just compare prices of server hard disks to those of consumer hard disks).

Solution 2

There are several products which will do a better and more efficient job of stress testing new hardware than simply running the CPU. Overclockers use these tools to test out how much they can overclock various pieces of hardware before the system becomes unstable. They do not run the software for a week, a few minutes will usually reveal a problem, maybe an hour if you want to really be sure.

• PassMark BurnInTest
• Prime95
• NovaBench
• 3DMark
• PCMark

I looked up some information on the brominated flame retardants the OP mentions, particularly Polybrominated Diphenyl Ethers (PBDEs), and there is concern about their bio accumulation. The CDC has a nice fact sheet about them. Note that the section entitled "What are the routes of exposure and the health effects of PBDE and PBB?" does not mention vapors from electronics. It does state that "PBDEs may enter the environment through emissions from manufacturing processes, volatilization from various products that contain PBDEs, recycling wastes and leach from waste disposal sites" citing ATSDR 2004. ATSDR's fact sheet on PBDEs in "How might I be exposed to PBDEs?" states these ways you might be exposed to PBDEs.

• The concentrations of PBDEs in human blood, breast milk, and body fat indicate that most people are exposed to low levels of PBDEs.
• You may be exposed to PBDEs from eating foods or breathing air contaminated with PBDEs.
• Workers involved in the manufacture of PBDEs or products that contain PBDEs may be exposed to higher levels than usual.
• Occupational exposure can also occur in people who work in enclosed spaces where PBDE-containing products are repaired or recycled.

In short, unless you're working directly with PBDEs and eWaste you're not going to be exposed any more than you already are.

Be sure not to confuse PBBs, which have been shown to be probably cancerous and have been banned since the 70s, with PBDEs which are still being studied and have only been partially banned.

As for "burning off" these problems, I would need a citation on that from a reputable study that A) it's even a problem to begin with and B) that running a laptop hot does anything to solve the problem. As you can see in the CDC fact sheet, the boiling point for all PBDEs is over 300C. Your CPU is not going to get much over 80C.

UPDATE @Fuhrmanator found an actual study about the chemicals coming out of computers and printers! Before you freak out about all the pretty graphs and scary chemical names, note that this is one study on a few computers and printers which is a sample size too small to draw general conclusions.

The study's conclusions about having computers and printers in a small (20m^3) room is "it's probably ok, but might accumulate, and we need more data". The relevant points are...

• "In almost all cases the computer and printer emissions give rise to concentrations well below guideline levels"
• "Formaldehyde emissions from computers and dibutylphthalate emissions from printers are possible exceptions—emissions are estimated to come close to or exceed the guideline limits"

Also note that printers are way worse than computers, laser printers worst of all.

And the report does conclude that "'Aging' of computers led to a reduction in chemical emissions". While it doesn't say anything about running at 100% CPU, running it for a couple days in a well ventilated place before putting it into cramped quarters might help.

Solution 3

The 7-day burn-in should not damage your laptop and if it does, it was defective in the first place. My laptop has been computing all sorts of distributed computing projects under BOINC for well over 7 years without any issues, and the CPU temperature has been between 70-80°C the whole time.

In fact, I think there could even be some benefits! Repeated heating and cooling decreases the lifespan of various electronic components and can exacerbate issues with cold solder joints, if there are any. The thermal paste on the CPU could degrade slower when the temperature is high, but stable. Having the CPU fan running at the same speed all the time might also be better for it than cycling it from 10% to 100% as the load changes.

I understand that this is all just anecdotal evidence, but my experience seems to suggest that there is some truth to it.

Edit: To address the question of SSD life, you probably don't need to worry too much. Modern SSDs can survive a huge amount of writes, the seven days of testing are tiny compared to what the disk can handle. The folks at The Tech Report decided to do an experiment and abuse several SSDs until they die. Some even survived over a whopping petabyte of written data.

http://techreport.com/review/27062/the-ssd-endurance-experiment-only-two-remain-after-1-5pb

Solution 4

In response to your edited question, notebook design is the result of compromises between a great many factors: size, weight, battery life, performance, cooling, etc. The average user will not hit 100% CPU usage on a non-gaming laptop for more than a second or two at a time*, so a common tradeoff is to reduce cooling capability in exchange for a thinner computer.

No, your laptop is not meant to be run at 100% CPU for hours on end. Because of other protective measures such as thermal throttling of the CPU this should not result in the computer failing within the warranty period. A warning about not maxing out the CPU likely would reduce sales and won't provide any benefits to the manufacturer, so they won't have one.

*As an example, I run system-monitoring software on my desktop that logs CPU usage. The largest non-gaming CPU spike in the past week lasted less than five seconds.

Solution 5

This is in response to your recent edits.

