| --- Log opened Sun Sep 30 00:00:41 2012 | ||
| -!- superkuh [~superkuh@unaffiliated/superkuh] has joined ##hplusroadmap | 00:01 | |
| -!- eridu [~eridu@gateway/tor-sasl/eridu] has quit [Remote host closed the connection] | 00:06 | |
| -!- eridu [~eridu@gateway/tor-sasl/eridu] has joined ##hplusroadmap | 00:07 | |
| -!- nmz787 [~Nathan@pool-108-39-145-80.pitbpa.fios.verizon.net] has quit [Quit: Leaving.] | 00:12 | |
| -!- lichen [~lichen@c-24-21-206-64.hsd1.or.comcast.net] has quit [Ping timeout: 255 seconds] | 00:21 | |
| -!- tav [~tav`@host86-145-157-241.range86-145.btcentralplus.com] has joined ##hplusroadmap | 00:24 | |
| -!- eridu [~eridu@gateway/tor-sasl/eridu] has quit [Ping timeout: 276 seconds] | 00:29 | |
| -!- Vicarious [diepfriet@CAcert/Vicarious] has joined ##hplusroadmap | 00:30 | |
| -!- strangewarp [~strangewa@c-76-25-200-47.hsd1.co.comcast.net] has quit [Read error: Connection reset by peer] | 00:32 | |
| -!- lichen [~lichen@c-24-21-206-64.hsd1.or.comcast.net] has joined ##hplusroadmap | 00:32 | |
| -!- strangewarp [~strangewa@c-76-25-200-47.hsd1.co.comcast.net] has joined ##hplusroadmap | 00:33 | |
| -!- eridu [~eridu@gateway/tor-sasl/eridu] has joined ##hplusroadmap | 00:43 | |
| @kanzure | nmz787: no | 01:01 |
|---|---|---|
| -!- eridu [~eridu@gateway/tor-sasl/eridu] has quit [Ping timeout: 276 seconds] | 01:22 | |
| -!- EnLilaSko [~Nattzor@unaffiliated/enlilasko] has joined ##hplusroadmap | 02:59 | |
| -!- sylph_mako [~mako@103-9-42-1.flip.co.nz] has quit [Ping timeout: 260 seconds] | 03:24 | |
| -!- strages_home [~strages@adsl-98-81-14-211.hsv.bellsouth.net] has quit [Remote host closed the connection] | 03:46 | |
| -!- chris_99 [~chris_99@unaffiliated/chris-99/x-3062929] has joined ##hplusroadmap | 04:46 | |
| -!- Mike_B [~nobody@cpe-72-229-39-72.nyc.res.rr.com] has joined ##hplusroadmap | 04:56 | |
| -!- Mike_B [~nobody@cpe-72-229-39-72.nyc.res.rr.com] has left ##hplusroadmap [] | 04:56 | |
| brownies | this looks interesting. http://pgbovine.net/burrito.html | 04:56 |
| -!- eudoxia [~eudoxia@r186-52-144-68.dialup.adsl.anteldata.net.uy] has joined ##hplusroadmap | 05:14 | |
| -!- eudoxia [~eudoxia@r186-52-144-68.dialup.adsl.anteldata.net.uy] has quit [Read error: Connection reset by peer] | 05:42 | |
| -!- eudoxia [~eudoxia@r186-52-144-68.dialup.adsl.anteldata.net.uy] has joined ##hplusroadmap | 05:42 | |
| -!- kirka [~Kirka@95-161-252-108.broadband.spb.TiERA.org] has joined ##hplusroadmap | 06:06 | |
| kirka | Hi | 06:06 |
| eudoxia | hey kirka | 06:07 |
| kirka | I see this CPU architecture in my dreams regularily: http://www.adapteva.com/introduction/ | 06:07 |
| kirka | network-on-chip of RISC CPUs | 06:08 |
| kirka | eudoxia ? | 06:09 |
| eudoxia | reading | 06:09 |
| kirka | They have a kickstarter: http://www.kickstarter.com/projects/adapteva/parallella-a-supercomputer-for-everyone | 06:10 |
| kirka | Here is a paper: http://74.220.215.219/~adapteva/wp-content/uploads/2011/10/adapteva_hpec11.pdf | 06:10 |
| eudoxia | 32 bit? | 06:10 |
| kirka | Yes | 06:10 |
| kirka | 1K-core chip should deliver 1.6TFLOPS per 20 watt. | 06:11 |
| kirka | That's very good | 06:11 |
| eudoxia | wouldn't that horribly restrict the memory, or am I thinking about this the wrong way? | 06:11 |
| kirka | No, they are small - these individual CPUS | 06:12 |
| kirka | There are 64+ of them | 06:12 |
| eudoxia | maybe if each processor has a certain alloted memory you could get past the 3.5 GB bound, but you'd only be able to access 3.5*(#Cores) GB of memory | 06:12 |
| eudoxia | that's 210 GB, while 64 bit lets you have 16 exabytes | 06:13 |
| kirka | It's hard to compare network on chip to standard multicore CPUs we have | 06:14 |
| kirka | I'd like to look at reference manual for their RISC arch, but seems there isn't any | 06:15 |
| kirka | External memory is a bottleneck as always | 06:16 |
| kirka | I think it's OK to access as network resource in this networ-on-a-chip | 06:17 |
