00:12 *** efc has joined #boinc
00:14 <PovAddict> "Actually writing public speeches is easy for me. The hard part is shutting up once I start speaking."
00:15 <efc> Moo
00:15 <efc> &wx 30625
00:15 <Romulus> efc: Temperature: 70°F / 21°C | Humidity: 83% | Pressure: 30.14in / 1020hPa (Falling) | Conditions: Haze | Wind Direction: ENE | Wind Speed: 5mph / 7km/h ; Rest of Tonight - Cloudy with a chance of showers and a slight chance of thunderstorms late...then a chance of showers. Lows in the lower 60s. East winds 10 to 15 mph. Chance of rain 40 percent.; Friday - Showers likely with isolated (1 more message)
00:15 <efc> &more
00:15 <Romulus> efc: thunderstorms. Highs around 80. East winds 10 to 15 mph. Chance of rain 60 percent.; Friday Night - Mostly cloudy with a chance of showers and isolated thunderstorms in the evening...then a chance of showers after midnight. Lows in the lower 60s. East winds 5 to 10 mph. Chance of rain 40 percent.;
00:15 <efc> weather feels weird here.. animals don't like it
00:25 <jasong> You know when the Linux disc says "starting install process. This may take several minutes."?
00:25 <jasong> It's been 36 minutes and it's not even halfway through that little progress bar
00:26 <jasong> The install hasn't even started yet
00:26 <PovAddict> I'd say it's stuck
00:26 <PovAddict> and there is no such thing as a "Linux disc"; what distro are you installing?
00:29 <jasong> Fedora 10, and it's been very slowly advancing, judging from the status bar
00:29 <PovAddict> how much RAM?
00:30 <PovAddict> last time I tried installing Fedora on a VM, it'd just hang on the installer
00:30 <PovAddict> so I rebooted, tried again, and it hanged even earlier -.-
00:30 <jasong> 8Gig
00:30 <PovAddict> installing on the physical machine?
00:30 <jasong> yeah
00:30 <PovAddict> 64-bit machine, 64-bit Fedora?
00:31 <jasong> yeah
00:31 <jasong> the first quad-core cpu Intel made
00:31 <jasong> I think
00:31 <jasong> e6600
00:31 <jasong> err
00:31 <jasong> q6600
01:26 *** yoyo[RKN] has joined #boinc
01:35 *** synapt has joined #boinc
01:38 *** MTughan is now known as MTughan_
01:39 *** MTughan has joined #boinc
02:13 *** [B^S]renemayer has joined #boinc
02:13 *** [B^S]renemayer has quit IRC
02:14 *** [B^S]renemayer has joined #boinc
03:38 <Tank_Master> gn ye all
03:54 <yoyo[RKN]> moin
03:55 <MTughan> 'lo yoyo
03:55 <yoyo[RKN]> for a new boinc project, should I download the stable branch or the latest one?
03:55 <MTughan> No idea.
03:56 <yoyo[RKN]> would be good to know how old the stable branch is
04:04 <yoyo[RKN]> ok, I'm using the stable branch
04:04 <yoyo[RKN]> save is save
04:04 <MTughan> That would be "safe".
04:04 <yoyo[RKN]> ahh, right
04:09 <yoyo[RKN]> at least it compiled
04:16 *** xionyc has quit IRC
04:17 *** XioNYC has joined #boinc
04:25 *** mejla_ has joined #boinc
04:49 *** desti_T2 is now known as desti
04:52 <FreeLarry58> humm what a good (mid range for cost) vid card for cuda processing?
04:53 <FreeLarry58> 9400 9600 9800 or other
05:01 *** [B^S]renemayer has quit IRC
05:02 *** [B^S]renemayer has joined #boinc
05:21 *** efc has quit IRC
05:53 *** mejla_ has quit IRC
06:26 <hawmps> mornin
06:26 <MTughan> 'lo Jay
06:42 *** Txt_file has joined #boinc
06:43 *** Txt_file has left #boinc
06:45 *** mejla has joined #boinc
07:16 *** mejla has quit IRC
07:17 *** mejla has joined #boinc
07:21 *** mejla has quit IRC
07:22 *** mejla has joined #boinc
07:23 *** mejla has quit IRC
07:24 *** mejla has joined #boinc
07:38 *** [B^S]renemayer has quit IRC
07:47 *** mejla has quit IRC
07:47 *** mejla has joined #boinc
07:55 *** RomW has joined #boinc
07:59 <MTughan> 'lo Rom
07:59 <MTughan> Up bright an early?
07:59 <MTughan> and*
08:16 <RomW> morning
08:17 <RomW> Was getting ready to put together a new private drop
08:17 <RomW> But...
08:17 <RomW> A URL is not handing out cookies anymore
08:19 *** Jackygrahamez has joined #boinc
08:20 <Jackygrahamez> anyone going BBQing this weekend? :-)
08:23 *** MTughan has quit IRC
08:25 *** MTughan has joined #boinc
08:27 *** wdsmia_w has joined #Boinc
08:39 <wdsmia_w> &wx 50319
08:39 <Romulus> wdsmia_w: Temperature: 58.9°F / 14.9°C | Humidity: 79% | Pressure: 30.18in / 1021.9hPa (Rising) | Conditions: Mostly Cloudy | Wind Direction: NE | Wind Speed: 0.0mph / 0.0km/h ; Today - Mostly cloudy with a 20 percent chance of showers. High in the mid 70s. East wind 10 to 15 mph.; Tonight - Partly cloudy until early morning then becoming mostly cloudy. A 20 percent chance of light showers after (1 more message)
08:39 <wdsmia_w> &more
08:39 <Romulus> wdsmia_w: midnight. Low in the upper 50s. Southeast wind near 5 mph.; Saturday - Mostly cloudy. A chance of showers and a slight chance of thunderstorms in the morning...then a chance of thunderstorms in the afternoon. High in the mid 70s. Southwest wind 5 to 10 mph shifting to the north in the afternoon. Chance of precipitation 50 percent.;
09:17 <hawmps> already 70 here..