CPU Issues

Even if I knew the specifications of your system, I could only speculate about your hardware. However, you can hit the Googles and find out if other people have had problems with your laptop model or its constituent parts...so I'll leave that as an exercise for you.

However, there is some general information I can give that's true for both AMD and Intel processors. When AMD or Intel release a new line of processors, they come out with different processors at different clock rates. For example, they might have 1.8 GHz, 2.3 GHz, 2.8 GHz, and 4 GHz. Now, people get the impression that these four types of processors are different, but they're really not. They all go through the exact same manufacturing process. In fact, a 2.3 GHz CPU could have been right behind a 4.0 GHz CPU on the conveyor belt at some point.

What's the difference, then, between the two? Intel tests their processors themselves - their own kind of stress-test - and they test all of the CPUs the same. The resulting frequencies are determined by the stability of the CPU under test. If the CPU can run at 4.0 GHz and remain stable, then Intel sell's it as a 4 GHz processor (in some cases, these are their 'extreme' edition processors, which sell at $1,000+). If the CPU has too many manufacturing defects, then it won't be stable at 4 GHz, but it might be stable at one of the lower frequencies, like 1.8 GHz.

A common mistake among overclockers who learn about this think they can buy the US$300 Intel processor, overclock it to the extreme edition and get more bang for their buck. In all likelihood, they can get away with it for a while, but it will succumb to instability eventually (instability in this case may or may not have anything to do with temperature). For non-overclockers, this has a different implication. If you buy a low-end processor, you're buying a processor that was in the 'not crappy enough to throw away, good enough to make some customers happy, and not good enough to make much profit' bin. If you stress out a low-end processor, it's not going to remain stable for very long. Heating it up will exacerbate its manufacturing defects sooner. Some low-end CPUs have proven to be very robust over the years; most are not so fortunate. It's really a question to ask the folks at Toms Hardware or other online hardware forums where you can have a proper discussion.

SSD Issues

This is even more speculative, since I'm not 100% sure about SSDs. The one thing that's resonating in the thread so far is that SSDs have a maximum life-time of writes. This is simply a constraint on the technology. It was a huge barrier when SSDs were starting out, and software had to be developed that would essentially load-balance writes to SSDs (completely transparent to system users, of course). I don't know the exact numbers, but let's say you can write to each memory location 1,000,000 times. During normal use, this probably means you have years if not decades to go before the physical device degrades. By doing a burn-in test, you're essentially forcing years of work on your physical devices. There's no way for me to tell you how much you degraded the SSD (if at all), but if you aged it even a year or two during that seven day test period, you would see the effects fairly quickly. Regardless of immediate degradation, you're just shooting yourself in the foot by testing your storage so rigorously. Even HDDs regularly fail. Ask a person who runs a lab, and they'll probably even be able to tell you the failure rate of their HDDs off the top of their heads.

As far as temperature goes, 40 °C sounds reasonable, but I don't know anything about SSD temperatures. You should probably search for answers (or ask here on superuser) from people who would know if there are issues with SSDs running for long periods at 40+ °C. SSDs are relatively new technology, and pushing them to their limits is bound to find issues. I'd be willing to bet that any issues you have with your SSD are more likely caused by too many writes rather than temperature. To give you a sense for the oddities of SSDs, just today on Slashdot there was an article from Samsung about SSDs taking a long time to access information that hasn't been touched in a month. On a HDD that would be a non-issue, but for SSDs it apparently is.

View more solutions

Comments

[Admin]

@TomStevens Where did you get the information that laptops need a long burn in period to remove plastic vapors or have toxic vapors in the first place?

[Admin]

Related: Too much processing able to kill MacBook Pro?

[Admin]

CPUs are intended to be run at 100% all day every day. If they couldn't be, then the product would be considered defective.

[Admin]

You should test exactly what your success condition is. If finding out that the notebook cannot withstand 100% CPU for seven days non-stop is going to cause you to return it to the manufacturer (good luck with that) then do the test. If you are prepared to accept a laptop that can do something less than this, then you should test that condition, rather than break it by aplying a load which you don't want it to stand up to.

[Admin]

Right, comments have been cleaned up. Obviously this topic generates a lot of discussion, which is all nice, but really not suited for this Q&A format. We have a nice Super User Chat for that as well.

[Admin]

Absolutely. But in my experience when dealing with cheap notepads it is more on the order of hours, not days (in a normal warm room sitting in a normal non-idea location). But it all depends on the temperature of the room and what it is sitting on. I would never run a laptop/notepad overnight without placing it on a hard solid basement floor, really helps with the overheating to have a gigantic cool heat-sink up against the device.

[Admin]

perhaps whatever software you were using caused unintended I/O or stress tested your SSD as well as the CPU.

[Admin]

What does 100% CPU mean? Different OS's have different definitions of 100% CPU

[Admin]

@Schwern arb.ca.gov/research/seminars/mckone/mckone.pdf shows how much of various types of chemicals come out of PCs. They do have toxic vapors, but the concentrations are probably low enough not to be harmful. See slide 55 for a summary of health risks.

[Isura Nirmal]

"I seriously question that notebooks have a life of one year", he said typing on his 2011 Macbook Pro.

[Peter]

I wasn't the downvoter, but I need to point out there are some inaccuracies in this answer. Laptops usually last 3-6 years, not 1. And in most parts of the world the manufacturer will not replace the device if sending it in for repair is cheaper.

[Gaurav Joseph]

This shoukd have been a comment, i think.

[Substantial]

A laptop may be on the market for about a year. The hardware can survive long past obsolescence. I have a decade-old Gateway M505x still trucking away as a media server.

[Him_Jalpert]

Maybe the warranty would last a year is what you're thinking? Every notebook I've had has lasted far longer than 1 year.

[v.oddou]

You didn't mention OCCT ? OCCT power supply test is the most stringent thing on a machine, and a machine should pass it before calling itself stable. even a little netbook. many machines can't pass this test though. especially some old ATI graphics cards that fried during that test.

[shortstheory]

Don't feed the troll people!

[Tamara Wijsman]

The longer the test, the more certainty about the less lifetime; for most consumers factory diagnostics (or own 15 - 45 min tests) suffice, those that want to be more sure can go over that for high reliability purposes. There are life and time critical matters that even go further than that; but in such cases, the idea is to use an extensively tested high quality component for at most a year and then replace it to avoid to reach the shortened lifespan. That is less than the quarter of a CPU's lifespan of the average consumer; for them a lot of testing can become a waste of time and money... :(

[Jason]

As long as you keep the temperature regulated (which, as you note, could require some external cooling if the laptop isn't up to it itself) then burning in a laptop today should be pretty much the same as burning in hardware 20 years ago. The harddrive (which you shouldn't be burning in, anyway) should actually fare better today than it would have back then.

[cHao]

The Dell laptops i've had tended to get really hot too.

[Daniel R Hicks]

This 6-year-old Sony laptop is still going strong.

[Daniel R Hicks]

(But it is true that everyone should have decent backup strategy, regardless of the age of the box.)

[Shaz]

The key difference between distributed computing and a benchmark/stress test is that the test is potentially using a lot of hard drive resources, in order to find things like peak and average read/write speeds. Distributed computing might use the hard drive, but not a lot. It'll stay inside memory for the most part. Since OP is having problems with an SSD, it seems most likely that his stress test pushed it beyond its max write limits, at least in some portions of the storage.

[Jonathon]

For point one, I think you are being overly optimistic on the quality of heat management some of these super cheap laptops come with. In my experience the parts that fail first, and never work all that well to begin with, are the heat management devices in this $100 notepads. I have had a graphics card overheat and fry an entire motherboard, while never slowing down enough or shutting down.

[JohnEye]

That's a good point, but I don't think this could realistically happen. SSDs use clever load balancing techniques to avoid writing to the same physical sectors over and over again to avoid pushing a specific part of it beyond its limits. Check out the edit I made to the answer, the amount of abuse SSDs can handle is truly impressive. It seems that the concerns about short SSD lifespan are a bit outdated.

[trpt4him]

For most consumers, it is not economical to plan on keeping a laptop or any device for only one year.

[Jason]

Don't be so quick to dismiss. In a previous job as a system builder, we would start memtest86 (looping) before we left for the night, and check the results in the morning. Every so often we'd get a machine that failed a test on say, the 39th pass. We would put it aside and run it again the next night, and it would fail again on some random pass. Even a 1-2 hour test would simply not find this.

[Jason]

Regarding gasses, I can tell you that if you turn on a brand new iMac with your head over it, you will inhale something akin to burning plastic until the fan control kicks in. The next day, not so much. Now I make sure to turn on new iMacs sitting down. :(

[Dan Is Fiddling By Firelight]

As a counter example, I ran a cheap 15" Acer laptop at near 100% load from 2006-2013 (distributed computing); and aside from a fan replacement at the ~5 year mark it was still working well when I gave it to a family member who needed an emergency replacement computer. AFAIK it's still running well today.

[Mark K Cowan]

I would type this from my 1994 Toshiba T2150CDT, but with only 24MB of RAM, no USB, no ethernet, it is a bit hard to get it on the internet.

[user]

1M writes is way off for write endurance of SSDs. 2-3 orders of magnitude less (1k-10k writes) is probably a more reasonable overall guesstimate, but it depends on the exact technology used. This is why wear leveling and write amplification are such big deals with SSDs, even though it's less of a problem in practice than it is often made out to be. With HDDs, write amplification costs performance; with SSDs, it literally costs lifetime.