# 00:00:43 |
S0-2 |
At a later point, it complains that termcap library not found, even though I installed it. |
# 00:02:58 |
S0-2 |
readelf -Ws /usr/lib/libtermcap.so | grep tgetent |
# 00:02:59 |
S0-2 |
34: 00000000000010c8 993 FUNC GLOBAL DEFAULT 11 tgetent |
# 00:03:20 |
S0-2 |
and yet |
# 00:03:23 |
S0-2 |
[ALL ] checking for library containing tgetent... no |
# 00:03:25 |
S0-2 |
[ERROR] configure: error: no termcap library found |
# 00:03:53 |
y_morin |
quits : Quit: leaving |
# 00:20:14 |
S0-2 |
Anyone? I have ncurses and termcap installed. |
# 00:24:48 |
S0-2 |
Also, cross-gdb's config.log ends with configure: exit 0 |
# 01:02:49 |
S0-2 |
quits : Ping timeout: 240 seconds |
# 01:09:06 |
S0-2 |
joins #crosstool-ng |
# 02:12:52 |
alan_o |
quits : Quit: Leaving |
# 02:27:51 |
ccole |
joins #crosstool-ng |
# 06:04:14 |
djerome |
quits : Remote host closed the connection |
# 06:05:13 |
smartin_ |
joins #crosstool-ng |
# 06:11:16 |
smartin_ |
quits : Quit: leaving |
# 09:21:06 |
sh4rm4 |
quits : Remote host closed the connection |
# 09:21:31 |
sh4rm4 |
joins #crosstool-ng |
# 09:49:20 |
y_morin |
joins #crosstool-ng |
# 11:00:11 |
SgrA |
joins #crosstool-ng |
# 11:02:58 |
S0-2 |
quits : Ping timeout: 240 seconds |
# 11:25:55 |
y_morin_ |
joins #crosstool-ng |
# 11:29:52 |
y_morin |
quits : *.net *.split |
# 11:30:56 |
memleak |
joins #crosstool-ng |
# 11:34:11 |
y_morin_ |
is now known as: y_morin |
# 11:37:34 |
doc2 |
joins #crosstool-ng |
# 16:21:16 |
mnt_real__ |
joins #crosstool-ng |
# 16:23:19 |
mnt_real_ |
quits : Ping timeout: 264 seconds |
# 16:41:35 |
diorcety |
joins #crosstool-ng |
# 17:39:28 |
doc2 |
quits : Remote host closed the connection |
# 17:52:44 |
doc2 |
joins #crosstool-ng |
# 18:09:32 |
S0-2 |
joins #crosstool-ng |
# 18:12:01 |
SgrA |
quits : Ping timeout: 240 seconds |
# 18:24:43 |
djerome |
joins #crosstool-ng |
# 18:51:13 |
djerome |
quits : Remote host closed the connection |
# 18:53:05 |
djerome |
joins #crosstool-ng |
# 19:31:13 |
djerome |
quits : Remote host closed the connection |
# 19:33:47 |
djerome |
joins #crosstool-ng |
# 19:54:15 |
sh4rm4 |
quits : Remote host closed the connection |
# 19:54:41 |
sh4rm4 |
joins #crosstool-ng |
# 19:57:34 |
sh4rm4 |
quits : Remote host closed the connection |
# 19:58:00 |
sh4rm4 |
joins #crosstool-ng |
# 20:11:24 |
Martell |
joins #crosstool-ng |
# 20:14:27 |
Martell |
Hey y_morin |
# 20:20:39 |
Martell |
Hey guys does anyone know how to add a new arch? |
# 20:21:05 |
Martell |
as in CT_ARCH_targetarch |
# 20:21:11 |
Martell |
in my case CT_ARCH_spu |
# 20:24:15 |
doc2 |
quits : Remote host closed the connection |
# 20:40:51 |
y_morin |
Martell: Look in config/arch/ for examples. |
# 20:41:29 |
y_morin |
Martell: Also, look at: "docs/8 - Internals.txt" |
# 20:42:13 |
y_morin |
Although that part of the doc is not really up-to-date. |
# 20:42:17 |
y_morin |
:-( |
# 20:42:38 |
memleak |
Martell, what arch are you going to be adding? jw |
# 20:43:12 |
y_morin |
Martell: 'spu' <-- what architecture is that? |
# 20:44:02 |
y_morin |
Oh, the Cell CPU in the PlayStation? |
# 20:45:10 |
Martell |
yup just the spu |
# 20:45:24 |
Martell |
its the cell processor |
# 20:45:28 |
memleak |
ps3 doesnt have otheros feature anymore though.. |
# 20:45:55 |
Martell |
I'm trying to create a file in config/arch |
# 20:46:00 |
Martell |
to match it |
# 20:46:32 |
Martell |
http://en.wikipedia.org/wiki/Cell_%28microprocessor%29#Synergistic_Processing_Elements_.28SPE.29 |
# 20:46:48 |
Martell |