09:31 *** ELGono has joined #boinc
09:44 *** MacG has joined #boinc
10:08 *** DerMeister has joined #boinc
10:38 <PovAddict> SETI screensaver graphics don't work correctly on a compositing window manager either
10:44 *** ELGono has quit IRC
10:46 <Jackygrahamez> ;-(
10:46 *** Aeternus has joined #boinc
10:51 *** Aruzo has joined #boinc
11:02 <Jackygrahamez> &wx 20001
11:02 <Romulus> Jackygrahamez: Temperature: 73.9°F / 23.3°C | Humidity: 64% | Pressure: 30.17in / 1021.6hPa (Falling) | Conditions: Mostly Cloudy | Wind Direction: SE | Wind Speed: 4.0mph / 6.4km/h ; Today - Sunny. Highs in the mid 80s. South winds around 5 mph.; Tonight - Partly cloudy. Lows in the lower 60s. South winds around 5 mph.; Saturday - Mostly sunny. Highs in the mid 80s. South winds 5 to 10 mph.;
11:02 <Jackygrahamez> &more
11:02 <Romulus> Jackygrahamez: Error: You haven't asked me a command; perhaps you want to see someone else's more. To do so, call this command with that person's nick.
11:08 <Jackygrahamez> Now I have finally seen a feature in Safari that makes me think it is better than any other browsers
11:08 <MTughan> And that would be?
11:09 <Jackygrahamez> I created a submission form for GROMACS that has some text boxes, there is a little thing on the lower right corner that allows me to make a bigger field :-)
11:11 <Jackygrahamez> the other interest thing about the mac platforms I have running autodock, they have peformed slightly better than Windows over the last 24 hours
11:12 <Jackygrahamez> as a mater of fact, we have zero errors from PowerPC
11:12 <Jackygrahamez> not many in that popution though
11:12 <Jackygrahamez> population :-)
11:13 <Jackygrahamez> so can't say anything with statistical signifcance
11:15 <Jackygrahamez> Someone want to get me a Tesla Super computer for my birthday?
11:23 <XioNYC> &wx LGA
11:24 <Romulus> XioNYC: Temperature: 75°F / 24°C | Humidity: 51% | Pressure: 30.13in / 1020hPa (Falling) | Conditions: Mostly Cloudy | Wind Direction: WSW | Wind Speed: 6mph / 9km/h ; Today - Mostly sunny. Highs from the upper 70s to the lower 80s. Southwest winds 10 to 15 mph.; Tonight - Mostly cloudy with a slight chance of showers and thunderstorms. Lows in the lower 60s. Southwest winds 10 to 15 mph. Chance of (1 more message)
11:24 <XioNYC> &more
11:24 <Romulus> XioNYC: rain 20 percent.; Saturday - Partly sunny with a slight chance of showers and thunderstorms. Highs in the lower 70s. East winds 5 to 10 mph... becoming southeast in the afternoon. Chance of rain 20 percent.;
11:26 <XioNYC> Jackygrahamez: I couldn't help but notice your "zero errors from PowerPC" in my text buffer... pray, tell: which project are you talking about?
11:28 <Jackygrahamez> Hydrogen@Home
11:29 <Jackygrahamez> How's NYC treating you?
11:29 <XioNYC> Employment issues aside, nice 'n good.
11:30 <Jackygrahamez> that's my home ground, living in DC is like being in acoma
11:32 <XioNYC> Yeah... once you've lived amongst the Eight Million, odds are that anywhere else will be "dead" by comparison ;-)
11:32 * XioNYC does not intend offense to anyone living in "dead" cities.
12:06 <PovAddict> X server hung again (nvidia driver), gotta reboot
12:06 <PovAddict> (sshing from ipod now)
12:10 <wdsmia_w> way to nice outside to be stuck here working
12:11 * wdsmia_w thinks he will start his vacation early
12:13 <PovAddict> whoa, it recovered
12:13 <PovAddict> first time I recover from this particular problem without a reboot
12:16 *** wdsmia_w has quit IRC
12:23 *** efc has joined #boinc
12:25 <efc> moo
12:45 <PovAddict> hmm nope, my X is still messed up...
12:45 *** PovAddict has quit IRC
12:48 <Tank_Master> arent most ex's messed up?
12:50 <wdsmia> re-hi
12:50 <efc> Moo
12:50 <Tank_Master> re-wb
12:51 <wdsmia> Its so nice I took the rest of the month off :)
12:59 <RomW> Howdy
13:06 <Tank_Master> howdy
13:12 *** PovAddict has joined #boinc
14:00 <PovAddict> ...is that a joke?
14:00 <PovAddict> even if the claim is true (that raising animals causes greenhouse emissions and global warming)...
14:00 <PovAddict> "save the planet - go vegetarian"? they can't be serious, that's the kind of thing I'd expect from The Onion!
14:02 <DerMeister> <PovAddict> even if the claim is true (that raising animals causes greenhouse emissions and global warming)... <= well, that is true. however, there are much bigger problems (like the whole fossil oil and coal industry)
14:02 <Tank_Master> or forest fires, volcanoes...
14:02 <DerMeister> Even though methan is fourty times (or something like that) as bad as CO2
14:02 <DerMeister> Tank_Master: true, but we don't have much influence on that ;)
14:03 <Tank_Master> I am suprised the envirmentels tone build a CO2 scrubber for volcanoes
14:03 <Tank_Master> dont*
14:03 <DerMeister> lol
14:03 <Tank_Master> or uh, allow us to actually put out forest fires before they get to be big...
14:03 <PovAddict> even IF animals were the main cause of global warming, which they're not, that activism page would still be ridiculous
14:03 <PovAddict> but all of peta2.com is about vegetarian activism, so...
14:04 <DerMeister> Well, the question would be: where to deposit the CO2? ;)
14:04 <Tank_Master> ship it to mars
14:04 <DerMeister> PovAddict: that's true
14:05 <FreeLarry58> they do tend to forget one small point - in the ecosystem man is also an animal
14:06 <PovAddict> and also breathes and farts
14:06 *** Jackygrahamez has quit IRC
14:06 <FreeLarry58> as all animals do or anything that lives and uses chemical dgestive system for fuel
14:07 <desti> but you are eating much more than one animal in your life
14:07 <FreeLarry58> so does everything of one type or another
14:08 <FreeLarry58> a belive the term food chain was coined somewhere in ancient history to discribe the process
14:09 <Romulus> Title: QDB: Quote #792343 (at bash.org)
14:11 <Tank_Master> page no worky
14:12 <FreeLarry58> worked for me Tank_Master
14:12 <Tank_Master> I get "page cannot be displayed"
14:13 <Tank_Master> huh
14:13 <Tank_Master> works in ff, but not IE
14:14 <DerMeister> Tank_Master: things like fossil oil and coal don't belong to the food chain and shouldn't be used as they are indeed real cliamte killers ;)
14:15 <Tank_Master> ?
