Existing PC is ASUS Vivobook 15 X512JA-F512JA with 256 GB nvme SSD.

Every once in a while I do a linux ddCopy of the Vivobook SSD onto an
external HDD and when I do that I write down the speed that is reported
upon completion. So I have a record of that speed for the last few
years. I use the same USB port and the same external HDD for target when
I record the speed. Also, more frequently, using Windows to back up a
large directory of many files from the Vivobook to a thumb drive,
however I don't have the speed data for that.

Recently I noticed that copying the large directory to thumb drive
seemed to be much much slower than it previously was. Now today, I did
the linux ddCopy of the Vivobook and I notice the reported speed is much
slower than before.

Previously it was 113 MB/sec but now it is 87 MB/sec.

I have heard that nvme SSDs somehow have substantial performance
degradation over time. Apparently it is bad to copy huge chunks of data
onto a nvme SSD. Actually I have done that kind of thing only once on
this Vivobook. When I first heard about the nvme problem, I read about
the reason but I decided it was beyond my comprehension (unlike HDDs).

I'm going to buy a new laptop before the drive on this one goes out. I
intend to avoid nvme SSDs, also I want a laptop that lasts longer than a
few years - unlike this Vivobook. Except for the degraded speed of
copying a large quantity of data to external storage I am happy with the
Vivobook features and performance. I am not a computer 'power' freak. It
seems like the nvme technology is questionable and I want to avoid it. I
will appreciate your suggestions.

Thanks. Bill S.

On 3/21/2024 2:43 PM, bilsch01 wrote:
> Existing PC is ASUS Vivobook 15 X512JA-F512JA with 256 GB nvme SSD.
>
> Every once in a while I do a linux ddCopy of the Vivobook SSD onto an external HDD and when I do that I write down the speed that is reported upon completion. So I have a record of that speed for the last few years. I use the same USB port and the same external HDD for target when I record the speed. Also, more frequently, using Windows to back up a large directory of many files from the Vivobook to a thumb drive, however I don't have the speed data for that.
>
> Recently I noticed that copying the large directory to thumb drive seemed to be much much slower than it previously was. Now today, I did the linux ddCopy of the Vivobook and I notice the reported speed is much slower than before.
>
> Previously it was 113 MB/sec but now it is 87 MB/sec.
>
> I have heard that nvme SSDs somehow have substantial performance degradation over time. Apparently it is bad to copy huge chunks of data onto a nvme SSD. Actually I have done that kind of thing only once on this Vivobook. When I first heard about the nvme problem, I read about the reason but I decided it was beyond my comprehension (unlike HDDs).
>
> I'm going to buy a new laptop before the drive on this one goes out. I intend to avoid nvme SSDs, also I want a laptop that lasts longer than a few years - unlike this Vivobook. Except for the degraded speed of copying a large quantity of data to external storage I am happy with the Vivobook features and performance. I am not a computer 'power' freak. It seems like the nvme technology is questionable and I want to avoid it. I will appreciate your suggestions.
>
> Thanks.    Bill S.

It doesn't matter what kind of computer you've got today,
benchmarking and making bold statements about speed, there
are a lot of ways you can get the wrong answer.

Both Windows and Linux have "System Write Cache". This interferes
with the dynamics of transfer. It means if "dd" prints a
result, the result is WRONG.

The NVMe is infinitely more reliable than a USB thumb drive.
At least, the average/cheap USB thumb drive.

Any flash devices which have static and dynamic wear leveling
are good. If they support TRIM, this allows the drive to
identify more of the NAND flash blocks that can be recycled.

There are a few thumb drives, which are SSD based and use
a USB to SATA adapter. And those can be better because the
wear leveling is full featured, and there might even be
TRIM passthru.

Because of the poor reliability of USB thumb drives (the average
ones), I do not recommend using them for archival storage.
They are for "sneakernet", for moving files from one machine
to another machine, without using a network cable. Then, if the
transfer fails, the master copy is still on the source machine.

You can attempt to degrade an NVME instantaneous performance, by
doing a 4KB random write test on the drive. This will cause the
future write rate to be slower than normal on the NVMe. It might
take a day for the drive internal routines, to functionally
defragment the 4K items and put them into flash blocks properly.
The free pool will then have more items in it, ready to use.

NVMe 3500MB/sec read, slightly less write. Many machines can do this much.
SATA SSD 530MB/sec
HDD 100-300MB/sec
USBflash Up to 2GB/sec for a USB3.2 2x2 interface NVME sled drive

I've not managed to locate any Thunderbolt or USB4 drives
with astonishing transfer rates. Some 14000MB/sec PCIe Rev5
NVMe are arriving now, and this means some day, we might see
4GB/sec or 8GB/sec portable devices. But the power such a thing
would consume while it was busy, would be pretty significant
(maybe 15 watts or more). You would term that "silly-fast"
because it is not all that practical.

90% of the "features" or speeds on a new computer, are unusable
on a day to day basis. It's hard to tell where all the performance went.
If you get your new computer, and are unimpressed with it, I will
not be surprised really.

Imagine owning a 5GHz CPU... and web pages are slow to scroll. Fuck!
Well, that's how it is.

Paul

bilsch01 <usenet@writer.com> wrote:

> Existing PC is ASUS Vivobook 15 X512JA-F512JA with 256 GB nvme SSD.
>
> Every once in a while I do a linux ddCopy of the Vivobook SSD onto an
> external HDD and when I do that I write down the speed that is reported
> upon completion. So I have a record of that speed for the last few
> years. I use the same USB port and the same external HDD for target when
> I record the speed. Also, more frequently, using Windows to back up a
> large directory of many files from the Vivobook to a thumb drive,
> however I don't have the speed data for that.
>
> Recently I noticed that copying the large directory to thumb drive
> seemed to be much much slower than it previously was. Now today, I did
> the linux ddCopy of the Vivobook and I notice the reported speed is much
> slower than before.
>
> Previously it was 113 MB/sec but now it is 87 MB/sec.
>
> I have heard that nvme SSDs somehow have substantial performance
> degradation over time. Apparently it is bad to copy huge chunks of data
> onto a nvme SSD. Actually I have done that kind of thing only once on
> this Vivobook. When I first heard about the nvme problem, I read about
> the reason but I decided it was beyond my comprehension (unlike HDDs).
>
> I'm going to buy a new laptop before the drive on this one goes out. I
> intend to avoid nvme SSDs, also I want a laptop that lasts longer than a
> few years - unlike this Vivobook. Except for the degraded speed of
> copying a large quantity of data to external storage I am happy with the
> Vivobook features and performance. I am not a computer 'power' freak. It
> seems like the nvme technology is questionable and I want to avoid it. I
> will appreciate your suggestions.
>
> Thanks. Bill S.

