| 00:27.45 | jabawok | ravelo: N900 would really benefit from a modern kernel with zswap support |
| 00:30.59 | jabawok | xman: i've had 2 N900's over 7 years (first one was preordered). They have both been overclocked to 1000mhz since day dot with zero problems |
| 00:31.39 | jabawok | you will NOT burn out a n900 by overclocking it at stock voltages |
| 00:52.55 | *** join/#neo900 goiken_ (~goiken@dslb-094-223-165-012.094.223.pools.vodafone-ip.de) |
| 01:34.18 | varu | mine got pretty good battery on a 900mhz oc with some careful undervolting. ram really killed it though, and i imagine compressed ram & swap probably wouldn't help the speed much |
| 01:36.39 | varu | the hardware's too far gone by this point, unfortunately. even neo900 is slowly going that route, not that it was particularly cutting edge to begin with, but that's OK with me. 1gb ram + some careful optimization is worth the tradeoff given a reasonable release date, for my use case anyway. i knew & accepted this going in :) |
| 01:38.24 | varu | seeing it come to completion and having its feature set available to users and developers will imo set the stage for a future project where a lot of the background work has already been done. it'll be somewhat easier, faster, and hopefully cheaper the second time around - and the docs will be (and are already) there for anyone to catch up with |
| 01:41.26 | Joerg-Neo900 | that's the idea, yes :-) |
| 01:44.10 | Joerg-Neo900 | jabawok: how many hours did the CPU actually in OP5++? |
| 01:47.47 | Joerg-Neo900 | just in case anybody didn't notice: this question was rhetorical. without special tools (which would have other adverse effects) you simply cannot answer it. Even a CPU "overclocked to 1000MHz" and power up 24/7/356 might not run a single CPU minute at OP5_overclocked aka maximum (1000MHz) frequency during years |
| 01:49.21 | Joerg-Neo900 | easily could spend 99% of time in suspend and 1% in 250MHz nevertheless |
| 01:50.00 | varu | (side-note and only somewhat related: not all n900s run 1ghz without issue; two of mine don't) |
| 01:50.23 | Joerg-Neo900 | no surprise |
| 01:51.10 | *** join/#neo900 xman (~xman@user-0cdft6f.cable.mindspring.com) |
| 01:53.05 | Joerg-Neo900 | anyway a lot of savy engineers think you may do whatever you like with your own device, no matter what poor idea it might be from a technical PoV. But *never* tell others you had _tested_ it and it was safe since it didn't explode into your face |
| 01:55.52 | Joerg-Neo900 | I'm smoking since 40 years, nevertheless I'd never tell anybody it's not harmful to your health since I'm still around. |
| 01:56.39 | varu | "pulling 30C from a 3000mAh lithium battery is *totally* safe!" </note 7 design team> |
| 01:56.58 | Joerg-Neo900 | haha |
| 01:57.03 | varu | note: i don't actually know if *that's* the reason behind the explosions, but it makes for a fun thought |
| 01:57.24 | Oksana | What is C? Not degrees Celsius, sure |
| 01:57.30 | Joerg-Neo900 | maybe 10C, and during charge |
| 01:57.47 | Joerg-Neo900 | C *here* is 'Capacity' |
| 01:58.19 | Joerg-Neo900 | 1C is the electrical current in Ampere that according to label would discharge the battery in 1h |
| 01:58.42 | Joerg-Neo900 | so 30C from a 3000mAh are a 90A |
| 01:58.56 | Oksana | Ah, I remember... Discharging the battery in 1h sounds like very fast |
| 01:59.12 | varu | people with e-cigs have gotten their hands intimately familiar with that equation |
| 01:59.25 | varu | the hospitals storing their records have as well |
| 02:00.26 | Joerg-Neo900 | 10C should be doable for a decent LiIon cell designed for heavy duty |