14:16 <DerMeister> Ah, sorry... FreeLarry58 mentioned the food chain, not you :)
14:16 <Tank_Master> a) I never said anything about fossil files, coal, or the food chain; and b) firest fires and other natural things like volcanoes produce for more greenhouse gasses than anything we do
14:17 *** Rebirther has joined #boinc
14:17 <DerMeister> ... still we produce too much ;)
14:17 <FreeLarry58> true Tank_Master and they lock up old co2 in them during thier life and death processes
14:17 <Tank_Master> its insignagent to what nature produces
14:17 <DerMeister> I don't think so.
14:17 <FreeLarry58> humm this may turn out to be a good machine after all http://boincstats.com/stats/boinc_host_graph.php?pr=bo&id=5474134
14:18 <PovAddict> "Hey Freud, I want to introduce you to 4chan."
14:18 * FreeLarry58 isn't that a p*rn channel name
14:18 <Tank_Master> lol
14:18 <PovAddict> FreeLarry58: no, it's not pr0n, but if you don't know what it is better for you
14:19 <FreeLarry58> k - just reminds me of something
14:19 <FreeLarry58> must be the chan ending
14:20 <PovAddict> well it does have a pr0n category or two, I guess...
14:20 <FreeLarry58> lol
14:21 *** saenger has joined #boinc
14:22 *** Aeternus has quit IRC
14:26 <FreeLarry58> more than a couple PovAddict
14:26 <FreeLarry58> but your right - not a site for me
14:28 *** si has joined #boinc
14:35 *** Ageless93 has joined #boinc
14:40 *** PovAddict has quit IRC
14:40 *** PovAddict has joined #boinc
15:29 *** Rebirther has quit IRC
15:30 *** diepchess has joined #boinc
15:32 <diepchess> hi
15:32 <diepchess> primegrid also here?
15:38 *** qnr has quit IRC
15:41 *** saenger has quit IRC
15:55 <diepchess> hi
16:05 <xcamel> not specifically..
16:05 <diepchess> ah
16:06 <diepchess> i can't login in primegrid
16:06 <xcamel> all projects welcome
16:06 <diepchess> not sure whether registration failed or whatever
16:06 <MTughan> What is BOINC saying?
16:06 <diepchess> Paul Underwood approached me
16:06 <diepchess> he wants to setup a new project
16:06 <diepchess> finding titanic AP9
16:07 <infinisoft> vincent?
16:07 <diepchess> y
16:08 <diepchess> you can do 'whois'
16:08 <Ageless93> logged out of primegrid and back in without a problem.
16:20 *** mejla has quit IRC
16:25 * FreeLarry58 wonders if he should ask opinions on vid cards for cuda
16:26 * PovAddict doesn't do CUDA
16:26 <wdsmia> only if you want 25 different answers
16:26 <Tank_Master> the higher the number on the card, the faster is crunches
16:27 <Tank_Master> go out and get yourself a 9800 and ill just keep using my poky old 295
16:27 <PovAddict> like GHz right?
16:27 <Tank_Master> :P
16:27 <PovAddict> (sarcasm)
16:28 <PovAddict> ah! I see what you did there
16:28 <Tank_Master> hehe
16:32 <efc> Anybody gonna buy that quad-card motherboard somebody posted earlier... An Asus I think
16:32 <FreeLarry58> nah just thinking upgrading one or two machines and looking for something in 100-150 range
16:33 <FreeLarry58> saw some 9600's going on sale for 125
16:33 <efc> I could use a new setup myself, old GF4 is getting long in the tooth
16:33 <FreeLarry58> but store trying to steer me to ati's told them nope - no cuda support
16:33 <FreeLarry58> yeah only have a couple of 8400/8500 here
16:34 <FreeLarry58> and a number a lot less
16:34 <efc> I imagine eventually there will be a more "open" api than CUDA but not yet that I know of
16:34 <DerMeister> I don't think this will happen any time soon
16:35 <PovAddict> OpenCL
16:35 <PovAddict> I think we're currently in the same mess that 3D graphics were before DirectX and OpenGL
16:35 <FreeLarry58> so best bet staying with nvidia for now - at least for crunching purposes -- ati i think has a little edge in gaming apps tho
16:35 <DerMeister> ... and it's still a mess
16:35 <Ageless93> am I the only one using his videocard for games and just generally seeing things happen on my monitor and TV? ;)
16:36 <FreeLarry58> nope Ageless93 most of mine to old to support crunching
16:36 <PovAddict> Ageless93: I use mine to show stuff on screen
16:36 <FreeLarry58> heck this machine still uses agp
16:36 <PovAddict> with efficient transparency and shadows and other effects :)
16:36 <Ageless93> so does mine. why I got a HD3850 at the time.
16:37 <PovAddict> even if this card supported CUDA, I think my desktop environment alone doesn't leave much processing time available for crunching
16:37 <PovAddict> and it heats enough as it is thank you very much
16:37 <Ageless93> not to go crunch with it at Milkyway. I tried that, didn't like that I couldn't use my computer and stopped doing that.
16:37 <PovAddict> "Nvidia released its first graphics card, the NV1, in 1995. [...] However, the NV1 struggled in a market-place full of several competing proprietary standards."
16:38 <FreeLarry58> true PovAddict have notices anytime i strike a key or move mouse gpu type wu stop till ikle for 3 minutes so its not for a work machine
16:38 <PovAddict> "Market interest in the product ended when Microsoft announced the DirectX specifications"
16:38 <efc> NV1 was a pretty big mess in general
16:38 <efc> fairly impressive that it didn't put them out of business
16:39 <PovAddict> efc: NV2 was worse: millions of dollars involved and never became a product
16:39 <PovAddict> although the money came from Sega, not from nVidia themselves
16:39 <efc> Don't think I heard of that one, but no suprise if it didn't get released hehe
16:39 * FreeLarry58 sees clock on wall saying commute time
16:39 <FreeLarry58> &tca
16:39 <Romulus> FreeLarry58: Are you daft?
16:39 <FreeLarry58> &bye ?
16:39 <Romulus> later ace
16:40 <PovAddict> "Sega eventually realized the flaws in implementing quadratic surfaces, and the NV2 was never fully developed"
16:40 <efc> Wonder where that quadratic/higher order stuff might have gone
16:42 <efc> I also find it amazing that the cards are or can be full floating point now. At the transistor level that is vastly more work.
16:43 <PovAddict> didn't they always use floating point?