| kirka | *access external memory | 06:17 |
| kirka | And 32 bits is more than enough to access local RAM | 06:17 |
| kirka | Making them 64 bit would increase silicon area by ~1.3-1.5 times, making the whole thing less attractive | 06:18 |
| -!- eudoxia [~eudoxia@r186-52-144-68.dialup.adsl.anteldata.net.uy] has quit [Read error: Connection reset by peer] | 06:18 | |
| -!- eudoxia [~eudoxia@r186-52-144-68.dialup.adsl.anteldata.net.uy] has joined ##hplusroadmap | 06:18 | |
| kirka | eudoxia Cluster of these chips would be very good at quantum chemistry and molecular dynamics simulations | 06:20 |
| kirka | Single 1024 core chip can easily cost 100$ | 06:21 |
| kirka | 130 mm^2 of silicon isn't much | 06:22 |
| eudoxia | i guess it's ok | 06:23 |
| eudoxia | it may not scale to a matrioshka brain like 64 bit does but then again who cares | 06:23 |
| kirka | Heh | 06:23 |
| kirka | Perspective of someone creating matrioshka brain is frightening | 06:24 |
| eudoxia | yeah, Nicoll-Dyson lasers and all that shit | 06:26 |
| @kanzure | on the diybio list someone just posted, | 06:27 |
| @kanzure | "какие-то агрессивно-высокомерные по отношению к непрофессионалу" | 06:27 |
| kirka | Heh | 06:27 |
| @kanzure | i think this is saying the person is very angry? | 06:27 |
| kirka | Nope | 06:27 |
| @kanzure | and nonprofessional? | 06:27 |
| eudoxia | i think i saw the word economy | 06:28 |
| kirka | In a nutshell, it says "You are too hard on me, the n00b" | 06:28 |
| @kanzure | bah | 06:28 |
| @kanzure | thank you for the ru2leet translation | 06:28 |
| @kanzure | kirka: https://github.com/kanzure/nanoengineer/commits/mocking | 06:29 |
| @kanzure | this version of nanoengineer will run on modern linux without trouble | 06:29 |
| @kanzure | however, you do have to run "sudo pip install mock" | 06:29 |
| kirka | Cool | 06:29 |
| kirka | >_metaclass_ | 06:30 |
| @kanzure | the __metaclass__ stuff is to fix some problems caused by mock | 06:31 |
| @kanzure | this is just a temporary solution :) | 06:31 |
| @kanzure | see, the problem is that the mock library was never intended to be used like this | 06:31 |
| @kanzure | and there are multiple classes that are being mocked in this branch | 06:31 |
| kirka | > mock is a library for testing in Python. | 06:31 |
| kirka | Ah | 06:31 |
| @kanzure | so when a class inherits a previously-mocked class, python doesn't know which metaclass to resolve to | 06:31 |
| @kanzure | so i just force it to make up a new one because i don't care | 06:32 |
| @kanzure | oops i mean when a "class inherits multiple previously-mocked classes" not "when a class inherits a previously-mocked class" | 06:32 |
| kirka | I don't understand python's metaclasses anyway, heh | 06:33 |
| kirka | Maybe it's similar to common lisp's though | 06:33 |
| gnusha | nanoengineer.git: 396f148 reinstate class GromacsProcess with mocking | 06:40 |
| @kanzure | http://publicdomainreview.org/2011/08/23/space-colony-art-from-the-1970s/ | 06:43 |
| eudoxia | the one with the bernal sphere always confused me | 06:44 |
| kirka | Cool, I haven't seen all of them | 06:44 |
| @kanzure | http://publicdomainreview.org/2012/07/09/arabic-machine-manuscript/ | 06:47 |
| @kanzure | there's some really weird stuff on here. | 06:47 |
| @kanzure | http://publicdomainreview.org/2012/06/30/france-in-the-year-2000-1899-1910/ | 06:47 |
| @kanzure | apparently, in the year 2000, frenchmen rode flying seahorses | 06:47 |
| Urchin | there looks to be couple of bernal sphere pictures there | 06:48 |
| eudoxia | Urchin: the one with the dudes positioning the mirrors | 06:48 |
| Urchin | yes, it confused me as well | 06:49 |
| eudoxia | http://publicdomainreview.org/2012/08/06/kaishi-hen-an-18th-century-japanese-anatomical-atlas/ | 06:49 |
| @kanzure | "battle cars" is also amusing | 06:49 |
| -!- Jaakko96 [~Jaakko@94-194-89-130.zone8.bethere.co.uk] has joined ##hplusroadmap | 06:49 | |
| @kanzure | http://publicdomainreview.org/2012/09/28/16th-century-prosthetics-1564/ | 06:53 |
