Catch up on stories from the past week (and beyond) at the Slashdot story archive

 


Forgot your password?
Close
typodupeerror
×
Data Storage Earth Science

The Dirty Carbon Secret Behind Solid State Memory Drives (discovermagazine.com) 146

Posted by msmash from the closer-look dept.
Solid state drives use far less power than hard disc drives. But a new study unexpectedly reveals that their lifetime carbon footprint is much higher than their hard disc cousins, raising difficult questions for the computer industry. From a report: The benefits of SSDs over HDDs are legion. They are smaller, mechanically simpler, faster to read and write data than their hard disc cousins. They are also more energy efficient. So with many computer manufacturers and datacenter operators looking to reduce their carbon footprints, it's easy to imagine that all this makes the choice of memory easy. But all is not as it seems, say Swamit Tannu at University of Wisconsin in Madison and Prashant Nair at the University of British Columbia in Vancouver. SSDs have a dirty secret. Tannu and Nair have measured the carbon footprint per gigabyte of these devices across their entire lifetimes and, unexpectedly, it turns out that SSDs are significantly dirtier. "Compared to SSDs, the embodied [carbon] cost of HDDs is at least an order of magnitude lower," say the researchers.

Tannu and Nair come to their conclusion by adding up the amount of carbon emitted throughout the estimated 10-year lifespans of these devices. This includes the carbon emitted during manufacture, during operation, for transportation and for disposal. The carbon emitted during operation is straightforward to calculate. To read and write data, HDDs consume 4.2 Watts versus 1.3W for SSDs. The researchers calculate that a 1 terabyte HDD emits the equivalent of 159 kilograms of carbon dioxide during a 10-year operating lifespan. By comparison, a 1 terabyte SSD emits just 49.2 kg over 10 years. But SSDs are significantly more carbon intensive to manufacture. That's because the chip fabrication facilities for SSDs operate at extreme temperatures and pressures that are energy intensive to maintain. And bigger memories require more chips, which increases the footprint accordingly. All this adds up to a significant carbon footprint for SSD manufacture.

Tannu and Nair calculate that manufacturing a 1 terabyte SSD emits the equivalent of 320 kg of carbon dioxide. By comparison, a similar HDD emits just 40 kg. So the lifetime footprint for a 1 terabyte SSD is 369.2 kg of carbon dioxide equivalent versus 199 kg for an HDD. So HDDs are much cleaner. That's a counterintuitive result with important implications. At the very least, it suggests that computer manufacturers and cloud data storage operators should reconsider the way they use SSDs and HDDs. For example, almost 40 per cent of the carbon footprint of a desktop computer comes from its SSD, compared to just 4 per cent from the CPU and 11 per cent from the GPU.
This discussion has been archived. No new comments can be posted.

The Dirty Carbon Secret Behind Solid State Memory Drives More

The Dirty Carbon Secret Behind Solid State Memory Drives

Comments Filter:

  • OMG! (Score:2, Insightful)

    by rotorbudd ( 1242864 )

    Who gives a Fuck?

    • Re: (Score:2, Funny)

      by Anonymous Coward

      Who gives a Fuck?

      SSDs are demonstrably better, just like HDDs are demonstrably better than storing data on a clay tablet (despite being more Carbon-costly than the clay tablet).
      Just like beef is demonstrably better tasting than crickets, despite being more carbon-costly.
      Use the better stuff.

      • Re: OMG! (Score:5, Insightful)

        by Z00L00K ( 682162 ) on Tuesday August 02, 2022 @08:30PM (#62757530) Homepage Journal

        What's better is depending on the purpose.

        A SSD has saved many man-hours of time just by allowing the computers to be more responsive.

        Boot in less than 20 seconds v.s. boot in 5 minutes - that's a lot of wasted time and emitted CO2 just in wait time.

        Platter HDDs are however still better at storing large amounts of data.

        • Exactly, was wondering if they considering the efficiency improvements which are provided by SSDs.