Has the size of unallocated space on the NVMe SSD been reduced? For
example, a recent Windows update to the Recovery (WinRE) partition
increased its size by 50% (for just a fix to a Bitlocker vulnerability).
The reduced unallocated size could reflect sufficient loss of usable
space for SSD overprovisioning. Some users will notice there is
unallocated space on their drive, and resize their partitions to use up
the unallocated space.

https://www.techtarget.com/searchstorage/definition/overprovisioning-SSD-overprovisioning
https://www.kingston.com/en/blog/pc-performance/overprovisioning

The default overprovisioning size by the drive maker is typically
sufficient for end-user workstations, and it is space neither you nor
the OS can access (it is preset hidden space not reported by the drive).
However, upping the unallocated space to increase the overprovisioning
space has benefits, like upping the endurance of the SSD, and faster
garbage collection, and reduced impact due to wear leveling. Rather
than go into the discussion of the benefits of boosting the
overprovisioning space beyond the factory default, first see if there is
unallocated space on the drive, and, if so, if it has changed in size.

Other possible reasons for preformance degradation is never optimizing
the SSD (defrag recognizes SSDs, and will optimize them instead of
moving sectors around with lots of write to get sequential the fragments
of a file), and not aligning on 4K boundaries for partitioning. In
addition, remapping is used to move contents of bad sectors (blocks) to
good sectors. Remapping slows access: the original sector gets
addressed only to get redirected to the remapped sector. Writes are
destructive to SSDs, so over time sectors go bad and have to get
remapped. The SSD slows down over time with more and more remapped
sectors. You could use defrag.exe, or a 3rd-party defragger, to only
optimize an SSD to force a TRIM to see if the SSD performs better. TRIM
just tells the SSD to do the operation when it decides, so allow some
time after optimize for the GC (garbage collection) to conclude. If the
SSD performs faster, you're doing something to consume the
overprovisioning space faster than the drive can erase the stale pages,
or worse is TRIM is disabled in the OS.

Also check in Device Manager on the properties of your SSD to ensure
write caching is enabled. RAM is still a hell of lot faster than NAND.
Since it is an internal (permanent) drive, no need to worry about not
ejecting (disconnecting) the drive before removing it as with removable
media. Since NAND, like RAM, is randomly addressable, no need to run
the Search Indexer on an SSD. That's to speed up searches on HDDs.
Disabling the Windows Search service will also eliminate the background
caching on HDDs, but how often do you actually use the Windows Search
feature? Instead I use [Search] Everything from voidtools which has its
own service for caching file entries. My C: drive (OS & apps, data is
elsewhere) was removed from Everything's indexing. I don't see [much
of] an advantage to index files on a randomly addressed storage device,
and add more writes to a database on a self-destructive drive; however,
Everything writes its everything.db (40 MB) and everything.ini (21 KB)
files only on its exit, and those files aren't huge, and the indexing is
a read-only operation, so I use the default config in Everything. Some
folks will install it on an HDD to eliminate its writes if installed on
an SSD. I've almost doubled the overprovisioning space from the hidden
default factory amount, so I'm not concerned about everything.db getting
updated whenever its service has to update its index database.

To reduce writes on the SSD, I enable pagefile space on my 2 TB D: drive
(HDD) that is system managed (16 MB to 4096 MB), but leave a 16 MB to
1024 pagefile on the C: drive (NVMe SSD); however, Windows will prefer
to first use pagefile space in a partition on a different physical drive
than the physical drive where is the OS partition. Some folks kill
pagefiling (set to zero), but there are apps that expect to preload some
data info pagefile space, like video games to preload textures and
object defines. How they handle the insufficient space error depends on
how they were coded. Having Windows do writes to its own partition on
one drive, and pagefile files writes to a different drive, allows
overlapping of write operations. There are whole bunch of tweaks to
reduce the number of writes on SSDs.

To further reduce writes to SSDs, I would disable hibernation; however,
your old and new computers will be laptops, and the savings in battery
power outweight the reduced stress from writes on the SSDs. Assuming
you're using your laptops as such (mobile devices) rather than in place
of a desktop PC. If the laptop never roams, sits atop a desk, and is
effectively your desktop PC, hibernation is a bit superfluous. You'll
never notice a minscule change in your electrical bill if you leave your
desktop, or laptop used as a fixed computer, always powered on versus
impinging all components with a surge current by powering off and on.
On portable devices, hibernation to reduce power makes sense. On
desktops, it's rather superfluous, and induces a wake time on storage
devices. Plus the hiberfil.sys file uses generates a lot of writes to
the SSD each time hibernation is used.

For SATA SSDs, S.M.A.R.T. will show attributes for reallocated sectors
count and uncorrectable count. NVMes may not show the same attributes
and instead showing something on available spare sectors. When the
reserve of remappable sectors has been consumed, no more bad sectors can
be remapped, and the device slows down even more.

TRIM was not available on Windows XP, and why SSDs would keep slowing
down unless they got optimized. As a result, most SSD makers have the
firmware in the drive do the optimize (trim), but those don't run at the
shorter intervals as when the OS issues a trim. It is possible on later
versions that do support TRIM to disable it.

https://recoverit.wondershare.com/harddrive-errors/enable-or-disable-trim-on-ssd.html

The result of running "fsutil behavior query DisableDeleteNotify" should
be 0 (zero) meaning TRIM is enabled on your SSD(s). However, Windows
issues TRIM only on NTFS-formatted drives. If you formatted the NVMe
using something else (e.g., exFAT) then Windows won't issue TRIM since
there is no journaling for recovery on non-NTFS drives. You would have
to optimize (issue a TRIM) on the SSD, like running defrag.exe on the
SSD (which optimizes, not do data writes). All an OS-issued TRIM does
is just notifying the SSD of LBA addresses the SSD is free to erase
(when the drive decides it is idle enough to perform the TRIM). It is
still possible the garbage collection cannot keep up with the amount of
changes on the drive (it ran out of overprovisioning space whether the
hidden factory space or additional unallocated space), so a huge copy
has to wait for the SSD to free up stale pages for the data you want to
write to it. That's why more overprovisioning space (by having more
unallocated space above the hidden factory default OP space) can speed
up write operations on a drive.

All writes or read are going to use the data bus. If you have other
processes that also do a lot of file I/O, they compete for the bus with
your copy operation. Some processes will lower their priority to allow
higher priority processes not only to get more CPU time, but also reduce
their impact on the data bus. For example, a backup program may give
you the option to run backups under a low(er) priority to reduce their
impact on responsiveness of the computer.

bilsch01 wrote on 3/21/24 11:43 AM:
> Existing PC is ASUS Vivobook 15 X512JA-F512JA with 256 GB nvme SSD.
>
> Every once in a while I do a linux ddCopy of the Vivobook SSD onto an
> external HDD and when I do that I write down the speed that is reported
> upon completion. So I have a record of that speed for the last few years.
> I use the same USB port and the same external HDD for target when I
> record the speed. Also, more frequently, using Windows to back up a large
> directory of many files from the Vivobook to a thumb drive, however I
> don't have the speed data for that.
>
> Recently I noticed that copying the large directory to thumb drive seemed
> to be much much slower than it previously was. Now today, I did the linux
> ddCopy of the Vivobook and I notice the reported speed is much slower
> than before.
>
> Previously it was 113 MB/sec but now it is 87 MB/sec.
>
> I have heard that nvme SSDs somehow have substantial performance
> degradation over time. Apparently it is bad to copy huge chunks of data
> onto a nvme SSD. Actually I have done that kind of thing only once on
> this Vivobook. When I first heard about the nvme problem, I read about
> the reason but I decided it was beyond my comprehension (unlike HDDs).
>
> I'm going to buy a new laptop before the drive on this one goes out. I
> intend to avoid nvme SSDs, also I want a laptop that lasts longer than a
> few years - unlike this Vivobook. Except for the degraded speed of
> copying a large quantity of data to external storage I am happy with the
> Vivobook features and performance. I am not a computer 'power' freak. It
> seems like the nvme technology is questionable and I want to avoid it. I
> will appreciate your suggestions.
>
> Thanks.    Bill S.
I'd be more inclined to lean towards the USB thumbdrive as the
controlling factor for the 113 vs. 87 rate.

NVMe SSD's are normally provide much higher speeds than SATA SSD's
- b/c of the PCIe interface connect directly to the CPU(or Server)
vs.the SATA which uses the SATA(serial ATA) Bus interface(slower)

There are 4 types of SSD's
M.2 SATA SSD M.2 connector, SATA connection protocol
M.2 NVMe SSD M.2 connector, PCIe connection protocol
SATA SSD Sata connector, SATA connection protocol
U.2 SSD U.2 connector, PCIe/SAS/SATA connection protocol

The choice is yours to make, as others have mentioned, the
rated(marketed) speeds are not necessarily the actual transfer speed when
copying data(i.e. the rated speeds may be peak, not continuous
sustainable). You may find that a good M.2 NVMe SSD has a higher
lifecycle than M.2 SATA SSD's)

If faster is better, then your choice really comes down to M.2 NVME and
M.2 SATA. U.2 is very uncommon in the consumer arena and primarily used
in specific use cases in enterprise servers and storage(expensive and
special SFF-8639 connector required)

--
....w¡ñ§±¤ñ

On 2024-03-21 19:43, bilsch01 wrote:
> Existing PC is ASUS Vivobook 15 X512JA-F512JA with 256 GB nvme SSD.
>
> Every once in a while I do a linux ddCopy of the Vivobook SSD onto an
> external HDD and when I do that I write down the speed that is reported
> upon completion. So I have a record of that speed for the last few
> years. I use the same USB port and the same external HDD for target when
> I record the speed. Also, more frequently, using Windows to back up a
> large directory of many files from the Vivobook to a thumb drive,
> however I don't have the speed data for that.
>
> Recently I noticed that copying the large directory to thumb drive
> seemed to be much much slower than it previously was. Now today, I did
> the linux ddCopy of the Vivobook and I notice the reported speed is much
> slower than before.
>
> Previously it was 113 MB/sec but now it is 87 MB/sec.

You know that this speed is typical of rotating rust?
nvme is much faster.

I assume that the laptop has nvme, and the backup is reading from there,
into an external hard disk, that is magnetic. The later is the limiting
factor, not the nvme. And possibly the USB bus.

I would do a SMART test of that rotating rust disk, using smartctl in
Linux. I'll make a guess: either it is full and fragmented, or it has
sectors in error.

--
Cheers, Carlos.

bilsch01 wrote:
> Existing PC is ASUS Vivobook 15 X512JA-F512JA with 256 GB nvme SSD.
>
> Every once in a while I do a linux ddCopy of the Vivobook SSD onto an
> external HDD and when I do that I write down the speed that is reported
> upon completion. So I have a record of that speed for the last few
> years. I use the same USB port and the same external HDD for target when
> I record the speed. Also, more frequently, using Windows to back up a
> large directory of many files from the Vivobook to a thumb drive,
> however I don't have the speed data for that.
>
> Recently I noticed that copying the large directory to thumb drive
> seemed to be much much slower than it previously was. Now today, I did
> the linux ddCopy of the Vivobook and I notice the reported speed is much
> slower than before.
>
> Previously it was 113 MB/sec but now it is 87 MB/sec.
>
> I have heard that nvme SSDs somehow have substantial performance
> degradation over time. Apparently it is bad to copy huge chunks of data
> onto a nvme SSD. Actually I have done that kind of thing only once on
> this Vivobook. When I first heard about the nvme problem, I read about
> the reason but I decided it was beyond my comprehension (unlike HDDs).
>
> I'm going to buy a new laptop before the drive on this one goes out. I
> intend to avoid nvme SSDs, also I want a laptop that lasts longer than a
> few years - unlike this Vivobook. Except for the degraded speed of
> copying a large quantity of data to external storage I am happy with the
> Vivobook features and performance. I am not a computer 'power' freak. It
> seems like the nvme technology is questionable and I want to avoid it. I
> will appreciate your suggestions.
>
> Thanks.    Bill S.

I'll never go back to hdd or ssd.
I installed a Samsung/500 970 Pro m.2 in 2018 and a Samsung/500 980 Pro
m.2 in 2020 in the gamer and ran Crystal Disk each time for a base. A
Crystal disk run tonight shows absolutely no degradation over 6 and 4 years.

bilsch01 <usenet@writer.com> wrote:
> Existing PC is ASUS Vivobook 15 X512JA-F512JA with 256 GB nvme SSD.
>
> Every once in a while I do a linux ddCopy of the Vivobook SSD onto an
> external HDD and when I do that I write down the speed that is reported
> upon completion. So I have a record of that speed for the last few
> years. I use the same USB port and the same external HDD for target when
> I record the speed. Also, more frequently, using Windows to back up a
> large directory of many files from the Vivobook to a thumb drive,
> however I don't have the speed data for that.
>
> Recently I noticed that copying the large directory to thumb drive
> seemed to be much much slower than it previously was. Now today, I did
> the linux ddCopy of the Vivobook and I notice the reported speed is much
> slower than before.
>
> Previously it was 113 MB/sec but now it is 87 MB/sec.
>
> I have heard that nvme SSDs somehow have substantial performance
> degradation over time. Apparently it is bad to copy huge chunks of data
> onto a nvme SSD. Actually I have done that kind of thing only once on
> this Vivobook. When I first heard about the nvme problem, I read about
> the reason but I decided it was beyond my comprehension (unlike HDDs).
>
> I'm going to buy a new laptop before the drive on this one goes out. I
> intend to avoid nvme SSDs, also I want a laptop that lasts longer than a
> few years - unlike this Vivobook. Except for the degraded speed of
> copying a large quantity of data to external storage I am happy with the
> Vivobook features and performance. I am not a computer 'power' freak. It
> seems like the nvme technology is questionable and I want to avoid it. I
> will appreciate your suggestions.