| 02:01.36 | varu | i always marvel at the people designing and running their own "mech mods" - basically connecting the battery straight to the heating coil without regulation. you can design for it, but, one mistake and... |
| 02:02.03 | Joerg-Neo900 | yeah |
| 02:02.17 | varu | there was a scooter song whose refrain accurately reflected the outcome |
| 02:02.25 | varu | likewise for the ensuing dance |
| 02:02.34 | Joerg-Neo900 | hehehe |
| 02:03.20 | Joerg-Neo900 | good night |
| 02:03.46 | Oksana | good night :-) |
| 02:06.23 | jabawok | Joerg-Neo900: re cpu pegged at 100% - lots, heaps and heaps of emulator gameplay |
| 02:08.18 | jabawok | varu: would still be an interesting experiment re zswap, considering when n900 lags, its heavily I/O bound on swap, with cpu almost idle |
| 02:11.02 | *** join/#neo900 galiven (~Andrew@50-205-116-131-static.hfc.comcastbusiness.net) |
| 02:11.14 | varu | do you think the compression would be significant enough? |
| 02:11.42 | varu | most of the swap issues, in my use case at least, come from the browser. a lot of the content the browser has to deal with is already compressed |
| 02:13.38 | varu | it may help with swapping system libs and parts of the browser itself, i'm really not sure how compressible most of that stuff is. i'm not sure whether someone has done an analysis of frequently ran programs / system services and their possible compression ratios |
| 02:16.32 | Joerg-Neo900 | not the canvas |
| 02:17.30 | Joerg-Neo900 | and I guess often it's exactly the canvas and pictures (also decompressed) that eats lots of memory |
| 02:17.48 | Joerg-Neo900 | re OC just bother to really read the stuff at |
| 02:17.52 | Joerg-Neo900 | +omap-oc |
| 02:17.57 | Joerg-Neo900 | ~omap-oc |
| 02:17.57 | infobot | it has been said that omap-oc is http://mg.pov.lt/maemo-irclog/%23maemo.2010-08-01.log.html#t2010-08-01T22:16:05 read that!, or and this http://mg.pov.lt/maemo-irclog/%23maemo.2011-03-11.log.html#t2011-03-11T03:04:11, or http://electronics.stackexchange.com/a/156130 |
| 02:19.38 | Joerg-Neo900 | http://mg.pov.lt/maemo-irclog/%23maemo.2011-03-11.log.html#t2011-03-11T03:04:11 is right from the TRM. And the at least exponential increase in deterioration of chip with each MHz overclocking is self evident from the physics behind it |
| 02:26.01 | *** join/#neo900 Xiaoman (~Xiaoman@dhcp-5-186-122-106.cgn.ip.fibianet.dk) |
| 02:31.42 | varu | i haven't read it, but probably some serious electromigration if you do it |
| 02:32.53 | Joerg-Neo900 | yep, among other effects |
| 02:32.55 | varu | i've seen pics of that stuff on higher-nm processors years ago, literally eats into the silicon and breaks all sorts of things |
| 02:33.33 | Joerg-Neo900 | also electrons shooting into the SiOx isolator layers |
| 02:34.17 | Joerg-Neo900 | and doping atoms moving around |
| 02:34.24 | varu | yeah.. no insulation == no transistor == byebye reliable proc |
| 02:34.35 | varu | then you're stuck diagnosing weird stuff that's just never reproducible |
| 02:34.43 | Joerg-Neo900 | exactly |
| 02:34.45 | varu | and.. why? for a few percent gain? |
| 02:35.01 | Joerg-Neo900 | and that's "the fault of the OS" then |
| 02:35.02 | varu | i overclocked like a madman in the socket 7 days, but since.. eh... |
| 02:35.36 | varu | not worth the hassle lol |
| 02:36.07 | varu | back then an extra 100mhz made a massive difference. now... (and tbh ever since c2d came out), who cares |
| 02:36.09 | *** join/#neo900 x29a (~x29a@unaffiliated/x29a) |
| 02:36.24 | varu | even more so for phones where battery life is #1 |
| 02:36.45 | Joerg-Neo900 | you most likely can get away with a 500h total of overclocked emulator game playing, but then it might start to behave strangley, in a maybe initially 5 to 10% of devices |
