info
Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
Claim: Cheap USB Sticks have fast memory only at beginning, rest is slooow!
1 view
Skip to first unread message
Java Jive
Jan 26, 2023, 7:03:00 PM1/26/23
to
Please excuse the Linux/Windows crosspost, this is a question about USB
hardware relevant to both OSs!
In a review raising the spectre of the VW emissions testing scandal all
over again, an Amazon customer for this 128GB USB 3.1 drive ...
https://www.amazon.co.uk/dp/B08LG94ZR8/ref=twister_B08R5S2ZWG?th=1
... claims that the manufacturer has put fast chips at the beginning so
that it passes review tests well, but fills up the memory beyond the
beginning with cheaper very slow chips.
Like everyone else, I've heard of USB sticks having fractions of the
memory that they are specified to have, with the memory wrapping back to
the beginning for address ranges beyond the physical limit actually
installed, but the above claim is a new con on me. Has anyone else
encountered this?
FTR, here's what he says in full (auto-translated from Spanish into
English) ...
"Cliente Amazon
1.0 out of 5 stars Satisfaction and disappointment mixed
Reviewed in Spain on 10 December 2019
Verified Purchase
As on previous occasions I had already happened with other Chinese
manufacturers that the external pendrives or SSD disks not only did not
fulfill what was promised but were also a complete deception, since they
mixed internally two types of memory (fast at first and very slow
afterwards), I decided to do a complete test for each of the 10
pendrives. I will try to explain the process I performed as best as
possible for less experienced users.
I proceeded to format each pendrive by unchecking the “Quick Format”
option, controlling the initial and final time to determine the actual
write speed (which is what formatting basically does).
The joys and disappointments in the process did not wait too long: some
took about 8 to format (writing speeds of 15~17 MB. /sec., reasonably
“high” for usb 2.0) and the rest between 17 and 21 minutes, which means
that at most they reached typical usbs 1.0 speeds, with “variable”
formatting times (nor did I want to do the speed calculations so as not
to get bitter).
Of course, with these pendrive it is true that “random access”...
What really annoyed me was, when I observed the speed of the process of
formatting the “slow” pendrives, that I could see that at first the
advance, in general, seemed “normal” for the first/second gigabytes and
that then there was, visually, almost a stoppage in the speed of that
advance until its completion. This, in my opinion, clearly indicates
that the manufacturer (again, another manufacturer) has mixed chips with
“fast” access along with very slow chips, placing access to the fastest
“at the beginning” so that the usual test tools give “good results”.
That is why I, personally, [no longer] use the typical test tools that
usually only verify a “minimum” part (just the “initial”) of the total
capacity of the external disks/pendrives. I prefer to take a little
longer and perform my writing/formatting tests more comprehensively due
to the bad experiences I have had with different memory manufacturers.
However, in my tests none of the pendrives gave write errors: a simple
method to know is to use chkdsk (in the command console) when the
formatting is finished and check that the text “bad sectors” does not
appear.
Of the 10 pendrives, 7 had an acceptable speed (one took 10 minutes, but
I “accept it with resignation” as passable) and 3 were horribly slow.
As the overall price is reasonable, I don't dare complain about it, but
I value it with only one star because of the “unexpected surprises” I
have encountered.
I'm sorry about the whole thing, but I think this is the only way to
show the problem and try to help other less experienced users perform
their own checks more guaranteably."
--
Fake news kills!
I may be contacted via the contact address given on my website:
www.macfh.co.uk
hardware relevant to both OSs!
In a review raising the spectre of the VW emissions testing scandal all
over again, an Amazon customer for this 128GB USB 3.1 drive ...
https://www.amazon.co.uk/dp/B08LG94ZR8/ref=twister_B08R5S2ZWG?th=1
... claims that the manufacturer has put fast chips at the beginning so
that it passes review tests well, but fills up the memory beyond the
beginning with cheaper very slow chips.
Like everyone else, I've heard of USB sticks having fractions of the
memory that they are specified to have, with the memory wrapping back to
the beginning for address ranges beyond the physical limit actually
installed, but the above claim is a new con on me. Has anyone else
encountered this?
FTR, here's what he says in full (auto-translated from Spanish into
English) ...
"Cliente Amazon
1.0 out of 5 stars Satisfaction and disappointment mixed
Reviewed in Spain on 10 December 2019
Verified Purchase
As on previous occasions I had already happened with other Chinese
manufacturers that the external pendrives or SSD disks not only did not
fulfill what was promised but were also a complete deception, since they
mixed internally two types of memory (fast at first and very slow
afterwards), I decided to do a complete test for each of the 10
pendrives. I will try to explain the process I performed as best as
possible for less experienced users.
I proceeded to format each pendrive by unchecking the “Quick Format”
option, controlling the initial and final time to determine the actual
write speed (which is what formatting basically does).
The joys and disappointments in the process did not wait too long: some
took about 8 to format (writing speeds of 15~17 MB. /sec., reasonably
“high” for usb 2.0) and the rest between 17 and 21 minutes, which means
that at most they reached typical usbs 1.0 speeds, with “variable”
formatting times (nor did I want to do the speed calculations so as not
to get bitter).
Of course, with these pendrive it is true that “random access”...
What really annoyed me was, when I observed the speed of the process of
formatting the “slow” pendrives, that I could see that at first the
advance, in general, seemed “normal” for the first/second gigabytes and
that then there was, visually, almost a stoppage in the speed of that
advance until its completion. This, in my opinion, clearly indicates
that the manufacturer (again, another manufacturer) has mixed chips with
“fast” access along with very slow chips, placing access to the fastest
“at the beginning” so that the usual test tools give “good results”.
That is why I, personally, [no longer] use the typical test tools that
usually only verify a “minimum” part (just the “initial”) of the total
capacity of the external disks/pendrives. I prefer to take a little
longer and perform my writing/formatting tests more comprehensively due
to the bad experiences I have had with different memory manufacturers.
However, in my tests none of the pendrives gave write errors: a simple
method to know is to use chkdsk (in the command console) when the
formatting is finished and check that the text “bad sectors” does not
appear.
Of the 10 pendrives, 7 had an acceptable speed (one took 10 minutes, but
I “accept it with resignation” as passable) and 3 were horribly slow.
As the overall price is reasonable, I don't dare complain about it, but
I value it with only one star because of the “unexpected surprises” I
have encountered.
I'm sorry about the whole thing, but I think this is the only way to
show the problem and try to help other less experienced users perform
their own checks more guaranteably."
--
Fake news kills!
I may be contacted via the contact address given on my website:
www.macfh.co.uk
iKook
Jan 26, 2023, 8:21:27 PM1/26/23
to
On 27/01/2023 00:02, Java Jive wrote:
>
>
> As on previous occasions I had already happened with other Chinese
> manufacturers that the external pendrives or SSD disks not only did
> not fulfill what was promised but were also a complete deception,
> since they mixed internally two types of memory (fast at first and
> very slow afterwards), I decided to do a complete test for each of the
> 10 pendrives. I will try to explain the process I performed as best as
> possible for less experienced users.
Can you explain to us why would Chinese manufacturers do what you
>
>
> As on previous occasions I had already happened with other Chinese
> manufacturers that the external pendrives or SSD disks not only did
> not fulfill what was promised but were also a complete deception,
> since they mixed internally two types of memory (fast at first and
> very slow afterwards), I decided to do a complete test for each of the
> 10 pendrives. I will try to explain the process I performed as best as
> possible for less experienced users.
suggest? How much do they save or to put it in positive terms, how much
do they make by doing that when the amount of work involved in doing
what you suggest they are doing. The item costs 14.38 (I believe it is
UK £) so for a few cents/pennies, are they so stupid to ruin their
business and reputation? These days everything is made in China so are
we going to spend time testing everything from that country? It's going
to be even more expensive for us when there are no alternative
manufacturers. European and American governments have destroyed their
factories by outsourcing everything to China and India.
Amazon Basic have similar items: <Amazon Basics - 128 GB, USB 3.1 Flash
Drive, Read Speed up to 130 MB/s : Amazon.co.uk: Computers & Accessories
<https://www.amazon.co.uk/dp/B0B6148YKN/>>
You get 128GB for £14.06 and 256GB for £21.74
Paul
Jan 27, 2023, 12:07:31 AM1/27/23
to
On 1/26/2023 7:02 PM, Java Jive wrote:
> Please excuse the Linux/Windows crosspost, this is a question about USB hardware relevant to both OSs!
>
> In a review raising the spectre of the VW emissions testing scandal all over again, an Amazon customer for this 128GB USB 3.1 drive ...
>
> https://www.amazon.co.uk/dp/B08LG94ZR8/ref=twister_B08R5S2ZWG?th=1
>
> ... claims that the manufacturer has put fast chips at the beginning so that
> it passes review tests well, but fills up the memory beyond the beginning
> with cheaper very slow chips.
The reviews I just glanced at, are for products other than the
> Please excuse the Linux/Windows crosspost, this is a question about USB hardware relevant to both OSs!
>
> In a review raising the spectre of the VW emissions testing scandal all over again, an Amazon customer for this 128GB USB 3.1 drive ...
>
> https://www.amazon.co.uk/dp/B08LG94ZR8/ref=twister_B08R5S2ZWG?th=1
>
> ... claims that the manufacturer has put fast chips at the beginning so that
> it passes review tests well, but fills up the memory beyond the beginning
> with cheaper very slow chips.
item in question. This is "the Amazon way". Amazon allows bozo traders
to pour every review they ever got, for tea towels and scone warmers,
to be poured into the reviews of a USB stick.
The "lowest tier" of USB3 sticks, reads at 100MB/sec and writes at 10MB/sec.
The write rate gets worse, as it ages. I have a 128GB stick, with this
characteristic. When an advert does not mention the speed, this is
what you're getting.
The write speed never normally matches the read speed. The advert
presenting 320 read 200 write, is not impossible.
What you will find, is a bit of poetic license. The write rate
will probably be around 100 or so. Best case.
*******
If capacity fraud has been committed, then yes, it's true
there is weird write rate behavior. The banana-crisps who test
these products they buy, are not clever enough to do a write-readverify
and determine they have been defrauded. "Oh, YES, I did get 2TB of
storage for £10." Yet, if you ask them to compare the hard drive copy,
to what is on the stick now, they suddenly become silent.
*******
On SSD and NVMe, the TLC flash uses a portion of the flash set up as
SLC. This is called the SLC cache. Writes happen in two stages.
Fast write into the SLC cache. Slow later transfer into the main TLC body
of the device. Once the SLC cache is full, device write rates drop to the
TLC rate. This behavior can be seen on TLC and QLC drives.
I'm not aware of USB sticks doing this. The controller is not typically
sophisticated enough to be doing this. The USB stick may not have proper
wear leveling, and may suffer premature failure after a year or two.
Whereas an MLC stick will "last forever", but of course, are hardly
ever made any more.
A few USB devices, use an SSD plus a USB to SATA adapter chip. But
the form factor of the device is not usually that of a pen drive.
The device should have better characteristics, but take up more
desk space and the form factor is a general nuisance.