Is your SSD very full, like >90%? Am overly full SSD can fail quicker than
normal.

Typically SSDs last just as long as HDDs and you could just replace your
NVME without getting a new laptop if you're otherwise happy with it. You'll
struggle to find a laptop without an SSD these days anyway.

bilsch01 wrote:

> I intend to avoid nvme SSDs

I would say that's a mistake ...

On 3/22/24 07:51 AM, Andy Burns wrote:
> bilsch01 wrote:
>
>> I intend to avoid nvme SSDs
>
> I would say that's a mistake ...
+1
That's my pref. I only want an HDD for bulk storage.
--
Linux Mint 21.3 Cinnamon 6.0.4 Kernel 6.5.0-26-generic
Al

On 3/21/2024 11:43 AM, bilsch01 wrote:
> Existing PC is ASUS Vivobook 15 X512JA-F512JA with 256 GB nvme SSD.
>
> Every once in a while I do a linux ddCopy of the Vivobook SSD onto an
> external HDD and when I do that I write down the speed that is reported
> upon completion.
Regarding accuracy of speed and time reported by ddCopy operation.
The linux ddCopy reports two values upon completion: seconds and MB/s.
Here are examples from yesterday:
a) 2951 s 86.8 MB/s
b) 5583 s 45.9 MB/s
Note: a) used USB 3.2 Gen1 port.
Note: b) used USB 2.0 port

Note: the SSD size is fixed: approx 256 GB
a) 2951 x 86.8 = 256.1 GB
b) 5583 X 45.9 = 256.3 GB

The numbers are pretty close to 256 GB. Close enough for me.

On 3/21/2024 4:39 PM, ...w¡ñ§±¤ñ wrote:
> bilsch01 wrote on 3/21/24 11:43 AM:
>> Existing PC is ASUS Vivobook 15 X512JA-F512JA with 256 GB nvme SSD.
>>
>> Every once in a while I do a linux ddCopy of the Vivobook SSD onto an
>> external HDD and when I do that I write down the speed that is
>> reported upon completion. So I have a record of that speed for the
>> last few years. I use the same USB port and the same external HDD for
>> target when I record the speed. Also, more frequently, using Windows
>> to back up a large directory of many files from the Vivobook to a
>> thumb drive, however I don't have the speed data for that.
>>
>> Recently I noticed that copying the large directory to thumb drive
>> seemed to be much much slower than it previously was. Now today, I did
>> the linux ddCopy of the Vivobook and I notice the reported speed is
>> much slower than before.
>>
>> Previously it was 113 MB/sec but now it is 87 MB/sec.
>>
>> I have heard that nvme SSDs somehow have substantial performance
>> degradation over time. Apparently it is bad to copy huge chunks of
>> data onto a nvme SSD. Actually I have done that kind of thing only
>> once on this Vivobook. When I first heard about the nvme problem, I
>> read about the reason but I decided it was beyond my comprehension
>> (unlike HDDs).
>>
>> I'm going to buy a new laptop before the drive on this one goes out. I
>> intend to avoid nvme SSDs, also I want a laptop that lasts longer than
>> a few years - unlike this Vivobook. Except for the degraded speed of
>> copying a large quantity of data to external storage I am happy with
>> the Vivobook features and performance. I am not a computer 'power'
>> freak. It seems like the nvme technology is questionable and I want to
>> avoid it. I will appreciate your suggestions.
>>
>> Thanks.    Bill S.
> I'd be more inclined to lean towards the USB thumbdrive as the
> controlling factor for the 113 vs. 87 rate.

there was no thumbdrive here.
Target drive is a spinning WD 2 TB HDD. Same for 113 and 87.
I did more tests yesterday using other targets and other USB ports.
Those results, combined with older results, show a sensible picture.

I AM NO LONGER SUSPICIOUS ABOUT A BAD NVME SSD.

There was a different setup when data points 113 and 87 were obtained.
The difference:
the WD 2 TB spinner was connected after booting. Result: 113 MB/s
the WD 2 TB spinner already connected when PC booted. Result: 87 MB/s

>
> NVMe SSD's are normally provide much higher speeds than SATA SSD's
>  - b/c of the PCIe interface connect directly to the CPU(or Server)
> vs.the SATA which uses the SATA(serial ATA) Bus interface(slower)
>
> There are 4 types of SSD's
>  M.2 SATA SSD   M.2 connector, SATA connection protocol
>  M.2 NVMe SSD    M.2 connector, PCIe connection protocol
>  SATA SSD    Sata connector, SATA connection protocol
>  U.2  SSD    U.2 connector, PCIe/SAS/SATA connection protocol
>
> The choice is yours to make, as others have mentioned, the
> rated(marketed) speeds are not necessarily the actual transfer speed
> when copying data(i.e. the rated speeds may be peak, not continuous
> sustainable). You may find that a good M.2 NVMe SSD has a higher
> lifecycle than M.2 SATA SSD's)
>
> If faster is better, then your choice really comes down to M.2 NVME and
> M.2 SATA. U.2 is very uncommon in the consumer arena and primarily used
> in specific use cases in enterprise servers and storage(expensive and
> special SFF-8639 connector required)

Thanks for good info about SSD types.
>

On 3/22/2024 10:05 AM, Big Al wrote:
> On 3/22/24 07:51 AM, Andy Burns wrote:
>> bilsch01 wrote:
>>
>>> I intend to avoid nvme SSDs
>>
>> I would say that's a mistake ...
> +1
> That's my pref.  I only want an HDD for bulk storage.

hard drive ----+---- performance degrade ---- clone --- carry on
| noticed over /
| /
+---- device failure, click --------- restore
of death, dropped on floor from backup

flash stick -------- Device drops to 2MB/sec ------X Flash sticks are *NOT*
$10 It's too late. Data loss storage devices.

NVMe, ----+---- ???? ---- clone --- carry on
SATA SSD | Detect degrade ??? over /
| /
+---- device failure, --------------- restore
no response, ransomware from backup

All devices need backups. USB flash sticks, being utter rubbish
to begin with, *should not even be thought of as storage devices* .
This happened with the introduction of TLC flash. I have an MLC
flash stick I actually trust (8GB). You can't buy those now.
Flash can be used for certain temporary tasks, or for write-once
read-multiple purposes. There are high-end USB sticks with
different internal design, which are OK. But if you paid $10, it's rubbish.

