00:04.09 | *** join/#htc-linux TheXev (~user@169.sub-174-252-197.myvzw.com) |
00:20.25 | *** join/#htc-linux swc|666 (~gecko@unaffiliated/swc666/x-4934821) |
00:22.11 | *** join/#htc-linux TheXev (~user@169.sub-174-252-197.myvzw.com) |
00:43.27 | *** join/#htc-linux drewis (~quassel@207-119-13-155.dyn.centurytel.net) |
00:55.37 | *** join/#htc-linux hemcha (7c29f17b@gateway/web/freenode/ip.124.41.241.123) |
01:24.27 | *** join/#htc-linux TheXev (~user@169.sub-174-252-197.myvzw.com) |
01:40.27 | *** join/#htc-linux arif-ali (~arif-ali@94-192-24-56.zone6.bethere.co.uk) |
02:02.14 | *** join/#htc-linux dr1337 (78931828@gateway/web/freenode/ip.120.147.24.40) |
02:50.56 | *** join/#htc-linux rob_w_ (~bob@ppp-188-174-65-73.dynamic.mnet-online.de) |
03:02.39 | *** join/#htc-linux jonpry (~jon@unaffiliated/jonpry) |
03:03.22 | jonpry | arrrghhh, any luck? |
03:07.23 | WisTilt2 | jonpry: i have oc'd memory working now. still testing for maximum before it becomes unstable but with ebi1/n oc'd it is noticeably snappier. haven't tried smi yet |
03:09.15 | rpierce99 | jonpry: i think his idea of brb is different than mine |
03:11.23 | arrrghhh | WisTilt2, nice |
03:11.33 | arrrghhh | jonpry, sorry went a little insane. still copying stuff back. |
03:11.57 | WisTilt2 | hi arrrghhh |
03:12.18 | arrrghhh | howdy WisTilt2. how goes it |
03:12.23 | WisTilt2 | rpierce99 you want to give this kernel a try and see if it is faster on cdma? |
03:12.47 | rpierce99 | sure, i still have that last tester installed with no 3d and camera :) |
03:13.05 | WisTilt2 | arrrghhh, just been going over some clock stuff comparing to how we have our device setup |
03:13.21 | arrrghhh | very cool |
03:13.22 | WisTilt2 | rpierce99 this one will get you all that working again and around 190mb memory |
03:14.51 | rpierce99 | i assume its already been upped? |
03:15.26 | WisTilt2 | not yet, changing the pmem stuff back for cdma |
03:15.38 | rpierce99 | k |
03:15.44 | *** join/#htc-linux TheXev (~user@13.sub-174-252-217.myvzw.com) |
03:15.46 | WisTilt2 | i havent finished the variant stuff yet so have to do it manually |
03:20.25 | WisTilt2 | doing a full rebuild rpierce99 from a clean tree so it will be a few mins. this kernel also has some other clock changes and should work fine on all rhods. curious if you notice it waking up from sleep quicker, panel wise |
03:21.28 | rpierce99 | k, i'll run it through its paces for a while then, probably won't be able to tell right away things like overall speed and panel wake |
03:22.05 | WisTilt2 | i think we also should be drawing less in sleep - the way the clocks are at least in .39, the dsp chip is still up when in sleep but its off now in this one. |
03:23.06 | arrrghhh | very cool1 |
03:23.07 | arrrghhh | damnit |
03:23.08 | WisTilt2 | yeah probably need to play with it awhile to see anything diff. you should have 192mb free with this also |
03:23.12 | arrrghhh | i meant to say nice find |
03:23.17 | arrrghhh | stupid up key |
03:23.38 | WisTilt2 | arrrghhh you will have to try this also since you have a nack for finding hidden issues:) |
03:23.48 | arrrghhh | jonpry, booting now :P |
03:24.03 | WisTilt2 | you guys still working on ics? |
03:24.06 | arrrghhh | hah |
03:24.16 | arrrghhh | jonpry threw me some libraries he built |
03:24.40 | arrrghhh | that i'm going to try on GB |
03:24.50 | WisTilt2 | that guy's a madman |
03:25.01 | arrrghhh | tell me about it |
03:25.34 | arrrghhh | i mentioned his work on ics on the tp2, and someone commented "isn't he working on the touchpad?" i responded "you don't know the animal that is jonpry" lol |
03:26.00 | rpierce99 | haha yeah, you know just doing some minor work rewriting gralloc, libgles, surfaceflinger, and kernel, no biggie, lol |
03:26.11 | *** join/#htc-linux dr1337 (3a606170@gateway/web/freenode/ip.58.96.97.112) |
03:26.17 | WisTilt2 | yeah no kidding |
03:26.40 | dr1337 | man hunting down textures in surfaceflinger is such a pain in the ass |
03:26.45 | arrrghhh | writing his own compositor. no biggie. |
03:26.50 | arrrghhh | lol |
03:27.37 | WisTilt2 | ok pierce99 kernel is up and ready |
03:28.43 | arrrghhh | heh i see permission denied in the console when i plug in adb |
03:28.51 | arrrghhh | i should probably run that unsecure rootfs... |
03:29.06 | arrrghhh | jonpry, i don't think it's booting tho. if hw3d isn't working, would i still see a bootani? |
03:30.23 | WisTilt2 | rpierce99 you should notice apps open fast right off the bat hopefully. memory is oc'd 20% in this build, im still looking for the upper limit. we run our device's at 20% on this chip and no problems reported by any of our users. |
03:30.39 | jonpry | oh no. i forgot. you have to delete /system/bin/bootanimation |
03:30.46 | arrrghhh | jonpry, check |
03:30.47 | jonpry | cause all 3d apps are broken |
03:30.51 | arrrghhh | lol |
03:31.06 | arrrghhh | i'll run that unsecure setting too so i can adb w/o gui |
03:36.15 | dr1337 | jonpry, do you know where surfaceflinger is calling a surfacetexture object? |
03:36.49 | dr1337 | trying to see if instead of replacing the enum for texture_external with texture_2d to try and modify it at the root |
03:37.15 | jonpry | i did that |
03:37.17 | dr1337 | this would have been so much easier if they released the source for the opengles drivers |
03:37.53 | jonpry | but it doesn't really get you anywhere with the need for the extension |
03:40.35 | dr1337 | gah |
03:40.44 | dr1337 | what do you reckon is the best way around that? |
03:42.06 | *** join/#htc-linux TheXev (~user@13.sub-174-252-217.myvzw.com) |
03:42.30 | jonpry | you haz kgsl? |
03:42.43 | dr1337 | or i suppose we can hope and wait for qcom to release proprietary drivers with the extensions |
03:42.50 | dr1337 | nah |
03:42.55 | dr1337 | i could build a kernel with it |
03:43.05 | jonpry | but your drivers don't use it? |
03:43.27 | dr1337 | oh actually the drivers do use kgsl |
03:43.32 | dr1337 | kgsl and mmu |
03:43.35 | dr1337 | so no need for pmem |
03:43.52 | dr1337 | i thought you were referring to the newer gnu drivers |
03:44.01 | rpierce99 | WisTilt2: moving around in launcher, launching and switching between apps does feel smoother, i also ran a 3d game and it seemed smoother, wish neocore textures were working so i could gauge approximate performance |
03:44.25 | arrrghhh | rpierce99, back to froyo with you |
03:44.33 | arrrghhh | i wonder why those textures never worked in gb |
03:44.35 | jonpry | your implementation will have to be different than mine because of the mmu. but basically i think you can catch a texture allocating in kgsl |
03:44.39 | arrrghhh | and why it only seemed to effect neocore... |
03:44.52 | rpierce99 | they worked for WisTilt2 in some configurations |
03:45.01 | WisTilt2 | rpierce99 nice. does it wake quicker? |
03:45.10 | arrrghhh | ok booting again |
03:45.35 | jonpry | dr1337, my plan is to create a texture. upload random data into it, find it in the pmem, and then create an android surface on it |
03:45.38 | dr1337 | jonpry so this is not going to happen in userspace? |
03:45.39 | WisTilt2 | arrrghhh for whatever reason i can only get textures when i move gpu1 into smi so might be gb issue who knows. minor problem in my book at this point |
03:45.50 | arrrghhh | hrm |
03:45.53 | arrrghhh | it is minor |
03:46.06 | arrrghhh | from what i've gleaned, games and other apps don't have missing textures |
03:46.22 | arrrghhh | jonpry, i haz lolcats |
03:46.32 | jonpry | dr1337, well mine is in userspace. yours is complicated from the mmu |
03:46.48 | jonpry | arrrghhh, but no lockscreen? |
03:46.59 | arrrghhh | well so far it's still booting |
03:47.12 | arrrghhh | this is the first attempt w/o /system/bin/bootanimation |
03:47.19 | arrrghhh | Oo |
03:47.23 | arrrghhh | some surfaceflinger errors |
03:47.24 | arrrghhh | juicy |
03:47.37 | rpierce99 | with my lights set to notification i can't tell when its fully sleeping, but at points when I *think* it is fully sleeping, it seems lightning fast, activity on the phone and weird timing (trying to wake it while its still trying to go to sleep) do still seem to cause slow or failed wakes |
03:47.45 | jonpry | that may be my coding style. LOGE ftw |
03:48.49 | dr1337 | jonpry yeah, did you also get the black screen problems like the n1? |
03:49.07 | jonpry | it just crashed the blob |
03:49.37 | arrrghhh | jonpry, http://pastebin.com/nXVQZHF8 |
03:50.09 | arrrghhh | seems to have stopped booting there. do you want a dmesg too? |
03:50.20 | jonpry | no |
03:50.27 | WisTilt2 | rpierce99 that wake while in process of sleep happens in the other kernels too, never looked into it but it does seem to get out of wack waking up again unless you wait 10secs or more. |
03:50.28 | jonpry | seems like the blob will not work |
03:50.47 | arrrghhh | aw |
03:50.48 | arrrghhh | ok |
03:50.50 | dr1337 | i wonder if it will be easier to try and revert back to how gingerbread dealt with textures |
03:50.57 | jonpry | can you build a new surfaceflinger? |
03:51.00 | arrrghhh | WisTilt2, usual link for that new .39 hotness? |
03:51.02 | rpierce99 | WisTilt2: i know we've tested wake performance a lot, and the phone will get into "good" cycles and "bad" cycles but this feel really really fast |
03:51.13 | dr1337 | afaik, gl_texture_2d and texture_external are interchangeable |
03:51.26 | dr1337 | except texture_external is more memory efficient |
03:51.38 | WisTilt2 | arrrghhh yep, but ill send it again, you might not have the latest one from past couple weeks |