          If one unit average work takes 1 hour using a HDD and one unit takes 55 minutes using a SSD, over 10 years (as per this study) it adds up. And that means less wear and tear / power usage on the other components (CPU, motherboard, RAM, and so on) for a similar amount of work performed.

          Not to mention savings on manpower, which can be used for other things or to perform more work.

    • > Tannu and Nair calculate that manufacturing a 1 terabyte SSD emits the equivalent of 320 kg of carbon dioxide.
      > By comparison, a similar HDD emits just 40 kg. So the lifetime footprint for a 1 terabyte SSD is 369.2 kg of
      > carbon dioxide equivalent versus 199 kg for an HDD. So HDDs are much cleaner.
      Much cleaner? I produce 1 kg per day by exhaling, so 12 times this amount even if I don't use an SSD.

      What twaddle.

  • Solar Panels (Score:5, Interesting)

    by WarJolt ( 990309 ) on Tuesday August 02, 2022 @02:58PM (#62756538)

    Put a bunch of solar panels on the roof of the TSMC fab they are building in Arizona. That's why the Sunbelt is awesome. Solar works great there. Domestic manufacturing allows us to better control the carbon footprint.

    • Exactly. This article is clickbait. Even if SSDs consume more raw energy, it doesn't have to be from a dirty source.

    • Comment removed (Score:5, Insightful)

      by account_deleted ( 4530225 ) on Tuesday August 02, 2022 @03:39PM (#62756692)
      Comment removed based on user account deletion

      • Re:Solar Panels (Score:4, Insightful)

        by burtosis ( 1124179 ) on Tuesday August 02, 2022 @04:21PM (#62756874)
        Less cars is one thing, but houses will need to be built or serviced, fire trucks need to get to fires, and disabled and elderly should not be confined in mobility. If the solution to a 15 minute drive is to walk 5 minutes to a stop where the bus comes every 30 minutes and then a 35 minute ride with a 5 minute walk to where you need to go it’s not only impractical, it’s downright harmful to society. Poor people taking 5 times longer or more to go anywhere is only making inequality worse. No roads and everyone walk or bike sounds good until the entire place burns down and no one can get anything rebuilt.

        Instead of the whole fk cars and roads mentality, maybe we should cut down by doing things like working from home when a physical presence isn’t required. That alone would make a truly massive impact.

        • It raises a good point that's usually forgotten. And quite honestly, yes, energy can be gotten from green sources, but the key word is "can", in reality we get energy from the cheapest legal source, which isn't always green.

          Depends on what country you live in, but for the US only about 20% of electricity comes from coal (the worst form) and it's dropping. Then there's NG, but solar and wind seems to be halting its growth, mostly. So the prospects are looking better.

          The enthusiasm, for example, for electric cars belies the fact that even outside of possible manufacturing CO2 outputs, there's issues with the fundamental concept of the car itself - the requirements of ever more concrete and tarmac to provide infrastructure for cars in an "everyone is forced to drive" policy to drive on which is way in excess of virtually every other means of transportation. Are average electric cars better for the environment than a Ford SUV? Sure! Are they significantly better? No. Not at all. Measuring one metric - fuel converted to CO2 - doesn't give you the full picture.

          Yes, EVs are significantly better. Have you ever looked at how gas comes to be near you? There's the initial drilling, the pump jacks (lots of electricity there), transportation to a refinery, refining, transportation to storage, transportation to the stations, pump

        • It doesn't make sense to take away existing roads, but it may make sense to scale back the number of lanes in a number of cases. It also doesn't make any sense to build more highways, or even highway lanes. I agree that we still need roads, but we should have less cars on them.

          Instead of the whole fk cars and roads mentality, maybe we should cut down by doing things like working from home when a physical presence isnâ(TM)t required.

          How about both?

          • It doesn't make sense to take away existing roads, but it may make sense to scale back the number of lanes in a number of cases. It also doesn't make any sense to build more highways, or even highway lanes. I agree that we still need roads, but we should have less cars on them.

