**** BEGIN LOGGING AT Sun Sep 18 02:59:58 2016
Sep 18 05:33:37 <sicelo> i sometimes have to reboot the whole modem ..
Sep 18 05:37:37 <Vajb> is that that pnatd thingy?
Sep 18 06:12:29 <sicelo> yes
Sep 18 07:50:16 <KotCzarny> maxd: i usually enable offline mode then reconnect
Sep 18 07:50:34 <KotCzarny> didnt have a situation yet when it didnt work, but probably it does what sicelo said
Sep 18 09:49:07 <Vajb> hehe. Announcement at flight "if u have samsung galaxy note 7, please contact staff member"
Sep 18 11:28:54 <sixwheeledbeast> I thought they had all been recalled over here
Sep 18 13:18:26 <Vajb> sixwheeledbeast: dunno 'bout that. That was domestic flight at finland.
Sep 18 15:12:50 <KotCzarny> they were recalled worldwide
Sep 18 15:12:51 <KotCzarny> :>
Sep 18 15:52:55 <sixwheeledbeast> I assumed so but wasn't sure
Sep 18 15:53:37 <sixwheeledbeast> There is always a chance someone kept it and ignored the recall
Sep 18 16:05:32 <kerio> how does samsung fuck up something like that
Sep 18 16:05:49 <KotCzarny> costly
Sep 18 16:06:22 <kerio> no i mean
Sep 18 16:06:36 <kerio> after so many phones, you'd think that they have the basic elements of engineering down
Sep 18 16:06:55 <KotCzarny> does samsung produce its own batteries?
Sep 18 16:07:15 <KotCzarny> also, fuckups happen
Sep 18 16:07:45 <kerio> is it just a matter of defective batteries?
Sep 18 16:09:18 <Vajb> after it explodes i guess it's whole device
Sep 18 16:09:30 <kerio> lel
Sep 18 16:11:10 <KotCzarny> some random comment: From 24 Million few Handsets went to an explosion! and samsung investigated on it and found few batches got battery from a local vendor! the quality of the battery pack! (the battery is maid by samsung! just the protecting cover!! is faulty) 
Sep 18 16:14:58 <KotCzarny> hmm, battery in note7 isnt user replaceable? lol
Sep 18 16:15:21 <KotCzarny> that would explain why recall on the whole handsets
Sep 18 16:16:08 <kerio> handsets should do what the nintendo 3ds does
Sep 18 16:16:23 <kerio> and have a user replaceable battery behind a screwed-in panel
Sep 18 16:16:32 <KotCzarny> https://imgur.com/a/RMBnv
Sep 18 16:36:20 <RST38h> Kot: They made it unreplaceable starting with Note 5
Sep 18 16:36:41 <RST38h> Kot: At Note 7, karma finally caught up with them
Sep 18 16:36:47 <KotCzarny> exactly
Sep 18 16:36:59 <RST38h> And in such an ironic way, hehe...
Sep 18 16:37:17 <KotCzarny> do you think it was apple's sabotage?
Sep 18 16:38:02 <RST38h> I think it was karmic justice.
Sep 18 16:38:59 <RST38h> Hopefully they will pay for replacing 1+ million Note7s and learn to make battery replaceable again in Note8
Sep 18 16:40:34 <KotCzarny> or just make design/testing/postproduction-testing better
Sep 18 16:41:17 <RST38h> Can't account for every fuckup, better have plans B, C, etc
Sep 18 16:41:39 <KotCzarny> my n900 didnt 'sploded yet
Sep 18 16:41:42 <RST38h> Meanwhile, Samsung got another, more interesting problem
Sep 18 16:41:44 <RST38h> http://www.theregister.co.uk/2016/09/13/arm_biglittle_gcc_bug
Sep 18 16:41:46 <KotCzarny> not going to change that
Sep 18 16:41:57 <KotCzarny> cache linesizes?
Sep 18 16:42:02 <KotCzarny> or something new?
Sep 18 16:42:17 <RST38h> Yep
Sep 18 16:42:25 <KotCzarny> saw it last week
Sep 18 16:42:37 <RST38h> Samsung's own cores are 64bit, so they come with wider cache lines
Sep 18 16:42:51 <KotCzarny> but its new thing, so as always, assumptions will fail sooner or later
Sep 18 16:42:56 <RST38h> GCC punks were either not informed or did not care enough
Sep 18 17:08:21 <DocScrutinizer05> LOL
Sep 18 17:11:13 <DocScrutinizer05> >>The faults involved code created just in time (JIT) on demand: Mono uses a JIT compiler to transform an app's portable bytecode to native ARM instructions on handheld devices.<< kill dat shite, with fire. It serves them right this is going to segfault
Sep 18 17:13:21 <kerio> the fuck
Sep 18 17:21:36 <DocScrutinizer05> ridiculous how they try to optimiue by determining I-cache width only once. Particularly since they were better off making thatbig-LITTLE transfer as rock solid as possible (purge cache byte by byte, worst case) and rather keep the frequency of such big-LITTLE transitions as low as possible
Sep 18 17:22:22 <DocScrutinizer05> again big fat LOL
Sep 18 18:14:14 <ShadowJK_> They don't do "transitions" anymore really, all cores can be active at same time
Sep 18 18:14:24 <ShadowJK_> So what you have is not SMP, but asymmetric multiprocessing :)
Sep 18 18:14:46 <KotCzarny> :)
Sep 18 18:15:06 <ShadowJK_> There are some 10-core SOCs now, where you have 4 low perf cores, 4 medium perf, and 2 high performance cores, and they can all be active at the same time
Sep 18 18:15:23 <KotCzarny> but its like running something on pentium-pro, then in an instant replacing cpu with pentium-mmx
Sep 18 18:15:51 <KotCzarny> ShadowJK_: are you sure they arent counting MALI cores? ;)
Sep 18 18:15:57 <ShadowJK_> yes
Sep 18 18:16:23 <KotCzarny> havent seen three-tier cpu, link/name?