03:51.55 | arrrghhh | k |
03:52.08 | jonpry | texture_external is much faster because it doesn't have to be converted and uploaded to gpu mem |
03:52.56 | jonpry | this is especially important when the actual screen update might only be part of the texture |
03:53.21 | dr1337 | it seems though that very few current vendor opengles drivers support that |
03:53.37 | dr1337 | i think only tegra currently supports it |
03:53.40 | jonpry | pmem hack will do it |
03:53.44 | dr1337 | not sure about any of the adrenos |
03:54.11 | WisTilt2 | rpierce99 the memory oc'd the way it is now is speeding things up but i also changed a couple other clocks that didn't match how we have on our device. im going to build another with another 5% and see if its stable. |
03:54.30 | dr1337 | how is pmem hack going to fix it? |
03:54.57 | jonpry | it will allow the android process to draw directly on the gpu texture |
03:55.10 | jonpry | so it can update as few bytes as it wants. and ditch like 3 copies |
03:55.21 | dr1337 | also, it seems like the msm7k in ICS is a little outdated compared to codeaurora |
03:55.35 | jonpry | or even GB |
03:55.51 | dr1337 | yeah |
03:56.38 | dr1337 | the new platform/hardware/qcom/display.git even supports ION |
03:57.08 | jonpry | what is that? |
03:57.14 | rpierce99 | it might just be the way i'm holding it or using it so much, but it feels warm |
03:57.23 | dr1337 | it's google's new memory management system |
03:57.56 | jonpry | hmm |
03:58.05 | WisTilt2 | rpierce99: probably using it too much:) depends on what you're doing. might let it sleep awhile and it should be nice and cold. |
03:58.30 | dr1337 | https://www.codeaurora.org/gitweb/quic/la/?p=kernel/msm.git;a=tree;f=drivers/gpu/ion;h=a4697619ba641f0c2a40e923f67bdbfa118ddfe4;hb=msm-3.0 |
03:58.34 | jonpry | arrrghhh, do you have a build setup for the xdandroid GB? |
03:58.41 | arrrghhh | jonpry, i did |
03:58.52 | arrrghhh | it's horridly outdated |
03:58.58 | arrrghhh | i haven't built anything from thar in a while |
03:59.03 | jonpry | repo sync? |
03:59.37 | arrrghhh | repo init -u git://gitorious.org/xdandroid/manifest.git -b gingerbread |
03:59.55 | jonpry | don't need to run that again |
03:59.59 | jonpry | just repo sync |
04:00.02 | jonpry | for new hotness |
04:00.09 | rpierce99 | ooh ooh are we playing the name a random repo command game, that's my fav |
04:00.17 | arrrghhh | lol |
04:00.22 | arrrghhh | git stash |
04:00.24 | arrrghhh | <3 |
04:00.42 | WisTilt2 | jonpry are you still remote building on arrrghhh's machine? i could certainly set you up with an ssh shell on one of our i7's if you're interested. |
04:01.03 | arrrghhh | heh. mine's only got 4 physical cores... |
04:01.05 | jonpry | i haven't built ics in a bit. trying to get this sorted out on GB first |
04:01.20 | arrrghhh | my laptop has an i7. how sad. |
04:01.22 | dr1337 | jonpry check this new git too |
04:01.23 | dr1337 | https://www.codeaurora.org/gitweb/quic/la/?p=platform/hardware/qcom/display.git;a=tree;f=libgralloc;h=2af4d8f92e8fadf78f17747011fec45fbab01caf;hb=refs/heads/gingerbread |
04:02.28 | jonpry | this ion is scary. not clear what it will do on msm7k, need userland changes. we have like 5 production kernel trees atm, and some number of userspaces as well |
04:04.19 | dr1337 | yeah |
04:04.29 | dr1337 | it's currently being used on the GN |
04:05.16 | jonpry | tegra? |
04:05.24 | rpierce99 | this thing is waking like winmo for me |
04:06.00 | dr1337 | and tegra too from that kernel |
04:06.03 | jonpry | WisTilt2, i could go for new machine. |
04:06.19 | dr1337 | but i think it's only the tegra tablets |
04:06.21 | arrrghhh | jonpry, his 'net is WAY faster too :P |
04:06.23 | jonpry | moving back to the states next week. probably get a new computer that doesn't overheat w/ ics build |
04:06.32 | rpierce99 | jonpry: i could get you ssh into a pogo plug :) |
04:06.39 | jonpry | lol |
04:07.00 | jonpry | i built linux kernel on the touchpad. its actually not even slow. just a few minutes |
04:07.46 | rpierce99 | i hear ICS is a whole different animal |
04:08.09 | jonpry | its just llvm afaict |
04:08.27 | rpierce99 | by that i mean slow as all hell |
04:08.30 | jonpry | takes at least half of the build time |
04:08.59 | jonpry | as usual the real problem is the android build system |
04:09.38 | jonpry | like one cannot choose to not build llvm. so unless you have changes arch's. why does it need to be rebuilt every time? |
04:12.31 | jonpry | i think adwlauncher is just kicking my flingers arse |
04:14.00 | jonpry | maybe not. maybe it just blows |
04:14.24 | *** join/#htc-linux mitsutaka (~mitsutaka@g1-27-253-203-144.bmobile.ne.jp) |
04:15.36 | rpierce99 | WisTilt2: is there any benefit to equally OCing the proc to what we do with mem on this bad boy, I generally don't OC on this thing, but I know on my pc they kind of go together |
04:16.28 | WisTilt2 | you can oc'd the cpu but tbo i dont see much increase. i run mine at 614400 |
04:16.55 | jonpry | its pretty well starved for memory bandwidth with the lack of processor cache |
04:21.53 | WisTilt2 | rpierce99 are you still showing 192mb free? |
04:22.28 | rpierce99 | 192568 total via free |
04:22.59 | WisTilt2 | after you kill whatever you can in systempanel how much does it show available? |
04:23.23 | WisTilt2 | should be up around 90mb on that device. im at 104mb with that kernel |
04:25.52 | rpierce99 | with stuff restarting itself all the time the highest i got it was 81 |
04:25.57 | rpierce99 | 82 |
04:26.12 | dr1337 | i think anything < adreno 200 is going to be really difficult to port |
04:26.18 | rpierce99 | but 17.2 of that is SPL |
04:26.29 | rpierce99 | so 82 + 17 = 99 |
04:33.13 | arrrghhh | wow |
04:33.21 | arrrghhh | downloading from the market over 3g t'was fast |
04:33.30 | arrrghhh | never seen 3mb go that fast before... |
04:33.38 | rpierce99 | my 3g has been slow lately :( |
04:33.46 | arrrghhh | usually it struggles on 300kb |
04:33.57 | arrrghhh | the wakes have been pretty slow |
04:34.00 | arrrghhh | but it's doing 15,000 things |
04:34.04 | arrrghhh | had to make a new data.img |
04:34.32 | arrrghhh | does feel quite a bit snappier so far |
04:34.40 | WisTilt2 | let it do its thing and set all settled down then it should be flying |
04:34.44 | arrrghhh | and this is on a first boot... that's saying something. |
04:35.01 | arrrghhh | yea i just setup sync and downloaded an app from the market while it was syncing |
04:35.08 | arrrghhh | and it was fast... i'm impressed. |
04:35.14 | WisTilt2 | ok i booted ok at 25% memory oc so we'll hammer it and see if it blows up |
04:35.21 | arrrghhh | lol |
04:35.48 | rpierce99 | is the intention at some point to make a startup.txt option for mem oc? never know when someone gets a lame chip |
04:36.15 | jonpry | +1 |
04:36.27 | WisTilt2 | yep for sure |
04:36.44 | arrrghhh | that would be cool |
04:36.48 | WisTilt2 | no telling with differences between hardware as we've seen so far |
04:36.51 | arrrghhh | different levels even |
04:36.54 | arrrghhh | just in case |
04:39.58 | *** join/#htc-linux detule (~oliver@pool-96-234-141-27.bltmmd.east.verizon.net) |
04:40.55 | arrrghhh | WisTilt2, Mem: 192568 |
04:40.56 | arrrghhh | nice |
04:42.14 | WisTilt2 | every little bit helps |
04:43.05 | arrrghhh | yup |
04:52.13 | *** join/#htc-linux detule (~oliver@pool-96-234-141-27.bltmmd.east.verizon.net) |
04:55.57 | *** join/#htc-linux Rob2222 (~Miranda@p4FFF2FB1.dip.t-dialin.net) |
04:57.29 | WisTilt2 | we're using PMDH not EMDH i hope |
05:02.14 | WisTilt2 | arrrghhh you haven't crashed it yet? |
05:02.37 | arrrghhh | not yet. restoring on tb... |
05:02.49 | arrrghhh | i guess i should've tested some stuff but i didn't feel like downloading a ton of apps :P |
05:02.58 | jonpry | probably time for some benches |
05:03.07 | rpierce99 | 25% still going well? |
05:03.14 | arrrghhh | gimmie 90 |
05:03.17 | arrrghhh | :D |
05:03.42 | WisTilt2 | 25% no probs so im trying 30% right now |
05:03.46 | arrrghhh | what about the GPU? |
05:04.00 | arrrghhh | i know some other devices got that and had a pretty good boost |
05:04.14 | arrrghhh | i think it works on kovs/blac/raph/diam |
05:04.42 | WisTilt2 | i need to do smi memory also but gpu's we'll see also |
05:05.51 | arrrghhh | cool |
05:05.53 | arrrghhh | oc all of it |
05:06.36 | arrrghhh | 1.21 gigawatts |
05:06.45 | WisTilt2 | jonpry whats a good app or method to benchmark this? if we are locked at 30fps from vsync i guess neocore wont tell us anything if we oc the gpus |
05:06.52 | WisTilt2 | lol doc |
05:07.00 | WisTilt2 | jigawatts |
05:07.11 | arrrghhh | ah, right. |
05:07.12 | arrrghhh | :P |
05:07.26 | arrrghhh | well there's a bunch of benchmark apps |
05:07.27 | jonpry | not unless neocore reports less that ~29 |
05:07.32 | arrrghhh | i'm not sure which one is "good" or not. |
05:07.45 | arrrghhh | fps2d is the one that seemed like we could get higher than 30fsp |
05:07.47 | arrrghhh | fps* |
05:07.48 | WisTilt2 | im getting 29.7 now on neocore |
05:07.52 | arrrghhh | but it never averaged more than that |
05:07.57 | arrrghhh | antutu? never used that one |
05:08.03 | arrrghhh | there's others that seem dubious |
05:08.05 | arrrghhh | linpack |
05:08.13 | jonpry | linpack is not 3d |
05:08.13 | arrrghhh | uhhh |
05:08.21 | arrrghhh | oh i thought we were testing overall perf |
05:08.25 | arrrghhh | not just 3d |
05:08.28 | jonpry | pfff |
05:08.32 | arrrghhh | lol |