*******
USB sticks have one or two flash chips. In previous generations,
the controller could be two channel, and the device ran faster
because it was accessing two flash. The flash now, is I/O is fast enough,
the two devices could share a single channel. This makes the
controller chip slightly cheaper to build.
A few USB sticks (the square-ish Patriot Extreme), could have
four flash chips. But you can't plug two of those in next to
one another, so it's one of those per stack.
While you could go to all the trouble, of using two different
flash SKUs on a single stick, the quality of the flash used
in general is so poor, who really cares ? If a TLC wears out in
a year or two, and receives light usage (is not used as an OS
boot drive or something), then does it really matter if the
floor sweepings had different things written on them ? Presumably
the configuration utility they use at the factory, could handle
this behavior, but it would likely be a nuisance when setting
up the devices.
Even reputable brands, have poor flash in them.
My OCZ Rally2, laughs at my other sticks, and their shenanigans.
Sure, the stick had small capacity, it went slow, but... it
has not degraded, it does not slow with age. It actually
fucking well works. We will never see another stick like it.
Summary: If the review you saw was for a 2TB capacity device,
the speed variation could be evidence of capacity fraud,
but only a simple write-readverify can confirm your data
is not being stored on the device.
Check through the reviews, and see if there are any for
a 128GB stick or not. So at least the review item is not
"a tea towel or a candle holder".
Verified Purchase
Ausgewiesen sind bis 350 MB/s Lesen und 200 MB/s Schreiben.
Gemessen wurde an einem Thunderbolt 3 Anschluß mit Rampow USB C Adapter, Test sequentiell:
Check Flash: Lesen 230 MB/s - Schreiben: 81 MB/s <=== so it's writing at 80... wot a surprise
HD Tune Pro: Lesen 300 MB/s - Schreiben: 83 MB/s
Win Explorer: Lesen 308 MB/s - Schreiben: 77 MB/s
Nur 3 Sterne, da die Schreibgeschwindigkeit doch stark vermindert ist.
[Only 3 stars, since the writing speed is greatly reduced]
As for the availability of branded sticks we recognize,
the availability of those is pretty weird. They do not
seem to be manufactured right now, at normal rates. Even though
there is an excess of flash chips around. I have no idea what
fab makes the controllers. Probably not TSMC. There are other
fabs and they make 22nm or 12nm chips. There are even fabs in
India.
https://www.amazon.ca/SanDisk-SDCZ880-128G-G46-Extreme-128GB-Solid/dp/B01MU8TZRV
"Mostly fantastic! A little misleading in a way, though. (Failed a year and half later btw)"
And that's the thing. You can actually get decent read/write rates, but life ? ...
Paul
Joerg Lorenz
Jan 27, 2023, 12:40:02 AM1/27/23
to
Am 27.01.23 um 01:02 schrieb Java Jive:
do with Linux or Windows even not with Mac which you forgot. And this
lengthy considerations do not reach the people you want to reach.
--
Gutta cavat lapidem (Ovid)
> Please excuse the Linux/Windows crosspost, this is a question about USB
> hardware relevant to both OSs!
>
> In a review raising the spectre of the VW emissions testing scandal all
> over again, an Amazon customer for this 128GB USB 3.1 drive ...
>
> https://www.amazon.co.uk/dp/B08LG94ZR8/ref=twister_B08R5S2ZWG?th=1
>
> ... claims that the manufacturer has put fast chips at the beginning so
> that it passes review tests well, but fills up the memory beyond the
> beginning with cheaper very slow chips.
This thread is completely OT in this group and has absolutely nothing to
> hardware relevant to both OSs!
>
> In a review raising the spectre of the VW emissions testing scandal all
> over again, an Amazon customer for this 128GB USB 3.1 drive ...
>
> https://www.amazon.co.uk/dp/B08LG94ZR8/ref=twister_B08R5S2ZWG?th=1
>
> ... claims that the manufacturer has put fast chips at the beginning so
> that it passes review tests well, but fills up the memory beyond the
> beginning with cheaper very slow chips.
do with Linux or Windows even not with Mac which you forgot. And this
lengthy considerations do not reach the people you want to reach.
--
Gutta cavat lapidem (Ovid)
Pancho
Jan 27, 2023, 6:20:59 AM1/27/23
to
On 27/01/2023 00:02, Java Jive wrote:
> Please excuse the Linux/Windows crosspost, this is a question about USB
> hardware relevant to both OSs!
>
> In a review raising the spectre of the VW emissions testing scandal all
> over again, an Amazon customer for this 128GB USB 3.1 drive ...
>
> https://www.amazon.co.uk/dp/B08LG94ZR8/ref=twister_B08R5S2ZWG?th=1
>
> ... claims that the manufacturer has put fast chips at the beginning so
> that it passes review tests well, but fills up the memory beyond the
> beginning with cheaper very slow chips.
>
> Like everyone else, I've heard of USB sticks having fractions of the
> memory that they are specified to have, with the memory wrapping back to
> the beginning for address ranges beyond the physical limit actually
> installed, but the above claim is a new con on me. Has anyone else
> encountered this?
>
I've often noticed, the initial speed writing a large file is fast, but
> hardware relevant to both OSs!
>
> In a review raising the spectre of the VW emissions testing scandal all
> over again, an Amazon customer for this 128GB USB 3.1 drive ...
>
> https://www.amazon.co.uk/dp/B08LG94ZR8/ref=twister_B08R5S2ZWG?th=1
>
> ... claims that the manufacturer has put fast chips at the beginning so
> that it passes review tests well, but fills up the memory beyond the
> beginning with cheaper very slow chips.
>
> Like everyone else, I've heard of USB sticks having fractions of the
> memory that they are specified to have, with the memory wrapping back to
> the beginning for address ranges beyond the physical limit actually
> installed, but the above claim is a new con on me. Has anyone else
> encountered this?
>
then slows dramatically after some time. I assumed some type of fast
cache memory was used. Caches have always been a thing, and are
generally sensible, although it is misleading to claim cache speed is
the total write speed.
Carlos E.R.
Jan 27, 2023, 6:55:28 AM1/27/23
to
is brutal.
Still, it would be better to use some software designed to test
thumbdrives for this. So, for now, it is only a suspicion. Needs further
testing/verification.
--
Cheers, Carlos.
Anton Shepelev
Jan 27, 2023, 7:03:46 AM1/27/23
to
Carlos E.R.:
> Still, it would be better to use some software designed to
> test thumbdrives for this. So, for now, it is only a
> suspicion. Needs further testing/verification.
The speed of bad-o]block tests in USB-imaging software may be
a reliable indicator. These tests include read-only and
read-write tests with various patterns.
--
() ascii ribbon campaign -- against html e-mail
/\ www.asciiribbon.org -- against proprietary attachments
> Still, it would be better to use some software designed to
> test thumbdrives for this. So, for now, it is only a
> suspicion. Needs further testing/verification.
a reliable indicator. These tests include read-only and
read-write tests with various patterns.
--
() ascii ribbon campaign -- against html e-mail
/\ www.asciiribbon.org -- against proprietary attachments
Andy Burns
Jan 27, 2023, 7:19:32 AM1/27/23
to
Java Jive wrote:
> claims that the manufacturer has put fast chips at the beginning so that
> it passes review tests well, but fills up the memory beyond the
> beginning with cheaper very slow chips.
The closest to that I've heard is some SSDs have a small area of SLC for
> claims that the manufacturer has put fast chips at the beginning so that
> it passes review tests well, but fills up the memory beyond the
> beginning with cheaper very slow chips.
initial writes and then the rest MLC/TLC/QLC, the SLC being faster and
more durable
VanguardLH
Jan 27, 2023, 10:51:10 AM1/27/23
to
Java Jive <ja...@evij.com.invalid> wrote:
> Please excuse the Linux/Windows crosspost, this is a question about USB
> hardware relevant to both OSs!
>
> In a review raising the spectre of the VW emissions testing scandal all
> over again, an Amazon customer for this 128GB USB 3.1 drive ...
>
> https://www.amazon.co.uk/dp/B08LG94ZR8/ref=twister_B08R5S2ZWG?th=1
>
> ... claims that the manufacturer has put fast chips at the beginning so
> that it passes review tests well, but fills up the memory beyond the
> beginning with cheaper very slow chips.
>
> Like everyone else, I've heard of USB sticks having fractions of the
> memory that they are specified to have, with the memory wrapping back to
> the beginning for address ranges beyond the physical limit actually
> installed, but the above claim is a new con on me. Has anyone else
> encountered this?
Of the USB flash drives that I've dismantled (after they
> Please excuse the Linux/Windows crosspost, this is a question about USB
> hardware relevant to both OSs!
>
> In a review raising the spectre of the VW emissions testing scandal all
> over again, an Amazon customer for this 128GB USB 3.1 drive ...
>
> https://www.amazon.co.uk/dp/B08LG94ZR8/ref=twister_B08R5S2ZWG?th=1
>
> ... claims that the manufacturer has put fast chips at the beginning so
> that it passes review tests well, but fills up the memory beyond the
> beginning with cheaper very slow chips.
>
> Like everyone else, I've heard of USB sticks having fractions of the
> memory that they are specified to have, with the memory wrapping back to
> the beginning for address ranges beyond the physical limit actually
> installed, but the above claim is a new con on me. Has anyone else
> encountered this?
catastrophically fail due to exceeding the maximum write cycles), there
is only one chip inside.
https://www.usbmemorydirect.com/blog/wp-content/uploads/2021/08/USB-flash-drive-motherboard-1.jpg
One chip is the for hardware interface between memory chip and the USB
protocol interface (aka mass storage controller). The other chip (just
one) is for the flash memory.
If a brand and model doesn't provide a spec sheet listing the sustained
read and write speeds, don't get that USB drive. There are lots of
cheap and promotional USB drives, and they're crap for speed. I usually
look at write speed, because read speed will be inherently faster.
However, if you intend to write once to use the USB drive as archival
storage, and since reads are non-destructive to flash memory, perhaps
fast write speed isn't much of a concern to you other than the first
time you add more files to the drive.
If there are multi-chip flash memory USB drives, I haven't seen them;
however, I don't buy crappy cheap no-name USB flash drives. Having to
wave solder multiple flash chips on a PCB with pads with multiple chips
would seem to be a more expensive manufacture process than having just
one flash chip to mount on the PCB. If you cannot find complete specs,
included read/write performances as bps or IOPs, on a drive, don't buy
that drive. You're getting an unknown that is sold solely on capacity,
and not on performance. Most consumers only look at capacity, and then
later find performance sucks.
Until the one making accusations provides proof, it looks more like
someone dissatisfied with their purchase spewing FUD. Also sounds like
someone that doesn't know the difference between burst or buffered mode
and sustained mode.
Carlos E.R.
Jan 27, 2023, 1:11:13 PM1/27/23
to
On 2023-01-27 13:03, Anton Shepelev wrote:
> Carlos E.R.:
>
>> Still, it would be better to use some software designed to
>> test thumbdrives for this. So, for now, it is only a
>> suspicion. Needs further testing/verification.
>
> The speed of bad-o]block tests in USB-imaging software may be
> a reliable indicator. These tests include read-only and
> read-write tests with various patterns.