16:48 *** Ageless93 is now known as Ageless93[Away]
16:48 * Ageless93[Away] is now away - Reason : Wouldn't you want me to be away?
16:49 <efc> I meant for the pixel format
16:49 <efc> floating point color
16:49 <infinisoft> diep: Just wondering, I'm a RWT lurker that's why :)
16:50 <efc> I'd assume the vertex processing has always been floating point, even on the card
16:52 <efc> double precision floating point pixels seems like overkill to me but of coarse it makes all the CUDA stuff much more useful
16:52 <PovAddict> I don't know exactly how the algorithms work for realtime 3d cards
16:52 <PovAddict> but I know for raytracing double precision is a must
16:55 <efc> it makes a lot of difference if say a color is processed iteratively so that errors build up
16:56 <PovAddict> I'd love to see a physical screen with more than 8 bits per color
16:58 <efc> Yeah right now, you are fairly lucky just to get 8, the cheaper panels only do 6
16:58 <efc> I think some "medical" displays might to 10
16:59 <efc> for major $$
17:46 *** DerMeister has quit IRC
17:52 *** yoyo[RKN] has quit IRC
18:01 *** RomW has quit IRC
18:06 *** Ageless93[Away] is now known as Ageless93
18:06 * Ageless93 is no longer away : Gone for 1 hour 17 minutes 56 seconds
18:29 *** Aruzo has quit IRC
18:40 *** Rookie_69 has joined #boinc
18:40 <Rookie_69> Is it friday yet?
18:40 <PovAddict> yes!
18:41 *** mejla has joined #boinc
18:41 <Rookie_69> Then I'm here on the right day.
18:41 <Ageless93> it's saturday here
18:42 <Rookie_69> Then I'm too late. :-(
18:44 <Rookie_69> Was 30C here today and calling for 15C tomorrow... It's the weekend. At least it isn't calling for rain... yet.
18:44 <Rookie_69> &wx yaw
18:44 <Romulus> Rookie_69: Temperature: 74.5Â°F / 23.6Â°C | Humidity: 46% | Pressure: 29.93in / 1013.4hPa | Conditions: Mostly Cloudy | Wind Direction: WNW | Wind Speed: 10.0mph / 16.1km/h | Updated: 7:38 PM ADT; Tonight - Clear. Increasing cloudiness this evening with 30 percent chance of showers overnight. Wind southwest 20 km/h (12 mph) gusting to 40 km/h (25 mph) becoming light this evening. Low 8C(46F).; (1 more message)
18:44 <Rookie_69> &more
18:44 <Romulus> Rookie_69: Saturday - Sunny with cloudy periods. High 15C(59F) except 9C(48F) along the coast. UV index 8 or very high. Saturday night increasing cloudiness. Showers beginning overnight. Fog patches overnight. Low 7C(45F).; Sunday - Cloudy with 60 percent chance of showers. High 13C(55F).;
18:45 <Rookie_69> Oh yeah... I'm along the coast.
18:46 <Rookie_69> I'll take my jacket with me.
18:54 <diepchess> hi
18:54 <diepchess> openCL will get more people into GPU sure
18:54 <diepchess> but it seems it is slower isn't it?
18:54 <diepchess> as it is already so hard to write a good proggie for CUDA or Brooke/CAL, of course a custom solution is faster and better and appreciated
18:55 <diepchess> if you write for SSE2 you don't 'try' to write it down in C code and pray a compiler is producing great SSE2
18:55 <PovAddict> there's no reason for OpenCL to be slower
18:55 <diepchess> the architecture is so so different AMD vs Nvidia
18:55 <diepchess> well it will be
18:55 <diepchess> the architectures are too different
18:56 <PovAddict> is DirectX slower than the native API of nvidia?
18:56 <diepchess> Nvidia has a gpu with 240 cores that are generic
18:56 <PovAddict> oh wait, there's no native API for graphics
18:56 <diepchess> each core can do multiplyadd or another instruction
18:56 <diepchess> AMD has 5 execution units which each can execute a different instruction
18:56 <diepchess> but not 5 multiplies at a time
18:57 <diepchess> so AMD is more like a 900Mhz 'core2' processor times 160 units (320 in the x2 edition)
18:57 <diepchess> nvidia is in the x2 edition 480 cores that all can execute similar code
18:57 <diepchess> nvidia has blocks of 30 cores and 8 'multicores' a gpu , so 16 in total in x2
18:57 <diepchess> in AMD they're all the same
18:57 <diepchess> openCL doesn't adress that very well
18:57 <diepchess> you can see this indeed in driver programming also
18:58 <diepchess> some stupid old cards that have 1.5Ghz * 256 cores are FASTER in some games in directx/opengl
18:58 <diepchess> than any other card
18:58 <diepchess> despite that these drivers have been optimized for those games very well
18:59 <diepchess> this where graphics parallellized excellent
18:59 <diepchess> parallellizes
18:59 <diepchess> so a 1600 execution unit AMD card or 480 stream processor nvidia card that's real kick butt now
18:59 <diepchess> it's slower than an oldie card as it's clocked higher
19:00 <diepchess> now imagine all kind of number crunching you want to run on those cards
19:00 <diepchess> especially the math guys
19:00 * Tank_Master detects rambeling
19:00 <diepchess> example is a chinese department in Beijing
19:00 <diepchess> they simply programmed in CUDA and get 40% efficiency (which is real high)
19:00 <diepchess> they buy dozens of cards when they can get their hands on them cheap
19:00 <diepchess> (power is for free seemingly)
19:01 <diepchess> when they can get their hands on AMD they'll program and optimize it for AMD
19:01 <diepchess> 'generic' code in openCL is so far away from how you program for multicore environment
19:01 <diepchess> that you EITHER get good performance in gpu's OR in multicores
19:01 <diepchess> not in both
19:02 <diepchess> but opencl will of course 'convince' governments that all the systems are compatible bla bla
19:02 <diepchess> and then that gives rise to more software
19:02 <diepchess> isn't it?
19:03 <Tank_Master> im not convinced that OCL willcatch on if it work slow
19:03 <diepchess> example
19:03 <Tank_Master> it probably wont be used much outside the science comunity anyway
19:03 <diepchess> i speak very premature about trying to get a project going
19:03 <diepchess> fighter jet flight simulator
19:03 <diepchess> at nvidia
19:04 <Tank_Master> will probably use OGL or DX as those are knowen and work already
19:04 <diepchess> government: "no way, cuda is not garantueed to work 4 years from now, amd might be bankrupt"
19:04 <diepchess> openCL will
19:04 <diepchess> in reality you then have extensions of course optimized for different manufacturers AMD, nvidia, etc
19:04 <diepchess> it doesn't pass the paper application
19:04 <diepchess> know what i mean?