| @kanzure | "winner, 5th olympiad, walking race 10,000 m" | 06:55 |
| eudoxia | the maple leaf looks a little suspicious | 06:58 |
| kirka | Maple leaf? | 07:01 |
| eudoxia | guy's shirt | 07:01 |
| * kirka Is reading about that many core architecture | 07:03 | |
| kirka | Local RAM instead of cache is cool | 07:03 |
| kirka | If I'd design a CPU for scientific computing, It'd be exactly like their one. | 07:03 |
| ParahSailin_ | which arch? | 07:04 |
| kirka | http://www.hpcwire.com/hpcwire/2011-05-03/startup_launches_manycore_floating_point_acceleration_technology.html | 07:04 |
| @kanzure | kirka: you should read the criticism. http://news.ycombinator.com/item?id=4583263 | 07:04 |
| ParahSailin_ | not a general cpu? | 07:05 |
| ParahSailin_ | ah it is | 07:06 |
| ParahSailin_ | why fucking go to in-q-tel | 07:07 |
| ParahSailin_ | thats a terrorist organization | 07:07 |
| kirka | kanzure Looks like there isn't a lot of chip|HPC guys on hackernews nowadays | 07:08 |
| kirka | >At the 28nm node, Olofsson thinks they can get up to 1,000 64-bit floating point cores on the die. | 07:12 |
| kirka | With this architecture it's possible to build exaflop machine @ 30MW. | 07:13 |
| ParahSailin_ | you always trade one thing for another on a die-- what does this arch sacrifice | 07:15 |
| kirka | Comptibility with old (x86) binary codes and large caches. | 07:16 |
| ParahSailin_ | ah, cache | 07:16 |
| kirka | Scratchpad RAM is better | 07:16 |
| kirka | Modern CPUs are very slow because they are required to support a lot of legacy software | 07:17 |
| kirka | Modern x86 CPUs are quite a mess | 07:17 |
| kirka | Threr is a unit that translates x86 code to internal RISC code | 07:18 |
| kirka | It takes space | 07:18 |
| kirka | Threr are x86 compatibility and real mode | 07:18 |
| kirka | And there are very large caches to help old software run faster | 07:18 |
| ParahSailin_ | doesn't translation to microcode take a small amount of the die? | 07:18 |
| kirka | I don't think so | 07:19 |
| kirka | Caches waste much bigger areas of course | 07:20 |
| ParahSailin_ | for many purposes, minimizing memory IO off-die is not a waste | 07:21 |
| kirka | I agree | 07:21 |
| kirka | But I'm interested in computing heavy applications | 07:22 |
| kirka | Physical modeling etc | 07:22 |
| ParahSailin_ | why not a gpu | 07:22 |
| kirka | GPUs are much less flexible, and they are more complex to program. | 07:22 |
| kirka | This "epiphany" architecture would be ideal for me | 07:23 |
| eudoxia | GPUs seem ideal for the mathematical applications you're interested in kirka | 07:24 |
| kirka | eudoxia They are good, but there are branch intensive algorithms that they cannot run effectively | 07:25 |
| kirka | For example multy-body physics engine | 07:25 |
| kirka | It's very hard to implement all sub-algorithms on GPU | 07:26 |
| kirka | GPU weren't meant for running such software | 07:28 |
| kirka | Ah, the machine of my dream | 07:29 |
| kirka | CLuster of these could be used to simulate c.Elegans in celullar resolution - in real time | 07:32 |
| kirka | These guys are creating such model: http://code.google.com/p/openworm/ | 07:32 |
| kirka | This CPU would be good for real-time computer vision and motion planning in robotics | 07:38 |
| kirka | UAVs with local image analysis for example | 07:39 |
| kirka | Military guys would be happy | 07:39 |
| -!- eudoxia [~eudoxia@r186-52-144-68.dialup.adsl.anteldata.net.uy] has quit [Quit: Lost terminal] | 07:43 | |
| ParahSailin_ | what is the trade this arch makes over gpu to be less parallel? | 08:09 |
| kirka | >less parallel ? | 08:09 |
| ParahSailin_ | less parallel as in 32 cores are not bound together on the same instructions | 08:10 |
| kirka | GPUs don't have separate control floe units near every FPU | 08:11 |
| kirka | *flow | 08:11 |
| ParahSailin_ | yes. what is the trade this new arch makes to give separate control for each core | 08:12 |