That burns the thing out. It is destructive.
> Carlos E.R.:
>
>> Still, it would be better to use some software designed to
>> test thumbdrives for this. So, for now, it is only a
>> suspicion. Needs further testing/verification.
>
> The speed of bad-o]block tests in USB-imaging software may be
> a reliable indicator. These tests include read-only and
> read-write tests with various patterns.
--
Cheers, Carlos.
Java Jive
Jan 27, 2023, 3:01:52 PM1/27/23
to
On 27/01/2023 00:02, Java Jive wrote:
>
>
> Please excuse the Linux/Windows crosspost, this is a question about USB
> hardware relevant to both OSs!
>
> In a review raising the spectre of the VW emissions testing scandal all
> over again, an Amazon customer for this 128GB USB 3.1 drive ...
>
> https://www.amazon.co.uk/dp/B08LG94ZR8/ref=twister_B08R5S2ZWG?th=1
>
> .... claims that the manufacturer has put fast chips at the beginning so
> hardware relevant to both OSs!
>
> In a review raising the spectre of the VW emissions testing scandal all
> over again, an Amazon customer for this 128GB USB 3.1 drive ...
>
> https://www.amazon.co.uk/dp/B08LG94ZR8/ref=twister_B08R5S2ZWG?th=1
>
> that it passes review tests well, but fills up the memory beyond the
> beginning with cheaper very slow chips.
Thanks for all the informative replies, all of which I've read.
> beginning with cheaper very slow chips.
It seems to me that most probably what the reviewer was actually
measuring was the difference between cache and memory write speeds.
Java Jive
Jan 27, 2023, 3:10:17 PM1/27/23
to
complexity of multiple instead of single chips possibly increasing
rather than reducing costs is reasonably convincing.
I too have noticed that many USB sticks seem slower at writing large
amounts of data, and I think the difference between cache and memory
write speeds is the most likely explanation to be correct.
Carlos E.R.
Jan 27, 2023, 4:28:42 PM1/27/23
to
On 2023-01-27 21:01, Java Jive wrote:
> On 27/01/2023 00:02, Java Jive wrote:
>>
>> Please excuse the Linux/Windows crosspost, this is a question about
>> USB hardware relevant to both OSs!
>>
>> In a review raising the spectre of the VW emissions testing scandal
>> all over again, an Amazon customer for this 128GB USB 3.1 drive ...
>>
>> https://www.amazon.co.uk/dp/B08LG94ZR8/ref=twister_B08R5S2ZWG?th=1
>>
>> .... claims that the manufacturer has put fast chips at the beginning
>> so that it passes review tests well, but fills up the memory beyond
>> the beginning with cheaper very slow chips.
>
> Thanks for all the informative replies, all of which I've read.
>
> It seems to me that most probably what the reviewer was actually
> measuring was the difference between cache and memory write speeds.
Dunno.
> On 27/01/2023 00:02, Java Jive wrote:
>>
>> Please excuse the Linux/Windows crosspost, this is a question about
>> USB hardware relevant to both OSs!
>>
>> In a review raising the spectre of the VW emissions testing scandal
>> all over again, an Amazon customer for this 128GB USB 3.1 drive ...
>>
>> https://www.amazon.co.uk/dp/B08LG94ZR8/ref=twister_B08R5S2ZWG?th=1
>>
>> .... claims that the manufacturer has put fast chips at the beginning
>> so that it passes review tests well, but fills up the memory beyond
>> the beginning with cheaper very slow chips.
>
> Thanks for all the informative replies, all of which I've read.
>
> It seems to me that most probably what the reviewer was actually
> measuring was the difference between cache and memory write speeds.
Cache write is orders of magnitude faster than actual write. It is easy
to notice.
He mentions a "writing speeds of 15~17 MB" initially", and that is not
cache speed, which would be at least in the hundreds, depending on
available RAM.
But he was not writing files, he was formatting (slow formatting). Is
the cache caching that?
I read the original text in Spanish, and he says that it goes reasonable
fast for the first gigabyte or two, and then it stalled. He says he did
not use "test tools" because they don't test the whole thumbdrive.
In any case, the information is only grounds for further, proper
investigation.
--
Cheers, Carlos.
Ant
Jan 27, 2023, 6:41:32 PM1/27/23
to
It's hard to find good reviews on these USB flash sticks. I still have
old 128 MB that STILL work today.
> I proceeded to format each pendrive by unchecking the ???Quick Format???
> that at most they reached typical usbs 1.0 speeds, with ???variable???
> What really annoyed me was, when I observed the speed of the process of
> formatting the ???slow??? pendrives, that I could see that at first the
> advance, in general, seemed ???normal??? for the first/second gigabytes and
> ???at the beginning??? so that the usual test tools give ???good results???.
> That is why I, personally, [no longer] use the typical test tools that
> usually only verify a ???minimum??? part (just the ???initial???) of the total
> As the overall price is reasonable, I don't dare complain about it, but
> I value it with only one star because of the ???unexpected surprises??? I
Note: A fixed width font (Courier, Monospace, etc.) is required to see this signature correctly.
/\___/\ Ant(Dude) @ http://aqfl.net & http://antfarm.home.dhs.org.
/ /\ /\ \ Please nuke ANT if replying by e-mail.
| |o o| |
\ _ /
( )
old 128 MB that STILL work today.
> option, controlling the initial and final time to determine the actual
> write speed (which is what formatting basically does).
> The joys and disappointments in the process did not wait too long: some
> took about 8 to format (writing speeds of 15~17 MB. /sec., reasonably
> ???high??? for usb 2.0) and the rest between 17 and 21 minutes, which means
> write speed (which is what formatting basically does).
> The joys and disappointments in the process did not wait too long: some
> took about 8 to format (writing speeds of 15~17 MB. /sec., reasonably
> that at most they reached typical usbs 1.0 speeds, with ???variable???
> formatting times (nor did I want to do the speed calculations so as not
> to get bitter).
> Of course, with these pendrive it is true that ???random access???...
> to get bitter).
> What really annoyed me was, when I observed the speed of the process of
> advance, in general, seemed ???normal??? for the first/second gigabytes and
> that then there was, visually, almost a stoppage in the speed of that
> advance until its completion. This, in my opinion, clearly indicates
> that the manufacturer (again, another manufacturer) has mixed chips with
> ???fast??? access along with very slow chips, placing access to the fastest
> advance until its completion. This, in my opinion, clearly indicates
> that the manufacturer (again, another manufacturer) has mixed chips with
> ???at the beginning??? so that the usual test tools give ???good results???.
> That is why I, personally, [no longer] use the typical test tools that
> capacity of the external disks/pendrives. I prefer to take a little
> longer and perform my writing/formatting tests more comprehensively due
> to the bad experiences I have had with different memory manufacturers.
> However, in my tests none of the pendrives gave write errors: a simple
> method to know is to use chkdsk (in the command console) when the
> formatting is finished and check that the text ???bad sectors??? does not
> longer and perform my writing/formatting tests more comprehensively due
> to the bad experiences I have had with different memory manufacturers.
> However, in my tests none of the pendrives gave write errors: a simple
> method to know is to use chkdsk (in the command console) when the
> appear.
> Of the 10 pendrives, 7 had an acceptable speed (one took 10 minutes, but
> I ???accept it with resignation??? as passable) and 3 were horribly slow.
> Of the 10 pendrives, 7 had an acceptable speed (one took 10 minutes, but
> As the overall price is reasonable, I don't dare complain about it, but
> have encountered.
> I'm sorry about the whole thing, but I think this is the only way to
> show the problem and try to help other less experienced users perform
> their own checks more guaranteably."
--
"But we have this treasure in jars of clay to show that this all-surpassing power is from God and not from us." --2 Corinthians 4:7. :) (L/C)NY 4721 [h2o black ????/(\_/)]! TGIF after passing out after 10:18 PM 4 >7h after a shreddy Th even wo new TV eps.
> I'm sorry about the whole thing, but I think this is the only way to
> show the problem and try to help other less experienced users perform
> their own checks more guaranteably."
--
Note: A fixed width font (Courier, Monospace, etc.) is required to see this signature correctly.
/\___/\ Ant(Dude) @ http://aqfl.net & http://antfarm.home.dhs.org.
/ /\ /\ \ Please nuke ANT if replying by e-mail.
| |o o| |
\ _ /
( )
Paul
Jan 28, 2023, 12:36:34 AM1/28/23
to
On 1/27/2023 6:41 PM, Ant wrote:
> It's hard to find good reviews on these USB flash sticks. I still have
> old 128 MB that STILL work today.
That's because it is SLC.
> It's hard to find good reviews on these USB flash sticks. I still have
> old 128 MB that STILL work today.
Paul
Ant
Jan 28, 2023, 3:16:57 AM1/28/23
to
was an interesting read.
--
"I pray that out of his glorious riches he may strengthen you with power through his Spirit in your inner being." --Ephesians 3:16. Wandering Earth 1 was a meh, but will 2 B better?
Paul
Jan 28, 2023, 3:43:27 AM1/28/23
to
On 1/28/2023 3:16 AM, Ant wrote:
> In alt.os.linux Paul <nos...@needed.invalid> wrote:
>> On 1/27/2023 6:41 PM, Ant wrote:
>>> It's hard to find good reviews on these USB flash sticks. I still have
>>> old 128 MB that STILL work today.
>
>> That's because it is SLC.
>
> https://www.kingston.com/en/blog/pc-performance/difference-between-slc-mlc-tlc-3d-nand
> was an interesting read.
>
Micron made a 32GB SLC flash chip.
> In alt.os.linux Paul <nos...@needed.invalid> wrote:
>> On 1/27/2023 6:41 PM, Ant wrote:
>>> It's hard to find good reviews on these USB flash sticks. I still have
>>> old 128 MB that STILL work today.
>
>> That's because it is SLC.
>
> https://www.kingston.com/en/blog/pc-performance/difference-between-slc-mlc-tlc-3d-nand
> was an interesting read.
>
But it appears to be a tease and does not
seem to ship in a USB stick.
Some person in China took a picture of a
USB3 stick, with the 32GB SLC flash chip
fitted to it. But it appeared to be for the
purpose of selling some other inferior product.
I doubt they ever had stock of those sticks,
just a prototype or a mockup of one.
There is a smaller company that makes SLC flash,
but their flash chips are the small capacity ones. And the pricing
is what you would expect for such a thing. No matter
what the capacity, they're $100+. You sometimes see
them listed as "industrial USB stick" on the electronics
company sites.
Paul
Ravi Kapoor
Jan 28, 2023, 3:21:23 PM1/28/23
to
On 28/01/2023 08:16, Ant wrote:
> In alt.os.linux Paul <nos...@needed.invalid> wrote:
>> On 1/27/2023 6:41 PM, Ant wrote:
>>> It's hard to find good reviews on these USB flash sticks. I still have
>>> old 128 MB that STILL work today.
>> That's because it is SLC.
> https://www.kingston.com/en/blog/pc-performance/difference-between-slc-mlc-tlc-3d-nand
> was an interesting read.