Sep 18 18:17:03 <ShadowJK_> Helio X20
**** BEGIN LOGGING AT Sun Sep 18 18:19:21 2016
Sep 18 18:23:08 <KotCzarny> unless its a way to sell wafers with unstable a53 cores that are stable at lower freq
Sep 18 18:28:31 <ShadowJK> http://www.anandtech.com/show/9227/mediatek-helio-x20
Sep 18 18:28:41 <ShadowJK> sorry if repeat I lost irc for a bit
Sep 18 18:32:01 <KotCzarny> i find it curious why they didnt make a53 cores in one cluster with cpufreq
Sep 18 18:32:19 <KotCzarny> unless the controller supports max 4 cores at once
Sep 18 18:39:21 <KotCzarny> seriously, it looks more like marketing, unless there are some more design details differentiating those two a53 clusters
Sep 18 19:38:20 <DocScrutinizer05> ShadowJK: thanks a lot
Sep 18 19:39:25 <DocScrutinizer05> ShadowJK: ((They don't do "transitions" anymore)) however that's exactly where their stuff explodes, see http://www.theregister.co.uk/2016/09/13/arm_biglittle_gcc_bug
Sep 18 19:41:28 <DocScrutinizer05> they create JIT code, so they need to purge I-cache to make the code really available, and that fails since they may clear 64bytes but the cache width of new core the process got reassigned to is 128byte
Sep 18 19:43:39 <DocScrutinizer05> if processes wouldn't get shifted/transferred from a small to a fat core and back, this bug would never happen
Sep 18 19:44:21 <ShadowJK> Oh, I thought you meant the old style big.little where you either had 1-4 big cores running, or 1-4 small cores running, but not both at the same time
Sep 18 19:45:12 <ShadowJK> KotCzarny, who knows if it's the design of the cores itselves, or details in manufacturing, but they do supposedly run at different efficiencies
Sep 18 19:45:17 <DocScrutinizer05> no, the ridiculous thing is the purge_cache function determines cache width only *once* and then keeps it in a sttaic variable
Sep 18 19:45:24 <ShadowJK> Beyond what you could do with just cpufreq
Sep 18 19:45:31 <ShadowJK> DocScrutinizer05, yeah
Sep 18 19:46:12 <ShadowJK> Interestingly that Helio SoC actually has an 11th CPU too :-)
Sep 18 19:47:27 <DocScrutinizer05> the sensor shite, yeah
Sep 18 19:48:09 <DocScrutinizer05> really weird how they throw mountains of etched silicon at such a relatively simple task like operating a smartphone
Sep 18 19:48:41 <ShadowJK> My current daily phone has 4 cores and I find I could use more ;-)
Sep 18 19:48:50 <DocScrutinizer05> instead of maybe decently isolating the modem, to start with
Sep 18 19:49:31 <DocScrutinizer05> pastebin top with 4 running processes or it didn't happen ;-)
Sep 18 19:50:40 <ShadowJK> It's mostly like open tab 1, 2, 3, 4, 5, 6 in browser, go to tab 1 and wait for it to load and hope the other ones have loaded before I'm done reading
Sep 18 19:50:41 <DocScrutinizer05> heck, even on my PC I hardly get all 4 cores (8 virtual) busy with concurrent processes
Sep 18 19:51:53 <DocScrutinizer05> and you think your browser really runs one thread per tab?
Sep 18 19:55:16 * DocScrutinizer05 prays for the day when Apple or Google will decide they kill/stop all JS and similar crap after it ate 0.1s CPU time on a webpage
Sep 18 19:56:00 <DocScrutinizer05> heck, I can *build* applications in less time
Sep 18 19:58:15 <DocScrutinizer05> a whole system-devele and hw industry makes a living from dealing with the turd so called "web programmers" declare as their result for solving what also they declared problems in web pages
Sep 18 19:59:04 <DocScrutinizer05> it's a sad joke
Sep 18 19:59:21 <kerio> DocScrutinizer05: nethack takes like 8 seconds on an 8T xeon to build
Sep 18 19:59:57 <DocScrutinizer05> I bet you can easily find a dozen webpages that take longer to render on same machine
Sep 18 20:02:46 <DocScrutinizer05> and the braindead sheep go "oh, this device is too old, the web pages take a 30s to render meanwhile. I need some stronger hardware" - instead if giving such webpages and their "web programmers" the finger and a loud "F*CK U!" and a kick with the heavy boot
Sep 18 20:03:15 <DocScrutinizer05> instead of*
Sep 18 20:04:24 <DocScrutinizer05> can't Mcaffee and firens simply tag all such incredibly fucked-up webpages as "DANGEROUS! Blocked it" KTNXBY
Sep 18 20:04:34 <DocScrutinizer05> friends*
Sep 18 20:05:14 <DocScrutinizer05> would teach them "web programmers" that we don't need animated menus and shit
Sep 19 02:29:36 <ShadowJK> Firefox is the only browser these days that don't use process/thread per tab, though firefox is going to change that soon
**** ENDING LOGGING AT Mon Sep 19 02:59:59 2016