19:04 <Tank_Master> anything 3d most likely will not be ported to OCL as that takes time and money
19:04 <diepchess> "not compatible"
19:04 <diepchess> dang
19:05 <diepchess> project not started
19:05 <PovAddict> Tank_Master: not 3D
19:05 <PovAddict> 3D graphics would use OpenGL or DX
19:05 <PovAddict> this is about general computing
19:05 <Tank_Master> agreed
19:05 <diepchess> tank master : YOU HAVE NO CLUE
19:05 <diepchess> it is about performance
19:05 <Tank_Master> no, he was talking about flight sim
19:05 <PovAddict> CUDA / OpenCL
19:05 <Tank_Master> thats 3d
19:05 <diepchess> i was speaking about simulator
19:05 <PovAddict> no
19:05 <diepchess> the ones that many companies use
19:05 <diepchess> army
19:05 <diepchess> and so on
19:05 <PovAddict> a flight sim, and any game really, has two parts:
19:05 <PovAddict> graphics, and simulation
19:06 <MTughan> Graphics and logic*
19:06 <diepchess> even if i would know i would not be allowed to comment on assumptions here
19:06 <diepchess> but all that code is 32 bits
19:06 <PovAddict> a flight sim probably has quite a lot of number crunching in the logic part
19:06 <diepchess> for example in adobe software
19:06 <diepchess> the entire planet is modelled
19:06 <diepchess> in 2 floats
19:06 <diepchess> that's enough for coordinates
19:07 <Tank_Master> I can see science projects using OCL, but not much beyond that
19:07 <Tank_Master> and only if its not slower
19:07 <diepchess> now those calculations are real generic in such software programs
19:07 <diepchess> you'd be amazed how much calculations just GPS is
19:07 <diepchess> now imagine you've got a missile
19:07 <diepchess> it doesn't ask its position like 1 time a second
19:08 <diepchess> aviation and car industry eat huge amounts of system time
19:08 <diepchess> a single tesla of 4 x 240 = 960 cores
19:08 <diepchess> just has a cost of 7500 dollar
19:08 <diepchess> each card can deliver, theoretic spoken many teraflops
19:08 <PovAddict> (get a blog ;) )
19:08 <diepchess> the equivalent 'supercomputer' delivering that is real huge
19:09 <diepchess> some processors here are crunching for wagstaffs
19:09 <diepchess> oldie K8 processors
19:09 <MTughan> Cores aren't everything. You can get 800 core ATI FireGL cards, but getting four of those @3600 cores probably won't do any more work than 4 Teslas.
19:09 <MTughan> Because the ATI cores are clocked slower.
19:09 <diepchess> they deliver, each core like 2.4Ghz * 2 = 4.8 gflop a core
19:09 <diepchess> 4 cores? ==> 9.6 gflop
19:09 <diepchess> a single gpu is alrady 2 Tflop
19:09 <diepchess> single precisioin
19:10 <diepchess> so for simulations or whatever fits in 32 bits floats, that is HUGE crucnhing power
19:10 <MTughan> A Tesla? No. They're only starting to break 1 Tflop theoretical.
19:10 <diepchess> i speak about single precision mtughan
19:10 <MTughan> Yes.
19:10 <diepchess> theoretic spoken :
19:10 <diepchess> 960 cores for a tesla (that's 4 gpu's)
19:11 <diepchess> 960 * 1.2 Ghz * 4 floats * 2 (multiplyadd) =
19:11 <diepchess> a lot
19:11 <MTughan> AMD advertised 1Tflop single precision estimated on their HD4870 cards.
19:11 <diepchess> that's a gpu
19:11 <diepchess> not a dedicated gpgpu
19:11 <diepchess> nvidia cards for gpgpu are worthless other than tesla
19:11 <MTughan> The workstation cards are based on the desktop chipsets.
19:12 <MTughan> AFAIK, anyway.
19:12 <diepchess> based yes
19:12 <diepchess> now i'm no big hardware expert, but there is pixel bla bla in the way
19:12 <diepchess> so you don't have any bandwidth in the gpu's
19:12 <diepchess> as that normally goes to graphics units
19:12 <diepchess> and in gpgpu you just want to use the RAM yourself not display it
19:13 <diepchess> so there is a huge bottleneck that total cripples the cards
19:13 <diepchess> of course if you don't need device RAm, you're safe :)
19:13 <diepchess> to some extend
19:13 <PovAddict> are you planning to do something with a GPU?
19:13 <Tank_Master> watch 3d porn
19:14 <diepchess> you mean if i had the money to do it at home?
19:14 <diepchess> yes
19:14 <diepchess> parameter optimization
19:14 <diepchess> i can need endless crunching power for that
19:14 <PovAddict> or are you just *ranting* about the general state of it?
19:14 <diepchess> ranting?
19:14 <diepchess> i'm telling you they are very powerful
19:14 <diepchess> scroll up
19:14 <diepchess> you pick out 1 sentence out of reality?
19:15 <diepchess> that's the type of guy you are?
19:15 <Tank_Master> o0
19:15 <PovAddict> I didn't say you said they were bad
19:15 <PovAddict> or anything like that
19:15 <PovAddict> I'm just wondering if you have any particular goal related to GPUs, like setting up your own BOINC project based on it
19:15 <diepchess> lookup the word 'ranting' in your dictionary
19:15 <PovAddict> or if you're just talking about it
19:15 <PovAddict> I will; i'll admit I'm not a native English speaker
19:15 <diepchess> yeah we want to setup something
19:16 <diepchess> i'm not native either but i'm not a moron :)
19:16 <MTughan> So you're calling PovAddict a moron?
19:16 <wdsmia> lol
19:16 <diepchess> there is people who like others who say a lot
19:16 <diepchess> there is people who dislike everyone who says a lot
19:16 <diepchess> pick your choice
19:16 <PovAddict> so far you called Tank_Master clueless, MTughan stupid, and me a moron, you won't get much help (or attention) if you keep up like that...
19:17 <MTughan> No, it just seems that you keep going on and on about something that may never see widespread acceptance.
19:17 <diepchess> help?
19:17 <Tank_Master> I call him long winded
19:17 <diepchess> you are telling me now that whatever i had in mind doing i could need you?
