[HN Gopher] Embodied Carbon: The Dirty Secret of SSDs (8x More T...
___________________________________________________________________
Embodied Carbon: The Dirty Secret of SSDs (8x More Than HDD)
Author : PaulHoule
Score : 25 points
Date : 2022-07-25 19:35 UTC (3 hours ago)
(HTM) web link (arxiv.org)
(TXT) w3m dump (arxiv.org)
| kerblang wrote:
| > digital data creation and consumption across the globe is snow
| bowling
|
| I think they meant "snowballing"
|
| Snow bowling would be a cool sport though
| Beltiras wrote:
| Solution was discussed on HN the other day. The process by which
| we displace putting more carbon in the air is quite well
| understood and will be economically viable if we can bring down
| and scale up energy production, cost by a little, production by a
| lot. The article talked about solar but I'd put my money on
| molten salt reactors for the energy production.
| jeffbee wrote:
| If you want to make that bet the world has provided you ample
| opportunities.
|
| https://www.google.com/finance/quote/GE:NYSE?comparison=SHE%...
|
| A faster way would be to just light the money on fire, though.
| hcarvalhoalves wrote:
| It is unlikely miraculous new technology alone will solve
| anything.
|
| https://en.wikipedia.org/wiki/Rebound_effect_(conservation)
| Beltiras wrote:
| MSRs are neither new nor miraculous.
| PaulHoule wrote:
| See moltex
| Beltiras wrote:
| Was unaware. There's also Thorcon. They have a test plant
| in Indonesia (under construction last I knew).
| eikenberry wrote:
| Sounds like this means it solved the problem of there not
| being enough of it being to expensive. The rebound effect
| doesn't mean it doesn't solve problems, it means you
| misunderstood the problem.
| golemotron wrote:
| World population is going to peak by 2050. The problem will be
| taking care of itself naturally at that point.
| Cyberdog wrote:
| Dare I ask, what is your source on this?
| Swenrekcah wrote:
| It will be to late by then if we don't start on the reduction
| path now.
| gumby wrote:
| Does this include the fuel of transporting a terabyte of hard
| drive vs a terabyte of flash?
| mjb wrote:
| Measured per gigabyte, which seems a little bit silly for a lot
| of applications (especially server applications). For example, a
| modern SSD can do as many random 4kB read IOs in a second as the
| average hard drive can do in a week! So if you're IO bound, then
| the cost is much much lower. Throughput is closer, but there's a
| nearly 10x and increasing gap there too with modern NVMe devices.
|
| If you're going to worry about SSD carbon footprint in servers,
| you need to include those factors too. May server applications
| are scaled on IO or throughput and not on storage, and its much
| harder to solve IO and throughput problems with tiering than it
| is to solve storage problems.
| er4hn wrote:
| FTA: Unfortunately, scaling technology can have a significant
| environmental impact. Across the globe, most semiconductor
| manufacturing use electricity that is generated from coal and
| natural gas.
|
| So the problem is the power plants in manufacturing environments.
| It's not that SSDs are packed with carbon that pollutes the
| environment when disposed of.
| wmf wrote:
| Two approaches:
|
| A. Death by a thousand guilt trips
|
| B. Tax carbon
|
| Let's stop accounting and start actually greening.
| lynndotpy wrote:
| This paper makes recommendations other than guilt trips and
| doesn't preclude carbon taxes.
|
| From the abstract:
|
| > First, this paper recommends directions that help select the
| right medium of storage (SSD vs HDD). Second, this paper
| proposes lifetime extension techniques for SSDs. Third, this
| paper advocates for effective and efficient recycling and reuse
| of high-density multi-level cell-based SSDs. Fourth,
| specifically for hand-held devices, this paper recommends
| leveraging elasticity in cloud storage.
|
| This seema reasonable to me. I don't think this paper claims to
| be much more than one item of research.
| jeffbee wrote:
| For real. The total global CO2 discussed here is less than 2%
| of the CO2 from the light vehicle fleet of the U.S.A. alone.
| Slap a price on carbon and let us choose whether we'd rather
| spend our carbon budget on IOPS or car rides.
| billfor wrote:
| Do you think China and Russia will go along with that?
| Arbortheus wrote:
| Does it matter? When the EU adopted GDPR the rest of the
| tech industry had to follow. If we introduce carbon levies
| then processes will improve everywhere. It's more economic
| to run one carbon efficient production line for everyone
| than an efficient line for one market and an inefficient
| line for another.
| mc32 wrote:
| One way if could work, if people were willing to put up
| with side effects (price increases) would be to tie it in
| to trade. Add import duties to imports that aren't carbon
| taxed, take that tax and use it to further incentivize
| alternate (non carbon emoting inputs).
| toolz wrote:
| In reasonably governed world this seems sensible. As it
| stands we have handfuls of congress in the US regularly
| beating all the hedge fund managers with their investments.
| Not exactly a governing body that's overly concerned with
| creating trust and protecting it's people. This doesn't even
| touch on the countries that produce the vast majority of CO2
| and don't seem to share as much popular sentiment as the west
| for reducing their emissions.
| Arbortheus wrote:
| Honestly, the fact SSDs represent 2% of the USA's light
| vehicle emissions is surprisingly high.
| agumonkey wrote:
| Indeed, that said I wonder if SSD latency reduction saved
| enough power from non wasted cpu cycles.
| Cyberdog wrote:
| Or the power it takes to continuously spin hard drive
| platters regardless of whether a data transfer is
| actually happening. Or, as stated in another subthread,
| the reduced energy usage per gigabyte involved in
| transporting SSDs versus larger, heavier HDDs.
|
| If we're going to stress the numbers this minutely, then
| let's really stress them.
| copx wrote:
| SSDs are produced in China, most things are these days, and the
| Chinese Communist Party certainly won't start taxing carbon..
|
| So you could only add the tax to SSDs when they are sold in a
| country where the government supports such taxes putting said
| countries at a competitive disadvantage in a globalized
| economy.
|
| Furthermore how are you going to get an honest assessment of
| the carbon footprint of their products out of the Chinese to
| begin with?
|
| The primary problem with reducing carbon emissions is that most
| of the world is unwilling to.
| parineum wrote:
| Place a carbon import tariff on anything imported without a
| carbon tax already imposed. Make that tariff 110% estimate of
| carbon production.
___________________________________________________________________
(page generated 2022-07-25 23:02 UTC)