| 00:51.59 | *** join/#brlcad Twingy (n=justin@74.92.144.217) | |
| 01:45.36 | Maloeran | Hey Erik, I understand you spoke to Lee about my "needs" for multi-cache boxes, who spoke to Mark, who came back to me speaking of sending a computer?... |
| 01:46.28 | Maloeran | Is there some hope of seeing that old box at home of yours put online? That would be easier and less wasteful |
| 02:36.47 | ``Erik | the sun? that'd take a few hundred bucks to get operational, I think |
| 02:36.56 | ``Erik | it's been gutted... no cpu, no ram, no drive |
| 02:38.12 | ``Erik | <-- can't give you an account on any of his work machines due to policy... if mark has access, I d'no if he'd be able to give you an account |
| 03:12.14 | Twingy | I have an ultra sparc that's operational |
| 03:15.29 | Maloeran | Neat Justin, how many processors in there? |
| 03:16.56 | Maloeran | Erik, it's apparently more expensive in man hours to type "adduser" on a box than purchase a new one. I haven't figured that one yet |
| 03:17.40 | Twingy | 200 MHz ultrasparc 1 w/ 128MB RAM |
| 03:18.50 | Maloeran | One processor? |
| 03:19.04 | Twingy | correct |
| 03:19.56 | Maloeran | Eheh, thanks though unfortunately the point is mostly to be able to test code on many cores |
| 03:20.33 | Maloeran | To find bugs due to non-shared cache, work on scalability, explore NUMA memory allocation, etc. |
| 03:27.55 | Maloeran | Ah, Bush considers "more marines" as the solution in Iraq. When will someone explain to him the difference between marines and peacekeepers, social workers? |
| 06:38.09 | *** join/#brlcad clock_ (i=clock@84-72-60-44.dclient.hispeed.ch) | |
| 06:42.11 | *** join/#brlcad clock__ (i=clock@84-72-60-44.dclient.hispeed.ch) | |
| 08:15.19 | *** join/#brlcad clock_ (n=clock@zux221-122-143.adsl.green.ch) | |
| 09:41.29 | *** join/#brlcad docelic (n=docelic@212.15.185.185) | |
| 09:53.34 | *** join/#brlcad clock_ (n=clock@zux221-122-143.adsl.green.ch) | |
| 10:40.06 | *** join/#brlcad docelic (n=docelic@212.15.185.185) | |
| 14:04.16 | *** join/#brlcad SWPadnos (n=Me@dsl245.esjtvtli.sover.net) | |
| 21:36.26 | ``Erik | must be one hell of a breakfast |
| 21:59.20 | *** join/#brlcad docelic (n=docelic@212.15.185.176) | |
| 23:32.45 | *** join/#brlcad Music_Shuffle (n=kumar@c-71-59-32-135.hsd1.ga.comcast.net) | |
| 23:32.48 | Music_Shuffle | Hi. |
| 23:32.53 | Music_Shuffle | Anyone here? |
| 23:33.13 | ``Erik | nope |
| 23:33.17 | Music_Shuffle | ./Ding. |
| 23:33.22 | Music_Shuffle | A quick CAD hardware question. |
| 23:33.53 | Music_Shuffle | Why is CAD so...demanding? |
| 23:33.59 | Music_Shuffle | Sorry, was wondering about the phrasing :P |
| 23:34.18 | Music_Shuffle | Hardware wise. |
| 23:34.18 | brlcad | you mean just in general? |
| 23:34.21 | Music_Shuffle | Yeah. |
| 23:34.40 | Music_Shuffle | Like workstations for CAD have huge amount of RAM, SCSI HDD's, Xeon's, etc, etc. |
| 23:35.00 | *** join/#brlcad Twingy (n=justin@74.92.144.217) | |
| 23:35.04 | Music_Shuffle | I've never used CAD software, but I've always wondered why it requires so much...power. For lack ofa better term. |
| 23:35.04 | brlcad | it's not incredibly demanding, imho, but in general the models are massive and the computation requirements to process those models are more than you have on a workstation |
| 23:35.30 | Music_Shuffle | Oh, so you're saying that it just ends up being so huge, that regardless of complexity, you need a lot of power just to be able to manipulate it? |
| 23:35.47 | brlcad | even compared to something like a game, CAD models (at least "real" ones) are intensely more complicated than any gaming model, plenty won't fit onto OpenGL memory |
| 23:36.06 | Music_Shuffle | Err...more complicated how? |
| 23:36.11 | brlcad | right, manipulation, visualization, consistency/topology checking |
| 23:36.37 | Music_Shuffle | Oh right, because the game physics don't have to actually WORK. |
| 23:36.39 | ``Erik | I d'no if I'd call a xeon a workstation |
| 23:36.39 | Music_Shuffle | Mm. |
| 23:36.48 | Music_Shuffle | Why not? >.> |
| 23:36.48 | brlcad | well in a game, you only have to care that something looks nice -- there's lots of tricks to get that result -- in CAD you have all that same external detail, if not more, as well as all of the internal geometry |
| 23:37.01 | ``Erik | in general, "workstations" historically have bunches of ram and scsi disk, been that way since workstations were kinda born... |
| 23:37.03 | brlcad | which might be wires, bolts, nuts, pipes, engine blocks, etc |
| 23:37.13 | ``Erik | wintel boxes and other "home toys" were HIGHLY abnormal in the grand scheme of things |
| 23:37.45 | Music_Shuffle | So most of today's CAD systems run what? |
| 23:38.01 | ``Erik | huh? |
| 23:38.20 | Music_Shuffle | CPU-wise. Is it mostly Xeons, Opterons, any specific? |
| 23:38.46 | ``Erik | most have a set of archs they run on... x86 tending to be included these days |
| 23:39.18 | brlcad | also when you're dealing with solid modeling CAD, which is a subset of CAD but one of the largest domains, the software has an additional workload of repeatedly evaluating the geometry to make sure it's topologically sound, that parametric contraints are upheld/updated if they exist, that overlaps/interferences are detected, that you don't end up with non-solid geometry |
| 23:39.47 | ``Erik | too many syllables, I'm on vacation o.O :D back to simpsons for me |
| 23:39.56 | Music_Shuffle | Lol! |
| 23:40.04 | archivist | and interference checking |
| 23:40.09 | Music_Shuffle | Oh ok. |
| 23:40.42 | Music_Shuffle | So pretty much its caused by needing to model reality in its entirety using software. |
| 23:41.20 | archivist | oops missed it |
| 23:41.37 | ``Erik | I mean, graphic artists (like the people making posters and books and stuff) tend to have gobs of ram on their desktop |
| 23:42.06 | brlcad | i wish these laptops went higher than 2GB.. i'm constantly out of memory |
| 23:42.12 | ``Erik | brlcad: how many gigs did you blow over with a typesetting program? |
| 23:43.40 | ``Erik | (and video games tend to punt on huge parts of the dataset to make it usable on lesser systems... I mean, take a look at 'roam' and 'vipm' and how they are designed to page off of disk so only the closest stuff is a high LOD) |
| 23:43.52 | ``Erik | especially roam o.O heh |
| 23:44.14 | Music_Shuffle | lol |
| 23:44.27 | ``Erik | not to mention other hacks like faking normals on highly decimated meshes so they kinda sorta appear more complex |
| 23:44.37 | brlcad | swap is currently at just under 15GB.. looks like the book i'm working on is using up a little over 3GB |
| 23:45.00 | Music_Shuffle | Fif...teen? |
| 23:45.04 | ``Erik | (it's all moving, so the human brain fixes the fuckups in videogames... in manufacturing or analysis, the fuckups are unacceptable) |
| 23:45.34 | Music_Shuffle | Right, that nasty bit about stuff breaking, falling, going boom, etc. |
| 23:46.20 | brlcad | yep, 15... and I don't even have a CAD model open right now .. that'd be a few more GB to worry about :) |
| 23:46.47 | Music_Shuffle | Mmm...what do you do with 15 gigs of swap? >_< |
| 23:46.53 | Twingy | <PROTECTED> |
| 23:47.35 | brlcad | though mged is pretty light on the memory usage overhead.. uses just a little more in memory than it takes to store it on disk, maybe 2X-10X disk size if it's going to be analyzed or raytraced, depending on the model |
| 23:48.38 | brlcad | Music_Shuffle: mostly a ton of image processing right now -- putting together a print-quality book, lots of high-resolution imagery, several image processing apps in the workflow |
| 23:48.48 | Music_Shuffle | Ahh ok |
| 23:48.51 | Music_Shuffle | You're writing a book? :o |
| 23:49.05 | brlcad | and of course this terminal running an irc client.. gotta be adding at least 10MB to the mix *ahem* |
| 23:49.06 | ``Erik | "20 years of playboy centefolds" |
| 23:49.13 | brlcad | woot |
| 23:49.19 | Music_Shuffle | Lol |
| 23:49.22 | ``Erik | that's the book, right? :D |
| 23:49.24 | brlcad | hmm.. 20 years.. |
| 23:49.45 | ``Erik | 240 girls too much? |
| 23:50.05 | ``Erik | I mean, a couple pages each... sounds like a lovely coffee table book... |
| 23:50.12 | ``Erik | perfect gift for friends and family |