| 02:36.54 | varu | well, #2 for neo900, ideology and correct implementation takes #1 |
| 02:37.00 | varu | but close enough |
| 02:37.05 | *** join/#neo900 knttl (~knttl@dyndsl-091-248-125-080.ewe-ip-backbone.de) |
| 02:37.53 | varu | yeah.. it's just not worth it. accept the hardware as it is and... well, buy something newer/faster if what you have isn't doing the trick |
| 02:38.05 | Joerg-Neo900 | would you like to review the schematics? |
| 02:38.59 | varu | the time wasted finding a reasonable OC level & the further time wasted debugging it could be spent productively enough to provide the finances of solving your problems through better hardware |
| 02:39.04 | Joerg-Neo900 | for N900 overclocking most usually isn't worth anything since the eMMC bandwidth (for swap) is the bottleneck |
| 02:39.28 | varu | if you're asking me, i don't have the expertise for it, not an EE |
| 02:40.33 | Joerg-Neo900 | and for swap, and memory shortage, a thumb enabled system and reswap (defragmenting) and slimming down the core system goes a long way |
| 02:41.15 | varu | the latter is very important, less ram is less of a problem if most of it is free to begin with |
| 02:42.05 | Joerg-Neo900 | yep, thus as much as possible in thumb code, and uninstall / don't start stuff you don't need |
| 02:42.28 | Joerg-Neo900 | like modest (email), depending on your usage pattern |
| 02:42.47 | varu | however, with 256mb and the general trend not to care too much about ram utilization (hardware'll make up for it, and that's consistently been true), optimizing doesn't make much sense. 256mb is for me a non-starter these days |
| 02:43.19 | Joerg-Neo900 | that's why we got a 4 times that :-) |
| 02:43.41 | Joerg-Neo900 | with the core system still biting out a ~200MB |
| 02:43.50 | varu | i'd rather find a way i can help neo900 development than waste hours on hardware that the modern development and environment has left behind |
| 02:44.03 | varu | with 1gb it's quite reasonable today |
| 02:44.29 | Joerg-Neo900 | you get 800MB free vs 50MB free |
| 02:44.31 | varu | not sure what it'll look like in 5 years, but again, once the project is completed, the next iteration will be quicker (at least so i expect) |
| 02:46.12 | Joerg-Neo900 | n8 again |
| 02:47.15 | varu | good night :) |
| 03:02.05 | jabawok | Joerg-Neo900 and varu: in my experience, the N900 overclock was not just "a few percent gain" - it was night and day for cpu intensive stuff like games and emulation - running the camera in video mode @720p, and overall responsiveness, including playing back flacs without skipping in mplayer when using the UI and task switching. |
| 03:02.52 | jabawok | also, in almost 3 decades of overclocking experience, i've never had a single breakdown of CPU or any reliability deterioration. |
| 03:03.23 | jabawok | what you're saying is pretty much just a bunch of wives tales as far as i can see. |
| 03:04.41 | jabawok | <Joerg-Neo900> also electrons shooting into the SiOx isolator layers |
| 03:04.59 | jabawok | ^^ those sorts of statements are all theoretical, not real-world |
| 03:05.17 | Joerg-Neo900 | aha |
| 03:05.19 | jabawok | anyway, just sayin, and trying to provide a balanced overall view for the channel readers |
| 03:05.52 | Joerg-Neo900 | https://en.wikipedia.org/wiki/Hot-carrier_injection |
| 03:06.36 | jabawok | https://en.wikipedia.org/wiki/Tomato |
| 03:06.37 | jabawok | :P |
| 03:06.49 | jabawok | i see "overclock" mentioned exactly 0 times on both those pages |
| 03:07.24 | Joerg-Neo900 | neither einstein |
| 03:07.38 | Joerg-Neo900 | or Bor |
| 03:07.49 | jabawok | my point is.. in *my* experience, the overclocking doomsayers are overstating things |