            It really depends. If you cut back the ability of traffic to flow on already busy streets without adding alternatives then travel times increase greatly. Maybe with electric cars this would only be bad for massively wasting peoples time but in ICE vehicles this causes massive increases in pollution as going very slowly and starting and stopping incessantly is hard on gas mileage. Even just getting the same pollution from more people giving up on moving around because of the time and expense is going to h

        • ...Instead of the whole fk cars and roads mentality, maybe we should cut down by doing things like working from home when a physical presence isn’t required. That alone would make a truly massive impact.

          OMG. Please don't mod this comment up or down. This is for the commenter quoted.

          burtosis, personal story and viewpoint; just don't read if you're not interested, please.

          First thing that popped into my mind when reading that was: Oh. My. God. Try telling the owner of the company that I work for that. Doesn't understand IT; thinks that everything he wants is magically possible in no time flat with any computer and a desktop is the same as a server but just is being used as a workstation. What's redundan

      • Re: (Score:2)

        by J-1000 ( 869558 )

        Should we encourage people to get electric cars? Probably not. Getting policy makers to legalize walkable neighborhoods and remove tax incentives against public transportation, allowing people who do not want to drive to not drive, would reduce emissions several times as much as replacing every car on the road with an electric one.

        Saying no to electric cars is simply saying yes to an equivalent number of gasoline cars, which is absurd. Should we encourage public transportation? Yes. Should we also encourage people to get electric cars? Absolutely.

        I'm guessing the limitations of public roads are partly what drove the people behind Tesla to push for full self driving, which would allow for greater traffic density (supposedly). If you are going to be the future of transportation you need to think about these things.

        • Let's also not lose sight of the fact that with EV cars you are concentrating power generation. Instead of 100s of thousands of ICE you end up with centralized power generation which is far easier to focus on. You could potentially even use coal because you focus all your energy on cleaning one spot instead of maintaining, upgrading, and cleaning all the engines out on the road.

          We all know EVs are going to replace everything in the near or even medium term but while they do create new problems are much mor

        • Re: (Score:3)

          by spitzak ( 4019 )

          Full self-driving would also mean all the cars can be autonomous taxis which would remove all the need for parking spaces which is actually a good deal more area than the roads in this country.

          • Full self-driving would also mean all the cars can be autonomous taxis which would remove all the need for parking spaces which is actually a good deal more area than the roads in this country.

            Really? So I drive 30+ minutes to get somewhere in my own car because auto-taxis don't service where I live, and I don't need a parking space? What do you expect my car to do, circle the building for the hour or two I'm inside taking care of business, watching a movie, whatever? Your statement comes across as if you think everyone lives in a city; we don't.

  • Forgetting something (Score:3, Interesting)

    by drw ( 4614 ) on Tuesday August 02, 2022 @02:58PM (#62756542)

    So they claim that the manufacturing of the ICs make SSDs worse. However, HDs also use multiple chips in their design so that nullifies their advantages.

    • bigger chips, lower yields on those chips, and the drives are being replaced more frequently.

      • Counting the mining process of the rust though?

        • silicon refineries are more energy intensive than production of iron oxide, which is a by product of the steel industry rather than mined directly by the hard drive and tape industry.

          but to your general point, I would also like to see a general accounting that compares solid state to spinning platters in terms of energy and materials. I suspect they are actually very close and that a 2.5" SATA form factor SSD is worse than a HDD, but a simple PCB form factor like mSATA or NVMe are better.

          There is quite a bi

          • There is quite a bit more aluminum, neodymium, and copper in a HDD than in SSD. Those materials take quite a bit of energy to refine and form.

            True, but they're also highly recyclable and take far less energy to reform - much of the chips in an NVMe are going to be far more energy intense to refine out the precious metals than just chucking the aluminium case, copper, and reuse the neodymium magnets in a new build; you could likely reuse the plastic bits as well if the form factor isn't changing.

            • The components in a HDD are not easy to recycle. Whose going to pay someone $15/hr to painstakingly separate and save less than $1/hr in materials back? If you have a big handful of gold dust in your hand it’s worth tens of thousands, but dump that from 5000 feet and it’s absolutely worthless as recovery would be many times that.

