Gentoo Archives: gentoo-user

From: Adam Carter <adamcarter3@×××××.com>
To: "gentoo-user@l.g.o" <gentoo-user@l.g.o>
Subject: Re: [gentoo-user] Slightly OT - My new SSD.
Date: Mon, 20 Mar 2017 06:09:24
Message-Id: CAC=wYCF+hgTRQ28-aa42PuTcsN7wyP4i_o982=bwzK8bGMA0BQ@mail.gmail.com
In Reply to: Re: [gentoo-user] Slightly OT - My new SSD. by Alan Mackenzie
1 > That, indeed, seems to to be the case. When I do cat /proc/interrupts |
2 > egrep '(CPU/nvm)', I get just the header line with one data line:
3 >
4 > CPU0 CPU1 CPU2 CPU3
5 > 17: 0 0 15 14605 IO-APIC 17-fasteoi
6 > ehci_hcd:usb1, nvme0q0, nvme0q1
7 >
8 > I'm kind of feeling a bit out of my depth here. What are the nvme0q0,
9 > etc.? "Queues" of some kind? You appear to have nine of these things,
10 > I've just got two. I'm sure there's a fine manual I ought to be
11 > reading. Do you know where I might find this manual?
12 >
13
14 Can't remember where i read up on this. Might have been troubleshooting
15 poor small packet performance on a firewall (some network drivers can have
16 multiqueue too). Maybe start with this;
17 https://www.thomas-krenn.com/en/wiki/Linux_Multi-Queue_Block_IO_Queueing_Mechanism_(blk-mq)
18
19 It looks like the nvme driver was made "mutliqueue" in kernel 3.19.
20
21 FWIW my system is 8 core (AMD 8350). Its odd having two queues on the same
22 interrupt, but I have the same for q0 and q1, but on your system i'd say
23 there should be some queues on other interrupts to they can be serviced by
24 other cores, so that doesnt look right.
25
26 Do you have MSI enabled? Bus options -> PCI Support -> Message Signaled
27 Interrupts (MSI and MSI-X)
28 If your system is not too old you may get more interrupts or a better
29 spread with that enabled.
30
31 When I look at the entire /proc/interrupts, there are just 30 lines
32 > listed, and I suspect there are no more than 32 interrupt numbers
33 > available. Is there any way I can configure Linux to give my SSD more
34 > than one interrupt line to work with?
35 >
36 > > FWIW
37 > > # hdparm -tT /dev/nvme0n1
38 >
39 > > /dev/nvme0n1:
40 > > Timing cached reads: 9884 MB in 2.00 seconds = 4945.35 MB/sec
41 > > Timing buffered disk reads: 4506 MB in 3.00 seconds = 1501.84 MB/sec
42 >
43 > I get:
44 >
45 > /dev/nvme0n1:
46 > Timing cached reads: 4248 MB in 2.00 seconds = 2124.01 MB/sec
47 > Timing buffered disk reads: 1214 MB in 3.00 seconds = 404.51 MB/sec
48 >
49 > So my "cached reads" speed is (a little under) half of yours. This is
50 > to be expected, since my PCIe lanes are only version 2 (and yours are
51 > probably version 3).
52
53
54 FWIW the motherboard manual says it has PCIe 2.0 x16 slots. Agree that
55 cache speed is likely a hardware issue.
56
57
58 > But the "buffered disk read" are much slower. Is
59 > this just the age of my PC, or might I have something suboptimally
60 > configured?
61 >
62
63 You look like you're getting SATA speeds, but since you have the nvme
64 device, i guess that implies you havent fallen back to SATA. Could well be
65 older hardware or less PCIe slots/lanes.

Replies

Subject Author
Re: [gentoo-user] Slightly OT - My new SSD. Alan Mackenzie <acm@×××.de>