| kirka | Yes | 08:12 |
| kirka | ACtually, tehre are no "cores" in GPU | 08:12 |
| kirka | There are lots of FPUs and some control units | 08:12 |
| kirka | *there | 08:12 |
| ParahSailin_ | the cuda manual calls them cores in a compute unit | 08:12 |
| kirka | So GPUs cannot execute branches | 08:12 |
| ParahSailin_ | ok i just use what nvidia calls them | 08:13 |
| kirka | "Epiphany" Is just an array of lightweight RISCs with FPUs | 08:13 |
| kirka | Every RISC has it's own instruction stream | 08:13 |
| kirka | That's very convenient | 08:13 |
| ParahSailin_ | to get that on the die, what's the trade they make? | 08:14 |
| kirka | No need too think about warps etc | 08:14 |
| ParahSailin_ | gpu already has little in the way of caching afaik | 08:14 |
| ParahSailin_ | and small amount of shared memory ~64kb dedicated to each compute unit | 08:15 |
| kirka | Well, GPUs still have texturing units, tesselation and a lot of specialized stuff | 08:15 |
| kirka | It takes space | 08:15 |
| kirka | http://www.adapteva.com/wp-content/uploads/2011/06/adapteva_mpr.pdf | 08:16 |
| kirka | Paper from MPR | 08:16 |
| ParahSailin_ | i think most of that stuff is general purpose fpu's that do that | 08:16 |
| kirka | Well, it's wide SIMD | 08:17 |
| kirka | It has advantages and disadvantages | 08:18 |
| kirka | I find GPUs complex to program | 08:18 |
| ParahSailin_ | they're harder than cpu, but nothing's ever free | 08:19 |
| kirka | I agree | 08:19 |
| ParahSailin_ | every design make choices, and this sounds like the best of both worlds | 08:19 |
| -!- eridu [~eridu@gateway/tor-sasl/eridu] has joined ##hplusroadmap | 08:23 | |
| ParahSailin_ | hm, save die space by cutting out int math, i could believe that | 08:26 |
| kirka | Note that there is no special FP functions | 08:27 |
| kirka | But it's important that this architecture is completely new | 08:28 |
| kirka | while both CPUs and GPUs are evolutionary developments | 08:28 |
| kirka | Nvidia and Ati have to seek compromise between graphics and general computation | 08:29 |
| ParahSailin_ | im reading the whitepaper | 08:29 |
| ParahSailin_ | so this architecture, they're taking it to a foundry, or still looking for money? | 08:30 |
| kirka | They are fabless, but they already have 4 generations of chips | 08:31 |
| kirka | http://www.adapteva.com/ | 08:31 |
| kirka | http://www.adapteva.com/products/silicon-devices/e64g401/ | 08:31 |
| ParahSailin_ | everyone's fabless except smc, samsung, intel, ibm | 08:31 |
| kirka | I mean they are IP company | 08:32 |
| ParahSailin_ | is it in production i mean | 08:32 |
| kirka | Looks like it is | 08:32 |
| kirka | But I have found it just a hour ago | 08:32 |
| kirka | I don't know it better than you | 08:32 |
| kirka | I'd like 4 channel DDR controller btw | 08:33 |
| -!- Jaakko96 [~Jaakko@94-194-89-130.zone8.bethere.co.uk] has quit [Quit: Nettalk6 - www.ntalk.de] | 08:35 | |
| -!- kirka1 [~Kirka@95-161-252-108.broadband.spb.TiERA.org] has joined ##hplusroadmap | 08:51 | |
| -!- kirka [~Kirka@95-161-252-108.broadband.spb.TiERA.org] has quit [Ping timeout: 240 seconds] | 08:53 | |
| -!- kirka1 [~Kirka@95-161-252-108.broadband.spb.TiERA.org] has left ##hplusroadmap [] | 08:53 | |
| -!- kirka1 [~Kirka@95-161-252-108.broadband.spb.TiERA.org] has joined ##hplusroadmap | 08:53 | |
| -!- marainein [~net@2001:388:608c:6cb5:613f:d749:c60c:a0db] has joined ##hplusroadmap | 09:00 | |
| -!- SDr [SDr@unaffiliated/sdr] has joined ##hplusroadmap | 09:01 | |
| -!- kirka1 [~Kirka@95-161-252-108.broadband.spb.TiERA.org] has left ##hplusroadmap [] | 09:15 | |
| -!- soylentbomb [~k@unaffiliated/soylentbomb] has quit [Quit: Leaving] | 09:21 | |
| -!- devrandom [~devrandom@gateway/tor-sasl/niftyzero1] has quit [Remote host closed the connection] | 09:30 | |
| -!- devrandom [~devrandom@gateway/tor-sasl/niftyzero1] has joined ##hplusroadmap | 09:32 | |
| -!- mensch [~mensch@c-24-128-26-165.hsd1.ma.comcast.net] has joined ##hplusroadmap | 09:41 | |