05:08.44 | jonpry | w/ compositor coming. there is only one metric |
05:08.51 | arrrghhh | i see |
05:09.25 | jonpry | dur. i try to tell cm7 not to go to sleep when usb is plugged in. just does it anyways |
05:13.32 | arrrghhh | quadrant |
05:13.35 | arrrghhh | that tests 3d... |
05:13.44 | arrrghhh | that's another one that seems to vary drastically tho |
05:13.59 | arrrghhh | NenaMark |
05:14.05 | arrrghhh | you might like this one sir jonpry |
05:14.13 | arrrghhh | Although Quadrant does perform a number of different tests and compiles them into a file score, in my experience its a bit tilted towards CPU performance. If youre interested in isolating the GPU, NenaMark is a great choice. |
05:15.07 | jonpry | nenamark is es 2.0 |
05:15.17 | arrrghhh | even the original? |
05:15.20 | arrrghhh | nenamark 1 |
05:15.58 | jonpry | dunno |
05:16.06 | jonpry | nenamark it is |
05:16.18 | arrrghhh | i'm trying to find out details about it... |
05:16.44 | arrrghhh | the v1 page still says it's 2.0 |
05:16.49 | arrrghhh | hrm |
05:17.17 | jonpry | doesn't quadrant give like individual scores? |
05:17.24 | arrrghhh | i believe so yes |
05:17.32 | jonpry | sounds perfect |
05:18.07 | arrrghhh | hrm |
05:18.17 | arrrghhh | sounds like you have to get the "advanced" version to have sub scores tho. |
05:21.43 | arrrghhh | jonpry, incoming pm |
05:23.09 | jonpry | atm all i have is broken cm7 |
05:24.19 | arrrghhh | you want a gb image? |
05:25.11 | jonpry | i have it. just only one sd card |
05:26.35 | jonpry | i think half the problem is that all of the layers are in 32bit argb |
05:27.04 | jonpry | sucking up oodles of ram and slowing the texture copies to shit |
05:27.24 | arrrghhh | you can put as many builds as your card can hold |
05:27.35 | jonpry | arrrghhh, its your kind of pixel format |
05:28.05 | WisTilt2 | rpierce99 give this one a try. this is crazy fast now and stable so far here. |
05:28.17 | jonpry | won't this partitioned thing not work? |
05:28.28 | arrrghhh | put no_partitions on your cmdline |
05:28.35 | arrrghhh | rootfs will ignore the partitions |
05:28.42 | jonpry | hmm |
05:32.37 | arrrghhh | Oo |
05:32.47 | arrrghhh | i see kexec in the menu - WisTilt2 is the .39 kernel ready to use that? |
05:32.54 | WisTilt2 | yep |
05:32.58 | arrrghhh | hells yea |
05:33.07 | arrrghhh | this is definitely fast. |
05:33.09 | WisTilt2 | guys this is like a different phone now. |
05:33.15 | arrrghhh | i just played a little words with friends |
05:33.17 | arrrghhh | zomg |
05:33.46 | WisTilt2 | arrrghhh get this next kernel, its much faster than the last one |
05:34.01 | WisTilt2 | im trying to find a decent benchmarking app |
05:34.03 | arrrghhh | got it |
05:34.07 | arrrghhh | kexec'ing :D |
05:35.01 | ellisway | what devices you guys playing with |
05:35.12 | arrrghhh | tp2 |
05:35.12 | rpierce99 | wish i could, but someone hasn't fixed adb on the mac :P |
05:35.13 | arrrghhh | rhodium |
05:35.33 | rpierce99 | cue the mac comment from arrrghhh |
05:35.49 | arrrghhh | use a real computer or something |
05:35.52 | arrrghhh | happy? |
05:36.31 | ellisway | just upgraded to a new handset but have many htc devices |
05:36.36 | rpierce99 | i just managed to remove the app drawer icon from adw |
05:36.41 | rpierce99 | no idea how to get it back now |
05:36.46 | arrrghhh | rpierce99, it's annoying |
05:36.56 | arrrghhh | i usually lock the desktop for that reason |
05:37.10 | arrrghhh | i don't know if there's an easier way, but i just long press on the desktop and create an icon for it |
05:37.14 | WisTilt2 | rpierce99 i still havent tried it on the mac at the office but can you look at dmesg in the terminal and see what errors out when you establish the connection and then try to use it |
05:37.18 | arrrghhh | then drag that icon onto the empty slot |
05:37.22 | arrrghhh | then lock the desktop :D |
05:37.43 | arrrghhh | kexec is p i m p |
05:37.49 | jonpry | you guys are using adw? |
05:37.50 | WisTilt2 | lol |
05:38.02 | arrrghhh | jonpry, i bounce back between adw and golauncher |
05:38.11 | arrrghhh | i kinda like golauncher's widgets but it feels slower. |
05:38.24 | jonpry | what kind of framerate you get on adw? |
05:38.30 | arrrghhh | zeam i've heard if fast, i haven't really used it much tbh |
05:38.44 | arrrghhh | i wouldn't know framerate |
05:38.50 | arrrghhh | 42 |
05:38.53 | jonpry | is it smooth? |
05:38.59 | arrrghhh | yea, it's much better than launcher2 |
05:39.07 | arrrghhh | but... pretty much anything is. |
05:39.19 | jonpry | i get like 5fps |
05:39.23 | arrrghhh | lol |
05:39.26 | arrrghhh | l2 sux |
05:39.33 | arrrghhh | replace it with _anything_ |
05:39.33 | jonpry | thats on adw |
05:39.36 | arrrghhh | oh |
05:39.43 | arrrghhh | you're on cm7 tho right? |
05:39.47 | jonpry | yeah |
05:40.12 | arrrghhh | the adw cm7 bundles seems... shitacular |
05:40.23 | arrrghhh | i don't know why |
05:40.26 | arrrghhh | it's still adw. |
05:40.45 | jonpry | btw i can manage to get some texture corruption |
05:41.03 | jonpry | so something may not be good with gpu0 |
05:44.54 | arrrghhh | my phone definitely seems warmer than usual, lol |
05:45.08 | rpierce99 | yeah i thought so too |
05:46.03 | jonpry | WisTilt2, have you hit 50% yet? |
05:46.11 | WisTilt2 | its on steroids so should be lol. it should be normal temp if you're not working it real hard though. that quadrant benchmark is nice. scored right below the nexus 1 |
05:46.55 | WisTilt2 | im at 35% right now and this sucker is flying crazy fast. dont know if pushing it higher will be noticeable but im going to try 40 |
05:47.21 | WisTilt2 | you guys run this new one yet? |
05:47.40 | rpierce99 | running it now |
05:47.49 | arrrghhh | ditto, have been for a little while |
05:48.02 | arrrghhh | when i was talkin 'bout kexec is when i started running it |
05:48.06 | WisTilt2 | after this 40% try im going to oc smi at the same 35% that i know is stable |
05:48.10 | arrrghhh | turnaround is soooo much faster when you skip winmo :D |
05:48.16 | rpierce99 | seems good, hoped angry birds rio was going to be like omg fast so we could impress all the noobs, but seems we are gpu limited on that one |
05:48.20 | WisTilt2 | then we move on to gpus |
05:48.27 | arrrghhh | rpierce99, lol |
05:48.46 | arrrghhh | ZOMG ANGRY BIRDZZZZ IS SOOOOOOOOOOOOOFASTOMGFAST!!11!! |
05:48.56 | rpierce99 | but with the free ram and OC launcher is like lightning |
05:52.02 | jonpry | i got it up to like 8 fps |
05:52.23 | WisTilt2 | well it runs at 40% and it might be a little faster but not such a big just as when i went from 20-35. 35% is amazingly fast |
05:59.43 | arrrghhh | having the RAM oc similar to the proc oc would be nice |
05:59.45 | jonpry | WisTilt2 i have some questions about msm_fb if you have time |
05:59.57 | arrrghhh | so then people could just push it as crazyfar or as low as they want. |
06:00.09 | WisTilt2 | sure but give me 1 |
06:00.34 | arrrghhh | np |
06:00.41 | arrrghhh | i'm just thinking when it's "finalized" |
06:00.48 | arrrghhh | i dun need it now :P |
06:00.50 | rpierce99 | he was talking to jonpry |
06:00.55 | arrrghhh | that too. |
06:00.57 | WisTilt2 | rpierce99 did you plug usb in about 30s after booting? |
06:01.10 | rpierce99 | hm, probably less than that, my batt is low |
06:01.18 | rpierce99 | it was still booting when i plugged it in |
06:01.25 | rpierce99 | IIRC |
06:01.32 | rpierce99 | it very well could have been 30s |
06:02.21 | WisTilt2 | ok ill have to try this on our mac at the office to figure it out. looks like the data channel never gets setup by the mac so kernel just waits |
06:02.34 | jonpry | oo, maybe 12 fps |
06:03.45 | WisTilt2 | arrrghhh that last kernel even better? |
06:04.53 | arrrghhh | yea, seemed to be |
06:05.01 | arrrghhh | i got stuck with something calling GPS on |
06:05.06 | arrrghhh | and it wouldn't shut down |
06:05.14 | arrrghhh | so i disabled it and rebooted. getting back now. |
06:05.14 | WisTilt2 | lol we need to fix or remove that from .39 |
06:05.29 | arrrghhh | heh |
06:05.38 | WisTilt2 | jonpry let me grab another beer and we'll talk fb:) |
06:05.46 | jonpry | k |
06:06.53 | jonpry | 12fps is pretty nice on the cm7. if i can double that it will be as good as touchpad |
06:07.21 | rpierce99 | you don't mess around when you set goals, yeah i'd like to double that... |
06:08.09 | jonpry | imho in software that is the only goal worth achieving. in 2 years a new processor will do it anyways |
06:09.30 | WisTilt2 | ok jonpry fire away |
06:09.42 | jonpry | so where exactly is the framebuffer? |
06:10.03 | jonpry | from what i can tell its definitely not in that pmem region |
06:11.01 | jonpry | er. do we have remotefb? |
06:11.49 | WisTilt2 | i thought pmem is where it is, or at least something fb related uses that. stick it in a dead memory area and no workie |
06:12.23 | WisTilt2 | ive never looked at msm_fb deep enough in that aspect to actually see |
06:12.25 | jonpry | i'm not saying that memory isn't used for anything. its just that if i write stuff in there. it does not automagically end up on the screen |
06:12.32 | arrrghhh | wow |
06:12.48 | arrrghhh | apps that never scroll smoothly are now scrolling at a pretty decent speed. |
06:13.00 | WisTilt2 | open faster than hell to dont they? |
06:13.05 | arrrghhh | that too :D |
06:13.09 | WisTilt2 | app menu especially |
06:13.09 | arrrghhh | this is awesome man |
06:13.28 | WisTilt2 | jonpry maybe pmem fb area is a pre buffer? |