19:17 <PovAddict> okay I get you don't need help (eg. with setting up a project)
19:17 <diepchess> or that you *would* be helpful
19:17 <diepchess> oh fly away
19:17 <Ageless93> monologuing
19:17 <Tank_Master> howdy Ageless93
19:18 <Ageless93> hey TM
19:18 <Tank_Master> welcome to today's ranting
19:18 <PovAddict> then kindly explain what your goal is
19:18 <Ageless93> I saw it
19:18 <Tank_Master> hehe
19:19 <diepchess> scroll up
19:19 <PovAddict> what was the purpose of your 110 lines of rambling about GPUs
19:19 <diepchess> and read
19:19 <diepchess> "ranting" "rambling"
19:19 <Tank_Master> same thing
19:19 <diepchess> you didn't even read it
19:19 <PovAddict> I read it, and you seemed to be long-windedly commenting on the performance of GPUs
19:19 <Ageless93> I didn't read it.
19:19 <diepchess> i explain about the power of gpu's compared to multicores
19:19 <Ageless93> not going to scroll up either
19:20 <Ageless93> got better ways to fall asleep
19:20 <PovAddict> yeah I know there's power
19:20 <diepchess> 2 Tflop vs 19.2 gflop here
19:20 <diepchess> and you call that "rambling and ranting about gpu's"
19:20 <PovAddict> I just didn't see anyone ask for an explanation
19:20 <diepchess> oh comeon
19:20 <MTughan> And GPUs are hard to program for ATM because they're geared towards graphics. OpenCL and Larrabee should help.
19:20 <Tank_Master> theres no question GPUs are faster at highly parrelled computing than CPUs, bu I just dont think OCL is the cure-al that some tout it to be
19:21 <xcamel> re-hi
19:21 <diepchess> i explain how projects work at governments and uni's
19:21 <Tank_Master> howdy xcamel
19:21 <MTughan> Correct, GPUs are only really good at embarrisingly parallel problems.
19:21 <diepchess> everything needs to be compatible
19:21 <diepchess> opencl offers that compatibility
19:21 <xcamel> &wx 12074
19:21 <Romulus> xcamel: Temperature: 71.9°F / 22.2°C | Humidity: 62% | Pressure: 30.09in / 1018.8hPa (Rising) | Conditions: Scattered Clouds | Wind Direction: NE | Wind Speed: 4.0mph / 6.4km/h ; Tonight - Partly cloudy. The low in the lower 50s. North winds 5 to 10 mph this evening...becoming light and variable.; Saturday - Mostly cloudy in the morning...then partly sunny with a chance of showers and thunderstorms (1 more message)
19:21 <xcamel> *more
19:21 <diepchess> so you pass the paper phase
19:21 <xcamel> &more
19:21 <Romulus> xcamel: in the afternoon. The high in the mid 70s. East winds 5 to 10 mph...becoming south in the afternoon. Chance of rain 40 percent.; Saturday Night - Mostly cloudy. A chance of showers and thunderstorms in the evening. The low in the mid 50s. South winds 5 to 10 mph...becoming southwest after midnight. Chance of rain 40 percent.;
19:21 <Tank_Master> why does everything need to be compatible with everything else?
19:21 <diepchess> understand?
19:21 <diepchess> that's requirement number 1 of governments
19:21 <PovAddict> how did we get to that topic? it was just you making a monologue of it; have you heard of blogs?
19:22 <Tank_Master> as long as the outputs are compatible, who cares if they ware they run are not?
19:22 <diepchess> governments have a checklist
19:22 <diepchess> projects must fullfill conditions 1 .. 99
19:22 <Ageless93> me too
19:22 <diepchess> and condition 1 is: it has to be compatible
19:22 <diepchess> otherwise no funding for project
19:22 <diepchess> so no apps get ported
19:22 * Ageless93 deply wonders if icechat has an ignore function
19:22 <diepchess> except 1 or 2 student projects
19:23 <diepchess> serious crunching requires serious funding
19:23 <efc> moo xcamel
19:24 <diepchess> for crunching programming at gpu's you need real good coders
19:24 <efc> I had some ideas on FPGA mass computing, but I gave it up 'cuz video cards were getting so fast, and seti@home appeared to be difficult to implement on them
19:24 <diepchess> example here in front of me is a FFT in integers
19:24 <diepchess> i studied how fast it would be at a nvidia card
19:25 <diepchess> (which of course is faster for floating point)
19:25 <diepchess> i concluded all those projects are fulltime work
19:25 <diepchess> yes efc, gpu's definitely replace crunching on fpga's for many
19:25 <diepchess> i see big future there
19:26 <diepchess> also those fpga development boards do not clock so high
19:26 <diepchess> 100Mhz if you're lucky
19:27 <diepchess> cheaper ones still are 30-31Mhz you achieve
19:27 <diepchess> tesla is 1.2Ghz and ati cards clock 900Mhz now
19:27 <diepchess> efc you said there is a mainboard supporting 4 cards now
19:28 <efc> yeah the cheaper FPGAs get up to about 200 Mhz
19:28 <diepchess> i thought the skulltrails already could have 5 ATI cards
19:28 <xcamel> 'lo Ed
19:28 <efc> More expensive ones are too expensive and lose the flops/$ argument
19:28 <efc> err I mean more powerful hehe
19:28 <diepchess> the compiler is $$$$$
19:28 <diepchess> good one
19:29 <efc> Still, they probably have the flops/watt advantage, and for all integer, probably have the $$/flop if you build a good number of them
19:29 <xcamel> I think a quad tesla running on a quad i7 would be quite a kick though..
19:29 <diepchess> well i checked price of production
19:29 <diepchess> but basically for a reasonable price you must produce 1000 cpu's
19:29 <diepchess> from fpga sample
19:29 <xcamel> I could live with one host then...
19:29 <diepchess> 50 dollar a chip
19:30 <diepchess> 100 dollar for pci card (and royalties for it)
19:30 <diepchess> 150 dollar bare product price
19:30 <diepchess> $150k for tiny project
19:30 <efc> I was willing to drop about $10k for 100 boards, but I could not figure that they would perform as well as I initially hoped
19:31 <diepchess> what type of chip you want to produce?
19:31 <diepchess> i mean what must it do?
19:31 <efc> Those were Spartan2s I think.. whichever the cheap one was
19:31 <efc> Seti@home
19:31 <diepchess> how big is the software that gets put on it?