Have the flash industry managed to sort out the size of 1GB yet? I
> In alt.os.linux Paul <nos...@needed.invalid> wrote:
>> On 1/27/2023 6:41 PM, Ant wrote:
>>> It's hard to find good reviews on these USB flash sticks. I still have
>>> old 128 MB that STILL work today.
>> That's because it is SLC.
> https://www.kingston.com/en/blog/pc-performance/difference-between-slc-mlc-tlc-3d-nand
> was an interesting read.
bought a 256GB flash drive sometime ago but it had only 231GB. People
told me that it is because nobody uses correct measurement yet. Is 1GB =
1000MB or is it 1024MB. I have not been able to reconcile how 256GB is
only 231GB when reformatted as NTFS or exFAT (default). I lost 25GB
unnecessarily.
Carlos E.R.
Jan 28, 2023, 4:56:32 PM1/28/23
to
learn the different units.
The industry has very clear that 1 GiB = 1024 MiB, and 1 GB = 1000 MB.
<https://en.wikipedia.org/wiki/Byte#Multiple-byte_units>
<https://en.wikipedia.org/wiki/Gigabyte#Consumer_confusion>
--
Cheers, Carlos.
Ravi Kapoor
Jan 28, 2023, 5:19:43 PM1/28/23
to
figures to demonstrate this, Mr Clever.
(256 x 1000) / 1024 = 250
How does 250 compare with 231? You can reconcile the figures for us to
understand better.
Ken Blake
Jan 28, 2023, 6:00:05 PM1/28/23
to
manufacturers define 1TB as 1,000,000,000,000 bytes, while the rest of
the computer world, including Windows, defines it as 2 to the 40th
power (1,099,511,637,776) bytes. So a 5 trillion byte drive is
actually around 4.5TB.
Some people point out that the official international standard defines
the "T" of TB as one trillion, not1,099,511,637,776. Correct though
they are, using the binary value of TB is so well established in the
computer world that I consider using the decimal value of a trillion
to be deceptive marketing on the part of the hard drive manufacturer.
Java Jive
Jan 28, 2023, 6:06:14 PM1/28/23
to
really matters is what people do, and the fact remains that I rarely see
GiB in practice, and OSs do one thing while disk manufacturers do another.
In OSs, and BTW this is also what I was taught in academia, ...
1KB = 1024 bytes
1MB = 1024 KB
1GB = 1024 MB
... but to storage manufacturers ...
1KB = 1000 bytes
1MB = 1000 of their KB
1GB = 1000 of their MB
Hence to convert disk manufacturers' GB figures to OS' GB figures, you
have to multiply by (1000/1024)^3, giving ...
256 (manufacturer) GB x 1000^3 / 1024^3 = 238 (OS) GB
Then there will be some file system overhead to deduct after that.
Carlos E.R.
Jan 28, 2023, 6:38:33 PM1/28/23
to
That is, 256 GB = 238 GiB. The rest is used in the filesystem overhead.
>
> How does 250 compare with 231? You can reconcile the figures for us to
> understand better.
>
Cheers, Carlos.
Paul
Jan 28, 2023, 7:17:48 PM1/28/23
to
so the blame is not always with the user. They switch between GB and GiB
in a willy-nilly fashion.
If I pipe dd to wc -c , I can count the bytes that way :-)
When I use slightly more efficient methods, this is what I get
"128GB" stick = 128,983,236,608 bytes [they are using the decimal method like HDD do]
"1GB" stick = 1,006,108,672 bytes [they are using the decimal method like HDD do]
Then the computer screen does the math for GiB, but prints "GB" next
to the resulting number. Unnecessarily scaring the user.
The above numbers, do not involve any file systems, so these sample
numbers are not a side effect of "formatting". These numbers
are collected from the physical layer.
Paul
Char Jackson
Jan 28, 2023, 7:27:22 PM1/28/23
to
On Sat, 28 Jan 2023 23:06:07 +0000, Java Jive <ja...@evij.com.invalid>
wrote:
Thanks for providing the actual formula. All these years I just use 93%
and call it good. The formula would put it at 93.1322574615478515625%.
wrote:
and call it good. The formula would put it at 93.1322574615478515625%.
Carlos E.R.
Jan 29, 2023, 7:26:43 AM1/29/23
to
two different unit multipliers, and that some use confusing legends, you
can figure out which they are using.
>
> If I pipe dd to wc -c , I can count the bytes that way :-)
>
> When I use slightly more efficient methods, this is what I get
>
> "128GB" stick = 128,983,236,608 bytes [they are using the decimal
> method like HDD do]
> "1GB" stick = 1,006,108,672 bytes [they are using the decimal
> method like HDD do]
>
> Then the computer screen does the math for GiB, but prints "GB" next
> to the resulting number. Unnecessarily scaring the user.
N and BYTES may be followed by the following multiplicative
suffixes: c=1, w=2, b=512, kB=1000, K=1024, MB=1000*1000, M=1024*1024,
xM=M, GB=1000*1000*1000, G=1024*1024*1024, and so on for T, P, E, Z, Y.
Binary prefixes can be used, too: KiB=K, MiB=M, and so on.
which is not what they were doing previously. The February 2018 edition
says instead:
N and BYTES may be followed by the following
multiplicative suffixes: c =1, w =2, b =512, kB =1000, K =1024, MB
=1000*1000, M =1024*1024, xM =M, GB =1000*1000*1000, G =1024*1024*1024,
and so on for T, P, E, Z, Y.
The new version does:
cer@Telcontar:~> dd if=/dev/zero of=test.dd count=1000 bs=1000
1000+0 records in
1000+0 records out
1000000 bytes (1,0 MB, 977 KiB) copied, 0,00394632 s, 253 MB/s
cer@Telcontar:~> dd if=/dev/zero of=test.dd count=1024 bs=1024
1024+0 records in
1024+0 records out
1048576 bytes (1,0 MB, 1,0 MiB) copied, 0,00561859 s, 187 MB/s
cer@Telcontar:~>
cer@Telcontar:~> dd if=/dev/zero of=test.dd count=1000 bs=1000000
1000+0 records in
1000+0 records out
1000000000 bytes (1,0 GB, 954 MiB) copied, 0,499733 s, 2,0 GB/s
cer@Telcontar:~>
The old version does:
cer@Isengard:~> dd if=/dev/zero of=test.dd count=1000 bs=1000
1000+0 records in
1000+0 records out
1000000 bytes (1.0 MB, 977 KiB) copied, 0.0131062 s, 76.3 MB/s
cer@Isengard:~> dd if=/dev/zero of=test.dd count=1024 bs=1024
1024+0 records in
1024+0 records out
1048576 bytes (1.0 MB, 1.0 MiB) copied, 0.0133418 s, 78.6 MB/s
cer@Isengard:~>
cer@Isengard:~> dd if=/dev/zero of=test.dd count=1000 bs=1000000
1000+0 records in
1000+0 records out
1000000000 bytes (1.0 GB, 954 MiB) copied, 2.66717 s, 375 MB/s
cer@Isengard:~>
At this moment I don't see how to choose units.
> The above numbers, do not involve any file systems, so these sample
> numbers are not a side effect of "formatting". These numbers
> are collected from the physical layer.
--
Cheers, Carlos.
Dan Purgert
Jan 29, 2023, 7:31:43 AM1/29/23
to
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512
["Followup-To:" header set to alt.os.linux.]
IEC standard (IEC 80000-13:2008) to use the base-2 powers (kibi, mebi,
etc.). This was in part due to general confusion that the SI prefixes
really do mean base-10; and the growing disparity between base-2 and
base-10 --- as I recall 2.5% at kilo/kibi; and increasing every marked
order of magnitude (so 5% at mebi/mega ; 7.5% at gibi/giga ; 10% at
tebi/tera, and so on).
"Marketing" will quite likely continue to use base-10, because it makes
the numbers bigger. I was hopeful that with the move to base-2
numbering for SSDs, we were gonna see a move in accuracy on the box, but
that didn't happen :(
-----BEGIN PGP SIGNATURE-----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=sWFx
-----END PGP SIGNATURE-----
--
|_|O|_|
|_|_|O| Github: https://github.com/dpurgert
|O|O|O| PGP: DDAB 23FB 19FA 7D85 1CC1 E067 6D65 70E5 4CE7 2860
Hash: SHA512
["Followup-To:" header set to alt.os.linux.]
On 2023-01-28, Java Jive wrote:
> On 28/01/2023 22:30, Ravi Kapoor wrote:
>>
>> On 28/01/2023 21:51, Carlos E.R. wrote:
>>>
>>> The industry has very clear that 1 GiB = 1024 MiB, and 1 GB = 1000 MB.
>>>
>>> <https://en.wikipedia.org/wiki/Byte#Multiple-byte_units>
>>>
>>> <https://en.wikipedia.org/wiki/Gigabyte#Consumer_confusion>
>>
>> So you are very clever here. How do you work out 256GB = 231GB? Use your
>> figures to demonstrate this, Mr Clever.
>>
>> (256 x 1000) / 1024 = 250
>>
>> How does 250 compare with 231? You can reconcile the figures for us to
>> understand better.
>
> 'The industry', whatever exactly that means, may say one thing, but what
> really matters is what people do, and the fact remains that I rarely see
> GiB in practice, and OSs do one thing while disk manufacturers do another.
>
> In OSs, and BTW this is also what I was taught in academia, ...
> 1KB = 1024 bytes
> 1MB = 1024 KB
> 1GB = 1024 MB
> ... but to storage manufacturers ...
> 1KB = 1000 bytes
> 1MB = 1000 of their KB
> 1GB = 1000 of their MB
Sure, up to about 20 years ago, or thereabouts, culminating in a 2008
> On 28/01/2023 22:30, Ravi Kapoor wrote:
>>
>> On 28/01/2023 21:51, Carlos E.R. wrote:
>>>
>>> The industry has very clear that 1 GiB = 1024 MiB, and 1 GB = 1000 MB.
>>>
>>> <https://en.wikipedia.org/wiki/Byte#Multiple-byte_units>
>>>
>>> <https://en.wikipedia.org/wiki/Gigabyte#Consumer_confusion>
>>
>> So you are very clever here. How do you work out 256GB = 231GB? Use your
>> figures to demonstrate this, Mr Clever.
>>
>> (256 x 1000) / 1024 = 250
>>
>> How does 250 compare with 231? You can reconcile the figures for us to
>> understand better.
>
> 'The industry', whatever exactly that means, may say one thing, but what
> really matters is what people do, and the fact remains that I rarely see
> GiB in practice, and OSs do one thing while disk manufacturers do another.
>
> In OSs, and BTW this is also what I was taught in academia, ...
> 1KB = 1024 bytes
> 1MB = 1024 KB
> 1GB = 1024 MB
> ... but to storage manufacturers ...
> 1KB = 1000 bytes
> 1MB = 1000 of their KB
> 1GB = 1000 of their MB
IEC standard (IEC 80000-13:2008) to use the base-2 powers (kibi, mebi,
etc.). This was in part due to general confusion that the SI prefixes
really do mean base-10; and the growing disparity between base-2 and
base-10 --- as I recall 2.5% at kilo/kibi; and increasing every marked
order of magnitude (so 5% at mebi/mega ; 7.5% at gibi/giga ; 10% at
tebi/tera, and so on).