[This device has internal NVMe SMART passthru, but the stats aren't all that useful.
This is not an endorsement. A poster bought one of these -- I don't know why.
The device overheats on really long transfers. Plastic casing. Other brands exist.
This should not fail exactly the same way as a $10 one. Arch is different.]

http://www.cdrlabs.com/reviews/patriot-supersonic-rage-prime-usb32-gen2-flash-drive/all-pages.html

The operating characteristic of a hard drive, is it's slow.
Even the drive with two arms is still slow. But, the degrade
characteristic, means the frequency of full backups can be
reduced a bit, according to your pain threshold on failure.

NVMe and SATA SSD, I want a little higher frequency of backup,
because I am unlikely to get a good warning in time, and the tools
provided, we can see how effective they are. What is alarming
or disturbing, is discovering that just about everything you
buy has issues, but it's pretty hard to track them down using
Google. Tie SSD I'm booted from, it turns out it has a
troubled past! Surprise! :-/

Microsoft has various mechanisms, as well as the backup companies,
to remind users about the need to back up (on a schedule), and
some other maintenance reminders are also presented. But some
people don't even know those GUIs exist, and I'm not sure
a Notification mechanism is even appropriate for a mission-critical
event such as pending device failure. If a Notification disappears
off the screen in five seconds, that does not count as Notification.

When I try to draw diagrams like the above ones, my assumption
usually, is the user is willfully ignorant (expects a miracle
to happen and someone here will click a button for them), or,
the user genuinely cannot find a tool to do an analysis. I have
my share of trouble, finding applications for things like NVMe.
Some SSD maker toolboxes are "rubbish" -- there are even cases
where an SSD they make, the toolbox refuses to acknowledge it
is their brand. This means you're looking for a CrystalDiskInfo
or a SmartMontool, looking for feedback about device health.

I've already seen disparaging remarks about "people trying to
help us from the command line", or "people trying to help us
using free tools they know to exist". Logically speaking,
their perspective is "you worry about it". Well, I'm not
worried about it, because it's not my problem :-) It's
*your* problem. I don't care if you lost 40 years of family
photos. I can suggest things you can do, with regard to
diagrams like the ones above, but I don't supply crying
towels here. Man up, and carry on.

I've lost data here. That's called "experience". Each event
teaches a lesson.

There is still some benefit from UPS behavior (uninterruptable
power supply). A laptop and its battery, are an intrinsic
UPS. Set the battery to charge to 80%, not 100%, and you
no longer have to think about the battery. The battery management
logic, insures a lot more "clean shutdowns" and preservation
of NVMe data. You're not constantly testing the integrity
of the device, by suddenly removing power.

On desktops, I would still recommend a UPS for people who can
afford one. They're not perfect. They introduce yet more failure
modes. But on balance, they should be included in your
computer room design. This removes some of the dice rolls
that influence NVME/SSD failures. Only you know how unreliable your
power provider is. The last time the protective fuse
blew on the pole out back of the house, it damaged my
carriage lamp controller. Power company protection
devices do not always have benign behavior. And a simple
UPS design can actually *amplify* a transient event. But
a UPS is definitely better than nothing.

Summary: NVMe ? Do backups on a schedule. We cannot
rely on the OS itself to warn us in time.
The backup can go on a hard drive, as long as
the dollars per gigabyte is the best. Current SSD
prices are too high, to use for backup.

NVMe and SATA SSD have "wear life", and using dd.exe towards
them as a destination, is a waste of write-life. You need to
adjust your practices while using the device. One of the
reasons I have a RAMDrive, is to use it for scratch writes.
when two machines have a RAM drive, you can pop a file over to
the second machine, while you reboot the first machine.

[Picture] Scratch storage space, that does not wear my SSD, speed=irrelevant
It just has to read and write.

https://i.postimg.cc/fy8FmKGs/HDTune-Benchmark-Pass-Markosfdisk.gif

Paul

On 3/22/24 02:39 PM, Paul wrote:
> On 3/22/2024 10:05 AM, Big Al wrote:
>> On 3/22/24 07:51 AM, Andy Burns wrote:
>>> bilsch01 wrote:
>>>
>>>> I intend to avoid nvme SSDs
>>>
>>> I would say that's a mistake ...
>> +1
>> That's my pref.  I only want an HDD for bulk storage.
>
> hard drive ----+---- performance degrade ---- clone --- carry on
> | noticed over /
> | /
> +---- device failure, click --------- restore
> of death, dropped on floor from backup
>
> flash stick -------- Device drops to 2MB/sec ------X Flash sticks are *NOT*
> $10 It's too late. Data loss storage devices.
>
> NVMe, ----+---- ???? ---- clone --- carry on
> SATA SSD | Detect degrade ??? over /
> | /
> +---- device failure, --------------- restore
> no response, ransomware from backup
>
> All devices need backups. USB flash sticks, being utter rubbish
> to begin with, *should not even be thought of as storage devices* .
> This happened with the introduction of TLC flash. I have an MLC
> flash stick I actually trust (8GB). You can't buy those now.
> Flash can be used for certain temporary tasks, or for write-once
> read-multiple purposes. There are high-end USB sticks with
> different internal design, which are OK. But if you paid $10, it's rubbish.
>
> [This device has internal NVMe SMART passthru, but the stats aren't all that useful.
> This is not an endorsement. A poster bought one of these -- I don't know why.
> The device overheats on really long transfers. Plastic casing. Other brands exist.
> This should not fail exactly the same way as a $10 one. Arch is different.]
>
> http://www.cdrlabs.com/reviews/patriot-supersonic-rage-prime-usb32-gen2-flash-drive/all-pages.html
>
> The operating characteristic of a hard drive, is it's slow.
> Even the drive with two arms is still slow. But, the degrade
> characteristic, means the frequency of full backups can be
> reduced a bit, according to your pain threshold on failure.
>
> NVMe and SATA SSD, I want a little higher frequency of backup,
> because I am unlikely to get a good warning in time, and the tools
> provided, we can see how effective they are. What is alarming
> or disturbing, is discovering that just about everything you
> buy has issues, but it's pretty hard to track them down using
> Google. Tie SSD I'm booted from, it turns out it has a
> troubled past! Surprise! :-/
>
> Microsoft has various mechanisms, as well as the backup companies,
> to remind users about the need to back up (on a schedule), and
> some other maintenance reminders are also presented. But some
> people don't even know those GUIs exist, and I'm not sure
> a Notification mechanism is even appropriate for a mission-critical
> event such as pending device failure. If a Notification disappears
> off the screen in five seconds, that does not count as Notification.
>
> When I try to draw diagrams like the above ones, my assumption
> usually, is the user is willfully ignorant (expects a miracle
> to happen and someone here will click a button for them), or,
> the user genuinely cannot find a tool to do an analysis. I have
> my share of trouble, finding applications for things like NVMe.
> Some SSD maker toolboxes are "rubbish" -- there are even cases
> where an SSD they make, the toolbox refuses to acknowledge it
> is their brand. This means you're looking for a CrystalDiskInfo
> or a SmartMontool, looking for feedback about device health.
>
> I've already seen disparaging remarks about "people trying to
> help us from the command line", or "people trying to help us
> using free tools they know to exist". Logically speaking,
> their perspective is "you worry about it". Well, I'm not
> worried about it, because it's not my problem :-) It's
> *your* problem. I don't care if you lost 40 years of family
> photos. I can suggest things you can do, with regard to
> diagrams like the ones above, but I don't supply crying
> towels here. Man up, and carry on.
>
> I've lost data here. That's called "experience". Each event
> teaches a lesson.
>
> There is still some benefit from UPS behavior (uninterruptable
> power supply). A laptop and its battery, are an intrinsic
> UPS. Set the battery to charge to 80%, not 100%, and you
> no longer have to think about the battery. The battery management
> logic, insures a lot more "clean shutdowns" and preservation
> of NVMe data. You're not constantly testing the integrity
> of the device, by suddenly removing power.
>
> On desktops, I would still recommend a UPS for people who can
> afford one. They're not perfect. They introduce yet more failure
> modes. But on balance, they should be included in your
> computer room design. This removes some of the dice rolls
> that influence NVME/SSD failures. Only you know how unreliable your
> power provider is. The last time the protective fuse
> blew on the pole out back of the house, it damaged my
> carriage lamp controller. Power company protection
> devices do not always have benign behavior. And a simple
> UPS design can actually *amplify* a transient event. But
> a UPS is definitely better than nothing.
>
> Summary: NVMe ? Do backups on a schedule. We cannot
> rely on the OS itself to warn us in time.
> The backup can go on a hard drive, as long as
> the dollars per gigabyte is the best. Current SSD
> prices are too high, to use for backup.
>
> NVMe and SATA SSD have "wear life", and using dd.exe towards
> them as a destination, is a waste of write-life. You need to
> adjust your practices while using the device. One of the
> reasons I have a RAMDrive, is to use it for scratch writes.
> when two machines have a RAM drive, you can pop a file over to
> the second machine, while you reboot the first machine.
>
> [Picture] Scratch storage space, that does not wear my SSD, speed=irrelevant
> It just has to read and write.
>
> https://i.postimg.cc/fy8FmKGs/HDTune-Benchmark-Pass-Markosfdisk.gif
>
> Paul
Is there any tool that will tell you the chips internal to a usb drive. Or is that just digging out
specs on the web?
--
Linux Mint 21.3 Cinnamon 6.0.4 Kernel 6.5.0-26-generic
Al