| 03:09.07 | jabawok | based on my real world experience of zero issues using reasonable overclock settings. (ie overclock upwards till unstable, then dial it back a bit and let it run for decades with zero issues) |
| 03:09.24 | Joerg-Neo900 | in MY experience you are teling me smoking is harmless |
| 03:09.43 | jabawok | fair enough - you've had some issues? |
| 03:12.49 | jabawok | anyway... i do know there's merrit in your argument. i'm not denying basic physics, but as i said, the doom is overstated in general, so i'm trying to pull the sentiment the other way a bit |
| 03:15.31 | Joerg-Neo900 | 70FIT after 10l Hours at OPP5 is a hard fact |
| 03:15.36 | Joerg-Neo900 | 10k |
| 03:16.23 | Joerg-Neo900 | and OC is for sure no therapy for that |
| 03:18.00 | Joerg-Neo900 | you're free to do whatever you want with all your appliances and devices, but please refrain from explaining others that you know bnetter than TI |
| 03:20.07 | Joerg-Neo900 | 10k is roughly one year. and obvously simple math applies, so even when OC only would halve the livespan, you could expect first significant failure rate increase after as short as 6 months |
| 03:21.41 | Joerg-Neo900 | I'm absolutely sure that an overclocking to almost twice the clock will do way more than just doubling the speed of deterioration |
| 03:22.04 | jabawok | so the fact that i have a 7 year old overclocked N900 thats still running beautifully after a *very* hard life run for weeks at a time plugged into a tv with emulator games.. does that mean i'm not running in OPP5? |
| 03:22.16 | Joerg-Neo900 | rather like an incandescent lamp where you run it with twice the power it's designed for |
| 03:22.30 | jabawok | or that it was just run for well less than 101 hours in OPP5 |
| 03:23.12 | Joerg-Neo900 | please understand the meaning of FIT (Failures In Thousand) |
| 03:23.42 | jabawok | yep but you're making my point for me |
| 03:23.47 | Joerg-Neo900 | you're just one of the maybe 60 happy ones out of 100 |
| 03:23.48 | jabawok | you're overstating the doom |
| 03:24.26 | jabawok | because my scenario is way worse than 101 hours in OPP5 |
| 03:24.38 | Joerg-Neo900 | 101??? |
| 03:24.55 | jabawok | oh typo |
| 03:24.59 | jabawok | i saw your 10l above |
| 03:25.03 | jabawok | assume k |
| 03:25.40 | jabawok | again, proving my point.. even with my n900 overclocking abuse, i doubt it would have been 10000h |
| 03:25.47 | Joerg-Neo900 | err you bought a "101 hours are roughly one year"??? |
| 03:26.18 | Joerg-Neo900 | I'm not interested in your doubts |
| 03:26.21 | jabawok | k |
| 03:26.30 | varu | jabawok: those statements are *not* theoretical, i've seen the electron microscope proof myself :) |
| 03:26.33 | Joerg-Neo900 | 10k hours are a tad more than one year |
| 03:26.50 | varu | also the anecdotal evidence of two ahtlon XPs i ruined by clocking them way too high for too long under 100% load |
| 03:26.52 | Joerg-Neo900 | and that's right from TI manuals |
| 03:26.55 | jabawok | varu: did it affect your n900? |
| 03:26.55 | varu | (converting videos) |
| 03:26.56 | Joerg-Neo900 | and datasheets |
| 03:27.20 | varu | i have my n900 at 900-950mhz, but my usage just doesn't run it at 100% for long enough to matter |
| 03:27.36 | varu | that and it's been sitting in a drawer for about a year ;) |
| 03:27.41 | jabawok | heh |
| 03:27.53 | jabawok | can you see my point about the doomsaying? |
| 03:27.56 | varu | i need browser and maps. if i can't do it, phone is useless |
| 03:28.01 | Joerg-Neo900 | no |
| 03:28.16 | varu | i can, but only within a particular use case |
| 03:28.31 | jabawok | i'd go so far as to say "almost all" use cases |
| 03:28.32 | varu | if you're not running 100% for much of the proc's uptime, you're probably fine for quite some time |