"Marketing" will quite likely continue to use base-10, because it makes
the numbers bigger. I was hopeful that with the move to base-2
numbering for SSDs, we were gonna see a move in accuracy on the box, but
that didn't happen :(
-----BEGIN PGP SIGNATURE-----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=sWFx
-----END PGP SIGNATURE-----
--
|_|O|_|
|_|_|O| Github: https://github.com/dpurgert
|O|O|O| PGP: DDAB 23FB 19FA 7D85 1CC1 E067 6D65 70E5 4CE7 2860
Ken Blake
Jan 29, 2023, 9:24:23 AM1/29/23
to
On Sat, 28 Jan 2023 23:06:07 +0000, Java Jive <ja...@evij.com.invalid>
wrote:
wrote:
>On 28/01/2023 22:30, Ravi Kapoor wrote:
>>
>> On 28/01/2023 21:51, Carlos E.R. wrote:
>>>
>>> The industry has very clear that 1 GiB = 1024 MiB, and 1 GB = 1000 MB.
>>>
>>> <https://en.wikipedia.org/wiki/Byte#Multiple-byte_units>
>>>
>>> <https://en.wikipedia.org/wiki/Gigabyte#Consumer_confusion>
>>
>> So you are very clever here. How do you work out 256GB = 231GB? Use your
>> figures to demonstrate this, Mr Clever.
>>
>> (256 x 1000) / 1024 = 250
>>
>> How does 250 compare with 231? You can reconcile the figures for us to
>> understand better.
>
>'The industry', whatever exactly that means, may say one thing, but what
>really matters is what people do, and the fact remains that I rarely see
>GiB in practice, and OSs do one thing while disk manufacturers do another.
Yes to everything in that paragraph.
>>
>> On 28/01/2023 21:51, Carlos E.R. wrote:
>>>
>>> The industry has very clear that 1 GiB = 1024 MiB, and 1 GB = 1000 MB.
>>>
>>> <https://en.wikipedia.org/wiki/Byte#Multiple-byte_units>
>>>
>>> <https://en.wikipedia.org/wiki/Gigabyte#Consumer_confusion>
>>
>> So you are very clever here. How do you work out 256GB = 231GB? Use your
>> figures to demonstrate this, Mr Clever.
>>
>> (256 x 1000) / 1024 = 250
>>
>> How does 250 compare with 231? You can reconcile the figures for us to
>> understand better.
>
>'The industry', whatever exactly that means, may say one thing, but what
>really matters is what people do, and the fact remains that I rarely see
>GiB in practice, and OSs do one thing while disk manufacturers do another.
Ken Blake
Jan 29, 2023, 9:27:40 AM1/29/23
to
On Sat, 28 Jan 2023 22:51:28 +0100, "Carlos E.R."
<robin_...@es.invalid> wrote:
>On 2023-01-28 21:30, Ravi Kapoor wrote:
>> On 28/01/2023 08:16, Ant wrote:
>>> In alt.os.linux Paul <nos...@needed.invalid> wrote:
>>>> On 1/27/2023 6:41 PM, Ant wrote:
>>>>> It's hard to find good reviews on these USB flash sticks. I still have
>>>>> old 128 MB that STILL work today.
>>>> That's because it is SLC.
>>> https://www.kingston.com/en/blog/pc-performance/difference-between-slc-mlc-tlc-3d-nand
>>> was an interesting read.
>> Have the flash industry managed to sort out the size of 1GB yet? I
>> bought a 256GB flash drive sometime ago but it had only 231GB. People
>> told me that it is because nobody uses correct measurement yet. Is 1GB =
>> 1000MB or is it 1024MB. I have not been able to reconcile how 256GB is
>> only 231GB when reformatted as NTFS or exFAT (default). I lost 25GB
>> unnecessarily.
>
>There is no confusion at all, except for you. You have to study and
>learn the different units.
>
>The industry has very clear that 1 GiB = 1024 MiB, and 1 GB = 1000 MB.
I don't agree. It may be clear to you and to me, and to others here,
<robin_...@es.invalid> wrote:
>On 2023-01-28 21:30, Ravi Kapoor wrote:
>> On 28/01/2023 08:16, Ant wrote:
>>> In alt.os.linux Paul <nos...@needed.invalid> wrote:
>>>> On 1/27/2023 6:41 PM, Ant wrote:
>>>>> It's hard to find good reviews on these USB flash sticks. I still have
>>>>> old 128 MB that STILL work today.
>>>> That's because it is SLC.
>>> https://www.kingston.com/en/blog/pc-performance/difference-between-slc-mlc-tlc-3d-nand
>>> was an interesting read.
>> Have the flash industry managed to sort out the size of 1GB yet? I
>> bought a 256GB flash drive sometime ago but it had only 231GB. People
>> told me that it is because nobody uses correct measurement yet. Is 1GB =
>> 1000MB or is it 1024MB. I have not been able to reconcile how 256GB is
>> only 231GB when reformatted as NTFS or exFAT (default). I lost 25GB
>> unnecessarily.
>
>There is no confusion at all, except for you. You have to study and
>learn the different units.
>
>The industry has very clear that 1 GiB = 1024 MiB, and 1 GB = 1000 MB.
but most people have never heard of GiB or MiB.
Carlos E.R.
Jan 29, 2023, 12:49:59 PM1/29/23
to
The computer industry have done it wrong for decades (looking at
Microsoft and others).
--
Cheers, Carlos.
Daniel65
Jan 30, 2023, 5:17:04 AM1/30/23
to
Carlos E.R. wrote on 29/1/23 8:42 pm:
<Snip>
> The October 2021 version of the dd manual says:
>
> N and BYTES may be followed by the following multiplicative
> suffixes: c=1, w=2, b=512, kB=1000, K=1024, MB=1000*1000, M=1024*1024,
> xM=M, GB=1000*1000*1000, G=1024*1024*1024, and so on for T, P, E, Z, Y.
> Binary prefixes can be used, too: KiB=K, MiB=M, and so on.
>
> which is not what they were doing previously. The February 2018 edition
> says instead:
>
> N and BYTES may be followed by the following
> multiplicative suffixes: c =1, w =2, b =512, kB =1000, K =1024, MB
> =1000*1000, M =1024*1024, xM =M, GB =1000*1000*1000, G =1024*1024*1024,
> and so on for T, P, E, Z, Y.
Hmm!! Might be Picky! Picky! but shouldn't ....
"N and BYTES may be followed by the following multiplicative
suffixes: ...."
really be ....
"N and BYTES may be *preceded* by the following multiplicative
suffixes: ...."
??
--
Daniel
<Snip>
> The October 2021 version of the dd manual says:
>
> N and BYTES may be followed by the following multiplicative
> suffixes: c=1, w=2, b=512, kB=1000, K=1024, MB=1000*1000, M=1024*1024,
> xM=M, GB=1000*1000*1000, G=1024*1024*1024, and so on for T, P, E, Z, Y.
> Binary prefixes can be used, too: KiB=K, MiB=M, and so on.
>
> which is not what they were doing previously. The February 2018 edition
> says instead:
>
> N and BYTES may be followed by the following
> multiplicative suffixes: c =1, w =2, b =512, kB =1000, K =1024, MB
> =1000*1000, M =1024*1024, xM =M, GB =1000*1000*1000, G =1024*1024*1024,
> and so on for T, P, E, Z, Y.
"N and BYTES may be followed by the following multiplicative
really be ....
"N and BYTES may be *preceded* by the following multiplicative
suffixes: ...."
??
--
Daniel
Daniel65
Jan 30, 2023, 5:27:19 AM1/30/23
to
Ken Blake wrote on 30/1/23 1:27 am:
I've heard/read of them both .... but am never confident which is
Decimal and which is Binary. At times I think it varies dependant on who
is speaking!!
--
Daniel
Decimal and which is Binary. At times I think it varies dependant on who
is speaking!!
--
Daniel
Carlos E.R.
Jan 30, 2023, 6:03:45 AM1/30/23
to
--
Cheers, Carlos.
Dan Purgert
Jan 30, 2023, 7:05:03 AM1/30/23
to
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512
["Followup-To:" header set to alt.os.linux.]
Hash: SHA512
["Followup-To:" header set to alt.os.linux.]
-----BEGIN PGP SIGNATURE-----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=g6s5
Daniel65
Jan 30, 2023, 7:28:35 AM1/30/23
to
Carlos E.R. wrote on 30/1/23 9:58 pm:
Oh!! Dear!! So you mean I may have been doing it wrong all these years!!
--
Daniel
--
Daniel
Ken Blake
Jan 30, 2023, 1:10:05 PM1/30/23
to
On Sun, 29 Jan 2023 18:43:57 +0100, "Carlos E.R."
Many English words, phrases, and abbreviations have been used
incorrectly for much longer than decades. But after a while their new
usage gets established, and nearly everyone uses it. What was once
wrong doesn't remain wrong forever. That's the nature of language; it
changes.
So as far as I'm concerned, despite the existing standards, KB mean
1024, MB means 1024 x 1024, GB means 1024 x 1024 x 1024, etc. and KiB,
MiB, Gib, etc. are almost never used and shouldn't be. A disk drive
that's called 2GB should have 2,147,483,648 bytes, not 2,000,000,000.
incorrectly for much longer than decades. But after a while their new
usage gets established, and nearly everyone uses it. What was once
wrong doesn't remain wrong forever. That's the nature of language; it
changes.
So as far as I'm concerned, despite the existing standards, KB mean
1024, MB means 1024 x 1024, GB means 1024 x 1024 x 1024, etc. and KiB,
MiB, Gib, etc. are almost never used and shouldn't be. A disk drive
that's called 2GB should have 2,147,483,648 bytes, not 2,000,000,000.
Richard Kettlewell
Jan 30, 2023, 1:36:35 PM1/30/23
to
1GB=2^30bytes, leading to the rather bizarre situation where a laptop
spec might claim, for example, 16GB RAM and 512GB SSD, with different
meanings for GB in each case.
--
https://www.greenend.org.uk/rjk/
Ken Blake
Jan 30, 2023, 2:18:19 PM1/30/23
to
Carlos E.R.
Jan 31, 2023, 5:15:32 AM1/31/23
to
--
Cheers, Carlos.
Daniel65
Jan 31, 2023, 5:27:13 AM1/31/23
to
Ken Blake wrote on 31/1/23 5:10 am:
times One Million ..... whereas in the U.S. of A. World "Billion" means
One Thousand times One Million??
--
Daniel
> On Sun, 29 Jan 2023 18:43:57 +0100, "Carlos E.R."
> <robin_...@es.invalid> wrote:
>> On 2023-01-29 15:27, Ken Blake wrote:
>>> On Sat, 28 Jan 2023 22:51:28 +0100, "Carlos E.R." <robin_...@es.invalid> wrote:
>
>>>> The industry has very clear that 1 GiB = 1024 MiB, and 1 GB = 1000 MB.
>>>
>>> I don't agree. It may be clear to you and to me, and to others here,
>>> but most people have never heard of GiB or MiB.
>>
>> Then learn.