On 3/22/24 03:03 PM, Big Al wrote:
> On 3/22/24 02:39 PM, Paul wrote:
>> On 3/22/2024 10:05 AM, Big Al wrote:
>>> On 3/22/24 07:51 AM, Andy Burns wrote:
>>>> bilsch01 wrote:
>>>>
>>>>> I intend to avoid nvme SSDs
>>>>
>>>> I would say that's a mistake ...
>>> +1
>>> That's my pref.  I only want an HDD for bulk storage.
>>
>>     hard drive ----+---- performance degrade ---- clone --- carry on
>>                    |          noticed              over        /
>>                    |                                          /
>>                    +---- device failure, click --------- restore
>>                          of death, dropped on floor      from backup
>>
>>     flash stick -------- Device drops to 2MB/sec ------X        Flash sticks are *NOT*
>>        $10               It's too late. Data loss               storage devices.
>>
>>     NVMe,      ----+----        ????         ---- clone --- carry on
>>     SATA SSD       |     Detect degrade ???        over        /
>>                    |                                          /
>>                    +---- device failure, --------------- restore
>>                          no response, ransomware      from backup
>>
>> All devices need backups. USB flash sticks, being utter rubbish
>> to begin with, *should not even be thought of as storage devices* .
>> This happened with the introduction of TLC flash. I have an MLC
>> flash stick I actually trust (8GB). You can't buy those now.
>> Flash can be used for certain temporary tasks, or for write-once
>> read-multiple purposes. There are high-end USB sticks with
>> different internal design, which are OK. But if you paid $10, it's rubbish.
>>
>>     [This device has internal NVMe SMART passthru, but the stats aren't all that useful.
>>      This is not an endorsement. A poster bought one of these -- I don't know why.
>>      The device overheats on really long transfers. Plastic casing. Other brands exist.
>>      This should not fail exactly the same way as a $10 one. Arch is different.]
>>
>>
>> http://www.cdrlabs.com/reviews/patriot-supersonic-rage-prime-usb32-gen2-flash-drive/all-pages.html
>>
>> The operating characteristic of a hard drive, is it's slow.
>> Even the drive with two arms is still slow. But, the degrade
>> characteristic, means the frequency of full backups can be
>> reduced a bit, according to your pain threshold on failure.
>>
>> NVMe and SATA SSD, I want a little higher frequency of backup,
>> because I am unlikely to get a good warning in time, and the tools
>> provided, we can see how effective they are. What is alarming
>> or disturbing, is discovering that just about everything you
>> buy has issues, but it's pretty hard to track them down using
>> Google. Tie SSD I'm booted from, it turns out it has a
>> troubled past! Surprise! :-/
>>
>> Microsoft has various mechanisms, as well as the backup companies,
>> to remind users about the need to back up (on a schedule), and
>> some other maintenance reminders are also presented. But some
>> people don't even know those GUIs exist, and I'm not sure
>> a Notification mechanism is even appropriate for a mission-critical
>> event such as pending device failure. If a Notification disappears
>> off the screen in five seconds, that does not count as Notification.
>>
>> When I try to draw diagrams like the above ones, my assumption
>> usually, is the user is willfully ignorant (expects a miracle
>> to happen and someone here will click a button for them), or,
>> the user genuinely cannot find a tool to do an analysis. I have
>> my share of trouble, finding applications for things like NVMe.
>> Some SSD maker toolboxes are "rubbish" -- there are even cases
>> where an SSD they make, the toolbox refuses to acknowledge it
>> is their brand. This means you're looking for a CrystalDiskInfo
>> or a SmartMontool, looking for feedback about device health.
>>
>> I've already seen disparaging remarks about "people trying to
>> help us from the command line", or "people trying to help us
>> using free tools they know to exist". Logically speaking,
>> their perspective is "you worry about it". Well, I'm not
>> worried about it, because it's not my problem :-) It's
>> *your* problem. I don't care if you lost 40 years of family
>> photos. I can suggest things you can do, with regard to
>> diagrams like the ones above, but I don't supply crying
>> towels here. Man up, and carry on.
>>
>> I've lost data here. That's called "experience". Each event
>> teaches a lesson.
>>
>> There is still some benefit from UPS behavior (uninterruptable
>> power supply). A laptop and its battery, are an intrinsic
>> UPS. Set the battery to charge to 80%, not 100%, and you
>> no longer have to think about the battery. The battery management
>> logic, insures a lot more "clean shutdowns" and preservation
>> of NVMe data. You're not constantly testing the integrity
>> of the device, by suddenly removing power.
>>
>> On desktops, I would still recommend a UPS for people who can
>> afford one. They're not perfect. They introduce yet more failure
>> modes. But on balance, they should be included in your
>> computer room design. This removes some of the dice rolls
>> that influence NVME/SSD failures. Only you know how unreliable your
>> power provider is. The last time the protective fuse
>> blew on the pole out back of the house, it damaged my
>> carriage lamp controller. Power company protection
>> devices do not always have benign behavior. And a simple
>> UPS design can actually *amplify* a transient event. But
>> a UPS is definitely better than nothing.
>>
>> Summary: NVMe ?   Do backups on a schedule. We cannot
>>           rely on the OS itself to warn us in time.
>>           The backup can go on a hard drive, as long as
>>           the dollars per gigabyte is the best. Current SSD
>>           prices are too high, to use for backup.
>>
>>           NVMe and SATA SSD have "wear life", and using dd.exe towards
>>           them as a destination, is a waste of write-life. You need to
>>           adjust your practices while using the device. One of the
>>           reasons I have a RAMDrive, is to use it for scratch writes.
>>           when two machines have a RAM drive, you can pop a file over to
>>           the second machine, while you reboot the first machine.
>>
>>           [Picture]  Scratch storage space, that does not wear my SSD, speed=irrelevant
>>                      It just has to read and write.
>>
>>            https://i.postimg.cc/fy8FmKGs/HDTune-Benchmark-Pass-Markosfdisk.gif
>>
>>     Paul
> Is there any tool that will tell you the chips internal to a usb drive.  Or is that just digging out
> specs on the web?
I know there is a program to tell me what memory chips my PC has.
--
Linux Mint 21.3 Cinnamon 6.0.4 Kernel 6.5.0-26-generic
Al