              • once you pull out the electronics and magnets you just throw the aluminum into the pot and melt that fucker. since aluminum melts at 1250 and steel at 2200, all the little steel inserts and shit are easy to separate.

                • It’s not cost effective nor is it practical. You can’t just have a whole bunch of trash contaminating everything and you cant simply fish out steel bits from molten aluminum as it’s going to cling to the surface of everything not melted. We don’t have the technology yet, we need better and cheaper robotic automation. What happens in real life is things like wire are often recycled by uncontrolled burning to get rid of the insulation creating massive amounts of pollution because t

                  • Itâ(TM)s not cost effective nor is it practical. You canâ(TM)t just have a whole bunch of trash contaminating everything and you cant simply fish out steel bits from molten aluminum as itâ(TM)s going to cling to the surface of everything not melted. We donâ(TM)t have the technology yet

                    O RLY [researchgate.net]?

                    • Separating the iron from aluminum by an ai driven robot might only consume a few tens of watt hours for processing and perhaps that again for the drive motors per pound of drives. You’re looking at tens of thousands of watt hours to melt it together and then separate it thus it’s not practical to do so. Further, it’s not just iron from the aluminum, you will have zinc, copper, lead, tin, and many other metals as well. It doesn’t take even a high school level of understanding to re

    • Yeah I don't seem to be able to access the original paper and I don't want to assume professors are being stupid here as I am not a professor but the gap in manufacturing resources seems way off, especially considering the HDD just has a die cast and machined aluminum/steel casing, metal covering, glass or metal platters that are precisely coated, a lot of screws and hardware and as you said a decent PCB with lots of components.

      Maybe there is something inherent to NAND memory production that makes it way mo

    • Here is what they are forgetting:
      before SSD: our bloated Windows systems endlessly thrash the drive for minutes at a time, wasting millions of man-hours world wide
      after SSD: wow how did we ever do without it?
    • Um... compare the number of VLSIs found on a 20TB rotating disk (there will be two, perhaps three: One drive controller/sas interface and a 256/512M flash cache) with the number found on 20TB of enterprise grade SSDs, namely that in three 8T NVME sticks (which will be about 9 each - one pcie interface and 8 1TB flash memories). Which is kind of the point.
    • I'm super-curious, what exactly do you think they did when they calculated the carbon footprint of manufacturing hard drives? Just ask the nearest hobo what he thought?

  • Everything seems to be a carbon emitter. (Score:3)

    by nightflameauto ( 6607976 ) on Tuesday August 02, 2022 @02:59PM (#62756546)

    Is this the point where we start discussing space as a viable manufacturing alternative? I mean, everything we build creates a carbon footprint down here. Up there? *SHRUG*. You get far enough out of the atmosphere that you aren't just dropping CO2 trails throughout the upper atmosphere, it'll dissipate. Is it the correct solution? Probably not forever, but us humans are short-sighted creatures. We'll find a way to end ourselves long before we corrupt the entire universe with our CO2 output.

    • Until the earth runs out of either carbon, or oxygen, sure. Why not just grow more plant-life to recycle the CO2 at the same rate as we are producing it? Massive sea-weed farms in the middle of the oceans, anyone?

    • In truth we should have been talking more about it a long time ago. We spent a whole lot of money and effort on the space race only to turn it into the cold war instead of exploiting that advantage to its fullest. The people who own everything on this planet were no doubt concerned that they wouldn't wind up owning everything in space. One might argue that treaties forbidding commercialization of space were a mistake in this regard — if the dominant paradigm is going to be capitalism, you have to figu

    • Stop reading so much sci-fi. Stuff doesn't magically get "up there". It takes lots of energy (aka carbon) to get anything to or from space. Meanwhile, we are manufacturing hundreds of millions of SSDs right here on the ground. It really does make a difference if they take so much more energy to produce.

    • This brings to mind something related :

      Even with space based manufacturing, what happens to the CO2 or other unwanted things produced? If you just vent them out in a low earth orbit factory, how much will reenter the planet's atmosphere?