>>
>> The computer industry have done it wrong for decades (looking at
>> Microsoft and others).
>
> Many English words, phrases, and abbreviations have been used
> incorrectly for much longer than decades. But after a while their new
> usage gets established, and nearly everyone uses it. What was once
> wrong doesn't remain wrong forever. That's the nature of language; it
> changes.
Back in the day, in the English World didn't "Billion" mean One Million
> <robin_...@es.invalid> wrote:
>> On 2023-01-29 15:27, Ken Blake wrote:
>>> On Sat, 28 Jan 2023 22:51:28 +0100, "Carlos E.R." <robin_...@es.invalid> wrote:
>
>>>> The industry has very clear that 1 GiB = 1024 MiB, and 1 GB = 1000 MB.
>>>
>>> I don't agree. It may be clear to you and to me, and to others here,
>>> but most people have never heard of GiB or MiB.
>>
>> Then learn.
>>
>> The computer industry have done it wrong for decades (looking at
>> Microsoft and others).
>
> Many English words, phrases, and abbreviations have been used
> incorrectly for much longer than decades. But after a while their new
> usage gets established, and nearly everyone uses it. What was once
> wrong doesn't remain wrong forever. That's the nature of language; it
> changes.
times One Million ..... whereas in the U.S. of A. World "Billion" means
One Thousand times One Million??
--
Daniel
Pancho
Jan 31, 2023, 5:58:17 AM1/31/23
to
We use decimal for most other stuff, why would we want to use binary for
this special case? K means 10^3 not 2^10, M means 10^6, G means 10^9.
Why introduce complexity, unnecessary special cases?
What advantage do you think 2^10, 2^20 offers?
Carlos E. R.
Jan 31, 2023, 9:09:57 AM1/31/23
to
of RAM" than "1024 bytes", and the difference is small, thus ignored.
Problem is, we now use gigas or teras, and the difference is notable.
As computers are recent and the unit prefixes like K, M, G... are older,
it is computerese which has to adapt and use new and different prefixes,
not usurp the old prefixes introducing confusion.
--
Cheers,
Carlos E.R.
Richard Kettlewell
Jan 31, 2023, 9:40:27 AM1/31/23
to
> What advantage do you think 2^10, 2^20 offers?
instead of 17.179869184GB RAM.
Persistent storage is less likely to be a power of 2 (even if the
underlying medium is 2^n bytes, there’s generally space reserved for
metadata, wear-levelling, error management, firmware, etc) but it’s
still generally divisible by at least 2^12 and often a larger power of
2. It’s usually not divisible by a nontrivial power of 10, so the
capacities in decimal units are usually not very precise.
--
https://www.greenend.org.uk/rjk/
Ken Blake
Jan 31, 2023, 10:05:25 AM1/31/23
to
On Tue, 31 Jan 2023 11:09:18 +0100, "Carlos E.R."
Yes, I know. But as I said, that's a decree I disagree with.
Ken Blake
Jan 31, 2023, 10:32:30 AM1/31/23
to
On Tue, 31 Jan 2023 10:58:15 +0000, Pancho <Pancho...@proton.me>
wrote:
>On 30/01/2023 18:10, Ken Blake wrote:
>> On Sun, 29 Jan 2023 18:43:57 +0100, "Carlos E.R."
>> <robin_...@es.invalid> wrote:
>>
>>> On 2023-01-29 15:27, Ken Blake wrote:
>>>> On Sat, 28 Jan 2023 22:51:28 +0100, "Carlos E.R." <robin_...@es.invalid> wrote:
>>
>>>>> The industry has very clear that 1 GiB = 1024 MiB, and 1 GB = 1000 MB.
>>>>
>>>>
>>>> I don't agree. It may be clear to you and to me, and to others here,
>>>> but most people have never heard of GiB or MiB.
>>>
>>> Then learn.
>>>
>>> The computer industry have done it wrong for decades (looking at
>>> Microsoft and others).
>>
>>
>> Many English words, phrases, and abbreviations have been used
>> incorrectly for much longer than decades. But after a while their new
>> usage gets established, and nearly everyone uses it. What was once
>> wrong doesn't remain wrong forever. That's the nature of language; it
>> changes.
>>
>> So as far as I'm concerned, despite the existing standards, KB mean
>> 1024, MB means 1024 x 1024, GB means 1024 x 1024 x 1024, etc. and KiB,
>> MiB, Gib, etc. are almost never used and shouldn't be. A disk drive
>> that's called 2GB should have 2,147,483,648 bytes, not 2,000,000,000.
>>
>
>But...
>
>We use decimal for most other stuff,
Most? Yes, but not for computers.
>why would we want to use binary for
>this special case?
Want to? It's not a matter of wanting to. We should accept it because
that's the way it's done with computers, except by drive
manufacturers.
>K means 10^3 not 2^10, M means 10^6, G means 10^9.
Except for computers, yes.
>Why introduce complexity, unnecessary special cases?
I don't want to introduce complexity, or have unnecessary special
cases. My point was that it's *already* been introduced and
established *in the computer world*. Going against what is established
is exactly what introduces unnecessary special cases, and confuses
people.
The number of people who buy a 1TB hard drive and are confused
because Windows tells them it's only around 900GB is enormous; look in
the Windows newsgroups and online forums and see how many people ask
"what happened to the other 100GB?"
Technically, you are right, of course. Using the established standard,
it's 1,000,000,000,000 bytes--1TB--and that's what drive manufacturers
call it. But according to Windows and most of the rest of the computer
word, 1TB is 1,099,511,637,776 bytes, so their "1TB" is only around
900 GB.
My point, once again, is that when drive manufacturers use the
established standard for disk drives when almost the rest of the
computer world does it differently, it confuses people and is a bad
thing to do. In my view, this is a case where consistency is more
important than standards.
>What advantage do you think 2^10, 2^20 offers?
That question is irrelevant to my point. I'm talking only about the
computer world.
wrote:
>On 30/01/2023 18:10, Ken Blake wrote:
>> On Sun, 29 Jan 2023 18:43:57 +0100, "Carlos E.R."
>> <robin_...@es.invalid> wrote:
>>
>>> On 2023-01-29 15:27, Ken Blake wrote:
>>>> On Sat, 28 Jan 2023 22:51:28 +0100, "Carlos E.R." <robin_...@es.invalid> wrote:
>>
>>>>> The industry has very clear that 1 GiB = 1024 MiB, and 1 GB = 1000 MB.
>>>>
>>>>
>>>> I don't agree. It may be clear to you and to me, and to others here,
>>>> but most people have never heard of GiB or MiB.
>>>
>>> Then learn.
>>>
>>> The computer industry have done it wrong for decades (looking at
>>> Microsoft and others).
>>
>>
>> Many English words, phrases, and abbreviations have been used
>> incorrectly for much longer than decades. But after a while their new
>> usage gets established, and nearly everyone uses it. What was once
>> wrong doesn't remain wrong forever. That's the nature of language; it
>> changes.
>>
>> So as far as I'm concerned, despite the existing standards, KB mean
>> 1024, MB means 1024 x 1024, GB means 1024 x 1024 x 1024, etc. and KiB,
>> MiB, Gib, etc. are almost never used and shouldn't be. A disk drive
>> that's called 2GB should have 2,147,483,648 bytes, not 2,000,000,000.
>>
>
>But...
>
>We use decimal for most other stuff,
>why would we want to use binary for
>this special case?
that's the way it's done with computers, except by drive
manufacturers.
>K means 10^3 not 2^10, M means 10^6, G means 10^9.
>Why introduce complexity, unnecessary special cases?
cases. My point was that it's *already* been introduced and
established *in the computer world*. Going against what is established
is exactly what introduces unnecessary special cases, and confuses
people.
The number of people who buy a 1TB hard drive and are confused
because Windows tells them it's only around 900GB is enormous; look in
the Windows newsgroups and online forums and see how many people ask
"what happened to the other 100GB?"
Technically, you are right, of course. Using the established standard,
it's 1,000,000,000,000 bytes--1TB--and that's what drive manufacturers
call it. But according to Windows and most of the rest of the computer
word, 1TB is 1,099,511,637,776 bytes, so their "1TB" is only around
900 GB.
My point, once again, is that when drive manufacturers use the
established standard for disk drives when almost the rest of the
computer world does it differently, it confuses people and is a bad
thing to do. In my view, this is a case where consistency is more
important than standards.
>What advantage do you think 2^10, 2^20 offers?
computer world.
Carlos E. R.
Jan 31, 2023, 10:54:59 AM1/31/23
to
the world is right.
Computer people have to adapt and say 1 GiB instead og 1 GB. Hard disk
people are doing it right since decades.
Microsoft, typically, hates standards and goes against.
--
Cheers,
Carlos E.R.
Ken Blake
Jan 31, 2023, 1:21:16 PM1/31/23
to
On Tue, 31 Jan 2023 16:54:57 +0100, "Carlos E. R."
<robin_...@es.invalid> wrote:
>On 2023-01-31 16:32, Ken Blake wrote:
>> On Tue, 31 Jan 2023 10:58:15 +0000, Pancho <Pancho...@proton.me>
>> wrote:
>
>
>> My point, once again, is that when drive manufacturers use the
>> established standard for disk drives when almost the rest of the
>> computer world does it differently, it confuses people and is a bad
>> thing to do. In my view, this is a case where consistency is more
>> important than standards.
>
>Mine is that those computer people are doing it wrong, and the rest of
>the world is right.
Going by the standards, you are of course correct.
<robin_...@es.invalid> wrote:
>On 2023-01-31 16:32, Ken Blake wrote:
>> On Tue, 31 Jan 2023 10:58:15 +0000, Pancho <Pancho...@proton.me>
>> wrote:
>
>
>> My point, once again, is that when drive manufacturers use the
>> established standard for disk drives when almost the rest of the
>> computer world does it differently, it confuses people and is a bad
>> thing to do. In my view, this is a case where consistency is more
>> important than standards.
>
>Mine is that those computer people are doing it wrong, and the rest of
>the world is right.
But it doesn't matter. What matters is what considered correct by most
people.
>Computer people have to adapt and say 1 GiB instead og 1 GB.
get consistency and that's for the drive manufacturers to use the
common understanding of KB, MB, GB etc. that the rest of the computer
world uses.
You will have course say that the drive manufacturers shouldn't change
and the rest of the computer world should. That might be a good choice
if it were possible, but it's not. There are only a handful of drive
manufacturers, but *millions* of computer users. You're not going to
change those millions.
>Hard disk
>people are doing it right since decades.
>
>Microsoft, typically, hates standards and goes against.
We agree on that. Is Microsoft responsible for the common meanings of
>Microsoft, typically, hates standards and goes against.
MB, GB, TB, etc. being different from the standards and being used the
way they are? Probably.
But it doesn't matter who is responsible. Whether you or I like it or
not (I also don't like it, but I have no real choice other than to
accept it), that's the way it is, and we are not going to change it.
Paul
Jan 31, 2023, 2:36:35 PM1/31/23
to
On 1/31/2023 5:58 AM, Pancho wrote:
>
> But...
>
> We use decimal for most other stuff, why would we want to use binary
> for this special case? K means 10^3 not 2^10, M means 10^6, G means 10^9.