19:31 <diepchess> there is cheapo boards of like $200
19:32 <efc> code size is not that significant, but I don't know exactly, however the data sets are large
19:32 <diepchess> oh dear
19:32 <efc> The kicker was memory bandwidth
19:32 <diepchess> means 1 card a node
19:32 <efc> they could do so much computing there was no way to keep them fed
19:32 <diepchess> PCI problems then
19:32 <efc> I was going to make them independent
19:33 <diepchess> how much data a second to keep cards get fed?
19:33 <efc> their own RAM etc
19:33 <diepchess> oh you want to build a memory controller?
19:33 <diepchess> wow that's very advanced
19:33 <efc> Open cores are available for them
19:34 <efc> the fast folding algorithm was the worst one for bandwidth
19:34 <efc> basically one multiplier even on the slow spartan at 133mhz would nearly saturate DDR2
19:34 <efc> and you could have 8 or so (sorry I'm fuzzy on details so approximating a lot)
19:35 <diepchess> if you have somethign that needs RAM, forget it yeah, unless you're a big hero
19:35 <diepchess> you can't get away with a tiny hashtable?
19:35 <diepchess> of a few hundreds of kilobyte?
19:35 <efc> nope, the data set is around 250k, no way around it
19:36 <diepchess> that should fit isn't it?
19:36 <PovAddict> found what I was looking for in the log
19:36 <xcamel> good one, PoV
19:36 <efc> and the fast folding algorithm (and unmodified FFT too) access pretty much all of it so its not really cacheable
19:37 <diepchess> yeah
19:37 <diepchess> this is for proving primes?
19:37 <diepchess> i thought about a way
19:37 <diepchess> to design 1 block
19:37 <efc> If that had worked I'd have a few percent of the total S@H production :P
19:37 <diepchess> which is doing a basic functionality
19:37 <diepchess> then just make a lot of those blocks and sequential link them somehow
19:37 *** desti has quit IRC
19:37 <diepchess> i wouldn't know how to feed data from 1 chip to the next
19:38 <diepchess> according to some OCZ dude (director technology) if distance is short of wiring
19:38 <diepchess> it should be transporting a LOT of data a second
19:38 *** desti has joined #boinc
19:38 <diepchess> then you can do it in unix style to say it simple
19:38 <diepchess> (i'm not hardware expert of course)
19:38 <diepchess> input and output
19:38 <efc> Yeah, that is the advantage of doing your own stuff, you can put just what you need very close together
19:38 <diepchess> so not much cacheable ram needed
19:39 <efc> If you are doing something needing less RAM you can use static ram up to a few megabits
19:39 <efc> and need no controller
19:39 <diepchess> yes i know
19:39 <PovAddict> efc: this is the highlevel structure of S@h algorithm? http://home.earthlink.net/~cottongim/Seti%40HomeFlowchart.jpg
19:39 <diepchess> but for example with wagstaff we're up to 3.3 mbit now
19:39 <PovAddict> I didn't really understand your other two graphs :]
19:39 <efc> That looks like the product of a software genius
19:39 <diepchess> primegrid also moved quickly through 321
19:40 <efc> Oh wait.. it mine!!
19:40 <PovAddict> yes it's yours
19:40 <PovAddict> :)
19:41 <Ageless93> I like the videocard though. too bad it only supports 2 monitors. :(
19:42 <efc> diepchess you might be interested in the BEE2, was a Stanford project, had 5 Virtex-2s on it, on a huge 22 layer board
19:42 <PovAddict> efc: how correct would you say your FlowchartHardware2 is? because it looks quite complete
19:42 <diepchess> well the big problem with most of those projects is the number of transistors/gates (or some more modern manner of counting it) you need for the logical blocks
19:42 <PovAddict> in particular, you have data sizes there, which may be relevant here :)
19:42 <diepchess> so it is real difficult to get it in 1 chippie
19:42 <diepchess> same problem with gpu's
19:42 <efc> It was fairly complete and accurate at the time I did it. S@H may have changed some since.
19:43 <PovAddict> any idea of date?
19:43 <PovAddict> nvm I can get that
19:43 <PovAddict> Last-Modified: Mon, 21 Jan 2008 05:41:11 GMT
19:43 <efc> last year, before the CUDA and multbeam came out, I think
19:43 <PovAddict> enhanced app was there already, wasn't it?
19:43 <xcamel> I sitll haven't figure out how they account for the bad 'fuzzy' math used in gpus
19:44 <PovAddict> there's no fuzzy math, it's just it used to be single-precision floats
19:44 <efc> Yeah I think so yes, enhanced just used wider search ranges, I beleive
19:45 <efc> At 32 bits per color per pixel you can be pretty fuzzy and not tell the difference visually.. mathmatically is a different story
19:45 <efc> I remember reading somewhere some s@h guy saying 16 bits would probably be enough for them
19:45 <PovAddict> maybe there are fuzzy maths in the particular algorithms for scanline rendering of graphics; but that's completely bypassed in CUDA
19:47 *** SJr|Work has quit IRC
19:48 <diepchess> efc regarding BEE2 : that cpu already exists for quite some years
19:48 <diepchess> just it is $$$$
19:48 <diepchess> interesting is when something like that gets done low budget
19:49 <efc> That Berkeley stuff had tons of donations from Xilinx. Must be nice.
19:49 <diepchess> they built a 0.13 cpu if i remember well that's gigantic and eating huge power
19:49 <diepchess> with huge RAM
19:49 <diepchess> that's 1 Tflop (double precision)
19:49 <diepchess> but i guess it disappeared, too expensive to sell
19:50 <efc> yeah the academics don't do cost-effective
19:50 <diepchess> yeah i knew a project that got 100 cards for free from xilinx
19:50 <efc> I mean 22 layers, holy PCB batman
19:50 <diepchess> just they were parallel not so genius
19:50 <diepchess> so didn't get it to work
19:50 <diepchess> oh i'm not expert on hardware details like layers
19:50 <diepchess> is that same layer like a processor layer?
19:51 <diepchess> like P4 initially 9 layers and opteron?
19:51 <efc> Layers of wire traces on the PCB
19:51 <diepchess> ah
19:51 <diepchess> wow that's complex
19:51 <efc> like your motherboard probably has 6
19:51 <PovAddict> on a PCB?!
19:51 <efc> Yeah
19:51 <PovAddict> they went a bit overboard
19:52 <efc> Probably it made the design easier
19:52 <efc> So if the production number was very small it worked out cheaper. I'm guessing.