06:13.44 | jonpry | remotefb would explain it |
06:14.02 | jonpry | but yes its somekind of not really the framebuffer thing |
06:14.02 | WisTilt2 | our of msm_fb or where would that be setup |
06:14.40 | jonpry | ? |
06:15.51 | arrrghhh | oh well. gonna go get some sleep. thanks for the test kernels WisTilt2, lookin forward to seein what you can do for the GPU ;) |
06:16.18 | WisTilt2 | ok arrrghhh. ill oc smi memory next then onto gpu and see what trouble we can get into |
06:16.40 | rpierce99 | i won't be satisfied until we have a personal heater |
06:16.45 | jonpry | we needz juice |
06:16.51 | WisTilt2 | lol |
06:18.41 | WisTilt2 | jonpry im looking at msm_fb and fb gets allocated there but i dont see where the pmem side comes into play |
06:19.21 | jonpry | it sets somekind of variable called fix |
06:19.38 | jonpry | it uses platform data and not pmem, but its the same region |
06:19.47 | WisTilt2 | ok that makes more sense |
06:20.18 | WisTilt2 | you sure we dont have 2 fb's going on here the way this is now? |
06:20.26 | jonpry | yes |
06:20.33 | WisTilt2 | a ghost in pmem that really never gets used |
06:20.44 | jonpry | i've attempted a blind write to all of it everywhere |
06:20.57 | jonpry | and nothing takes effect unless i make an ioctl |
06:21.12 | WisTilt2 | writing to the 2mb pmem region directly you mean? |
06:21.28 | jonpry | yes |
06:21.52 | jonpry | and the ioctl causes a mdp dma operation to something |
06:23.03 | WisTilt2 | if you write to pmem region and make the ioctl does it then appear in fb? |
06:23.52 | jonpry | yes |
06:29.25 | jonpry | if its using remotefb, then that is fine. just wondering if there was a way to avoid the copy |
06:30.10 | WisTilt2 | jonpry: is line 756 where the fb in pmem is setup? |
06:31.26 | WisTilt2 | looks like it so if the resource isnt there then what happens? |
06:31.26 | jonpry | i have 762 but mine is patched |
06:31.33 | jonpry | static int setup_fbmem(struct msmfb_info *msmfb, struct platform_device *pdev) |
06:31.44 | WisTilt2 | resource = platform_get_resource(pdev, IORESOURCE_MEM, 0); |
06:31.53 | jonpry | yeah thats it |
06:32.20 | WisTilt2 | ok the way that code is there is a test if the resource isnt there to do what? |
06:32.22 | jonpry | it aborts if there is no fb |
06:33.26 | jonpry | maybe all mddi panels are remote fb |
06:33.40 | WisTilt2 | right, so that would be the only fb area right? so the ioctl triggers the copy from that buffer. |
06:34.05 | jonpry | probably |
06:34.33 | WisTilt2 | i know once its passed to the panel it stays on the panel unless fb updates it again |
06:34.59 | jonpry | so the other question i have is about using very strange framebuffer formats |
06:35.34 | jonpry | like 565 rgb, then 16bits of nothing |
06:35.56 | jonpry | is that going to be possible with our old mdp? |
06:37.08 | jonpry | fairly certain i could get it done on mdp4 |
06:37.48 | WisTilt2 | i need to look at this but should be possible. |
06:38.38 | jonpry | this is the price of using the adreno output as the framebuffer |
06:38.56 | jonpry | most stuff will want depth buffer which is interleaved with the rgb |
06:39.36 | WisTilt2 | we already have 565 in mdp |
06:39.46 | WisTilt2 | case 16 |
06:39.49 | jonpry | this is 56516 though |
06:39.56 | WisTilt2 | oh |
06:39.59 | jonpry | 32 bits per pixel |
06:40.04 | WisTilt2 | does our panel support that? |
06:40.31 | jonpry | my understanding is that mdp is an overgrown dma engine |
06:40.44 | jonpry | so you tell it, like take this buffer packed like this, and output it like that |
06:41.07 | jonpry | panel probably only takes one format, but mdp takes all sorts of them. like 8888 |
06:41.37 | WisTilt2 | so mdp does all the reformatting |
06:41.48 | jonpry | yeah |
06:42.09 | jonpry | so can it be programmed to essentially skip every other 16bit pixel |
06:43.03 | WisTilt2 | should be doable. you trying to run 32bpp on 16bpp or what |
06:43.14 | jonpry | depth buffer |
06:44.49 | jonpry | gotta pretend its not there |
06:45.23 | jonpry | currently i have pmem_fb == pmem_gpu1 |
06:48.05 | WisTilt2 | overlapped? so part of gpu1 base is fb then |
06:48.19 | jonpry | yeah the bottom |
06:48.30 | jonpry | causes thats where gpu puts its output buffers |
06:48.43 | jonpry | very convenient when using opengl composition |
06:49.28 | jonpry | i can even sort of get page flipping to work. but there is a problem |
06:49.45 | WisTilt2 | that almost sounds like the ioctl that copies fb copies it to the base 2mb of gpu1 then |
06:50.39 | jonpry | msm_fb has no idea bout gpu1 |
06:52.33 | jonpry | i can't figure out if egl is using the top or bottom buffer at any instant. android must know somehow to do the blit. urg |
06:55.22 | WisTilt2 | rats nest |
06:56.21 | jonpry | i think this compositor is going to bring new life to our phones |
06:57.26 | WisTilt2 | bring it on! if we can get the gpu's oc'd we'll have state of the art new old devices |
06:58.05 | jonpry | for sure :p |
07:00.10 | WisTilt2 | brb |
07:22.28 | *** join/#htc-linux mitsutaka (~mitsutaka@g1-27-253-203-144.bmobile.ne.jp) |
07:30.24 | WisTilt2 | jonpry: panel supports 8,16, and 24 bit depth with rgb 565,666,888 formats |
07:30.56 | WisTilt2 | sounds like you need to write a 56516 converter in mdp:) |
07:32.53 | jonpry | well working on the buffer flipping problem for now |
07:32.57 | WisTilt2 | rpierce99 if you're still awake, was that last kernel you tried the 30% oc one? |
07:33.18 | WisTilt2 | ok. im going to oc smi then have a go at gpu's before i hit the sack |
07:33.36 | rpierce99 | 9:27p your time |
07:33.59 | rpierce99 | my chat log doesn't go back that far |
07:35.33 | WisTilt2 | yeah that was at 30%, 35 works nice so does 40 but i dont see much diff from 35-40 thats worth it. let me oc smi to 35% and ill upload for you to try if you want. |
07:35.46 | jonpry | what about 50? |
07:35.51 | jonpry | or 100? |
07:36.01 | WisTilt2 | lol, dont want to melt my phone |
07:36.15 | WisTilt2 | ill try 50 real quick |
07:36.41 | WisTilt2 | it will die at 100 i can pretty much guarantee |
07:38.37 | *** join/#htc-linux bzo (~chatzilla@c-174-62-79-238.hsd1.ca.comcast.net) |
07:38.50 | rpierce99 | 10" honeycomb tablet for $229, one of you guys should buy one so you can make android awesome for me :) |
07:39.37 | bzo | I got my 10" ICS (to be) tablet for $100 :P |
07:39.46 | rpierce99 | what kind? |
07:39.53 | bzo | touchpad |
07:39.59 | rpierce99 | that's 10? |
07:40.06 | bzo | yeah |
07:40.17 | rpierce99 | is still mad about that |
07:40.18 | bzo | of course that's moot since there are no more |
07:40.30 | rpierce99 | unless you buy an hp laptop |
07:41.36 | bzo | I got really lucky to run across that deal exactly when it was available |
07:42.06 | rpierce99 | yeah i talked to a guy yesterday that was standing in a best buy store when someone came up to the employee he was talking to and was like dude we can sell the tablets now |
07:42.43 | bzo | I ordered mine from the hp site, so I didn't even have to leave the house |
07:43.21 | bzo | that whole touchpad debacle got everyone else offering their tabs at a reasonable price now |
07:43.40 | bzo | it seems like the going rate for a 10" tab is down to like 300, instead of 500-600 |
07:44.05 | bzo | everyone but apple that is |
07:46.34 | jonpry | but none are as good as touchpad |
07:47.03 | bzo | why? aside from it running your kernel :P |
07:47.51 | jonpry | the 8660 is the best processor out there |
07:48.36 | rpierce99 | not a tegra 2 fan? |
07:48.48 | jonpry | tegra can't play 1080p |
07:49.29 | jonpry | and its got no neon. like wtf |
07:49.43 | bzo | is the 8660 an a8 or a9 like generation chip? |
07:49.59 | jonpry | a9 |
07:50.25 | bzo | so it's not just a faster snapdragon then? |
07:50.36 | bzo | I mean higher clocked, that is |
07:51.07 | jonpry | it still is a snapdragon |
07:51.22 | jonpry | more or less same as 8260 without the radio |
07:51.29 | jonpry | much faster than 8x50 |
07:51.44 | bzo | I thought the snapdragon was like a slightly better a8? |
07:52.33 | jonpry | afaict a9 is just dual core a8 |
07:54.31 | bzo | I think the a9 has much better ipc, shorter pipeline and OOE |
08:03.29 | WisTilt2 | rpierce99: 35% kernel is up |
08:03.53 | bzo | WisTilt2: what is the default speed of memory? |
08:04.01 | WisTilt2 | jonpry above 50% is no good and 50 is not stable |
08:04.40 | WisTilt2 | as our kernels have it you mean? |
08:04.52 | bzo | right, out of the box |
08:05.49 | rpierce99 | so this is the first thing you OCed, and then smi to 35? |
08:06.31 | WisTilt2 | dont know for sure, im using our stuff from our device right now. i think .39 is 132mhz |
08:06.43 | WisTilt2 | rpierce99 yes both are at 35 now |
08:07.22 | rpierce99 | why on earth am i watching the crocodile hunter |
08:07.40 | rpierce99 | i stayed up to catch this sale |
08:07.55 | rpierce99 | and apparently ignored the tv |
08:07.55 | bzo | for the tablet? |
08:07.59 | rpierce99 | yeah |
08:08.05 | bzo | which one is it? |
08:08.12 | rpierce99 | iconia |
08:08.31 | bzo | did that come out at $500? |
08:08.40 | rpierce99 | the 16gb was up there |
08:08.46 | rpierce99 | this 8gb is a new sku |
08:08.50 | rpierce99 | just released today |
08:08.57 | rpierce99 | er, like 20 minutes ago |
08:09.20 | bzo | ah, ic. Though the difference in BOM for 8gb must be like $4 |
08:09.41 | rpierce99 | yeah probably, sd card slot anyways, i don't care either way |
08:10.09 | bzo | yeah, that's the only downside of the touchpad, no sd |