      And what of toxic stuff created as an unwanted byproduct - will that get burned up upon reentry and become harmless or will it be adding toxic gases into the atmosphere? And that's assuming it all burns up and does not reach the surface - and if it reaches the surface, will

    • Re: (Score:2)

      by AmiMoJo ( 196126 )

      There's no advantage to making SSDs in space. Most of the emissions are from gathering the raw materials, which only exist on Earth in the quantities needed, and from power generation. Power generation in space means solar panels, but they are much cheaper to install on top of your terrestrial factory.

      When you factor in the emissions created launching and de-orbiting stuff, it's going to be worse than doing it on the ground.

    • ... We'll end ourselves long before we corrupt the entire universe with our CO2 output. (-find a way)

      Tidied that up a hair for you, you know, to reduce the carbon footprint of generating it on more readers' computers. Pay no attention to the fact that this comment offsets that 100% (or more). ;)

  • 10 year life span? (Score:4, Insightful)

    by pecosdave ( 536896 ) on Tuesday August 02, 2022 @03:01PM (#62756554) Homepage Journal

    I can't get a heavy use SSD past a year or two. Seriously, I'm changing the damned things out all the time at work. I've literally had to replace a CFast three times an an SSD twice on a system that still has the original spinning disk in it. Yes, some of our industrial systems have all three in them, and the solid state stuff just doesn't last like the spinners do.

    They're great in my laptop, but damn, I don't put too much trust in them.

    • Are you exceeding the write allotment or something?

    • Re: (Score:2)

      by King_TJ ( 85913 )

      Are you doing SQL database related things on the SSDs?

      I was always told not to use solid state drives for SQL databases, because although the performance is great, it always causes you to exceed the lifetime read/write capabilities of the media.

      I've heard the same thing to a lesser extent about putting SSDs into RAID arrays (other than a RAID 0 mirror anyway).

      • Yes.

        We are "sort of" doing RAID with them. We are using DRBD across two systems so the database is ready in case of fail over. I'm trying to convince my company to allow me to put in spinning disks as replacements. Some engineer before I was hired decided we needed SSDs everywhere and it's proving to be a costly decision. We have to use "wide temperature range" industrial SSDs. Regular laptop HDDs survive just fine in the the rugged (lot of heat but little to no vibration) environments.

        There are some S

    • Re: (Score:2, Insightful)

      by Anonymous Coward

      I can't get a heavy use SSD past a year or two. Seriously, I'm changing the damned things out all the time at work. I've literally had to replace a CFast three times an an SSD twice on a system that still has the original spinning disk in it. Yes, some of our industrial systems have all three in them, and the solid state stuff just doesn't last like the spinners do.

      They're great in my laptop, but damn, I don't put too much trust in them.

      There are a lot of counter-intuitive realities when it comes to SSD vs HDD.

      In database applications that are not write limited we always recommend sticking with HDDs and investing in more ram. The nature of wear leveling adds risks of unrelated non-logged data loss during uncontrolled shutdown, unpredictable write amplifications and inherently reduced durability of data. There are some very nasty feedback loops as cells age out due to writes as oxide is eaten away they will leak more reducing their reliab

    • That's why optane should have been a success.

    • If you're going through them that fast, then you should either be using an array of them, or you should have a lot more RAM, and be doing a lot more caching. And if all that's not enough, then it's back to disk arrays for you.

  • SSDs Are Not Aways Writing (Score:4, Interesting)

    by BrendaEM ( 871664 ) on Tuesday August 02, 2022 @03:04PM (#62756566) Homepage
    The comparison is a big fail because you don't always write to a SSD, so the article fails to consider reads and quiescent time. Additionally, SSDs writes are much quicker than mechanical hard drives. Still, HDDs still have a place with large volumes with low write counts.
    • Did you (and, presumably, the people modding this up) not even read the summary, let alone the article itself? The point is that the huge disparity in the carbon footprint of manufacture overwhelms any energy savings. Presuming their hard disk number is accurate, unless the SSD is actually converting CO2 to O2 during operation, it's worse.
  • I find this hard to believe. Why exactly are SSD semis so much more energy intensive to produce? I don't think they are built on smaller nodes than the cpu or gpus, and they aren't really any bigger than them either. Sure there are more chips per device, but that holds true for memory DIMMs as well. I'd be interested to hear a response from Micron or another manufacturer. By my quick back of the envelope, this would imply that it takes 800kWh to produce their 1TB SSD. Note that this also does not take