Synergistic_Processing_Element is the exact name |
# 20:47:07 |
Martell |
memleak: there are many custom firmwares with otheros available |
# 20:47:22 |
Martell |
there is even a linux distro still being updated for it |
# 20:47:48 |
Martell |
http://sourceforge.net/projects/redribbon/ |
# 20:49:55 |
Martell |
y_morin: I successfully built the powerpc64 toolchain with the PS3 specifics on ct-ng |
# 20:50:23 |
Martell |
so now I just need the spu to have a full 4.8 chain that should be better than the homebrew chains out there |
# 21:06:30 |
memleak |
heh what do those noobs use? codesourcery and shit? |
# 21:16:24 |
Martell |
no they still use gcc |
# 21:16:37 |
Martell |
but the versions are 3.x |
# 21:16:41 |
Martell |
or 4.3 etc |
# 21:17:02 |
Martell |
4.8 is a major jump ;) |
# 21:25:02 |
djerome |
quits : Remote host closed the connection |
# 21:26:52 |
djerome |
joins #crosstool-ng |
# 21:49:38 |
ansiwen_ |
quits : Remote host closed the connection |
# 21:50:31 |
ansiwen |
joins #crosstool-ng |
# 22:23:39 |
alan_o |
joins #crosstool-ng |
# 22:41:38 |
memleak |
at this point i'd wait for 4.8.3 before building a ton of stuff with it |
# 22:43:56 |
ovf |
ibm dumped cell, right? |
# 22:51:23 |
memleak |
yes |
# 22:51:47 |
memleak |
if it wasn't, PS4 would have it |
# 22:51:50 |
memleak |
:P |
# 22:54:59 |
ovf |
i kind of liked it (and the promotional free access to ibm qs22) |
# 23:01:01 |
memleak |
i liked the cell too |
# 23:01:11 |
memleak |
thing was a beast |
# 23:01:49 |
memleak |
still kinda is..... i mean the SPE can actually access GDDR5 memory, normally CPUs cannot access VRAM |
# 23:02:02 |
memleak |
rather only SDRAM |
# 23:02:48 |
memleak |
how they integrated the memory controller and all that stuff is just fascinating, plus it being an 8 core isn't bad either to say the least |
# 23:03:19 |
memleak |
the GPU on the other hand is another story, that's actually outdated... |
# 23:06:34 |
ovf |
the amd thing on this year's consoles also talks to gddr5 i think? |
# 23:07:08 |
memleak |
well besides those two of course |
# 23:07:20 |
ovf |
(anyone running custom code on them yet?) |
# 23:07:40 |
memleak |
no |
# 23:07:57 |
memleak |
well xbox one doesnt even use GDDR5 |
# 23:08:01 |
memleak |
i dont keep up with xbox at all |
# 23:08:02 |
memleak |
fuck microsoft |
# 23:08:52 |
memleak |
PS4 has 8GB GDDR5 dedicated + 2GB DDR3 for swap or some shit |
# 23:09:06 |
ovf |
well, even if i get to run my code on ps4, there isn't much i could do with 8gb |
# 23:09:31 |
memleak |
sony wont say exactly what the 2GB is for but hackers claim to have the answer |
# 23:09:37 |
memleak |
as always... elitists... |
# 23:10:46 |
ovf |
ddr3 is like a hard drive these days, cheap and everywhere |
# 23:10:53 |
memleak |
indeed |
# 23:10:57 |
memleak |
thats why i run 32GB |
# 23:11:05 |
ovf |
i think my consumer grade ssd comes with a bunch built-in |
# 23:11:31 |
memleak |
how long are the latest SSDs lasting now? |
# 23:11:39 |
memleak |
i remember at first the flash kept wearing out |
# 23:11:55 |
memleak |
still pieces of garbage or are they reliable? |
# 23:12:35 |
ovf |
don't know, i don't do any server hardware stuff |
# 23:12:38 |
y_morin |
memleak: I've been using a Samsung 840 Pro 128GiB for more than a year, mounted on my development dir, without issues. |
# 23:12:45 |
ovf |
but i think they solved the wear... with software |
# 23:13:21 |
memleak |
software as in drivers, or software as in built in (firmware) |
# 23:13:23 |
y_morin |