08:10.43 | bzo | that and the somewhat lo-res xga display |
08:10.57 | bzo | hp really knocked themselves out copying the ipad |
08:11.51 | bzo | WisTilt2: I would think that at the time the tp2 was manufactured, the min speed of ram would be like 166mhz |
08:12.10 | bzo | so, the 25% from 133 to 166 is still probably in spec |
08:13.46 | WisTilt2 | you'd think so, ddr ram from a few years ago was good up to around 200 i think |
08:14.25 | *** join/#htc-linux toer (~tore@179.81-166-86.customer.lyse.net) |
08:14.59 | WisTilt2 | at 35% oc this puppy is really flying and i dont see enough difference above that to risk it |
08:15.43 | bzo | from my experience, OCing desktop 166 ram to about 180 is pretty safe (i.e. your 35% oc) |
08:15.56 | WisTilt2 | rpierce check and make sure you still have 192mb free with this, i may have dropped that in this build |
08:15.59 | rpierce99 | yeah this is noticeably faster than the last, which is way faster than the normal, this is silky smooth |
08:16.11 | WisTilt2 | sweet isn't it |
08:16.11 | bzo | a lot of stick wouldn't hit 200 reliably though (your 50% oc) |
08:16.26 | rpierce99 | only concern for me is this heat = energy = lower battery life |
08:16.41 | rpierce99 | nope, 192568 |
08:16.46 | WisTilt2 | bzo that sounds right on with what im seeing |
08:17.09 | WisTilt2 | ok good, thats the max memory on cdma i can squeeze out at this time i guess |
08:17.26 | bzo | heh, I guess phh was right - ocing the memory is a big win |
08:17.33 | WisTilt2 | huge |
08:17.43 | WisTilt2 | this is like a totally different device now |
08:19.01 | WisTilt2 | bzo:found another power saver in sleep also. we weren't shutting down the dsp. dont have the fluke and test battery to measure here though but should be less drain |
08:19.14 | bzo | nice |
08:19.35 | WisTilt2 | 18ma is where im at as is with scbs running, thats still pretty nice |
08:20.12 | bzo | I may be able skip a phone upgrade for another year |
08:20.54 | WisTilt2 | rpierce if you cant blow it up i think ill set 35% as the number for now. i need to hit the sack so ill try oc'ing gpus tomorrow |
08:21.24 | WisTilt2 | bzo true, i have 15 rhod300's so i dont need to upgrade all those now either:) |
08:21.52 | bzo | I ought to measure the drain on the 400, must be higher than the 300s |
08:22.14 | bzo | did you hack a battery to add a test point or something? |
08:22.24 | WisTilt2 | would be curious myself. it seems cdma users dont get the same bat life we do on gsm |
08:22.41 | WisTilt2 | yes, spare battery wired to it and have fluke in line |
08:23.32 | rpierce99 | go over 30 fps on neocore :) |
08:23.38 | rpierce99 | with no textures :) |
08:23.46 | WisTilt2 | you got over 30? |
08:23.47 | rpierce99 | s/go/got/ |
08:23.51 | rpierce99 | 30.2 |
08:24.29 | WisTilt2 | its probably higher than that but jonpry said vsync is locked to 30fps so we never see higher. |
08:24.47 | bzo | rpierce99: have you tried fps2d? |
08:24.59 | rpierce99 | not on this kernel no |
08:25.00 | WisTilt2 | i cant get over how fast app menu opens not to mention opening apps now |
08:25.25 | rpierce99 | yeah, it's hard to find apps that don't have some network component to test with though |
08:25.31 | rpierce99 | neocore loads up fast |
08:25.36 | bzo | might not see better fps2d result until the smi is OCed also |
08:25.51 | WisTilt2 | it is in this kernel |
08:26.11 | bzo | ah nice, you got to both of them |
08:26.25 | WisTilt2 | so is gpu0 2d then? |
08:26.44 | bzo | I was thinking about the framebuffer memory |
08:27.00 | WisTilt2 | i dont get why i can set gpu0 down to 1mb and nothing changes with anything. its like that gpu never gets used |
08:27.25 | bzo | winmo has gpu0=2mb and gpu1=13mb |
08:27.50 | bzo | we have our 7mb/8mb config because that's what the qcom lib is hardcoded for |
08:27.51 | WisTilt2 | really... why do we have gpu0-7mb |
08:27.55 | WisTilt2 | oh |
08:28.35 | bzo | back in the day, we tried hacking the lib for th 2+13 config, but it didn't seem to do anything |
08:28.40 | bzo | so we didn't put any more work into it |
08:29.24 | WisTilt2 | any idea how much oc we can do to the gpu's? |
08:29.45 | bzo | no clue, didn't alex play with that a while back? |
08:30.11 | WisTilt2 | dont know but if he did he might know the upper limit and save me some time |
08:30.41 | WisTilt2 | i just went in 5% steps with memory and hammered it for stability |
08:30.54 | bzo | lol, is there anything we use the gpu except for necore? |
08:31.08 | WisTilt2 | camera |
08:31.12 | rpierce99 | avg=30 stdev=6.14 |
08:31.42 | bzo | of course, when jonpry gets the compositor working the gpu will be huge |
08:32.16 | WisTilt2 | yep. tp2's out perform the new stuff would be nice |
08:32.21 | bzo | rpierce99: hmm, not much change then |
08:32.42 | rpierce99 | we're capped at 30 though |
08:33.10 | bzo | tru, though I would hope one would be pegged closer to 30 and have less deviation |
08:33.33 | bzo | just ran on mine and got the same result |
08:35.41 | rpierce99 | some of the crappier 3d games that aren't quite as gpu intensive run much better |
08:35.49 | rpierce99 | paper plane is smooth now |
08:36.25 | bzo | have you tried the all important angry birds? |
08:36.39 | rpierce99 | only rio |
08:36.42 | rpierce99 | i don't have regular |
08:36.44 | rpierce99 | rio is still slow |
08:36.51 | WisTilt2 | i need to hit the sack guys. rpierce99 you might run in sleep mode overnight and see what kind of battery drain you get now. im seeing .8% hr before the dsp fix so hoping for less tonight. |
08:37.07 | rpierce99 | i don't have enough batt to do a drain test |
08:37.26 | rpierce99 | it's been off charger all day |
08:37.36 | rpierce99 | so i'll charge it overnight and then let it drain tomorrow |
08:37.36 | *** join/#htc-linux kiozen (~kiozen@ppp-93-104-76-47.dynamic.mnet-online.de) |
08:37.40 | WisTilt2 | np, when you can. nite guys, gpu oc tomorrow:) |
08:37.44 | rpierce99 | cant wait |
08:38.15 | bzo | I'm out as well, gn |
08:38.22 | rpierce99 | nite |
09:16.12 | *** join/#htc-linux drewis (~quassel@71.20.84.142) |
09:16.53 | *** join/#htc-linux [acl] (~abel@cpe-69-203-141-229.si.res.rr.com) |
09:25.51 | *** join/#htc-linux paulk (~paulk@lib33-1-82-233-88-171.fbx.proxad.net) |
09:43.25 | *** join/#htc-linux helicopter88 (~helicopte@host177-94-dynamic.22-79-r.retail.telecomitalia.it) |
10:16.11 | *** join/#htc-linux dr1337 (~textual@CPE-120-147-24-40.lflg2.win.bigpond.net.au) |
10:30.55 | *** join/#htc-linux drewis_ (~quassel@71.20.84.142) |
11:21.24 | *** join/#htc-linux shaky153 (5eacbfea@gateway/web/freenode/ip.94.172.191.234) |
11:21.49 | *** join/#htc-linux DJ_Jedi (975f0450@gateway/web/freenode/ip.151.95.4.80) |
11:23.45 | shaky153 | really quiet in here... |
11:28.47 | *** join/#htc-linux Sandvold (~Sandvold@207.80-202-111.nextgentel.com) |
11:28.53 | *** part/#htc-linux Sandvold (~Sandvold@207.80-202-111.nextgentel.com) |
11:37.47 | *** join/#htc-linux GNUtoo (~gnutoo@host40-58-dynamic.116-80-r.retail.telecomitalia.it) |
11:40.02 | *** join/#htc-linux mastermerlin (~Adium@p4FEE5C6C.dip.t-dialin.net) |
11:46.57 | *** part/#htc-linux DJ_Jedi (975f0450@gateway/web/freenode/ip.151.95.4.80) |
12:16.02 | *** join/#htc-linux ttols (~ttols@221.147.250.191) |
12:24.02 | *** join/#htc-linux LjL (~ljl@unaffiliated/ljl) |
12:25.12 | LjL | hi. uhm, i was thinking of getting a used HTC Universal, but apparently the wiki in the topic doesn't list it. do you know whether i could run a decent version of Linux on it? |
12:54.32 | *** join/#htc-linux rajkosto (~rajkosto@2001:470:d76b:da7a:e466:66e3:84ff:8b75) |
12:56.54 | *** join/#htc-linux Rob2222 (~Miranda@p4FFF2FB1.dip.t-dialin.net) |
13:34.09 | *** join/#htc-linux ali1234 (~ajbuxton@robotfuzz.co.uk) |
13:37.04 | *** join/#htc-linux helicopter88 (~helicopte@host177-94-dynamic.22-79-r.retail.telecomitalia.it) |
13:54.52 | *** join/#htc-linux rzk (~rzk@95-27-103-72.broadband.corbina.ru) |
14:09.21 | *** join/#htc-linux detule (~oliver@pool-96-234-141-27.bltmmd.east.verizon.net) |
14:20.23 | *** join/#htc-linux LordDeath (~LordDeath@cable-81-173-164-108.netcologne.de) |
14:22.36 | *** join/#htc-linux chaowang (~chaowang@2001:da8:215:8103:21e:ecff:fe66:40d7) |
14:39.59 | *** join/#htc-linux helicopter88 (~helicopte@host177-94-dynamic.22-79-r.retail.telecomitalia.it) |
14:52.51 | *** join/#htc-linux ImCoKeMaN (~imcokeman@pool-96-249-152-166.hrbgpa.fios.verizon.net) |
14:53.50 | *** join/#htc-linux defendthecommons (~defendthe@50-48-20-8.drr05.mdtw.ny.frontiernet.net) |
14:57.13 | *** join/#htc-linux jonpry (~jon@unaffiliated/jonpry) |
15:19.27 | *** join/#htc-linux Alex[sp3dev] (~alexander@178.176.153.58) |
15:32.39 | *** join/#htc-linux rpierce99 (~rpierce99@96-42-102-103.dhcp.stcd.mn.charter.com) |
15:46.48 | *** part/#htc-linux Alex[sp3dev] (~alexander@178.176.153.58) |
16:00.20 | *** join/#htc-linux mitsutaka (~mitsutaka@g1-27-253-206-140.bmobile.ne.jp) |
16:02.47 | jonpry | ouch it hurts |
16:16.57 | *** join/#htc-linux detule (~detule@pool-96-234-141-27.bltmmd.east.verizon.net) |
16:22.09 | arrrghhh | jonpry, the htc universal? lol |
16:22.23 | arrrghhh | LjL, i wouldn't. it seems it *can* run linux, but from what i've gleaned it doesn't run it well. |
16:22.49 | jonpry | mostly my hangover |
16:22.59 | arrrghhh | ah |
16:23.10 | arrrghhh | wha'd you drink last night? |
16:23.41 | jonpry | something roughly pronounced glasgo |
16:23.54 | arrrghhh | hrm |
16:24.05 | arrrghhh | my parents were feeding me this stuff they called "gluvine" |