    • I find it very hard to believe it takes the energy equivalent of burning 161 kg of carbon to produce a chip which is, at best, no more than a few grams of silicon. I could easily understand that for the entire crystal, or even perhaps a wafer, but there are hundreds, if not thousands of chips on a wafer. The estimate sounds off by a few orders of magnitude.

      By way of comparison, two pounds of charcoal is sufficient to melt about pound of aluminum.

  • Another option to consider (Score:3)

    by Waffle Iron ( 339739 ) on Tuesday August 02, 2022 @03:19PM (#62756616)

    Never underestimate the carbon footprint of a station wagon full of tapes hurtling down the freeway.

    (Especially if it's something like a '71 Chrysler Town & Country with a big block V8.)

    • (Especially if it's something like a '71 Chrysler Town & Country with a big block V8.)

      Man, what I would give to have one of those land yachts today!

      • (Especially if it's something like a '71 Chrysler Town & Country with a big block V8.)

        Man, what I would give to have one of those land yachts today!

        Mercedes has you covered if all you want is a big wagon with respectable power. Sadly, they quit bringing over the E63, which was a big wagon with the ability to shred its own tires.

  • There are a ton of carbon costs when making drive parts. If a SSD outlasts hard drives by an order of magnitude, then that is one thing. Similar, if a SSD's performance is better than a bunch of HDD spindles that are being striped.

    Wear comes to mind. SSDs have a definite wear life, and when they are done, they are done. HDDs don't have as definite a life, although they do definitely wear out, be it bearings, motors, head actuators, and other mechanical parts. If a drive takes more carbon to make, but l

  • Energy differential calculation wrong (Score:3)

    by JoeyRox ( 2711699 ) on Tuesday August 02, 2022 @03:28PM (#62756650)
    They calculated the difference in energy consumption but failed to factor in the orders-of-magnitude difference of NAND performance via HDDs. To wit, an HDD will spend much more if its time at peak power consumption during I/O vs SSDs, by function of how much longer it takes complete the I/O. And that doesn't even factor in the additional power consumption from keeping the system running longer to complete I/O-bound workloads for a given user task.

  • Delta of HDD v SSD = 10kg / year
    1 hectre of trees sequesters 4,000-40,000kg / year

  • Climate change is very important. Reducing our "carbon footprint" is a significant part of that. But there is a point where things go beyond practical and become obsessive. I'm not going to live my life having every decision I make dictated by potential the "carbon footprint" of what I use. I assure you, when trying to decide between a HDD or SSD, the "carbon footprint" of the drive is the absolute last thing on my mind...

    • I believe the entire point of the article is that we need to start thinking about carbon footprints in more areas of our existence - and that there may be some things we do that are unexpectedly carbon-heavy compared to alternatives.

      This is the way the future (preferably, the extremely near future) is going to look. The need to consider the environmental impact of our choices is going to have to trickle down from large infrastructure (power plants) into smaller and smaller domains (what food do I buy).

  • According to TFA, IT accounts for 2% of global carbon output and is on the rise (big surprise). Basically, this particular thing is a rounding error. Switch from coal to nuclear. Next!

  • The majority of the carbon footprint they stated was in the fabrication of the chips used in the SSD.

    This same fabrication method is used in all high density chips to much of the same extent and to less of an extent all chips in general. They all require super clean, high temp or at least very controlled temp and high energy processes to be produced. Though that's spread out across however many hundreds or thousands they produce at the same time using that energy.

