| mup | Bug #1455840 was opened: network-bridge compatibility with Open vSwitch <network> <openvswitch> <ovs> <juju-core:New> <https://launchpad.net/bugs/1455840> | 00:49 |
|---|---|---|
| === kadams54 is now known as kadams54-away | ||
| === kadams54 is now known as kadams54-away | ||
| === kadams54-away is now known as kadams54 | ||
| === kadams54_ is now known as kadams54-away | ||
| davecheney | mwhudson: do arm put "superceeded" on the background of every page unconditionally | 21:27 |
| davecheney | or are there actually some pages they have not yet depricated ? | 21:28 |
| mwhudson | davecheney: heh | 21:28 |
| mwhudson | davecheney: only the pages that have been around long enough to be found by google, i think | 21:28 |
| mwhudson | there will be a version that does not have superceded written on | 21:28 |
| mwhudson | but it can be pretty hard to find | 21:28 |
| davecheney | mwhudson: http://paste.ubuntu.com/11193890/ | 23:32 |
| davecheney | sad NEON is sad | 23:32 |
| mwhudson | fun times | 23:34 |
| davecheney | for limited values of fun | 23:34 |
| mwhudson | is there no neon at all? | 23:34 |
| davecheney | my guess is it's disabled in the kernel | 23:35 |
| davecheney | is that a thing ? | 23:35 |
| davecheney | model name: ARMv7 Processor rev 10 (v7l) | 23:35 |
| davecheney | BogoMIPS: 790.52 | 23:35 |
| davecheney | Features: swp half thumb fastmult vfp edsp neon vfpv3 tls vfpd32 | 23:35 |
| davecheney | compared to my freescale imx6 | 23:35 |
| mwhudson | oh yeah, no noen in features | 23:36 |
| mwhudson | *neon | 23:36 |
| davecheney | but is that something you can disable in ther kernel | 23:36 |
| davecheney | or does their SoC not have that feature ? | 23:36 |
| mwhudson | it is also something you can not include the silicon for | 23:36 |
| mwhudson | no idea, i don't know much about marvell designs | 23:36 |
| davecheney | as in , it's not optional in cortex-A ? | 23:36 |
| mwhudson | not sure about the answer to that specific question | 23:38 |
| mwhudson | it's optional in armv7 | 23:38 |
| * mwhudson asks @scaleway on twitter | 23:40 | |
| mwhudson | ... and people ask why arm hasn't taken over the data centre | 23:42 |
| davecheney | mwhudson: i also dm'd the guy | 23:44 |
| davecheney | we'll see what happens | 23:44 |
| mwhudson | cool | 23:44 |
| davecheney | i'd be amazed that | 23:44 |
| davecheney | a. they go their own silicon, rather than just taking a marvell or freescale soc | 23:44 |
| mwhudson | i guess 2am on monday morning is not a good time to expect quick replies | 23:44 |
| mwhudson | it's a slightly custom marvell part | 23:44 |
| davecheney | o_O! | 23:44 |
| davecheney | why do people do things like this | 23:44 |
| mwhudson | i did manage to dig that out of their site | 23:44 |
| mwhudson | "We developed with Marvell a Cortex A9 ARM SoC offering excellent performances for web applications." | 23:45 |
| mwhudson | https://www.scaleway.com/faq/server#why-did-you-choose-arm- | 23:45 |
| davecheney | blah blah | 23:46 |
| mwhudson | it's also hinted at in the cpuinfo | 23:46 |
| mwhudson | Processor : Marvell PJ4Bv7 Processor rev 2 (v7l) | 23:46 |
| davecheney | the number of things you can tweak on a custom silicon that actually make a difference | 23:46 |
| davecheney | are effectively zero | 23:46 |
| mwhudson | it's probably just form factor or something | 23:46 |
| mwhudson | cortex a9 means it's arm's design for the actual cpu bit | 23:47 |
| mwhudson | s | 23:47 |
| davecheney | yeah | 23:47 |
| davecheney | oh | 23:47 |
| davecheney | a9 | 23:47 |
| davecheney | booh | 23:47 |
| mwhudson | although it's always possible that scaleway's marketing material is using the terms wrong :) | 23:49 |
| davecheney | inconceivable | 23:49 |
| mwhudson | the hints i see around the net sort of imply that they are | 23:56 |
| mwhudson | i _think_ it's based on the armada 370, which is not a cortex a9 | 23:57 |
| mwhudson | and does not have neon | 23:57 |
| mwhudson | (and i think cortex a9s always do have neon) | 23:57 |
| davecheney | yeah, looks like armada 370 | 23:57 |
| davecheney | that is oooooooooooooold | 23:57 |
| mwhudson | yeah | 23:57 |
| mwhudson | at least it has better IO than a panda :) | 23:58 |
| davecheney | how they got a 4 core version of that is a great mistery | 23:58 |
| davecheney | a yugo has better io than a pandaboard | 23:58 |
| mwhudson | oh yeah | 23:58 |
| mwhudson | cortex a9 is also dual core | 23:58 |
| mwhudson | oh no, 1-4 | 23:59 |
| mwhudson | lying there | 23:59 |
| mwhudson | ah, and neon is optional | 23:59 |
| mwhudson | specifications on http://www.arm.com/cortex-a9.php | 23:59 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!