16:25.00 | jonpry | which is just the lazy pronunciation of clark's court |
16:25.31 | jonpry | not big into the consonants here |
16:25.47 | jonpry | what is gluvine? |
16:26.07 | arrrghhh | basically wine with boos and some other flavorings cooked on a stove |
16:26.13 | arrrghhh | so it warms you as you drink it |
16:26.17 | arrrghhh | plus, boos. |
16:26.18 | jonpry | lol |
16:26.20 | detule | sounds slavic |
16:26.25 | arrrghhh | german IIRC |
16:26.28 | jonpry | like spiced wine |
16:26.34 | arrrghhh | http://encyclopedia.thefreedictionary.com/Gluvine |
16:26.36 | arrrghhh | jonpry, exactly |
16:28.21 | jonpry | i tried that once. not bad |
16:28.28 | arrrghhh | yea, good stuff |
16:28.36 | arrrghhh | my dad put a shitton of hard liquor in it tho |
16:28.48 | jonpry | mine didn't have that extra kick |
16:28.51 | arrrghhh | you couldn't really taste it, but damn. 2 little mugs of it and i'm good |
16:30.00 | jonpry | i am not feeling that good atm |
16:31.40 | *** join/#htc-linux Ondalf (~ondalf@cable-roi-fffddd00-9.dhcp.inet.fi) |
16:32.03 | detule | fight fire with fire? |
16:32.27 | jonpry | lol |
16:32.35 | detule | or as my people would say, "use a nail to hammer out a nail" |
16:32.43 | arrrghhh | a little hair of the dog |
16:34.31 | jonpry | just need a little rhodium hacking |
16:35.40 | jonpry | arrrghhh, i was thinking that xdandroid might do something w/ stock surfaceflinger. if you just use that kernel and the libgles |
16:36.05 | arrrghhh | jonpry, sure |
17:45.23 | *** join/#htc-linux mastermerlin (~Adium@p4FEE5C6C.dip.t-dialin.net) |
18:01.00 | *** join/#htc-linux riotz (riotz@unaffiliated/riotz) |
18:06.19 | *** join/#htc-linux LargePrime (~LargePrim@173-110-181-129.pools.spcsdns.net) |
18:08.27 | *** join/#htc-linux ftoz (~root@214-151.gtt-net.cz) |
18:12.02 | *** join/#htc-linux Alex[sp3dev] (~alexander@178.176.130.231) |
18:46.43 | *** join/#htc-linux swc|666 (~gecko@unaffiliated/swc666/x-4934821) |
18:58.28 | arrrghhh | jonpry, hrm. i'm noticing perms might be screwed for the test i ran |
18:58.39 | arrrghhh | i should try this with the same perm's. d'oh. |
18:59.11 | arrrghhh | the one i put in has executable rights... that's the only difference. |
18:59.21 | arrrghhh | oh well, i'll try it with just your GLES and kernel. |
19:00.37 | *** join/#htc-linux Blista (~nel@209.59.220.47) |
19:01.30 | *** join/#htc-linux riotz (riotz@unaffiliated/riotz) |
19:15.41 | *** part/#htc-linux Blista (~nel@209.59.220.47) |
19:16.07 | *** join/#htc-linux Blista (~nel@209.59.220.47) |
19:25.02 | *** join/#htc-linux LargePrime_ (~LargePrim@173-122-37-142.pools.spcsdns.net) |
19:25.58 | *** join/#htc-linux WisTilt2 (~wisgreg@wireless248.wirelesstcp.net) |
19:55.31 | arrrghhh | jonpry, it appears to be working |
19:55.41 | arrrghhh | i have a blank screen where i assume the bootanimation normally would be :P |
19:56.08 | *** join/#htc-linux surge (surge@pool-72-88-82-28.bflony.fios.verizon.net) |
20:34.50 | *** join/#htc-linux dr1337 (~textual@CPE-120-147-24-40.lflg2.win.bigpond.net.au) |
20:35.47 | *** join/#htc-linux helicopter88 (~helicopte@host177-94-dynamic.22-79-r.retail.telecomitalia.it) |
20:49.39 | *** part/#htc-linux Alex[sp3dev] (~alexander@178.176.130.231) |
20:57.23 | *** join/#htc-linux xromiuz (~xN@bas1-toronto06-2925211125.dsl.bell.ca) |
21:00.11 | *** part/#htc-linux riotz (riotz@unaffiliated/riotz) |
21:09.40 | *** join/#htc-linux mitsutaka (~mitsutaka@g1-27-253-206-140.bmobile.ne.jp) |
21:17.12 | jonpry | arrrghhh, thats a good start. what about real graphics? |
21:18.21 | arrrghhh | jonpry, i didn't have any benchmark apps |
21:18.27 | arrrghhh | should i just try to run neocore and see what i get? |
21:18.35 | jonpry | no 3d stuff won't work |
21:18.52 | jonpry | fps2d probably will. but is it smooth? |
21:19.02 | arrrghhh | oh right |
21:19.11 | arrrghhh | yea, seems smooth |
21:19.19 | arrrghhh | whether or not it's a huge improvement, i can't really say... |
21:19.24 | jonpry | hmm. wonder why its not on mine. have logcat? |
21:19.25 | arrrghhh | just flipping around the UI |
21:19.29 | arrrghhh | i can get one |
21:19.41 | arrrghhh | you're running this on gb? |
21:19.42 | arrrghhh | or cm7? |
21:19.48 | jonpry | cm7 |
21:19.53 | arrrghhh | that might be why |
21:20.01 | arrrghhh | cm7 is kind of a pig for our phones |
21:20.12 | jonpry | still not smooth. and i have speed up patches in surface flinger |
21:20.54 | arrrghhh | well... gb wasn't like dramatically smoother |
21:20.59 | jonpry | i saw a cm7 video for rhod on youtube. looked pretty smooth. tbh i didn't pay much attention to it before i started hacking away |
21:21.14 | arrrghhh | lol |
21:21.44 | arrrghhh | we really need a better way to empirically tell if it's better or not. |
21:21.53 | jonpry | assuming it is actually working. then any level of smoothness is pretty amazing |
21:22.04 | arrrghhh | lol |
21:22.08 | arrrghhh | yea, that's the other thing |
21:22.09 | arrrghhh | is it working |
21:22.14 | arrrghhh | i'll get a lolcat here shortly |
21:22.16 | arrrghhh | just want one from boot? |
21:22.20 | arrrghhh | cuz it's booting now |
21:22.24 | jonpry | sure |
21:22.48 | jonpry | you get any funny stuff on the screen? |
21:23.10 | jonpry | thats a sure sign of it "working" |
21:23.19 | arrrghhh | funny stuff? |
21:23.31 | arrrghhh | other than weird issues when the phone would try to sleep, nothing out of the ordinary |
21:23.43 | jonpry | hmm |
21:23.59 | jonpry | should be like occasional messed up textures |
21:24.23 | arrrghhh | other than the normal android ones |
21:24.24 | arrrghhh | lol |
21:24.26 | arrrghhh | i didn't see any |
21:24.31 | arrrghhh | i loaded a data.img with all my poop in it |
21:24.31 | jonpry | and sometimes it gets confused about which is the front and back buffer. then everything starts looking half drawn |
21:24.34 | arrrghhh | i'll see what i can break |
21:24.42 | arrrghhh | wow |
21:24.46 | arrrghhh | never saw anything like that |
21:26.31 | jonpry | yeah there is probably at least one patch to flinger you need for it to startup |
21:35.56 | arrrghhh | jonpry, it does do some crazy wacky things in sleep |
21:36.01 | arrrghhh | i'll randomly see the console |
21:36.12 | arrrghhh | and it doesn't want to behave |
21:36.21 | arrrghhh | but other than that... unfortunately i don't notice any tangible benefits. |
21:41.19 | jonpry | i doubt its working |
21:42.17 | arrrghhh | probably not |
21:42.25 | jonpry | lolcat? |
21:42.30 | arrrghhh | o right |
21:42.34 | arrrghhh | i never grabbed one o dem |
21:50.01 | *** join/#htc-linux shaky153 (5eacbfea@gateway/web/freenode/ip.94.172.191.234) |
21:55.55 | *** join/#htc-linux Sandvold (~Sandvold@207.80-202-111.nextgentel.com) |
21:56.53 | *** join/#htc-linux [acl] (~abel@cpe-69-203-141-229.si.res.rr.com) |
21:58.53 | arrrghhh | [acl], what's up |
21:58.58 | arrrghhh | jonpry, do i have to boot with usb plugged in? |
21:59.01 | [acl] | arrrghhh: pissed at the rhod 100 |
21:59.13 | [acl] | arrrghhh: fuck that thing.. im not gonna bother with it |
21:59.15 | jonpry | arrrghhh, shouldn't |
21:59.17 | arrrghhh | i don't remember if there's any hoops i have to jump thru for adb in .39 |
21:59.25 | jonpry | works for me |
21:59.26 | arrrghhh | [acl], lol. pissed because it's so much different? |
21:59.37 | arrrghhh | jonpry, damnit. i think it was working before. it's not now for some reason. |
21:59.44 | jonpry | [acl] that is not the spirit |
21:59.46 | [acl] | arrrghhh: yeah bro.. shit has completely diff way of doing things |
21:59.58 | arrrghhh | yea |
22:00.08 | arrrghhh | it does. with the front cam and different panel |
22:00.13 | [acl] | arrrghhh: prox for them wont work.. so they will have to suffer like dogs |
22:00.15 | arrrghhh | i'm sure there's a lot of poop different under the hood |
22:00.18 | arrrghhh | lol |
22:00.20 | arrrghhh | just rhod100? |
22:00.24 | arrrghhh | other gsm devices are fine? |
22:00.25 | [acl] | yeah |
22:00.27 | arrrghhh | wacky |
22:00.29 | [acl] | works on 300 |
22:00.43 | [acl] | granted 300 is also very retarded but workable |
22:00.44 | jonpry | 210? |
22:00.52 | [acl] | i dunno im assuming 210 is the same |
22:00.53 | [acl] | :-p |
22:00.55 | arrrghhh | ;p; |
22:00.57 | arrrghhh | lol |
22:01.03 | [acl] | where is cotulla |
22:01.13 | [acl] | cotulla has a 300 or 210 ? |
22:01.15 | arrrghhh | somewhere in russia |
22:01.22 | arrrghhh | AFAIK he has a 100 |
22:01.24 | arrrghhh | but i could be wrong |
22:01.30 | [acl] | i need to compare some notes with him i guess save each other some time |
22:01.49 | jonpry | [acl], i have hardware composition in gingerbread |
22:02.09 | [acl] | jonpry: explain thyself .. |
22:02.15 | [acl] | you got my attention :-) |
22:02.33 | jonpry | surface flinger uses a hardware gl context |
22:02.37 | [acl] | correct |
22:02.48 | jonpry | i mean it does now. it didn't used to |
22:03.14 | [acl] | jonpry: some parts were gl. I did some work on the electron beam |
22:03.25 | [acl] | jonpry: but some older shit wasnt |
22:03.38 | jonpry | yeah but its using the software renderer. if you check the lolcat. it says Render: pixel flinger |
22:03.42 | jonpry | er renderer |
22:04.03 | jonpry | but nows it says Renderer: Q3Dimension |
22:04.18 | [acl] | jonpry: i was always confused about that. The gl calls are made to do all the fun stuff but then copied back via software right ? |
22:04.45 | [acl] | jonpry: i think because natively they used frame buffer objects on gles 2 |