    Ignoring that though, how is it that a h

    • > the power usage of ssd's and hdds in this post also dont seem to consider cooling costs either. How much additional energy is needed to cool the hdds vs the ssd? In the comments they made about data centers, the difference here is not negligible.

      Correct. You usually figure 3x power for cooling as a rule of thumb (4x power draw total).

      They cannot leave that out and claim to have done a serious analysis.

  • To assume more power usage means more CO2 is the same flawed thinking that BEV critics use to justify the continued use of ICEVs.

    For the study to mean anything they need to run the numbers at the two extremes, 100% coal powered fabs and 100% clean energy powered fabs, and present the two sets of numbers. Their summary should then probably point out that new fabs should be required to use clean energy sources to operate if we want to benefit from moving from HDD to SSD.

  • "Order of magnitude" abuse (Score:5, Insightful)

    by DJBurgie ( 679629 ) on Tuesday August 02, 2022 @04:40PM (#62756944)

    To say something is big these days it seems like you must include the term "order of magnitude!!!!!" to sound awesome, but that term actually has a meaning. It's like people who abuse "literally" in (not-literally) every single sentence. "I laughed so hard I literally died!" No, you didn't.

    From the first paragraph quoting the researchers:

    "Compared to SSDs, the embodied [carbon] cost of HDDs is at least an order of magnitude lower"

    From the third paragraph:

    "...lifetime footprint for a 1 terabyte SSD is 369.2 kg of carbon dioxide equivalent versus 199 kg for an HDD"

    Since all of the numbers cited appear to be decimal (base 10), an order of magnitude means that that something is ten (10) times something else, e.g. SSDs are 10x worse, or HDDs are 1/10 as bad. Unfortunately for the writer, 199 kg * 10 is nowhere near 369.2 kg; the 369.2 number isn't even double, much less 10x worse.

    I appreciate that somebody did some number crunching to show differing technologies in a light I had not considered before, but the inability of the researchers to either do basic math or understand the words that quantify numeric differences makes me wonder if they could really do the real math required to calculate the carbon footprint of these processes in a meaningful way.

    • Re: (Score:3)

      by splutty ( 43475 )

      Amen, brother. That literally pisses me off.

    • You just decimated their editorial team! By which I mean they've reduced their headcount by 10%.

      Okay, you haven't literally done so, but clearly their editors are doing a piss-poor job. Is it not supposed to be a scientific publication?

  • 159 kilograms of carbon dioxide. Doesn't that imply roughly the same mass of fuel? 159 liters of gasoline would cost well over $200 US at current prices, making the price of the SSD prohibitive.

    • Re: (Score:2)

      by spitzak ( 4019 )

      Carbon is only 12/44 of the mass of CO

    • I calculate $40 - $60 of electricity (in the US) to make that much carbon. Energy is a big part of the cost to make semiconductors so this seems plausible.

      • Re: (Score:2)

        by MrL0G1C ( 867445 )

        Sounds completely implausible to me considering the cut the retailer will take and that energy will still only be a fraction of the cost taking into account wages, profits, building costs etc.

        A new 1TB starts from around $60 so the electricity used much be a small fraction of that. The authors have fucked up, the numbers are wrong, perhaps very very out of date.

        • Note the post you are responding to:

          I calculate $40 - $60 of electricity (in the US) to make that much carbon.

          Just curious, when was the last time you bought an SSD fabbed in the US? Was it NEVER?

          • I was just using US consumer energy prices as a proxy because they are easy to find. I'd be hard pressed finding out what Samsung, Hynix or Micron actually pay for electricity (probably less). Flash is a commodity so margins are thin and energy costs are a bound to be a big part. This seems like a better back of the envelope estimate than the post I was responding to ($200 based on gasoline prices).