memleak: As report by smarctl, it is only 3% weared out. |
# 23:13:30 |
ovf |
firmware |
# 23:13:31 |
y_morin |
memleak: firmware, not drivers. |
# 23:13:35 |
memleak |
ah thats good! |
# 23:13:58 |
y_morin |
memleak: No, that's not good: we can't audit firmware, and we cant fix it, and we can't enhance it. |
# 23:14:29 |
memleak |
well yes but drivers suck up CPU resources and whatnot, it shouldnt be in userspace, not for a hard drive anyway. |
# 23:14:41 |
ovf |
well, yes, that's a mixed blessing. but at least i can just implement the basic sata stuff and have decent storage in my os |
# 23:14:58 |
y_morin |
memleak: I prefer to trade a few CPU cycles for that. |
# 23:15:01 |
memleak |
and there is really a percentage meter in the SMART data telling you how broken your drive is? |
# 23:15:19 |
y_morin |
memleak: yes, it is reported in the SMART info. |
# 23:15:24 |
memleak |
wow thats sad. |
# 23:15:51 |
memleak |
that you need a meter for such a thing |
# 23:15:51 |
ovf |
y_morin: by the way the samsung 840 is funny business. bigger models (256gb+) come with built-in raid0 |
# 23:16:16 |
y_morin |
memleak: Why? It's pretty important to know when the drive is getting weared out so one can replace it. |
# 23:16:37 |
ovf |
i think memleak wants things to work forever |
# 23:16:38 |
memleak |
"just so you know, you're drive is this much worn out! when you decide enough is enough, buy a new SSD" |
# 23:16:52 |
y_morin |
ovf: I gues all models come with some kind of "RAID" in it: they are parallelising the acces to the flash anyway. |
# 23:17:11 |
y_morin |
*access |
# 23:17:22 |
ovf |
they claim something like 1.5x speed increase starting with 256gb model |
# 23:17:52 |
y_morin |
ovf, now, that the 256 is just two 128 side-by-side I don't care much, as long as it appears asa single drive to the OS. |
# 23:18:25 |
memleak |
my non-SSD drive lasted me 15 years.. |
# 23:18:40 |
memleak |
and im totally happy with that |
# 23:18:50 |
y_morin |
ovf: My 128GiB one gives me roughly 540MiB per second, which is close to the maximum bandwidth SATA3 can give. |
# 23:19:00 |
memleak |
woah thats fast... |
# 23:19:17 |
memleak |
would you two advise to buy an SSD or no? |
# 23:19:22 |
y_morin |
memleak: SSD power, yes. And it sucks about 480MiB per second when writing. |
# 23:19:40 |
y_morin |
memleak: I am *very* happy with the Samsung 840 Pro. |
# 23:20:07 |
y_morin |
memleak: I got it for ~120⬠last years, it is now around 115â¬. |
# 23:20:27 |
ovf |
y_morin: huh, how do you measure? :-) |
# 23:20:30 |
y_morin |
memleak: So if you have the bucks, it is a very good investment. |
# 23:20:51 |
ovf |
yes, at least for consumer stuff you really aren't losing anything going to ssd |
# 23:20:56 |
ovf |
'cept money |
# 23:21:12 |
y_morin |
ovf: that's linear transfer: dd if=/dev/sdb bs=1048576 >/dev/null |
# 23:21:23 |
y_morin |
ovf: and get the results from dd. |
# 23:21:44 |
memleak |
oooo.. a pci express SSD drive... |
# 23:21:52 |
ovf |
and also in non-cost-sensitive business areas people generally say "forget hdd" |
# 23:21:52 |
y_morin |
ovf: random transfers of "small" files is a bit slower, but still much, much faster than spiniing rust. |
# 23:21:58 |
memleak |
kinda expensive.. 4,000 for an 800GB |
# 23:22:45 |
ovf |
y_morin: ah yes, dd gets me 548gb/s |
# 23:23:00 |
ovf |
er, mb/s |
# 23:23:28 |
ovf |
let's reassert the gb/s version in 10 years |
# 23:23:54 |
y_morin |
memleak: There's an OCZ RevoDrive 960GiB, pci-express 4x, for ~1400⬠|
# 23:24:26 |
y_morin |
memleak: but otherwise, the SATA3 Samsung 840 Pro is just SSD-gold. |
# 23:24:49 |
memleak |
oh thats not terrible |
# 23:25:19 |
ovf |
memleak: if you ever feel you need to spend more, there are pci-e dram cards... |
# 23:26:09 |
memleak |
no no no no im good so far thanks :P |
# 23:28:07 |
ovf |
ah! i just remembered i had an on-topic question |
# 23:28:41 |
memleak |
:D |
# 23:29:38 |
ovf |
i have some (production) code i need to compile for linux i686, x86_64, arm and (less of a requirement) windows i686 and x86_64 |
# 23:29:58 |
ovf |
i'd also like a fairly recent version of gcc |
# 23:30:37 |
ovf |
right now we use our distribution packages for gcc, which has an unpleasant side-effect that everyone has different versions |
# 23:31:39 |
ovf |
i also want things relatively stable (which contradicts the "fairly recent" point), at least i don't want big regressions. |
# 23:32:40 |
ovf |
what are my options? ctng offers me too many options to choose wisely, and it looks like commercial support for gcc for i686/x86_64 is mostly nonexistent except for maybe redhat |
# 23:34:22 |
y_morin |
ovf: Go for latest of all: gcc-4.8.2, binutils-2.24, glibc-2.19. |
# 23:35:59 |
ovf |
and also update to keep it latest? |
# 23:38:38 |
ovf |
so basically you suggest i maintain my own ctng toolchain, right? this makes sense. |
# 23:39:13 |
y_morin |
ovf: Well, you're in #crosstool-ng so expect we suggest you use crosstool-NG. ;-) |
# 23:39:33 |
ovf |
i still wonder why there are no third-party options. if i wanted to do arm, i could pick linaro, or codesourcery, or ... but nothing for x86 |
# 23:39:51 |
ovf |
it's not like i have some very specific requirements |
# 23:39:52 |
y_morin |
ovf: But since you want newer tools, and don't want to depend on your distro or any other vendor, you don't have much choice. |
# 23:41:07 |
ovf |
well, maybe i would have been content with a vendor, but i'm not quite sure how that relationship could have looked like. people say bad things about intel, for instance. |
# 23:42:44 |
ovf |
speaking of glibc, i know this is probably bad practice, but if i really want, can i have say gcc 4.8 with glibc 2.12? |
# 23:43:19 |
ovf |
since the code is often going to run in such environments. |
# 23:44:48 |
ovf |
should i take crosstool-NG 1.19.0 or tip? |
# 23:45:43 |
y_morin |
ovf: If you plan to compile stuff with a crosstol-NG toolchain, and run it in an existing environment, this is not supported. |
# 23:46:07 |
y_morin |
ovf: crosstool-NG builds toolchain for you to then build a complete system from scratch. |
# 23:46:18 |
y_morin |
ovf: ... not target an existing system. |
# 23:46:41 |
y_morin |
ovf: That is best handled by using the tools provided sith your system. |
# 23:46:47 |
y_morin |
s/sith/with/ |
# 23:47:30 |
ovf |
that's what i feared: you're sending to me to go talk to redhat to use their ancient and dubiously patched toolchain |
# 23:47:40 |
ovf |
and then go to canonical, rinse and repeat |
# 23:48:32 |
y_morin |
ovf: Yes, basically. |
# 23:48:32 |
ovf |
by "not supported", how big will be my problems? :-) |
# 23:48:58 |
y_morin |
ovf: depends, hard to say. Maybe none. Maybe corner cases hard to debug. |
# 23:49:31 |
ovf |
actually the only real issue with environment we had so far was the memcpy/memmove reality show |
# 23:50:22 |
ovf |
we're still having it, kind of, since memmove performance is absolutely unpredictable across glibc versions |
# 23:50:41 |
ovf |
thus i wanted to be able to choose a build-side glibc at will |