From: Michael <confabulate@kintzios.com>
To: gentoo-user@lists.gentoo.org
Subject: Re: [gentoo-user] m.2 nvme stick not what I was expecting at all.
Date: Wed, 26 Feb 2025 16:38:34 +0000 [thread overview]
Message-ID: <3865089.kQq0lBPeGt@rogueboard> (raw)
In-Reply-To: <f14b8557-8bb6-e0f6-abe0-ff6b6dd82891@gmail.com>
[-- Attachment #1: Type: text/plain, Size: 3867 bytes --]
On Wednesday, 26 February 2025 14:43:41 Greenwich Mean Time Dale wrote:
> Rich Freeman wrote:
> > On Tue, Feb 25, 2025 at 12:26 PM Dale <rdalek1967@gmail.com> wrote:
> >> I'm pretty sure you mentioned this once before in one of my older
> >> threads. I can't find it tho. I use PCIe x1 cards to connect my SATA
> >> drives for my video collection and such. You mentioned once what the
> >> bandwidth was for that setup and how many drives it would take to pretty
> >> much max it out. Right now, I have one card for two sets of LVs. One
> >> LV has four drives and the other has three. What would be the limiting
> >> factor on that, the drives, the PCIe bus or something else?
> >
> > It depends on the PCIe revision, and of course whether the controller
> > actually maxes it out.
> >
> > 1x PCIe v3 can do 0.985GB/s total. That's about 5 HDDs if they're
> > running sequentially, and again assumes that your controller can
> > actually handle all that data. For each generation of PCIe
> > forward/backwards either double/halve the transfer rate. The
> > interface works at the version of PCIe supported by both the
> > motherboard+CPU and the adapter card.
> >
> > If you're talking about HDDs in practice the HDDs are probably still
> > the bottleneck. If these were SATA SSDs then odds are that the PCIe
> > lane is limiting things, because I doubt this is an all-v5 setup.
> >
> > https://en.wikipedia.org/wiki/PCI_Express#History_and_revisions
> >
> > The big advantage of NVMe isn't so much the bandwidth as the IOPS,
> > though both benefit. Those run at full PCIe 4x interface speed per
> > drive, but of course you need 4 lanes per drive for this, which is
> > hard to obtain on consumer motherboards at any scale.
>
> This I think is what I needed. As it is, I'm most likely not maxing
> anything out, yet. The drives for Data, torrent stuff, stays pretty
> busy. Mostly reading. My other set of drives, videos, isn't to busy
> most of the time. A few MBs/sec or something, playing videos type
> reading. Still, next time I power down, I may stick that second card in
> and divide things up a bit. Might benefit if those cards aren't to great.
>
> I did copy this info and stuck in in a text file so I don't have to dig
> for it again, or ask again. ;-)
>
> Thanks.
>
> Dale
>
> :-) :-)
The other thing to straighten out, already hinted at by Rich et al., is an
NVMe M.2 card in a USB 3 enclosure won't be able to maximise its SSD transfer
rates. To do this it will require a Thunderbolt connector and a corresponding
Thenderbolt PC port, which will connect it internally to the computer's PCIe
bus, rather than USB/SATA. Hence a previous comment questioning the perceived
value of paying for a NVMe SSD M.2 form factor within a USB enclosure. It
won't really derive much if any performance benefit compared to a *good*
quality USB 3 flash drive (UFD), which can be sourced at a much lower price
point.
External storage medium, transfer protocol, device controller, PC bus and
cables/connectors/ports, will all have to have aligned generations of
technology and standards, if you expect to make most of their advertised
transfer speeds. Otherwise you'll be stuck at some component of a lower
performance providing a bottleneck to your aspirations. ;-)
Sometime ago I bought a SanDisk 1TB Extreme Portable SSD, which has a USB-C
connector and a USB 3.2 Gen 2 controller as a replacement for a flaky USB 3.0
stick. It is slightly bigger than a small UFD and more expensive than the
cheaper UFD offerings, but the faster speeds more than compensate for it. At
the time I bought it, external NVMe M.2 drives were too expensive and I only
had USB 3.0 ports anyway. So in my use case it was offering the best bang for
my buck.
[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 833 bytes --]
next prev parent reply other threads:[~2025-02-26 16:40 UTC|newest]
Thread overview: 60+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-02-24 18:58 [gentoo-user] m.2 nvme stick not what I was expecting at all Dale
2025-02-24 19:20 ` Jack
2025-02-24 19:49 ` Dale
2025-02-24 19:40 ` Mark Knecht
2025-02-24 19:41 ` Rich Freeman
2025-02-24 19:53 ` Dale
2025-02-24 20:00 ` Rich Freeman
2025-02-24 20:34 ` Dale
2025-02-24 20:02 ` eric
2025-02-24 20:11 ` Mark Knecht
2025-02-24 20:44 ` Dale
2025-02-24 22:48 ` Frank Steinmetzger
2025-02-25 3:56 ` Dale
2025-02-25 10:08 ` Michael
2025-02-25 11:00 ` Dale
2025-02-25 11:16 ` Michael
2025-02-25 15:57 ` Dale
2025-02-25 17:05 ` Rich Freeman
2025-02-25 19:00 ` Michael
2025-02-25 19:18 ` Rich Freeman
2025-02-25 20:39 ` Michael
2025-02-25 22:16 ` Frank Steinmetzger
2025-02-25 14:48 ` Peter Humphrey
2025-02-25 17:00 ` Michael
2025-02-24 22:13 ` Frank Steinmetzger
2025-02-25 4:20 ` Dale
2025-02-25 8:18 ` Wols Lists
2025-02-25 14:07 ` [gentoo-user] " Grant Edwards
2025-02-25 8:20 ` [gentoo-user] " Wols Lists
2025-02-25 11:09 ` Dale
2025-02-25 12:26 ` Dale
2025-02-25 15:04 ` Rich Freeman
2025-02-25 15:32 ` Dale
2025-02-25 16:29 ` Rich Freeman
2025-02-25 17:26 ` Dale
2025-02-25 17:41 ` Rich Freeman
2025-02-26 14:43 ` Dale
2025-02-26 16:38 ` Michael [this message]
2025-02-26 20:34 ` Dale
2025-02-26 20:48 ` Mark Knecht
2025-02-25 20:19 ` Wol
2025-02-25 20:21 ` Michael
2025-02-25 21:34 ` Frank Steinmetzger
2025-02-26 7:52 ` Wols Lists
2025-02-26 16:44 ` Michael
2025-02-26 18:03 ` Wol
2025-02-26 18:05 ` Rich Freeman
2025-02-26 18:42 ` Wol
2025-02-26 19:26 ` Rich Freeman
2025-02-26 19:47 ` Wols Lists
2025-02-26 19:56 ` Rich Freeman
2025-02-26 21:25 ` Wol
2025-02-26 22:37 ` Michael
2025-02-26 16:41 ` Michael
2025-02-28 5:43 ` Dale
2025-02-28 16:40 ` [gentoo-user] " Grant Edwards
2025-03-02 16:01 ` Dale
2025-03-02 22:06 ` Frank Steinmetzger
2025-03-02 22:39 ` Dale
2025-03-03 3:01 ` Grant Edwards
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=3865089.kQq0lBPeGt@rogueboard \
--to=confabulate@kintzios.com \
--cc=gentoo-user@lists.gentoo.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox