01:11.01 | *** join/#bzflag infobot (ibot@rikers.org) |
01:11.01 | *** topic/#bzflag is http://BZFlag.org || http://ohloh.net/p/bzflag || https://wiki.bzflag.org/Getting_Help || Channel Logs: http://infobot.rikers.org/%23bzflag/ || 2.4.4 released! https://forums.bzflag.org/viewtopic.php?f=62&t=19186 |
01:11.01 | *** mode/#bzflag [+v infobot] by ChanServ |
01:16.05 | macsforme | what is the accepted method for running servers with alternative protocols on the list server? |
01:17.28 | macsforme | starting with something other than "BZFS" seems to result in the server not being listed, and the client expects an exactly 8-character protocol string, and I don't want to overlap with standard client protocol versions |
01:22.31 | BulletCatcher | Have you tried something like BZFSxNNN where "x" is literal and NNN are some digits? |
01:27.07 | BulletCatcher | In the set of officially used protocols in the BZFSNNNN format, the first digit has always been a zero or one. It is a safe bet that we'll never make it up to BZFS9NNN. |
01:35.36 | blast007 | macsforme: it's supposed to support things that don't start with 'BZFS'. It should work for anything AAAANNNN (A = alpha, N = numeric). |
01:42.57 | blast007 | and it does indeed work just fine: |
01:43.05 | blast007 | 192.169.83.30 BARK0221 00000010000100000000000000000000c800c800c800c800c800c800c8 192.169.83.30 test |
01:43.29 | blast007 | perhaps you were not providing a public key? or you have some other configuration issue? |
02:33.07 | macsforme | blast007: how did you modify it? just changing the return value of getServerVersion()? |
02:42.31 | blast007 | yes |
02:42.49 | macsforme | the exact same configuration fails under protocol "GCW00221" |
02:43.28 | macsforme | I get 2016-06-12 02:41:57: Got line: "MSG: ADD trinity.fairserve.net:5162 GCW00221 ... Grassroots by Babel" but then nothing else |
02:44.08 | macsforme | maybe it's the number in the first four characters |
02:45.01 | rodie | .g 248 |
02:45.54 | macsforme | yeah, it does work with "GCWZ0221"... problem solved |
02:49.57 | the_map | use "GCWO0221" instead of "GCW00221" :) |
02:51.51 | macsforme | that will be for the GCW One :P |
02:52.48 | the_map | I have no clue what that means |
02:53.43 | macsforme | the successor to the GCW Zero |
02:54.39 | the_map | yeah, no clue on that either :P |
02:54.46 | the_map | googles it |
02:55.18 | blast007 | it's an underpowered linux gaming handheld with poor manufacturer support |
02:55.32 | the_map | blast007: so what's your opinion on it? :P |
02:55.33 | macsforme | the_map: http://fairserve.net/~josh/bzflag/embedded/gcwzero5.png |
02:56.03 | the_map | macsforme: ooh, I can count the pixels! :) |
02:56.20 | blast007 | 320x240 resolution |
02:57.48 | blast007 | the_map: that isn't an opinion, that's just fact |
02:57.56 | the_map | yeah |
02:57.59 | the_map | wait |
02:58.18 | the_map | on the resolution, or the previous statement? |
02:58.22 | blast007 | previous |
02:58.26 | blast007 | it's a single core 1GHz MIPS processor with 512MB DDR2 RAM |
02:58.40 | the_map | gtg |
02:58.52 | blast007 | and they haven't released an update to the firmware (the OS) since 2014 |
02:59.17 | macsforme | interestingly, it seems to generally be CPU-bound when running bz |
02:59.40 | blast007 | macsforme: it's probably CPU bound on desktop too |
03:00.08 | macsforme | true... not very efficient |
03:00.36 | blast007 | I'd think that would be *less* surprising on an tiny mobile platform since they generally rely more heavily on the GPU to carry the load |
03:00.39 | macsforme | but on the raspberry pi it seemed to be GPU bound, at least when running at high resolutions |
03:03.00 | macsforme | a lot of stuff we do like sorting geometry for draw order consumes a lot of CPU with no practical benefit on modern hardware |
03:03.16 | blast007 | keep in mind that the Pi hasn't upgraded the GPU on their different hardware releases |
03:03.19 | macsforme | oh, and tessellating faces for lighting |
03:03.27 | blast007 | it's the same thing that came with their original 700MHz single core |
03:04.00 | blast007 | might be clocked a bit faster and have access to more memory (since they're packing more RAM on), but still the same GPU |
03:05.41 | JeffM | the low end GPUs also don't do transforms on the GPU just triangle fills, they will be CPU bound by geometry. |
03:07.11 | macsforme | even transformations in the vertex shader? |
03:07.28 | JeffM | yeah, if the GPU can't do it, the driver has to have the CPU do it |
03:07.43 | JeffM | some low end GPUs ONLY accept screen space triangles |
03:08.04 | JeffM | the driver provides abstraction for all of GL, even shaders if it has to |
03:08.24 | JeffM | that's how intel drivers got by with shaders when the hardware couldn't do it all. |
03:08.55 | JeffM | that's the nice thing about GL, you arn't supposed to care about who's doing the work of the drawing ;) |
03:09.21 | JeffM | you'll notice a lot of the low end linux drivers use parts of mesa, that's for the stuff the CPU has to do. |
03:09.46 | JeffM | just swap out the stuff your GPU can do with hardware calls, leave the rest up to the CPU. |
03:12.23 | JeffM | You've never used a one of the 3D "Decellerator cards" have you? |
03:13.39 | blast007 | my S3 Savage IX was one of those :D |
03:13.49 | JeffM | yeah S3 had a line of them |
03:13.52 | blast007 | and not only that, but it made stuff *look* worse too! |
03:14.00 | JeffM | if you had a CPU better then a Pentium Pro, it was faster to use the CPU |
03:14.42 | blast007 | they managed to somehow screw up texture indices, because I recall in BZ the gound was white, there was grass in place of one mountain, and a mountain was flying overhead |
03:14.58 | JeffM | yeah, it wasn't a compatible GL impleentation |
03:15.16 | JeffM | first card I had that worked well with BZ was a permedia2 |
03:15.39 | JeffM | Hardware T&L FTW |
03:16.57 | blast007 | S3 Virge 3-D, one of the first 3D decellerators |
03:17.31 | blast007 | looks like I could have upgraded my copy of Descent II to the "accelerated" version for only $4.95 shipping and handling! |
03:17.50 | JeffM | allow 4 to 6 weeks for delivery |
03:17.58 | blast007 | yes, it does say that too :) |
04:17.37 | rodie | cmos battery included! |
04:25.41 | rodie | oh my, my cfg had kierra silenced. must have accidently hit the k key;) |
04:26.10 | kierra | :( |
04:26.17 | kierra | I"m a nice girl, I am |
04:26.18 | rodie | sorry about that! |
04:26.24 | kierra | no worries |
04:26.40 | kierra | on bzflag or here? |
04:27.10 | rodie | i'll script it to install a fresh cfg file each invocation. sheesh;) |
04:27.46 | rodie | how do you know you're nice? |
04:27.48 | rodie | how do we know? |
04:27.49 | rodie | ;) |
04:29.11 | rodie | watching those vids now, between kitty feedings. |
05:24.31 | kierra | hmm, how do i know? |
05:24.41 | kierra | cuz i try to be |
05:26.28 | kierra | "I'm a nice girl, I am" a quote from My Fair Lady |
08:21.21 | *** join/#bzflag zuii (~cauchy@BC06BF93.catv.pool.telekom.hu) |
10:28.48 | *** join/#bzflag Leppp (~Leppp@cpe-24-94-29-105.san.res.rr.com) |
11:23.42 | *** join/#bzflag zuii (~cauchy@BC06BF93.catv.pool.telekom.hu) |
12:11.43 | *** join/#bzflag zuii (~cauchy@BC06BF93.catv.pool.telekom.hu) |
12:55.12 | *** join/#bzflag zuii (~cauchy@BC06BF93.catv.pool.telekom.hu) |
14:12.11 | *** join/#bzflag Void7 (~Void7@205.157.147.198) |
15:10.17 | *** join/#bzflag JeffM (uid48618@gateway/web/irccloud.com/x-bgifllheehjibpyh) |
15:20.37 | the_map | <<blast007>> it's a single core 1GHz MIPS processor with 512MB DDR2 RAM |
15:20.58 | the_map | I just looked up the specs of a laptop I got lubuntu running on |
15:21.35 | the_map | 1.6 GHz processor, 512 MB DDR SDRAM |
15:22.21 | the_map | it's a dell inspiron 600m |
15:22.45 | the_map | released 3/12/03 |
15:25.33 | the_map | the processor is single core |
15:41.43 | *** join/#bzflag zuii (~cauchy@BC06BF93.catv.pool.telekom.hu) |
16:07.23 | blast007 | a MIPS core isn't as fast as an x86 core at the same frequency |
16:15.25 | *** join/#bzflag KaadmY (uid135503@gateway/web/irccloud.com/x-nvbfqgfozrpzqgrk) |
16:15.50 | blast007 | at least I think - can't actually find any CPU benchmarks for that SoC |
16:16.21 | JeffM | 1.6 is probalby a celly or an atom, it's going to have a lot more power then a mips, and use a lot more power |
16:17.02 | blast007 | based on the age, might even be an early Pentium M |
16:17.11 | JeffM | those were good CPUs |
16:17.17 | JeffM | way better then the P4s |
16:17.22 | blast007 | yep |
16:17.29 | the_map | of course, the gcw isn't trying to run a desktop OS, either |
16:17.34 | JeffM | they are what become the core line |
16:18.34 | the_map | the laptop is a dell inspiron 600m |
16:18.45 | the_map | it has a pentium m |
16:19.07 | JeffM | that's going to have a lot more power than a SOC |
16:19.15 | JeffM | and a better GPU |
16:19.33 | JeffM | unless you are comparing it to modern phones |
16:19.37 | the_map | and DDR1 memory :) |
16:19.48 | JeffM | better then SDR memory |
16:20.04 | blast007 | the_map: might have a wider memory bus than the SoC though |
16:20.20 | the_map | JeffM: it has 512 MB DDR SDRAM |
16:20.25 | JeffM | ahh |
16:20.35 | the_map | it isn't *better*, it *is* :P |
16:20.47 | JeffM | yeah, lappy probaby has a 64 bit bus at a higher clock |
16:21.02 | JeffM | SOC of that age proalby has a 32 bit bus |
16:21.23 | blast007 | the Pentium M's also had a lot of cache |
16:21.38 | JeffM | also lappy will have dedicated hardware for dealing with 64 bit floats, SOC probalby has to do it in software. |
16:21.49 | the_map | I would just like to know why I can install lubuntu on a 13 yr old laptop, but not ubuntu on a couple year old computer with an SSD |
16:22.04 | JeffM | FPU performance on SOC has not been good traditionally. |
16:23.14 | blast007 | the 700MHz ARM single-core Raspberry Pi had performance similar to a Pentium II 300MHz (but a much better GPU) |
16:24.19 | the_map | if you guys are wondering, specs for the laptop are at http://www.cnet.com/products/dell-inspiron-600m-for-home-pentium-m-1-4ghz-512mb-30gb-series/specs/ and http://www.notebookreview.com/notebookreview/dell-inspiron-600m-review-pics-specs/ |
16:25.03 | JeffM | yeah that's a generic craptop |
16:26.32 | blast007 | https://www.raspberrypi.org/help/faqs/#topCost |
16:27.20 | JeffM | I have one of the 9$ computers, it's hella slow |
16:27.43 | blast007 | the Pi Zero? or you talking about the CHIP thing? |
16:27.49 | JeffM | the CHIP thing |
16:28.10 | JeffM | the video out sucks |
16:29.31 | blast007 | don't worry, for another $15 you can get HDMI, or $10 for VGA |
16:30.25 | JeffM | didn't buy it for video, so I don't really care. I was just surprised how flaky the analog signal was. |
16:43.40 | macsforme | someone in the GCW channel did mention that there's no hardware support for doubles, so that math takes twice as long as with floats... that may indeed be one of the big performance barriers |
16:44.24 | macsforme | but not something I can easily verify |
16:47.59 | *** join/#bzflag Void7 (~Void7@205.157.147.198) |
16:53.04 | blast007 | "XBurst® FPU instruction set supporting both single and double floating point format which are IEEE754 compatible" |
16:57.26 | *** join/#bzflag bryjen (~bryjen@cpe-76-92-80-32.mi.res.rr.com) |
17:01.37 | macsforme | at the same speed? maybe I didn't explain it correctly, but he said doubles are twice as slow as floats |
17:01.50 | blast007 | could be |
17:06.04 | blast007 | hi bryjen |
17:07.20 | bryjen | hola |
17:19.42 | JeffM | Yeah, you can support doubles on hardware that only does floats but it takes more code and cycles |
17:20.10 | JeffM | You can even fake an flu in software on an integer only cpu |
17:20.28 | JeffM | Fpu not flu ð¬ |
19:52.11 | *** join/#bzflag Void7 (~Void7@205.157.147.198) |
21:28.33 | *** join/#bzflag zuii (~cauchy@BC06BF93.catv.pool.telekom.hu) |