On 3/22/2024 11:39 AM, Paul wrote:
> On 3/22/2024 10:05 AM, Big Al wrote:
>> On 3/22/24 07:51 AM, Andy Burns wrote:
>>> bilsch01 wrote:
>>>
>>>> I intend to avoid nvme SSDs
>>>
>>> I would say that's a mistake ...
>> +1
>> That's my pref.  I only want an HDD for bulk storage.
>
> hard drive ----+---- performance degrade ---- clone --- carry on
> | noticed over /
> | /
> +---- device failure, click --------- restore
> of death, dropped on floor from backup
>
> flash stick -------- Device drops to 2MB/sec ------X Flash sticks are *NOT*
> $10 It's too late. Data loss storage devices.
>
> NVMe, ----+---- ???? ---- clone --- carry on
> SATA SSD | Detect degrade ??? over /
> | /
> +---- device failure, --------------- restore
> no response, ransomware from backup
>
> All devices need backups. USB flash sticks, being utter rubbish
> to begin with, *should not even be thought of as storage devices* .
> This happened with the introduction of TLC flash. I have an MLC
> flash stick I actually trust (8GB). You can't buy those now.
> Flash can be used for certain temporary tasks, or for write-once
> read-multiple purposes. There are high-end USB sticks with
> different internal design, which are OK. But if you paid $10, it's rubbish.
>
> [This device has internal NVMe SMART passthru, but the stats aren't all that useful.
> This is not an endorsement. A poster bought one of these -- I don't know why.
> The device overheats on really long transfers. Plastic casing. Other brands exist.
> This should not fail exactly the same way as a $10 one. Arch is different.]
>
> http://www.cdrlabs.com/reviews/patriot-supersonic-rage-prime-usb32-gen2-flash-drive/all-pages.html
>
> The operating characteristic of a hard drive, is it's slow.
> Even the drive with two arms is still slow. But, the degrade
> characteristic, means the frequency of full backups can be
> reduced a bit, according to your pain threshold on failure.
>
> NVMe and SATA SSD, I want a little higher frequency of backup,
> because I am unlikely to get a good warning in time, and the tools
> provided, we can see how effective they are. What is alarming
> or disturbing, is discovering that just about everything you
> buy has issues, but it's pretty hard to track them down using
> Google. Tie SSD I'm booted from, it turns out it has a
> troubled past! Surprise! :-/
>
> Microsoft has various mechanisms, as well as the backup companies,
> to remind users about the need to back up (on a schedule), and
> some other maintenance reminders are also presented. But some
> people don't even know those GUIs exist, and I'm not sure
> a Notification mechanism is even appropriate for a mission-critical
> event such as pending device failure. If a Notification disappears
> off the screen in five seconds, that does not count as Notification.
>
> When I try to draw diagrams like the above ones, my assumption
> usually, is the user is willfully ignorant (expects a miracle
> to happen and someone here will click a button for them), or,
> the user genuinely cannot find a tool to do an analysis. I have
> my share of trouble, finding applications for things like NVMe.
> Some SSD maker toolboxes are "rubbish" -- there are even cases
> where an SSD they make, the toolbox refuses to acknowledge it
> is their brand. This means you're looking for a CrystalDiskInfo
> or a SmartMontool, looking for feedback about device health.
>
> I've already seen disparaging remarks about "people trying to
> help us from the command line", or "people trying to help us
> using free tools they know to exist". Logically speaking,
> their perspective is "you worry about it". Well, I'm not
> worried about it, because it's not my problem :-) It's
> *your* problem. I don't care if you lost 40 years of family
> photos. I can suggest things you can do, with regard to
> diagrams like the ones above, but I don't supply crying
> towels here. Man up, and carry on.
>
> I've lost data here. That's called "experience". Each event
> teaches a lesson.
>
> There is still some benefit from UPS behavior (uninterruptable
> power supply). A laptop and its battery, are an intrinsic
> UPS. Set the battery to charge to 80%, not 100%, and you
> no longer have to think about the battery. The battery management
> logic, insures a lot more "clean shutdowns" and preservation
> of NVMe data. You're not constantly testing the integrity
> of the device, by suddenly removing power.
>
> On desktops, I would still recommend a UPS for people who can
> afford one. They're not perfect. They introduce yet more failure
> modes. But on balance, they should be included in your
> computer room design. This removes some of the dice rolls
> that influence NVME/SSD failures. Only you know how unreliable your
> power provider is. The last time the protective fuse
> blew on the pole out back of the house, it damaged my
> carriage lamp controller. Power company protection
> devices do not always have benign behavior. And a simple
> UPS design can actually *amplify* a transient event. But
> a UPS is definitely better than nothing.
>
> Summary: NVMe ? Do backups on a schedule. We cannot
> rely on the OS itself to warn us in time.
> The backup can go on a hard drive, as long as
> the dollars per gigabyte is the best. Current SSD
> prices are too high, to use for backup.
>
> NVMe and SATA SSD have "wear life", and using dd.exe towards
> them as a destination, is a waste of write-life. You need to
> adjust your practices while using the device. One of the
> reasons I have a RAMDrive, is to use it for scratch writes.
> when two machines have a RAM drive, you can pop a file over to
> the second machine, while you reboot the first machine.
>
> [Picture] Scratch storage space, that does not wear my SSD, speed=irrelevant
> It just has to read and write.
>
> https://i.postimg.cc/fy8FmKGs/HDTune-Benchmark-Pass-Markosfdisk.gif
>
> Paul

Thanks for very useful info about various USB sticks
Bill S.

bilsch01 wrote on 3/22/24 11:31 AM:
> On 3/21/2024 4:39 PM, ...w¡ñ§±¤ñ wrote:
>> bilsch01 wrote on 3/21/24 11:43 AM:
>>> Existing PC is ASUS Vivobook 15 X512JA-F512JA with 256 GB nvme SSD.
>>>
>>> Every once in a while I do a linux ddCopy of the Vivobook SSD onto an
>>> external HDD and when I do that I write down the speed that is
>>> reported upon completion. So I have a record of that speed for the
>>> last few years. I use the same USB port and the same external HDD for
>>> target when I record the speed. Also, more frequently, using Windows
>>> to back up a large directory of many files from the Vivobook to a
>>> thumb drive, however I don't have the speed data for that.
>>>
>>> Recently I noticed that copying the large directory to thumb drive
>>> seemed to be much much slower than it previously was. Now today, I did
>>> the linux ddCopy of the Vivobook and I notice the reported speed is
>>> much slower than before.
>>>
>>> Previously it was 113 MB/sec but now it is 87 MB/sec.
>>>
>>> I have heard that nvme SSDs somehow have substantial performance
>>> degradation over time. Apparently it is bad to copy huge chunks of
>>> data onto a nvme SSD. Actually I have done that kind of thing only
>>> once on this Vivobook. When I first heard about the nvme problem, I
>>> read about the reason but I decided it was beyond my comprehension
>>> (unlike HDDs).
>>>
>>> I'm going to buy a new laptop before the drive on this one goes out. I
>>> intend to avoid nvme SSDs, also I want a laptop that lasts longer than
>>> a few years - unlike this Vivobook. Except for the degraded speed of
>>> copying a large quantity of data to external storage I am happy with
>>> the Vivobook features and performance. I am not a computer 'power'
>>> freak. It seems like the nvme technology is questionable and I want to
>>> avoid it. I will appreciate your suggestions.
>>>
>>> Thanks.    Bill S.
>> I'd be more inclined to lean towards the USB thumbdrive as the
>> controlling factor for the 113 vs. 87 rate.
>
> there was no thumbdrive here.

Ok.
The comment in your post "Recently I noticed that copying the large
directory to *thumb drive*" was the reason for my reply. Directly below
your 'thumb drive' reference were that stats for the 113 vs 87 speeds.

> Target drive is a spinning WD 2 TB HDD. Same for 113 and 87.
> I did more tests yesterday using other targets and other USB ports.
> Those results, combined with older results, show a sensible picture.
>
> I AM NO LONGER SUSPICIOUS ABOUT A BAD NVME SSD.
>
> There was a different setup when data points 113 and 87 were obtained.
> The difference:
> the WD 2 TB spinner was connected after booting.      Result: 113 MB/s
> the WD 2 TB spinner already connected when PC booted. Result:  87 MB/s
I guess the connection sequence could be a factor, but I suspect in the
long run, the results(for the same exact data copy if done enough times
for more data points) would probably normalize to similar speeds.

>
>>
>> NVMe SSD's are normally provide much higher speeds than SATA SSD's
>>   - b/c of the PCIe interface connect directly to the CPU(or Server)
>> vs.the SATA which uses the SATA(serial ATA) Bus interface(slower)
>>
>> There are 4 types of SSD's
>>   M.2 SATA SSD   M.2 connector, SATA connection protocol
>>   M.2 NVMe SSD    M.2 connector, PCIe connection protocol
>>   SATA SSD    Sata connector, SATA connection protocol
>>   U.2  SSD    U.2 connector, PCIe/SAS/SATA connection protocol
>>
>> The choice is yours to make, as others have mentioned, the
>> rated(marketed) speeds are not necessarily the actual transfer speed
>> when copying data(i.e. the rated speeds may be peak, not continuous
>> sustainable). You may find that a good M.2 NVMe SSD has a higher
>> lifecycle than M.2 SATA SSD's)
>>
>> If faster is better, then your choice really comes down to M.2 NVME and
>> M.2 SATA. U.2 is very uncommon in the consumer arena and primarily used
>> in specific use cases in enterprise servers and storage(expensive and
>> special SFF-8639 connector required)
>
> Thanks for good info about SSD types.
>>
>
You're welcome.

--
....w¡ñ§±¤ñ

On 3/22/2024 3:03 PM, Big Al wrote:

>> Is there any tool that will tell you the chips internal to a usb drive.
> Or is that just digging out specs on the web?
> I know there is a program to tell me what memory chips my PC has.

You're thinking of ChipGenius.

https://www.softpedia.com/get/System/System-Info/ChipGenius.shtml

I asked Copilot:

Is the program ChipGenius safe to use?

And it couldn't give a plain yes or no answer.
Uses "UPX v1.25 (Delphi) Stub"
FileVersion 4.19.0319

https://www.virustotal.com/gui/file/b58f240840e89587769db5b989e12a91dca5b216631a7f13c60143b6d1baa477/community

https://www.hybrid-analysis.com/sample/b58f240840e89587769db5b989e12a91dca5b216631a7f13c60143b6d1baa477/5cab23160388384a76bcebca

Paul

On 2024-03-22 19:31, bilsch01 wrote:
> On 3/21/2024 4:39 PM, ...w¡ñ§±¤ñ wrote:
>> bilsch01 wrote on 3/21/24 11:43 AM:

....

> I AM NO LONGER SUSPICIOUS ABOUT A BAD NVME SSD.

Ah. Makes sense.

> There was a different setup when data points 113 and 87 were obtained.
> The difference:
> the WD 2 TB spinner was connected after booting.      Result: 113 MB/s
> the WD 2 TB spinner already connected when PC booted. Result:  87 MB/s

I would compare the output of the "mount" command issues in both cases
(the line corresponding to the spinner).

If they are exactly the same, then I would have to suspect the usb
settings somehow.

--
Cheers, Carlos.

On Fri, 22 Mar 2024 11:51:02 +0000, Andy Burns <usenet@andyburns.uk> wrote:

>bilsch01 wrote:
>
>> I intend to avoid nvme SSDs
>
>I would say that's a mistake ...

+1