19:53 <diepchess> wow monthly pay $7500 for that project
19:53 <diepchess> that's like other salaries than here
19:53 <diepchess> based upon 50% work
19:53 <diepchess> jeez
19:53 <diepchess> $15k a month
19:53 <diepchess> all in
19:53 *** sigmund has joined #boinc
19:53 <diepchess> you can hire half the nation for that here
19:53 <diepchess> i guess they didn't get a grant for that?
19:53 <diepchess> assuming a salary of $15k a month?
19:54 <diepchess> just for *research* of a PHD ???
19:54 <efc> Who's making 15k a month for BEE2??
19:54 <diepchess> project application of BEE@ let me get you link:
19:54 <Romulus> Title: Proposal for Extending and Enhancing GSRC Research Through the Use of a Common Experimental Platform (at www.gigascale.org)
19:54 <efc> I guess that might not be too unusual for California (insert terminator accent)
19:54 <diepchess> page 3
19:55 <diepchess> most phd's here are guys and women who are real bad
19:55 <efc> to keep the Pros from jumping ship and going private
19:55 <diepchess> because it pays so bad
19:55 <diepchess> the good guys and women directly go commercial
19:55 <diepchess> i remember a professor lurking around at a place whre i was
19:55 <Rookie_69> Reboot coming up...
19:55 <diepchess> looking for a phd for 650 euro a month
19:56 <diepchess> which is less than the rent a month over there of a small apartment :)
19:56 <efc> That may include benefits, overhead, etc
19:56 <diepchess> average salary in IT here is 52k euro a year
19:56 <diepchess> that's excluding social taxes and excluding pension
19:56 <PovAddict> &math calc 52/12
19:56 <Romulus> PovAddict: 4.33333333333
19:56 <diepchess> yes of course that 15k is including
19:56 *** Rookie_69 has quit IRC
19:57 <efc> dunno about their industry but in mine the employee's personal pay is around 1/3 of the bill rate
19:57 <diepchess> it's from 2006 so before US was bankrupt and dollar still strong
19:57 <diepchess> oh in governments that's world wide similar, in this case that 15k is not far away from pay to the researcher in question
19:58 <diepchess> as this is subsidy on salary, not on materials
19:58 <diepchess> so the expensive part for companies is of course fact you need a room, a chair, computer etc
19:58 <diepchess> which is 50% additional
19:58 <diepchess> but in government that's, as far as i saw in different cases not the case
19:58 <efc> so bill $100/hour, employee might get $33, benefits are i dunno $20 more, then there's some tax, and the rest is overhead (ie, pay the big fat VP)
19:59 <efc> No profit included in that
19:59 <diepchess> wait wait i speak of bruto payments
19:59 <diepchess> not netto (after tax) to employee
19:59 *** Ageless93 has left #boinc
19:59 <diepchess> so for example most projects here is 50 euro an hour
19:59 <diepchess> based upon 40 hours a week and say usually 4 weeks
19:59 <diepchess> that's 160 * 50 = 8000
20:00 <efc> heck the labor rate at a decent auto shop is about $75/hour
20:00 <diepchess> wrong
20:00 <diepchess> 20 working days a week
20:00 <diepchess> 20 days a month * 50 euro an hour * 8 hours = 160 * 50 = 8000
20:00 <diepchess> yes that results then in a salary of 5300 that's including pension in that salary
20:00 <diepchess> but ex social taxes and insurances
20:01 <diepchess> so roughly 5/8
20:01 *** Rookie_69 has joined #boinc
20:01 <efc> moo Rookie
20:01 <diepchess> work shop is not working for months nonstop for you
20:02 <Rookie_69> Hi, Eddie!
20:02 <xcamel> ge's baaaaack..
20:02 <diepchess> civil servant salaries here are quite low
20:02 <diepchess> compared to commercial work
20:03 <diepchess> entire netherlands (16.5 million inhabitants) has at universities and institutes directly falling under government
20:04 <diepchess> 1500 researchers from which 1000 professor
20:04 <diepchess> so 500 'independant' researchers'
20:04 <diepchess> on their paybill
20:04 <diepchess> this is report on 2007
20:04 <diepchess> so basically you get a professor title for free :)
20:04 <diepchess> in order to get *some* pay
20:05 *** sigmund_ has quit IRC
20:32 <wdsmia> &ping
20:32 <Romulus> pong
20:32 <Rookie_69> Oops... I would have said Hi, Jay earlier, but i just noticed your ge remark.
20:33 <Tank_Master> heh
20:33 <Tank_Master> rookie
20:33 <Rookie_69> Yes, I am.
20:34 <Tank_Master> :D
20:35 <xcamel> laterz..
20:35 <Rookie_69> 2U2.
20:35 <Tank_Master> tc
21:12 *** diepchess has quit IRC
21:40 <wdsmia> &wx 50319
21:40 <Romulus> wdsmia: Temperature: 70.7°F / 21.5°C | Humidity: 66% | Pressure: 30.05in / 1017.5hPa (Steady) | Conditions: Scattered Clouds | Wind Direction: SW | Wind Speed: 0.0mph / 0.0km/h ; Tonight - Partly cloudy. A 20 percent chance of light showers after midnight. Low in the lower 60s. Southeast wind near 10 mph.; Saturday - Partly sunny. A slight chance of light showers in the morning...then a chance of (1 more message)
21:41 <wdsmia> &more
21:41 <Romulus> wdsmia: thunderstorms in the afternoon. High in the mid 70s. Northeast wind near 10 mph. Chance of precipitation 30 percent.; Saturday Night - Mostly cloudy with a slight chance of light showers and isolated thunderstorms. Low in the upper 50s. East wind 5 to 10 mph. Chance of precipitation 20 percent.;
21:49 *** si has quit IRC
22:22 *** infinisoft has quit IRC
22:23 <Rookie_69> NN.
22:23 *** Rookie_69 has quit IRC
22:33 <MTughan> Oops...
23:01 * KathrynM is watching the Cav's game. Interestingly enough, it's better when she can't understand a word of the commentary aside from 'jum-puh shah-tuh'.
23:18 *** kathryn_ has joined #boinc
23:23 *** KathrynM has quit IRC
23:26 *** kathryn_ is now known as KathrynM
23:30 <KathrynM> this is the Clevleand I know and love. blowing a huge lead
23:33 <KathrynM> apparantly miracles still happen. James hits a 3 pointer at the buzzer.
23:34 <MTughan> What sport?
23:34 <MTughan> Basketball?
23:36 <KathrynM> yeah