> Why introduce complexity, unnecessary special cases?
>
> What advantage do you think 2^10, 2^20 offers?
Decoding logic is simpler when you use powers-of-two.
>
> But...
>
> We use decimal for most other stuff, why would we want to use binary
> for this special case? K means 10^3 not 2^10, M means 10^6, G means 10^9.
> Why introduce complexity, unnecessary special cases?
>
> What advantage do you think 2^10, 2^20 offers?
This was important... a long time ago.
In this example, someone uses a '139 to decode an address and select a device with it.
https://blog.idorobots.org/media/upnod3/ram.png
Paul
Carlos E.R.
Jan 31, 2023, 3:46:56 PM1/31/23
to
On 2023-01-31 19:21, Ken Blake wrote:
> On Tue, 31 Jan 2023 16:54:57 +0100, "Carlos E. R."
> <robin_...@es.invalid> wrote:
>
>> On 2023-01-31 16:32, Ken Blake wrote:
>>> On Tue, 31 Jan 2023 10:58:15 +0000, Pancho <Pancho...@proton.me>
>>> wrote:
>>
>>
>>> My point, once again, is that when drive manufacturers use the
>>> established standard for disk drives when almost the rest of the
>>> computer world does it differently, it confuses people and is a bad
>>> thing to do. In my view, this is a case where consistency is more
>>> important than standards.
>>
>> Mine is that those computer people are doing it wrong, and the rest of
>> the world is right.
>
> Going by the standards, you are of course correct.
>
> But it doesn't matter. What matters is what considered correct by most
> people.
>
>> Computer people have to adapt and say 1 GiB instead og 1 GB.
>
> "Have to"? Not a chance. It will never happen. There's only one way to
> get consistency and that's for the drive manufacturers to use the
> common understanding of KB, MB, GB etc. that the rest of the computer
> world uses.
Not going to happen :-)
> On Tue, 31 Jan 2023 16:54:57 +0100, "Carlos E. R."
> <robin_...@es.invalid> wrote:
>
>> On 2023-01-31 16:32, Ken Blake wrote:
>>> On Tue, 31 Jan 2023 10:58:15 +0000, Pancho <Pancho...@proton.me>
>>> wrote:
>>
>>
>>> My point, once again, is that when drive manufacturers use the
>>> established standard for disk drives when almost the rest of the
>>> computer world does it differently, it confuses people and is a bad
>>> thing to do. In my view, this is a case where consistency is more
>>> important than standards.
>>
>> Mine is that those computer people are doing it wrong, and the rest of
>> the world is right.
>
> Going by the standards, you are of course correct.
>
> But it doesn't matter. What matters is what considered correct by most
> people.
>
>> Computer people have to adapt and say 1 GiB instead og 1 GB.
>
> "Have to"? Not a chance. It will never happen. There's only one way to
> get consistency and that's for the drive manufacturers to use the
> common understanding of KB, MB, GB etc. that the rest of the computer
> world uses.
>
> You will have course say that the drive manufacturers shouldn't change
> and the rest of the computer world should. That might be a good choice
> if it were possible, but it's not. There are only a handful of drive
> manufacturers, but *millions* of computer users. You're not going to
> change those millions.
>
>> Hard disk
>> people are doing it right since decades.
>
>
> Technically, yes. Practically, no.
>>
>> Microsoft, typically, hates standards and goes against.
>
> We agree on that. Is Microsoft responsible for the common meanings of
> MB, GB, TB, etc. being different from the standards and being used the
> way they are? Probably.
>
> But it doesn't matter who is responsible. Whether you or I like it or
> not (I also don't like it, but I have no real choice other than to
> accept it), that's the way it is, and we are not going to change it.
Cheers, Carlos.
Pancho
Jan 31, 2023, 5:12:15 PM1/31/23
to
On 1/31/23 14:40, Richard Kettlewell wrote:
> Pancho <Pancho...@proton.me> writes:
>> On 30/01/2023 18:10, Ken Blake wrote:
>>> So as far as I'm concerned, despite the existing standards, KB mean
>>> 1024, MB means 1024 x 1024, GB means 1024 x 1024 x 1024, etc. and KiB,
>>> MiB, Gib, etc. are almost never used and shouldn't be. A disk drive
>>> that's called 2GB should have 2,147,483,648 bytes, not 2,000,000,000.
>>
>> But...
>>
>> We use decimal for most other stuff, why would we want to use binary
>> for this special case? K means 10^3 not 2^10, M means 10^6, G means
>> 10^9. Why introduce complexity, unnecessary special cases?
>
> Well, it’s hardly ‘introduce’ any more, the convention is decades old.
>
>> What advantage do you think 2^10, 2^20 offers?
>
> Being able to talking about 16GB RAM (or 16GiB if you really must)
> instead of 17.179869184GB RAM.
>
RAM is never 16 GiB either, areas will be reserved by the OS. For
> Pancho <Pancho...@proton.me> writes:
>> On 30/01/2023 18:10, Ken Blake wrote:
>>> So as far as I'm concerned, despite the existing standards, KB mean
>>> 1024, MB means 1024 x 1024, GB means 1024 x 1024 x 1024, etc. and KiB,
>>> MiB, Gib, etc. are almost never used and shouldn't be. A disk drive
>>> that's called 2GB should have 2,147,483,648 bytes, not 2,000,000,000.
>>
>> But...
>>
>> We use decimal for most other stuff, why would we want to use binary
>> for this special case? K means 10^3 not 2^10, M means 10^6, G means
>> 10^9. Why introduce complexity, unnecessary special cases?
>
> Well, it’s hardly ‘introduce’ any more, the convention is decades old.
>
>> What advantage do you think 2^10, 2^20 offers?
>
> Being able to talking about 16GB RAM (or 16GiB if you really must)
> instead of 17.179869184GB RAM.
>
instance, 4 GiB on Windows XP 32 only had about 3.1 GiB available for my
use.
Floating point numbers cannot represent rational numbers like 1/3
precisely/concisely. We live with them for simplicity.
My point was that when I do a calculation for how much RAM I need, I use
decimal, ~17.2 * 10^9 B is the number I want to know, not the prettier
16 GiB.
If you support both systems, you encourage confusion.
Other than that, I agree with what Carlos is saying.
Pancho
Jan 31, 2023, 5:12:34 PM1/31/23
to
Forty years ago I did that stuff, but not recently. I use high level
languages, and the kids I worked with understood far less than I did,
about that type of stuff, and didn't care.
Just as I don't know how to double declutch a car.
Richard Kettlewell
Feb 1, 2023, 3:35:34 AM2/1/23
to
Pancho <Pancho...@proton.me> writes:
> On 1/31/23 14:40, Richard Kettlewell wrote:
>> Pancho <Pancho...@proton.me> writes:
>>> On 30/01/2023 18:10, Ken Blake wrote:
>>>> So as far as I'm concerned, despite the existing standards, KB mean
>>>> 1024, MB means 1024 x 1024, GB means 1024 x 1024 x 1024, etc. and KiB,
>>>> MiB, Gib, etc. are almost never used and shouldn't be. A disk drive
>>>> that's called 2GB should have 2,147,483,648 bytes, not 2,000,000,000.
>>>
>>> But...
>>>
>>> We use decimal for most other stuff, why would we want to use binary
>>> for this special case? K means 10^3 not 2^10, M means 10^6, G means
>>> 10^9. Why introduce complexity, unnecessary special cases?
>> Well, it’s hardly ‘introduce’ any more, the convention is decades
>> old.
>>
>>> What advantage do you think 2^10, 2^20 offers?
>> Being able to talking about 16GB RAM (or 16GiB if you really must)
>> instead of 17.179869184GB RAM.
>
> RAM is never 16 GiB either, areas will be reserved by the OS. For
> instance, 4 GiB on Windows XP 32 only had about 3.1 GiB available for
> my use.
Why did you bother to ask the question when you’re not going to accept
> On 1/31/23 14:40, Richard Kettlewell wrote:
>> Pancho <Pancho...@proton.me> writes:
>>> On 30/01/2023 18:10, Ken Blake wrote:
>>>> So as far as I'm concerned, despite the existing standards, KB mean
>>>> 1024, MB means 1024 x 1024, GB means 1024 x 1024 x 1024, etc. and KiB,
>>>> MiB, Gib, etc. are almost never used and shouldn't be. A disk drive
>>>> that's called 2GB should have 2,147,483,648 bytes, not 2,000,000,000.
>>>
>>> But...
>>>
>>> We use decimal for most other stuff, why would we want to use binary
>>> for this special case? K means 10^3 not 2^10, M means 10^6, G means
>>> 10^9. Why introduce complexity, unnecessary special cases?
>> Well, it’s hardly ‘introduce’ any more, the convention is decades
>> old.
>>
>>> What advantage do you think 2^10, 2^20 offers?
>> Being able to talking about 16GB RAM (or 16GiB if you really must)
>> instead of 17.179869184GB RAM.
>
> RAM is never 16 GiB either, areas will be reserved by the OS. For
> instance, 4 GiB on Windows XP 32 only had about 3.1 GiB available for
> my use.
the answer?
--
https://www.greenend.org.uk/rjk/
Pancho
Feb 1, 2023, 5:43:46 AM2/1/23
to
When I see something I consider bad practice, I always ask why? What
benefit does it offer?
Quite often I have missed something important, other times I haven't.
Don't you do that? How else do we understand stuff?
Anyway, I wasn't even rejecting your answer. I just didn't feel the
concise notation merited the complexity of a special case.
Richard Kettlewell
Feb 1, 2023, 8:36:23 AM2/1/23
to
so.
--
https://www.greenend.org.uk/rjk/
Ken Blake
Feb 1, 2023, 10:25:21 AM2/1/23
to
On Tue, 31 Jan 2023 22:12:13 +0000, Pancho <Pancho...@proton.me>
wrote:
>On 1/31/23 14:40, Richard Kettlewell wrote:
>> Pancho <Pancho...@proton.me> writes:
>>> On 30/01/2023 18:10, Ken Blake wrote:
>>>> So as far as I'm concerned, despite the existing standards, KB mean
>>>> 1024, MB means 1024 x 1024, GB means 1024 x 1024 x 1024, etc. and KiB,
>>>> MiB, Gib, etc. are almost never used and shouldn't be. A disk drive
>>>> that's called 2GB should have 2,147,483,648 bytes, not 2,000,000,000.
>>>
>>> But...
>>>
>>> We use decimal for most other stuff, why would we want to use binary
>>> for this special case? K means 10^3 not 2^10, M means 10^6, G means
>>> 10^9. Why introduce complexity, unnecessary special cases?
>>
>> Well, it’s hardly ‘introduce’ any more, the convention is decades old.
>>
>>> What advantage do you think 2^10, 2^20 offers?
>>
>> Being able to talking about 16GB RAM (or 16GiB if you really must)
>> instead of 17.179869184GB RAM.
>>
>
>RAM is never 16 GiB either,
Yes it is.
> areas will be reserved by the OS. For
>instance, 4 GiB on Windows XP 32 only had about 3.1 GiB available for my
>use.