| @kanzure | prediction: people will want to go to mars to hunt for bacteria fossils | 09:43 |
| -!- eudoxia [~eudoxia@r186-54-224-107.dialup.adsl.anteldata.net.uy] has joined ##hplusroadmap | 09:51 | |
| streety | I suspect people will want to go to mars for a variety of reasons | 09:52 |
| jrayhawk | woo another massively parallel chipset company that places seemingly zero emphasis on compilation | 10:01 |
| jrayhawk | soon to be added to the pile of defunct companies that thought they could do the same thing | 10:01 |
| eudoxia | but it says it's fully ANSI-C/C++ programmable :(((( | 10:02 |
| jrayhawk | "we're just as much of a pain in the ass to parallelize from the software end as existing architectures" is not exactly an encouraging claim | 10:06 |
| ParahSailin_ | i bet one could do some interesting haskell tools for it | 10:06 |
| eudoxia | someone should write an LLVM backend for it | 10:06 |
| eudoxia | it's got SIMD and vectorization and stuff I suppose must speed things up | 10:07 |
| ParahSailin_ | should be the designer's job to do that stuff though | 10:09 |
| -!- eridu [~eridu@gateway/tor-sasl/eridu] has quit [Ping timeout: 276 seconds] | 10:27 | |
| -!- eudoxia [~eudoxia@r186-54-224-107.dialup.adsl.anteldata.net.uy] has quit [Quit: Lost terminal] | 10:28 | |
| @kanzure | jrayhawk: so the sha256 asic i mentioned to you is apparently being marketed to bitcoin miners, so i don't anticipate that being a "well kept secret". | 10:32 |
| @kanzure | if somenoe was serious about bitcoin mining, i think a custom asic run would be in order. not using what everyone else is going to be using. | 10:32 |
| ParahSailin_ | i think there are a couple asics coming out for bitcoin hashing | 10:35 |
| ParahSailin_ | some might even be out | 10:35 |
| @kanzure | *someone | 10:36 |
| @kanzure | ParahSailin_: i'm surprised there's no "rainbow tables as a service" company yet. | 10:36 |
| @kanzure | md5.gromweb.com only has 65 million entries, which is pathetic | 10:36 |
| @kanzure | i'd gladly pay a few bucks to run something in the cloud for a few hours to fill out some rainbow tables, as long as i had access to a few billion entries | 10:37 |
| ParahSailin_ | is sha256 vulnerable to rainbow tables? | 10:37 |
| @kanzure | don't rainbow tables work for anything that digests something? | 10:38 |
| ParahSailin_ | i only have vague understanding | 10:39 |
| -!- eridu [~eridu@gateway/tor-sasl/eridu] has joined ##hplusroadmap | 10:39 | |
| @kanzure | ParahSailin_: well, for md5 rainbow tables, you just iterate through all possible combinations of bytes and dump out md5 hashes | 10:40 |
| @kanzure | then you store the original text with the hash in a giant table | 10:40 |
| ParahSailin_ | yeah, i think sha256 is just too big to be practical | 10:40 |
| @kanzure | what do you mean? | 10:40 |
| jrayhawk | Well, you don't compute the whole space since that's cosmologically impossible, you just compute within the constraints you care about (typically [a-zA-Z0-9]{6-12}) | 10:41 |
| @kanzure | sure, sure. | 10:41 |
| @kanzure | but 65 million is super pathetic | 10:41 |
| @kanzure | and the fact that i have to setup my own ec2 instances for oshashcat or w/e gpgpu craptastic software, is sorta annoying | 10:42 |
| ParahSailin_ | what exactly do you want to do with hashing? | 10:42 |
| @kanzure | i have lots of passwords in md5 format that i'd be interested in having in plaintext | 10:43 |
| @kanzure | hah these guys will ship you some hard drives for $600 http://www.freerainbowtables.com/tables/ | 10:43 |
| ParahSailin_ | whose passwords? | 10:44 |
| ParahSailin_ | some consulting job? | 10:44 |
| @kanzure | nope.. just stuff like siai's database and other hilariously insecure things | 10:44 |
| jrayhawk | hot damn, that's a pretty cool service | 10:45 |
| ParahSailin_ | isnt the standard way to just tell people their password is expired to force migration of user database? | 10:45 |