| 03:28.37 | Joerg-Neo900 | you're asuming arbitrary random stuff and blaming others for quoting facts as "doomsaying" |
| 03:29.18 | varu | like, all of these things we mention can, do, and *will* happen. 100%, no ifs ands or buts. the only question is when, and how it'll manifest itself to you from an end-user perspective |
| 03:29.50 | varu | yes, you can argue that "when" is large enough for you (or most) to OC and be ok |
| 03:29.52 | varu | and maybe it's so |
| 03:29.56 | jabawok | well i'm just here to affirm that overclocking an N900 after all these years with people doomsaying in the forums about it, i've had zero issues. in my books i'm completely vindicated |
| 03:29.58 | varu | feel free. it's *your* device :) |
| 03:30.00 | Joerg-Neo900 | there's no question "when", TI clearly says after ~1y for 7 of 100 devices |
| 03:30.17 | varu | but to say that you'll absolutely be fine is false |
| 03:30.21 | varu | you most likely will be |
| 03:30.26 | jabawok | and were |
| 03:30.28 | varu | but if you're not, be ready to pay the price |
| 03:30.33 | jabawok | sure |
| 03:30.34 | varu | yes, *you* were :) |
| 03:30.36 | jabawok | we can aggree on that |
| 03:30.51 | varu | the problem from a development standpoint is, let's say the entire community does it |
| 03:31.03 | varu | let's say they go crazy and encode x264 with their n900s all day |
| 03:31.11 | varu | theoretical but stick with me |
| 03:31.33 | varu | now a developer that targets n900 has to deal with *random* reports of *random* issues |
| 03:31.52 | varu | weird behaviour, app crashes, no logs or cooperation from most users.. |
| 03:32.38 | varu | because if you OC and you run x264 24/7, you WILL wreck your cpu. matter of time. and the way you'll wreck yours will likely not be how someone else's will be wrecked |
| 03:32.50 | varu | making development for the platform and bughunting an absolute *nightmare* |
| 03:33.06 | jabawok | yeah i see your point. its a tough one. you'd hope i guess that if someone is savvy enough to overclock, they'll realise the potential pitfalls. I'm guessing thats not always the case though |
| 03:33.11 | varu | ...making developers scurry away to platforms where that stuff is either locked down or where enough people don't do it |
| 03:33.21 | varu | no, its not, so you can't blanket-recommend it |
| 03:33.52 | varu | the forum thread (if there is one) recommending this sort of stuff should have a relatively in-depth description of the risks involved |
| 03:34.18 | varu | it took me 10min on irc to lay it out (to my satisfaction at least), shouldn't be too hard for an intro post in a forum thread :) |
| 03:35.18 | Joerg-Neo900 | forget it, 95% of OC fans will simply blame you (and me) for "doomsaying" |
| 03:35.25 | jabawok | yeah all the overclocking discussions have plenty of warnings etc in forums, and enough people like yourself saying "please dont" |
| 03:35.34 | varu | well jabawok at least seems reasonable in the end :P |
| 03:35.43 | varu | but yeah, few are :( |
| 03:35.57 | varu | yes, but "please don't" has no explanatory power |
| 03:36.02 | varu | people always love to think they know better |
| 03:36.07 | Joerg-Neo900 | nobody said "please don't overclock" - I said "please NEVER suggest it to others" |
| 03:36.17 | Joerg-Neo900 | your answer was "you're doomsaying" |
| 03:36.23 | jabawok | hopefully have an open mind. i consider myself schooled on this one - for reasons other than initially touted though. I didn't factor in the development point of view |
| 03:36.57 | varu | all this stuff matters in the survival of a platform. it's all intertwined. miss one and... the platform has a high chance of dying :( |
| 03:37.31 | varu | looks over at his amiga |