22:04.47 | [acl] | which we dont have |
22:05.01 | [acl] | so newer devices dont ever use software |
22:05.11 | jonpry | the wormhole and electron beam require FBO's |
22:05.15 | [acl] | yeah |
22:05.24 | [acl] | well beam i switch to use gl1 calls |
22:05.33 | [acl] | but the copy back to the framebuffer |
22:05.41 | [acl] | would still be done software wise |
22:05.43 | [acl] | i think |
22:05.44 | [acl] | not sure |
22:05.55 | jonpry | all of surface flinger is done using software atm |
22:06.08 | [acl] | even for newer devices? |
22:06.28 | jonpry | all gles2.0 probably use hardware |
22:06.35 | [acl] | ahh yeah thats what i mean |
22:06.37 | [acl] | ok |
22:06.41 | [acl] | yeah i get you |
22:06.44 | [acl] | we are on the same page |
22:06.54 | jonpry | so i've got it running on gles 1 now |
22:06.55 | [acl] | so you got the copy back to fb done via hardware ? |
22:07.19 | [acl] | damn i might be able to bump the beam to 24 frames again :-) |
22:07.21 | jonpry | skipping that whole process. just put fb at the bottom of gpu1 |
22:07.26 | [acl] | the fill rate was killing me |
22:07.44 | [acl] | put fb bottom of gpu1 ? woa... |
22:07.50 | [acl] | didnt wistilt try this in the past |
22:07.52 | [acl] | many moons ago |
22:07.57 | jonpry | thats where the output surface is |
22:08.13 | [acl] | haha |
22:08.16 | [acl] | fuck.. |
22:08.31 | jonpry | i haven't got page flipping to work quite right. sometimes it the back and front buffers are flipped. i can't sync it |
22:08.49 | [acl] | yeah i can imagine that being a beotch |
22:09.20 | jonpry | brb |
22:09.23 | [acl] | coo |
22:09.51 | arrrghhh | but [acl] - wistilt2's been overclocking the RAM |
22:09.57 | [acl] | haha |
22:10.02 | arrrghhh | fancy fancy. got a pretty good boost |
22:10.07 | [acl] | that dood has 100 rhods to fry |
22:10.22 | arrrghhh | 15 i think, but yea. |
22:10.22 | [acl] | maybe mre |
22:10.24 | [acl] | haha |
22:10.29 | arrrghhh | might as well be 100 :P |
22:10.34 | [acl] | i know right |
22:10.39 | [acl] | took me 3 years to fry 1 |
22:10.46 | [acl] | so he has a long time |
22:10.49 | arrrghhh | heh |
22:11.47 | [acl] | well jonprys work is interesting |
22:11.52 | [acl] | cotulla and i talked about it in the past |
22:11.57 | [acl] | but both too lazy to do it |
22:12.05 | WisTilt2 | hi guys. arrrghhh im about to put up a test for you with gpu overclocked once i test this build |
22:12.23 | [acl] | WisTilt2: sup dood.. heard you are over clocking stuff |
22:12.36 | arrrghhh | WisTilt2, sweet |
22:12.48 | WisTilt2 | yeah compared clock stuff with how we are doing it on our device, many changes:) |
22:13.15 | WisTilt2 | oc'd ebi and smi really woke these tp2 up bigtime |
22:13.15 | [acl] | love innovation.. |
22:13.42 | WisTilt2 | jonpry get composition going and we'll have some nice "old" devices |
22:13.46 | [acl] | break the 30 fps yet ? |
22:14.00 | WisTilt2 | we need to fix that vsync or whatever locks it at 30 |
22:14.04 | [acl] | yeah |
22:14.18 | [acl] | either lower the rate or change it to something usable |
22:14.41 | [acl] | it starts at 60hz, so unless u match that, 30 is the next best thing :-( |
22:15.04 | WisTilt2 | where is this? |
22:15.04 | [acl] | or do what alex did, just lower the resolution |
22:15.12 | [acl] | WisTilt2: panel init |
22:15.20 | [acl] | we went over this dood remember.. about a year ago |
22:15.21 | shaky153 | hey guys |
22:15.32 | WisTilt2 | hmm, so we're locking vsyn in panel? that should be easy to change |
22:15.40 | [acl] | WisTilt2: the evo had the same issue |
22:15.52 | [acl] | WisTilt2: so their patch changes a few values around |
22:16.23 | [acl] | thing is, not sure if all the panels have the same init value. I know evo may have had epson ? not sure but we have tons to worry about |
22:16.59 | jonpry | [acl] so the when it starts up stuff has 50% chance of being all in sync or all out of sync, and it will stay that way. some things seem to roll the dice again. i figure it call fbPost without eglswapbuffers under some circumstances |
22:17.00 | WisTilt2 | [acl]: i need to look at panel vsync then. after i get gpu's oc'd ill see what i can do with that. we can work out the different panel type once we get it working on these |
22:17.03 | [acl] | WisTilt2: if you search back at the chat logs you should be able to find it. I know you and i spent some time reading up on it. |
22:17.14 | WisTilt2 | yeah we did |
22:18.30 | [acl] | if our rate can go higher, then thats a possible solution. Or we can lower it to the point where we can match the vsync |
22:18.59 | [acl] | so if we lower it at 45 and our fill rate can match that, well then boys our devices will be the msm7k to envy |
22:19.21 | jonpry | isn't the eye kind of limited? |
22:19.22 | WisTilt2 | whats a good benchmark app that only tests gpu? this quadrant is good but checks everything |
22:19.34 | jonpry | why waste so much memory bandwidth on invisible frames? |
22:19.38 | [acl] | jonpry: yeah .. we will def notice 45..lol |
22:21.38 | [acl] | jonpry: so how much of memory are you over lapping? |
22:21.58 | arrrghhh | WisTilt2, other than neocore or neamark... not sure. |
22:22.06 | [acl] | wait, that made no sense. how much mem are you allocating to 3d now ? |
22:23.27 | *** join/#htc-linux MethoS- (~clemens@134.102.106.250) |
22:25.01 | jonpry | same old 8mb |
22:25.35 | jonpry | it doesn't seem to work with anything else. could probably hack the driver somehow |
22:25.59 | [acl] | jonpry: hack the oem driver ? we tried but there were no gains |
22:26.11 | [acl] | bumped it up to like 13 i think |
22:26.16 | [acl] | useless |
22:26.16 | jonpry | this all started because i can't get it to copy its output buffers anywhere when its running in surface flinger. yes oem |
22:26.21 | arrrghhh | jonpry, did you get that lolca? |
22:26.22 | arrrghhh | cat* |
22:26.40 | jonpry | yeah working on it |
22:26.48 | WisTilt2 | jonpry are you overlapping fb @ 8mb to fill all of gpu1? |
22:26.48 | arrrghhh | ok np |
22:26.52 | arrrghhh | just wanted to make sure you got it ;) |
22:27.43 | jonpry | WisTilt2, no. i think it doesn't matter what size you set fb at as long as its larger than the 1.2MB or whatever is required |
22:28.12 | WisTilt2 | im still puzzled about that 1mb hw3d |
22:30.28 | [acl] | aite fellas.. gotta go to the gym. the ladies dont care about my android work.. they just care about my arms :-( |
22:30.39 | [acl] | bitches :-( |
22:30.49 | arrrghhh | lol |
22:30.54 | jonpry | they don't know about nerds? |
22:31.16 | [acl] | they do but they arent exactly chasing after them |
22:31.25 | [acl] | lol |
22:31.27 | WisTilt2 | later [acl] |
22:31.45 | [acl] | jonpry: let me know how it goes dodo.. sounds interesting .. |
22:31.53 | jonpry | alright |
22:31.54 | [acl] | WisTilt2: dont fry your phone :-p |
22:31.59 | WisTilt2 | got backups:) |
22:32.03 | [acl] | nice nice |
22:34.02 | jonpry | i can't find a definition for egl_native_window_t anywhere in the gingerbread source |
22:39.01 | *** join/#htc-linux rpierce99 (~rpierce99@96-42-102-103.dhcp.stcd.mn.charter.com) |
22:41.07 | WisTilt2 | jonpry: are you using the board panel file from .39 for your tests? |
22:42.03 | *** join/#htc-linux kaoruko (~nickls@statusfailed.com) |
22:42.06 | *** join/#htc-linux drewis (~quassel@71.20.84.142) |
22:42.08 | *** join/#htc-linux pigeon (~pigeon@eth5284.nsw.adsl.internode.on.net) |
22:42.11 | *** join/#htc-linux sado1 (~sado1@v0ll.de) |
22:42.14 | jonpry | WisTilt2, yes |
22:42.14 | *** join/#htc-linux infernix (nix@unaffiliated/infernix) |
22:42.14 | *** join/#htc-linux dcordes (~oedevel@unaffiliated/dcordes) |
22:42.15 | *** join/#htc-linux Mavy-bnc (mavfree@91.196.169.2) |
22:42.15 | *** join/#htc-linux bukington (~bukington@fac34-2-82-228-151-145.fbx.proxad.net) |
22:42.16 | *** join/#htc-linux T-Junk (~TechnoJun@chaosarmy.org) |
22:42.19 | *** join/#htc-linux ElFinLazz (~elfinlazz@112.154.189.103) |
22:42.20 | *** join/#htc-linux nrirclog (~nrirclog@netripper.com) |
22:42.20 | *** join/#htc-linux hasi (hasi@mausi.hasicon.org) |
22:42.48 | WisTilt2 | well, i dont know when or how stuff got removed from init but there is stuff missing so im fixing it. lots of it is has to do with vsync and hsync. all that timing stuff i did last year is not in here now |
22:43.06 | rpierce99 | WisTilt2: I've been doing sort of a mini drain test today, purposefully leaving the phone off, even when i get alerts. I took it off the charger 7 hours ago and I am at 56% battery |
22:44.24 | WisTilt2 | rpierce99 was it at 100% when you started? thats way too much drain. overnight i averaged just under .8%hr so dsp dropped it a small amount. |
22:44.40 | rpierce99 | well my drain has never been as much as yours |
22:44.54 | WisTilt2 | 7hrs you should only drop 7% at worst in sleep unless its not really going into full sleep from some app keeping it awake |
22:45.00 | rpierce99 | only dropping 45% over 7+ hours is good for this build |
22:45.10 | arrrghhh | lol |
22:45.18 | arrrghhh | WisTilt2, remember he's got a lot of stuff syncing' |
22:45.20 | WisTilt2 | cat /proc/msm_pm_stats and look at the idle-sleep numbers and see how high it is |
22:45.46 | arrrghhh | i've noticed a few apps i've added auto-sync and there's seemingly no setting to enable to disable that 'feature' |
22:46.07 | arrrghhh | s/to/or/ |
22:46.14 | arrrghhh | lol |
22:46.16 | arrrghhh | oh well. |
22:46.21 | arrrghhh | enable or disable :P |
22:46.31 | arrrghhh | auor-sync, win |