| jrayhawk | i don't think you understand just what a bastard kanzure is | 10:45 |
| @kanzure | i just like having these things. | 10:46 |
| -!- eridu [~eridu@gateway/tor-sasl/eridu] has quit [Remote host closed the connection] | 10:46 | |
| -!- SDragon [SDr@cpc10-dals18-2-0-cust809.hari.cable.virginmedia.com] has joined ##hplusroadmap | 10:47 | |
| @kanzure | jrayhawk: i'm not sure why they have to ship you hard drives | 10:47 |
| @kanzure | why can't they just make you pay for online access? | 10:47 |
| @kanzure | i suppose there's situations where having the entire table available is important | 10:47 |
| jrayhawk | i assume privacy is important to a lot of people doing rainbow tabling | 10:47 |
| @kanzure | if it's just for passwords, it's not like it's a big surprise that the passwords follow standard distribution curves :p | 10:48 |
| streety | this is also presumably a cost to doing the lookup | 10:48 |
| streety | *there is | 10:48 |
| jrayhawk | yeah, but that cost would presumably be lower than $600 | 10:48 |
| @kanzure | hrm, i'm not sure how much time a single lookup would take on that much data | 10:48 |
| streety | depends on the scale | 10:48 |
| @kanzure | you could also conceivably use multiple machines to look on different parts of the data simultaneously | 10:48 |
| @kanzure | or multiple processes or whatever the right scaling approach is. | 10:49 |
| jrayhawk | yeah, you have to scale the costs of operating the hardware, which don't really scale | 10:49 |
| jrayhawk | i.e. watt-hours | 10:49 |
| @kanzure | and, since most password lookups follow the standard distribution, you could just memcache the most popular. for non-password things, you're shit out of luck and have to wait around a bit. | 10:49 |
| -!- SDr [SDr@unaffiliated/sdr] has quit [Ping timeout: 246 seconds] | 10:50 | |
| -!- SDragon is now known as SDr | 10:50 | |
| -!- SDr [SDr@cpc10-dals18-2-0-cust809.hari.cable.virginmedia.com] has quit [Changing host] | 10:50 | |
| -!- SDr [SDr@unaffiliated/sdr] has joined ##hplusroadmap | 10:50 | |
| @kanzure | so, i might be wrong, but shouldn't it be possible to just do a billion copies of md5 on a single asic die | 10:53 |
| @kanzure | and just load up your target into a register and let it spin around until it finds a hit? | 10:53 |
| ParahSailin_ | spend millions on an asic to crack an obsolete hash function | 10:54 |
| jrayhawk | asics aren't expensive | 10:54 |
| ParahSailin_ | if you dont mind 500 um process | 10:54 |
| @kanzure | not so obsolete.. everyone who is hilariously insecure happens to also use md5. | 10:54 |
| @kanzure | if you are thoughtful enough to use something other than md5, you're probably thoughtful enough to patch your ancient version of php or whatever i use to get those hashes in the first place. | 10:55 |
| ParahSailin_ | er, 500nm process | 10:57 |
| @kanzure | so, i haven't actually done a custom asic run before, but for some reason i have the crazy idea that it would cost maybe $10k for a single run | 10:58 |
| jrayhawk | MOSIS and CMP have prices in about that range, yeah. | 10:58 |
| @kanzure | http://electronics.stackexchange.com/questions/7042/how-much-does-it-cost-to-have-a-custom-asic-made | 10:59 |
| ParahSailin_ | http://cmp.imag.fr/products/ic/?p=prices | 11:00 |
| @kanzure | "MOSIS doesn't publish their rates, but CMP's cheapest rate on a 0.35 micron process for 650 Euros/mm^2. A non-trivial design will probably cost $3000 or more for 40 chips." | 11:00 |
| @kanzure | shit that's cheap | 11:00 |
| @kanzure | does cmp also do packaging? | 11:03 |
| @kanzure | they've only done 800 runs? | 11:04 |
| jrayhawk | http://cmp.imag.fr/products/packaging/ | 11:04 |
| jrayhawk | keep in mind that a 'run' is potentially enormous | 11:04 |
| @kanzure | one wafer? | 11:04 |
| AdrienG | 3k for 40 chips? | 11:04 |
| AdrienG | thats dirt cheap | 11:04 |
| jrayhawk | one set of masks is usually used to make many wafers | 11:04 |
| AdrienG | packaged or raw wafers? | 11:05 |
