Skip to main content

software RAID on old vs. new CPUs


The Linux kernel has several software RAID algorithms, and selects the one that is fastest on your CPU. Isn't that always the same algorithm then ? No, definitely not. Newer CPUs have additional instructions that help speed things up. And it's not just clock speed that matters, memory bandwidth plays an important role too.

  • On an old Pentium II Xeon 450 MHz, raid5 uses p5_mmx, and raid6 uses mmxx2. Software raid6 calculations are 72% slower than raid5.
  • On a Pentium IV Xeon 1.5 GHz, raid5 using pIII_sse, and raid6 uses sse2x2. Software raid6 calculations are 12% slower than raid5.
  • On an AMD Athlon XP2000+ (1.6 GHz), raid5 uses pIII_sse, raid6 uses sse1x2. Software raid6 calculations are 42% faster than raid5.
On 64-bit systems, no relevant instructions are different between generations so far:
  • On a AMD Athlon64 XP3400 (2.4 GHz), raid5 uses generic_sse, raid6 uses sse2x4 (raid6 44% slower than raid5).
  • On a Xeon 5160 3GHz, raid5 uses generic_sse, raid6 uses sse2x4 (raid6 15% slower than raid5).
  • Same algorithms on a Xeon X5450 3GHz (raid6 20% slower than raid5).
  • Same algorithms on a Xeon E5430 2.66GH (raid6 18% slower than raid5).
  • Same algorithms again on a Xeon X5650 2.66GHz (raid6 15% slower than raid5).

Comments

deinoscloud said…
Very interesting!
Do you know what it's been used when running on an Intel® Atom™ Processor D510 1.66GHz (Dual-Core) 32Bit??

Cheers,
Didier
bert said…
@deinoscloud: the QNAP Linux kernel doesn't report it in dmesg, apparently. I did try to find that info, of course :-) What did you expect ? ;-)
bert said…
@deinoscloud: found it: on the Atom CPU in my QNAP raid 5 uses pIII_sse, measured at 5120 MB/s, and raid6 uses sse2x2, measured at 1082 MB/s.
These are the same algorithms as on the 32-bit Pentium IV CPU I tested. Performance for raid5 is 2.5x higher for comparable clock speed, but raid6 performance is only half.
The raid5 performance is far above
Of course, being a NAS device, the CPU in the QNAP is basically free to do this, so the CPU overhead is not slowing down other tasks, as long as it doesn't become a bottleneck. And with 4 cores available to spread other tasks over, I don't see it becoming a bottleneck soon.
deinoscloud said…
Awesome. Thx for the info!

Popular posts from this blog

How to solve "user locked out due to failed logins" in vSphere vMA

In vSphere 6, if the vi-admin account get locked because of too many failed logins, and you don't have the root password of the appliance, you can reset the account(s) using these steps:

reboot the vMAfrom GRUB, "e"dit the entry"a"ppend init=/bin/bash"b"oot# pam_tally2 --user=vi-admin --reset# passwd vi-admin # Optional. Only if you want to change the password for vi-admin.# exitreset the vMAlog in with vi-admin These steps can be repeated for root or any other account that gets locked out.

If you do have root or vi-admin access, "sudo pam_tally2 --user=mylockeduser --reset" would do it, no reboot required.

Volkswagen UHV bluetooth touch adapter & its problems

My Volkswagen car has the "universal cellphone preparation" UHV built-in. This is the main part of a car kit, but requires an additional adapter for connecting to a cellphone. At first, I was using an adapter for my good old Nokia 6310, even after I changed to the Nokia E71. Connecting was easy: pair the phone with the "VW UHV" bluetooth entity, and done. This has the phone connected to the car kit at all times, so even non-call-related functions use the car audio system (e.g. voice recognition).
But progress will have its way, no matter what happens. So in comes the "bluetooth touch adapter". Instead of a phone-specific adapter, this is a small touchscreen device that slots into the UHV dashboard mount. Connecting a phone is very different now:
the Bluetooth Touch Adapter connects to the "VW UHV" device via bluetooth
the phone connects to "Touch Adapter" device, also via bluetoothThe device doesn't allow step 2 if step 1 didn't s…

Multiple VLANs on a Synology NAS

Synology, like other SOHO/SMB NAS vendors, touts VLAN functionality with their current DSM 4.1 software. However, the web interface just lets you specify one VLAN tag to use over each eth interface (or bond interface).
Manual approachIn the busybox environment that you can ssh into as root (after enabling ssh through the webinterface), there's all the tools you need to use multiple VLANs over one link (eth or bond), however:
First you insert the 802.1q module into the Linux kernel:
 /sbin/lsmod | /bin/grep -q 8021q || /sbin/insmod /lib/modules/8021q.koThen you add each VLAN you need to every interface (bond0 in this example)
 /sbin/vconfig add bond0 4And finally you can configure IP addresses on every interface.vlan combination (bond0.4 in this example)
 /sbin/ifconfig bond0.4 192.168.4.1 broadcast 192.168.4.255 netmask 255.255.255.0The same type of script would work on a QNAP NAS too, by the way. They offer 8021q.ko and vconfig in their commandline environment as well.
Packets from…