22:46.40 | WisTilt2 | well if apps are waking it up a bunch that idle-sleep counter should be pretty high over 7hrs, if its low then its really going into sleep most of the time. |
22:46.49 | rpierce99 | idle-sleep: count: 10194, total time: 646.69 |
22:46.59 | arrrghhh | is that time in seconds? |
22:47.02 | WisTilt2 | yes |
22:47.12 | WisTilt2 | have you rebooted at all since this 7hr time? |
22:47.15 | arrrghhh | that's only 10 mins or so |
22:47.17 | rpierce99 | no |
22:47.25 | rpierce99 | idle-request is ridiculously high |
22:47.35 | WisTilt2 | yeah that was my next question |
22:47.48 | rpierce99 | 29551992.20 |
22:47.55 | WisTilt2 | ok let me ask you this... |
22:48.27 | WisTilt2 | did you have charger plugged in until full charge then just start the test from there or did you reboot first after charging? |
22:48.58 | *** join/#htc-linux CIA-110 (~CIA@cia.atheme.org) |
22:49.32 | rpierce99 | i had it plugged in all night (3am-9:30am), then just took it off |
22:49.40 | rpierce99 | and used it, no reboots |
22:50.05 | rpierce99 | so it has been on/running since you provided the 35/35 kernel last night |
22:50.11 | WisTilt2 | thats why idle-request is so high. look at idle-abort and it should be very high also. see what idle-failed is also |
22:50.34 | rpierce99 | idle-abort is only20379.17 |
22:50.35 | WisTilt2 | i have it prevented from idle-sleep with any usb plugged in |
22:51.17 | WisTilt2 | what is the counter value for idle-abort compared to counter on idle-request? |
22:51.23 | WisTilt2 | top number |
22:52.40 | rpierce99 | 174,537,411 - 176,922,748 |
22:52.46 | WisTilt2 | yeah thats good |
22:53.01 | WisTilt2 | idle-failed count and time is what? |
22:53.26 | WisTilt2 | idle-sleep failed i think its called |
22:53.33 | rpierce99 | idle-failed-sleep: count: 132, total_time: 0.16 |
22:54.16 | WisTilt2 | yeah thats good also. so this means it is going to sleep most of the time. you didnt use it at all during this 7hr period? |
22:54.47 | rpierce99 | nope, screen off, just syncing, i received about 10 emails, so a few alerts went off, and it rang once or twice |
22:55.42 | rpierce99 | android says cell standby is 51% of usage, phone idle is 48%, display is 3% |
22:55.43 | WisTilt2 | it was in idle-sleep for almost 170mins thats good but it does look like you're awake for 4hrs out of the 7 then so apps are hogging battery waking it up too much now that i look at it |
22:55.58 | rpierce99 | and actual time unplugged is 7h22m |
22:56.46 | WisTilt2 | if you run in winmo with same apps syncing do you get around same drain? |
22:56.59 | rpierce99 | can't say, i've never used winmo on this phone |
22:57.08 | rpierce99 | threw xdandroid on there the same week i got it |
22:57.09 | arrrghhh | can't really get the same apps to sync either |
22:57.13 | rpierce99 | and there's no GV app |
22:57.16 | arrrghhh | ^^ |
22:57.23 | rpierce99 | or gtalk |
22:57.26 | rpierce99 | or latitude |
22:57.27 | rpierce99 | lol |
22:57.29 | arrrghhh | there's nothing in winmo |
22:57.36 | arrrghhh | why do you think we're here WisTilt2 :P |
22:57.40 | WisTilt2 | you might try running scbs so you get a more accurate battery reading also. cant imagine that much drain |
22:57.52 | arrrghhh | rpierce99, original battery? |
22:57.53 | WisTilt2 | yeah good to here no winmo) |
22:57.58 | rpierce99 | yeah original |
22:58.01 | arrrghhh | hrm |
22:58.20 | arrrghhh | battery might just be old too. i would say we should compare drain with our own batteries/devices/usages |
22:58.28 | arrrghhh | unless we're running radio off no syncs |
22:58.36 | arrrghhh | can't really compare apples to apples |
22:58.57 | WisTilt2 | scbs would tell battery age condition better than the standard battery algo too |
22:59.07 | rpierce99 | scbs means new rootfs right? |
22:59.33 | WisTilt2 | yes, i can send you emwe's that im using that jonpry added the daemon to if you want |
22:59.44 | arrrghhh | rpierce99, IIRC it's a 1-2 line chane in rootfs |
22:59.46 | rpierce99 | yeah that'd be great |
22:59.54 | rpierce99 | i don't have the binaries either though arrrghhh |
22:59.56 | arrrghhh | change* |
22:59.59 | arrrghhh | oic |
23:01.31 | WisTilt2 | rpierce99 its on the server same path but rootfs.img - i didnt compress it sorry so its 15mb |
23:01.37 | rpierce99 | no prob |
23:02.20 | WisTilt2 | you'll have to make the model so you'll have to do the charge discharge thing again while it stats so you can make the model. |
23:02.55 | WisTilt2 | unless you have an old scbs.conf file from back in the testing days |
23:03.00 | rpierce99 | i'll get the low end of the charge tonight, charge it up again overnight, then let it drain again |
23:03.08 | rpierce99 | i probably do still have all of my old scbs files on my sd |
23:03.41 | WisTilt2 | just take the most recent .log file and make the model from it and that should be close |
23:11.17 | *** join/#htc-linux detule (~oliver@pool-96-234-141-27.bltmmd.east.verizon.net) |
23:17.09 | *** part/#htc-linux Sand_away (~Sandvold@207.80-202-111.nextgentel.com) |
23:17.19 | detule | jonpry, if you send me those surfaceflinger patches i can probably build a suitable xdandroid image |
23:18.15 | jonpry | sounds good. let me get it working again though. got it all screwed up atm |
23:18.52 | arrrghhh | nice. all i can do is run them and look dumb detule :D |
23:27.43 | detule | WisTilt2, the panel code in the .3x trees is as you left it in June/July....which clock/bus are you oc-ing btw? |
23:33.53 | WisTilt2 | i cant believe those panel inits got missed but i guess i left them out. this oc'ing, im using what we do in our device. need to integrate it into the current clock code when we're happy with the values |
23:34.29 | rpierce99 | 35/35 is great |
23:34.41 | rpierce99 | on mine at least |
23:34.52 | jonpry | detule, you need a fb==gpu1 kernel? |
23:35.00 | WisTilt2 | still snappier than standard and it hasn't slowed down over time? |
23:35.30 | detule | jonpry, no i can do that |
23:35.42 | rpierce99 | no i had one slowdown right before bed but i'm sure it was just doing something, it's super snappy again now |
23:36.28 | WisTilt2 | did you try running quadrant benchmark on it? mine right now is just below nexus 1 at 453rating |
23:36.50 | rpierce99 | no i''ll d/l quadrant |
23:37.38 | detule | WisTilt2, so mysterious :) |
23:40.24 | jonpry | detule, i think you got a kernel anyways. this one is funky when it works. the buffer situation seems to get swapped whenever a new program takes full screen kind of view. so opening and closing the settings or something has a good chance of getting it working |
23:40.42 | jonpry | or breaking it |
23:41.25 | rpierce99 | now that i've been playing with it for a bit i'm being told battery is <15% |
23:42.40 | WisTilt2 | you need scbs. standard batt stuff seems to go all wrong below 40% thinking battery is lower than it really is. |
23:42.59 | arrrghhh | standard batt stuff seems wrong no matter where it is :P |
23:43.13 | arrrghhh | but yea, above 90 and below 40 seem really erratic. |
23:43.51 | WisTilt2 | yes i notice after a short call, maybe 10mins talk time, it thinks its over 20% lower but recovers after a few mins. scbs tracks and reports pretty spot on |
23:44.32 | rpierce99 | this 2nd run of quadrant is going much better than the first, first one was laggy because the phone was busy i think |
23:44.46 | rpierce99 | placed the same though, hmm |
23:44.56 | rpierce99 | 420, just below xperia x10 |
23:45.38 | WisTilt2 | 420 hmm, im 453 but you dont have this kernel yet. im testing the panel changes then ill upload for you guys to try new one |
23:45.51 | rpierce99 | whats new? |
23:46.24 | WisTilt2 | more clock changes in axi and gpu1 oc'd 10% |
23:46.35 | rpierce99 | haha gpu oc, nice |
23:46.50 | detule | i recall starfox overclocking the axi clock |
23:47.23 | WisTilt2 | did he see any improvements? |
23:47.28 | detule | but then emwe pointed something out that the axi clock never gets used, this is all from very vague recollection |
23:47.39 | detule | rpierce99, and arrrghhh perhaps remember better |
23:47.53 | rpierce99 | never worked with starfox, never tested his kernel |
23:47.56 | detule | and this was on .27 which still uses JB's clocks |
23:48.07 | WisTilt2 | axi is used! wouldnt run without it |
23:48.21 | WisTilt2 | unless he's changed to run ahb |
23:50.03 | WisTilt2 | arrrghhh/rpierce99 kernel is up |
23:50.22 | WisTilt2 | this also has cpu oc'd at 614400 |
23:51.46 | detule | this is what i am thinking about : emweXofrats: btw... can you please verify... ebi1_clk doesn't work. so increasing the axiclk rate does just nothing IIRC. tried that 2 years ago ;) |
23:51.49 | arrrghhh | WisTilt2, CPU OC built-in...? |
23:51.56 | WisTilt2 | can you guys tell me if its my imagination but stuff on the screen is much cleaner looking now. panel changes should be only thing that would have affected that |
23:52.10 | WisTilt2 | arrrghhh yes for now. oc from cmd line wasnt working anyway |
23:52.24 | arrrghhh | oh right, i forgot 'bout that |
23:52.53 | jonpry | this stuff has meltdown written all over it |
23:52.59 | arrrghhh | heh |
23:53.02 | WisTilt2 | detule yeah ebi1_clk gets set (structure anyway) but its never used again. axi is something totally different |
23:54.07 | WisTilt2 | my phone actually only gets slightly warm after pounding it all day. cdma devices seems to be getting hotter though, still not that warm imo |
23:58.13 | *** join/#htc-linux surge (surge@pool-72-88-82-28.bflony.fios.verizon.net) |