| jrayhawk | and whatever design you give to them is but one of many in a given mask | 11:05 |
| jrayhawk | which is why they're able to offer you decent prices | 11:05 |
| @kanzure | packaging setup fee of €370 per lot | 11:06 |
| @kanzure | hrm. | 11:06 |
| ParahSailin_ | 18kGates/mm^2 at .35um -- thats a lot more than i expected | 11:06 |
| ParahSailin_ | i figured it would be like 1 kGate | 11:07 |
| -!- strages_home [~strages@adsl-98-81-14-211.hsv.bellsouth.net] has joined ##hplusroadmap | 11:13 | |
| AdrienG | how do they test custom chips? | 11:22 |
| @kanzure | maybe they don't | 11:23 |
| AdrienG | then its more like 3k for 20 chips. | 11:27 |
| -!- eudoxia [~eudoxia@r186-54-224-107.dialup.adsl.anteldata.net.uy] has joined ##hplusroadmap | 11:35 | |
| -!- chris_99 [~chris_99@unaffiliated/chris-99/x-3062929] has quit [Remote host closed the connection] | 11:38 | |
| -!- chris_99 [~chris_99@unaffiliated/chris-99/x-3062929] has joined ##hplusroadmap | 11:39 | |
| -!- Mokstar1 [~Nate@c-76-115-136-13.hsd1.or.comcast.net] has joined ##hplusroadmap | 11:41 | |
| -!- Mokstar [~Nate@c-76-115-136-13.hsd1.or.comcast.net] has quit [Ping timeout: 244 seconds] | 11:43 | |
| -!- Mokstar [~Nate@c-76-115-136-13.hsd1.or.comcast.net] has joined ##hplusroadmap | 11:48 | |
| -!- Mokstar1 [~Nate@c-76-115-136-13.hsd1.or.comcast.net] has quit [Ping timeout: 255 seconds] | 11:50 | |
| @kanzure | jrayhawk: i'd like to add wiki.iimarck.us to gnusha to show a certain piny ikiwiki instance. i'm ok with using secure.diyhpl.us for editing. how do i setup this virtual host with piny properly? | 11:55 |
| -!- Mokstar1 [~Nate@c-76-115-136-13.hsd1.or.comcast.net] has joined ##hplusroadmap | 11:57 | |
| -!- Mokstar [~Nate@c-76-115-136-13.hsd1.or.comcast.net] has quit [Ping timeout: 260 seconds] | 11:59 | |
| @kanzure | jrayhawk: i got it. | 12:09 |
| -!- strages_home [~strages@adsl-98-81-14-211.hsv.bellsouth.net] has quit [Quit: Lost terminal] | 12:19 | |
| -!- eudoxia [~eudoxia@r186-54-224-107.dialup.adsl.anteldata.net.uy] has quit [Quit: Lost terminal] | 12:20 | |
| -!- strages_home [~strages@adsl-98-81-14-211.hsv.bellsouth.net] has joined ##hplusroadmap | 12:41 | |
| jrayhawk | oh good | 12:44 |
| -!- sylph_mako [~mako@103-9-42-1.flip.co.nz] has joined ##hplusroadmap | 12:57 | |
| @kanzure | so the pokemon rom hackers want me to make them a map editor in javascript | 13:01 |
| @kanzure | but they want me to make it parse the rom or source code to find addresses | 13:01 |
| @kanzure | instead of just editing files. how do i convince them that this is a terrible idea. | 13:01 |
| -!- mensch [~mensch@c-24-128-26-165.hsd1.ma.comcast.net] has quit [Quit: Lost terminal] | 13:12 | |
| jrayhawk | aren't those ROMs static anyway | 13:13 |
| @kanzure | jrayhawk: there's a large community of "rom hackers" that stare at hex editors all day to change bits around | 13:14 |
| @kanzure | http://www.rijon.com/prism/ | 13:14 |
| jrayhawk | oh, well, i can see the need for a dynamic solution, at least, but i don't see that needing to be done clientside | 13:15 |
| @kanzure | it's like a developer continuously losing source code to his game, and then deciding 'welp, might as well parse the binary instead of keeping track of my source code' | 13:15 |
| jrayhawk | well, i guess it depends on exactly what you want to put in and what exactly you want to get out of your map editor | 13:16 |
| @kanzure | also, prism is worth trying if you ever played the earlier games | 13:16 |
| jrayhawk | if you want to input whole ROMs and get back out whole ROMs then you can avoid the distribution of derived protected works by making it all clientside | 13:16 |
| chris_99 | i love the look of this debugger for C64 http://icu64.blogspot.co.uk/ | 13:16 |
| @kanzure | jrayhawk: yeah, exactly. the rom hackers tend to just distribute patches that they apply to the roms. | 13:17 |
| @kanzure | but i dumped all the maps into pokecrystal.git, and so far nobody from nintendo is sending me angry letters | 13:17 |