  • A surprising result demands a second look (Score:5, Insightful)

    by LightStruk ( 228264 ) on Tuesday August 02, 2022 @05:23PM (#62757062)

    The authors used Life Cycle Analyses (LCAs) from 3rd parties [arxiv.org] - they did not do their own. So when they get results like:

    But SSDs are significantly more carbon intensive to manufacture. That's because the chip fabrication facilities for SSDs operate at extreme temperatures and pressures that are energy intensive to maintain. And bigger memories require more chips, which increases the footprint accordingly. All this adds up to a significant carbon footprint for SSD manufacture... For example, almost 40 per cent of the carbon footprint of a desktop computer comes from its SSD, compared to just 4 per cent from the CPU and 11 per cent from the GPU.

    They should have known they were comparing apples to kumquats. This result is bonkers. There's no way that a single SSD has 4 times the carbon footprint of a GPU. Have the authors ever looked at a GPU? It's on a huge PCB, has an IC with literally hundreds of millions of transistors, surrounded by 4, 8, or 16 DRAM chips with even more hundreds of millions of transistors, and consumes hundreds of watts in operation.

    There's nothing unique about the carbon footprint of IC fabrication for SSD flash memory chips compared to DRAM chips, CPUs, or GPUs. A HDD contains chips too - DRAM cache, interface controller, etc. Did they consider those? The CPU consumes dozens or hundreds of watts in operation, with a huge IC die of over 9 billion transistors.

    • Nothing was given to credibly calibrate extraordinary claims about carbon sunk into SSD chip production. It might be pointing the way toward a real and valuable study, but a meta-analysis of a collection of works written to satisfy or deflect a regulatory intent, that do not themselves give supporting data, is not scientifically based engineering. The most credible sunk carbon analysis that I have seen, came from toilet manufacturer Toto. That industry is likely not as clouded with technological and comme

      • toilet manufacturer... sunk carbon... I had to chuckle. Boom-whoosh. Can't post a turd with eyes on here, but yeah, imagine that [here]. /humor

  • Heart surgery. Totally carbon intensive. Not only are hospitals very carbon intensive, but every heart surgery extends the residence time of a homo sapien, which is a species with a VERY high carbon footprint.

    Come on people. Priorities. Let’s make our cars electric and require heavy industries to be cleaner, for sure. Tell the coal industry to go f*&k itself and build tons of solar, wind and nuckear power. But can we please make a few allowances for the hospitals and maybe we can tolerate the
  • Assuming TFA's numbers are accurate it sounds like the difference in manufacturing cost is pretty decisive; but I'd be curious if operating costs that factor in more than just the drive itself change the equation any: in the HDD context you are much more likely to see really big arrays that are overkill in terms of capacity; but necessary to provide enough spindles to keep your random I/O from being utterly crippling. Systems really, really, loaded with DRAM, because if it's not in memory it's glacially slo
  • Recycling electronics is a sham, and this article really illustrates why that's more important than ever. I'm still running an i5-2500k and haven't upgraded my PC significantly in 10 years. Only now am I seeing the CPU actually being strained by some of the more intensive games. If this was a smartphone, it would have been burned down for precious metals years ago... not because it's obsolete but just so certain companies can sell more hardware. We need to get a handle on this toxic cycle of forced obsolesc

  • False equivalence... C != W (Score:3)

    by OneOfMany07 ( 4921667 ) on Tuesday August 02, 2022 @10:06PM (#62757650)

    The truth is SSD's use more power than HDD for the same period of time. OK. And how much more? 3W... We can't even get rid of incandescent bulbs that use 150W. We're selling GPU's that expect to draw 100's of watts. I'd rather have my GPU be 10% more efficient (I'm looking at you AMD/ATI) and keep my SSD's random access speeds.

    • I don't know if you saw this line in the article, but if you didn't get ready for a good laugh:

      "For example, almost 40 per cent of the carbon footprint of a desktop computer comes from its SSD, compared to just 4 per cent from the CPU and 11 per cent from the GPU."

      At first you might think, "oh no, I was wrong about wanting a more efficient GPU", and if you are, stop. Why? Because your next thought will be, "Wait a sec, that's obvious bullhonkey. This must not have been worth the time it took to read

  • So what?

    Who cares?

    And of course: Why are you bothering me with this?

Slashdot Top Deals

It is easier to write an incorrect program than understand a correct one.

Close