Yes, but how much RAM there is and how much is available for your use
are two different things. The amount of RAM installed on your computer
is 4GB, not 3.1GB.
For example If I go to System>About, here under Windows 11, it says
Installed RAM 32.0 GB (31.8 usable). How much is installed and how
much is usable are two different things.
wrote:
>On 1/31/23 14:40, Richard Kettlewell wrote:
>> Pancho <Pancho...@proton.me> writes:
>>> On 30/01/2023 18:10, Ken Blake wrote:
>>>> So as far as I'm concerned, despite the existing standards, KB mean
>>>> 1024, MB means 1024 x 1024, GB means 1024 x 1024 x 1024, etc. and KiB,
>>>> MiB, Gib, etc. are almost never used and shouldn't be. A disk drive
>>>> that's called 2GB should have 2,147,483,648 bytes, not 2,000,000,000.
>>>
>>> But...
>>>
>>> We use decimal for most other stuff, why would we want to use binary
>>> for this special case? K means 10^3 not 2^10, M means 10^6, G means
>>> 10^9. Why introduce complexity, unnecessary special cases?
>>
>> Well, it’s hardly ‘introduce’ any more, the convention is decades old.
>>
>>> What advantage do you think 2^10, 2^20 offers?
>>
>> Being able to talking about 16GB RAM (or 16GiB if you really must)
>> instead of 17.179869184GB RAM.
>>
>
>RAM is never 16 GiB either,
> areas will be reserved by the OS. For
>instance, 4 GiB on Windows XP 32 only had about 3.1 GiB available for my
>use.
are two different things. The amount of RAM installed on your computer
is 4GB, not 3.1GB.
For example If I go to System>About, here under Windows 11, it says
Installed RAM 32.0 GB (31.8 usable). How much is installed and how
much is usable are two different things.
Ken Blake
Feb 1, 2023, 10:28:00 AM2/1/23
to
On Tue, 31 Jan 2023 21:44:02 +0100, "Carlos E.R."
<robin_...@es.invalid> wrote:
>On 2023-01-31 19:21, Ken Blake wrote:
>> On Tue, 31 Jan 2023 16:54:57 +0100, "Carlos E. R."
>> <robin_...@es.invalid> wrote:
>>
>>> On 2023-01-31 16:32, Ken Blake wrote:
>>>> On Tue, 31 Jan 2023 10:58:15 +0000, Pancho <Pancho...@proton.me>
>>>> wrote:
>>>
>>>
>>>> My point, once again, is that when drive manufacturers use the
>>>> established standard for disk drives when almost the rest of the
>>>> computer world does it differently, it confuses people and is a bad
>>>> thing to do. In my view, this is a case where consistency is more
>>>> important than standards.
>>>
>>> Mine is that those computer people are doing it wrong, and the rest of
>>> the world is right.
>>
>> Going by the standards, you are of course correct.
>>
>> But it doesn't matter. What matters is what considered correct by most
>> people.
>>
>>> Computer people have to adapt and say 1 GiB instead og 1 GB.
>>
>> "Have to"? Not a chance. It will never happen. There's only one way to
>> get consistency and that's for the drive manufacturers to use the
>> common understanding of KB, MB, GB etc. that the rest of the computer
>> world uses.
>
>Not going to happen :-)
We agree on that.
<robin_...@es.invalid> wrote:
>On 2023-01-31 19:21, Ken Blake wrote:
>> On Tue, 31 Jan 2023 16:54:57 +0100, "Carlos E. R."
>> <robin_...@es.invalid> wrote:
>>
>>> On 2023-01-31 16:32, Ken Blake wrote:
>>>> On Tue, 31 Jan 2023 10:58:15 +0000, Pancho <Pancho...@proton.me>
>>>> wrote:
>>>
>>>
>>>> My point, once again, is that when drive manufacturers use the
>>>> established standard for disk drives when almost the rest of the
>>>> computer world does it differently, it confuses people and is a bad
>>>> thing to do. In my view, this is a case where consistency is more
>>>> important than standards.
>>>
>>> Mine is that those computer people are doing it wrong, and the rest of
>>> the world is right.
>>
>> Going by the standards, you are of course correct.
>>
>> But it doesn't matter. What matters is what considered correct by most
>> people.
>>
>>> Computer people have to adapt and say 1 GiB instead og 1 GB.
>>
>> "Have to"? Not a chance. It will never happen. There's only one way to
>> get consistency and that's for the drive manufacturers to use the
>> common understanding of KB, MB, GB etc. that the rest of the computer
>> world uses.
>
>Not going to happen :-)
Carlos E.R.
Feb 1, 2023, 4:39:10 PM2/1/23
to
--
Cheers, Carlos.
Pancho
Feb 1, 2023, 5:21:24 PM2/1/23
to
> For example If I go to System>About, here under Windows 11, it says
> Installed RAM 32.0 GB (31.8 usable). How much is installed and how
> much is usable are two different things.
>
exactly 15.8 GiB or only accurate to 1 dp. Either way, the prettiness of
exactly 16 has gone.
The prettiness is gone, but I'm still left with the problem that if I
want to calculate the amount of usable RAM my software data structures
require, in GiB, I have to convert my natural decimal calculations to a
binary format, to avoid the 7.4% difference between the GiB, and the
more orthodox decimal GB. Maybe other programmer don't estimate memory
requirements, don't use algorithms that require a lot of memory? It
wouldn't surprise me, innumeracy is surprisingly high in IT.
Clinging to unnecessary complexity reminds me of the metric martyrs and
their insistence on using imperial weights and measures. When automating
some business process, you often see veterans of the industry try to
cling to unnecessary complexity. I guess if you remove the complexity,
the competitive advantage they have in understanding it disappears, they
are diminished.
As far as I can see, the US government and the standards organizations
have agreed on the decimal GB.
Ken Blake
Feb 1, 2023, 5:41:36 PM2/1/23
to
On Wed, 1 Feb 2023 22:21:22 +0000, Pancho <Pancho...@proton.me>
As do I. But what either of us cares about has nothing to do with how
much RAM 4GB is.
much RAM 4GB is.
Daniel65
Feb 2, 2023, 3:51:37 AM2/2/23
to
Pancho wrote on 2/2/23 9:21 am:
How much of that 3.1GB would be useful for you if the other 0.9GB were
not being used?? ;-P I'm guessing NONE!!
--
Daniel
not being used?? ;-P I'm guessing NONE!!
--
Daniel
Carlos E.R.
Feb 2, 2023, 4:38:53 AM2/2/23
to
standards organizations, there would be no doubts about what that "16 GB
and 15.8 GB" of yours actually means. If some one writes GB it is
decimal, or else he writes GiB. No need to second guess.
--
Cheers, Carlos.
Ken Blake
Feb 2, 2023, 10:41:35 AM2/2/23
to
stick to the powers of two definitions of GB, etc. there would be no
doubt about what KB, MB, GB, TB, etc. meant. What's most important is
consistency in the computer world, not what definitions are used.
I don't think we'll ever get consistency.
Carlos E. R.
Feb 2, 2023, 11:12:40 AM2/2/23
to
have to deduce from context if this is 1000 or 1024.
Like disc manufacturers, always using 10^
>
> I don't think we'll ever get consistency.
calculations. But now that there is a standardization by the
organizations that do standards, I decided to accept it in full.like it
or not.
--
Cheers,
Carlos E.R.
Daniel65
Feb 3, 2023, 6:24:04 AM2/3/23
to
Carlos E. R. wrote on 3/2/23 3:12 am:
>> I completely agree. But alternatively if all software and docs would
>> stick to the powers of two definitions of GB, etc. there would be no
>> doubt about what KB, MB, GB, TB, etc. meant. What's most important is
>> consistency in the computer world, not what definitions are used.
>
> But there would be confusion to the many people for which K is 1000, and
> have to deduce from context if this is 1000 or 1024.
Should we all go on strike until Society accepts that the smaller
letter, 'k', represents the smaller number, 1000, and the larger letter
. 'K', represents the larger number, 1024?? Etc., etc.!
Who's with me?? ;-P
--
Daniel
> On 2023-02-02 16:41, Ken Blake wrote:
<Snip>
>> I completely agree. But alternatively if all software and docs would
>> stick to the powers of two definitions of GB, etc. there would be no
>> doubt about what KB, MB, GB, TB, etc. meant. What's most important is
>> consistency in the computer world, not what definitions are used.
>
> But there would be confusion to the many people for which K is 1000, and
> have to deduce from context if this is 1000 or 1024.
letter, 'k', represents the smaller number, 1000, and the larger letter
. 'K', represents the larger number, 1024?? Etc., etc.!
Who's with me?? ;-P
--
Daniel
Ken Blake
Feb 3, 2023, 9:12:46 AM2/3/23
to
recognize the difference and too hard to remember what's what.
K and k is essentially no different from KB and KiB, just spelled
differently.
Carlos E. R.
Feb 3, 2023, 9:48:28 AM2/3/23
to
And the SI already says it is "k", lower case, which means "kilo". There
is no "K". So that would be a new change on the standards.
B and b are used for byte and bits, respectively. And we forget.
--
Cheers,
Carlos E.R.
jjb
Feb 3, 2023, 10:22:35 AM2/3/23
to
rather silly).
Eric Pozharski
Feb 4, 2023, 1:33:09 PM2/4/23
to
with <ZE9DL.424202$Tcw8....@fx10.iad> jjb wrote:
> On 03-02-2023 15:48, Carlos E. R. wrote:
>> On 2023-02-03 15:12, Ken Blake wrote:
>>> On Fri, 3 Feb 2023 22:24:01 +1100, Daniel65
>>> <dani...@nomail.afraid.org> wrote:
>>>> Carlos E. R. wrote on 3/2/23 3:12 am:
>>>>> On 2023-02-02 16:41, Ken Blake wrote:
*SKIP*
> On 03-02-2023 15:48, Carlos E. R. wrote:
>> On 2023-02-03 15:12, Ken Blake wrote:
>>> On Fri, 3 Feb 2023 22:24:01 +1100, Daniel65
>>> <dani...@nomail.afraid.org> wrote:
>>>> Carlos E. R. wrote on 3/2/23 3:12 am:
>>>>> On 2023-02-02 16:41, Ken Blake wrote:
>> And the SI already says it is "k", lower case, which means "kilo".
>> There is no "K". So that would be a new change on the standards.
>> B and b are used for byte and bits, respectively. And we forget.
> Furthermore, a lowercase m stands for milli (which, for bytes, seems
> rather silly).
Unless speeds or densities.
>> There is no "K". So that would be a new change on the standards.
>> B and b are used for byte and bits, respectively. And we forget.
> Furthermore, a lowercase m stands for milli (which, for bytes, seems
> rather silly).
Also, </usr/share/misc/units.dat> is worth checking. Turns out 'K' is
already taken (it's Kelvin).
Also, I was musing about nice tangent: is byte primitive or derived?
If derived then does it come from mass and/or temperature? Turns out --
dead end, 'bit' is kinda primitive, 'byte' is derived. Such a loss :(
--
Torvalds' goal for Linux is very simple: World Domination
Stallman's goal for GNU is even simpler: Freedom
0 new messages