| 03:39.18 | jabawok | Joerg-Neo900: apologies if that discussion was more heated than it was meant to be. I do see your point of view. |
| 03:45.06 | *** join/#neo900 Kabouik (~quassel@65.76.93.92.rev.sfr.net) |
| 04:36.19 | *** join/#neo900 cybiko123 (~cybiko123@unaffiliated/cybiko123) |
| 04:59.54 | *** join/#neo900 goiken_ (~goiken@dslb-094-223-165-012.094.223.pools.vodafone-ip.de) |
| 05:19.21 | *** join/#neo900 neo900 (~office@neo900/coreteam/joerg) |
| 05:19.21 | *** mode/#neo900 [+v Joerg-Neo900] by ChanServ |
| 05:19.23 | *** join/#neo900 DocScrutinizer05 (~saturn@openmoko/engineers/joerg) |
| 05:19.23 | *** mode/#neo900 [+v DocScrutinizer05] by ChanServ |
| 06:14.31 | *** join/#neo900 goiken_ (~goiken@dslb-094-223-165-012.094.223.pools.vodafone-ip.de) |
| 07:26.11 | *** join/#neo900 goiken_ (~goiken@dslb-094-223-165-012.094.223.pools.vodafone-ip.de) |
| 07:46.31 | *** join/#neo900 radekp (~radek@82.113.39.213) |
| 08:25.42 | *** join/#neo900 mzki (~koza@89-76-126-150.dynamic.chello.pl) |
| 08:45.58 | *** join/#neo900 jabawok (~Jabawok__@110-175-170-78.static.tpgi.com.au) |
| 09:40.43 | *** join/#neo900 jonsger (~Thunderbi@HSI-KBW-046-005-000-166.hsi8.kabel-badenwuerttemberg.de) |
| 09:50.23 | *** join/#neo900 Xiaoman (~Xiaoman@unaffiliated/xiaoman) |
| 10:22.02 | *** join/#neo900 M-DzzzzzzR (dzzzzzzrma@gateway/shell/matrix.org/x-hehhfcwwjqsqqtxe) |
| 11:17.21 | *** join/#neo900 SylvieLorxu (~TheLastPr@541B7AAC.cm-5-4b.dynamic.ziggo.nl) |
| 12:19.30 | *** join/#neo900 paulk-collins (~paulk@gagarine.paulk.fr) |
| 13:17.47 | *** join/#neo900 qwazix (~qwazix@Maemo/community/ex-council/qwazix) |
| 13:35.03 | *** join/#neo900 radekp (~radek@82.113.39.213) |
| 14:02.00 | *** join/#neo900 arossdotme (~zxy@host-78-151-242-106.as13285.net) |
| 14:18.20 | *** join/#neo900 Zero_Chaos (~zerochaos@gentoo/developer/pentoo/zerochaos) |
| 14:34.50 | *** join/#neo900 jonsger (~Thunderbi@HSI-KBW-046-005-000-005.hsi8.kabel-badenwuerttemberg.de) |
| 15:01.04 | *** join/#neo900 SylvieLorxu (~TheLastPr@145.102.91.159) |
| 15:36.46 | *** join/#neo900 SylvieLorxu (~TheLastPr@541B7AAC.cm-5-4b.dynamic.ziggo.nl) |
| 15:44.24 | *** join/#neo900 jonsger (~Thunderbi@HSI-KBW-046-005-000-005.hsi8.kabel-badenwuerttemberg.de) |
| 15:50.23 | *** join/#neo900 louisdk (~louisdk@static-5-103-130-65.seas-nve.net) |
| 15:55.24 | *** join/#neo900 jonsger (~Thunderbi@HSI-KBW-046-005-000-005.hsi8.kabel-badenwuerttemberg.de) |
| 16:32.53 | *** join/#neo900 Pali (~pali@Maemo/community/contributor/Pali) |
| 16:33.46 | *** join/#neo900 jonsger (~Thunderbi@HSI-KBW-046-005-000-005.hsi8.kabel-badenwuerttemberg.de) |
| 17:25.11 | *** join/#neo900 chomwitt (~chomwitt@ppp-94-67-221-230.home.otenet.gr) |
| 17:35.40 | *** join/#neo900 xes_ (~xes@unaffiliated/xes) |
| 17:46.14 | *** join/#neo900 _whitelogger (~whitelogg@uruz.whitequark.org) |
| 17:46.14 | *** mode/#neo900 [+v _whitelogger] by ChanServ |
| 19:18.30 | *** join/#neo900 chomwitt (~chomwitt@ppp-94-67-221-230.home.otenet.gr) |
| 20:20.43 | *** join/#neo900 xman (~xman@user-0cdft6f.cable.mindspring.com) |
| 20:25.47 | *** join/#neo900 drrz (~drrrrz@104.200.151.76) |
| 20:50.15 | *** join/#neo900 neo900 (~office@neo900/coreteam/joerg) |
| 20:50.15 | *** mode/#neo900 [+v neo900] by ChanServ |
| 21:05.26 | *** join/#neo900 qws-user-1228 (~quassel@cpe-76-181-123-141.columbus.res.rr.com) |
| 22:29.22 | *** join/#neo900 Oksana (~chatzilla@Maemo/community/ex-council/Wikiwide) |
| 23:21.00 | *** join/#neo900 galiven_ (~Andrew@50-205-116-131-static.hfc.comcastbusiness.net) |
| 23:34.01 | *** join/#neo900 goiken_ (~goiken@dslb-094-223-165-012.094.223.pools.vodafone-ip.de) |
| 23:55.24 | *** join/#neo900 jonwil (~jonwil@27-33-80-219.tpgi.com.au) |