| -!- ElixirVitae [~Shehrazad@unaffiliated/shehrazad] has joined ##hplusroadmap | 13:21 | |
| -!- ThomasEgi [~thomas@p54BAD0EC.dip.t-dialin.net] has joined ##hplusroadmap | 13:26 | |
| -!- ThomasEgi [~thomas@p54BAD0EC.dip.t-dialin.net] has quit [Changing host] | 13:26 | |
| -!- ThomasEgi [~thomas@panda3d/ThomasEgi] has joined ##hplusroadmap | 13:26 | |
| Sanqui | jrayhawk, kanzure: I don't think anybody should care about "protected works", the sonic community has had perfect disassemblies for like ten years | 13:27 |
| ParahSailin_ | kanzure, the people who work at nintendo who would understand what pokecrystal.git is don't care because they're on salary, and the lawyers who get bonuses for harassment don't know what github is yet | 13:31 |
| AdrienG | poke her crystals | 13:36 |
| Sanqui | okay, I lied. the first sonic disassembly: 0930223514» <+MainMemory> sometime before 2005 | 13:37 |
| -!- augur [~augur@208.58.5.87] has quit [Remote host closed the connection] | 13:48 | |
| jrayhawk | no way is kanzure going to let the sonic community upstage the pokemon community | 13:55 |
| -!- yashgaroth [~f@cpe-66-27-117-179.san.res.rr.com] has joined ##hplusroadmap | 14:29 | |
| -!- chris_99 [~chris_99@unaffiliated/chris-99/x-3062929] has quit [Quit: Leaving] | 14:36 | |
| -!- eudoxia [~eudoxia@r186-54-224-107.dialup.adsl.anteldata.net.uy] has joined ##hplusroadmap | 14:37 | |
| strangewarp | http://www.foreignaffairs.com/articles/138154/neil-gershenfeld/how-to-make-almost-anything?page=show | 14:43 |
| strangewarp | "One of the assemblers we are developing works with components that are a bit bigger than amino acids, cluster of atoms about ten nanometers long (an amino acid is around one nanometer long). These can have properties that amino acids cannot, such as being good electrical conductors or magnets. The goal is to use the nanoassembler to build nanostructures, such as 3-D integrated circuits." | 14:43 |
| strangewarp | Any thoughts about this guy? | 14:43 |
| strangewarp | (afk, will read the backlog when I get back) | 14:43 |
| eudoxia | I sure hope he's not hyping up a 3d printed atomic force microscope | 14:44 |
| @kanzure | strangewarp: gershenfeld is well known because of fablabs and mit media lab | 14:47 |
| @kanzure | also, he wrote cad.py | 14:48 |
| @kanzure | i should port the biocurious wiki to diyhpluswiki | 15:07 |
| @kanzure | there's no reason to be using pbwiki | 15:07 |
| -!- EnLilaSko [~Nattzor@unaffiliated/enlilasko] has quit [Quit: - nbs-irc 2.39 - www.nbs-irc.net -] | 15:07 | |
| -!- eudoxia [~eudoxia@r186-54-224-107.dialup.adsl.anteldata.net.uy] has quit [Quit: Lost terminal] | 15:36 | |
| @kanzure | hrm https://github.com/defunkt/hub | 15:44 |
| @kanzure | git pull-request "I've implemented feature X" -b defunkt:master -h mislav:feature | 15:45 |
| @kanzure | i guess that might be useful, maybe. | 15:45 |
| strangewarp | Hmm, so Gershenfield might not be talking out his ass, then.. that's pretty rad | 15:59 |
| -!- sylph_mako [~mako@103-9-42-1.flip.co.nz] has quit [Quit: Leaving] | 16:02 | |
| -!- augur [~augur@c-69-250-16-111.hsd1.md.comcast.net] has joined ##hplusroadmap | 16:31 | |
| -!- ThomasEgi [~thomas@panda3d/ThomasEgi] has quit [Remote host closed the connection] | 16:45 | |
| -!- soylentbomb [~k@unaffiliated/soylentbomb] has joined ##hplusroadmap | 16:46 | |
| -!- augur [~augur@c-69-250-16-111.hsd1.md.comcast.net] has quit [Read error: Connection reset by peer] | 16:50 | |
| -!- augur [~augur@c-69-250-16-111.hsd1.md.comcast.net] has joined ##hplusroadmap | 16:52 | |
| -!- soylentbomb [~k@unaffiliated/soylentbomb] has quit [Quit: Leaving] | 16:53 | |
| @kanzure | i need to find someone who was an actual member of the "homebrew computer club" to write an angry rant about how people should stop comparing shit to them. | 17:36 |
| ParahSailin_ | transistors:computers::molecules:singularity | 17:57 |
| @kanzure | well, i just saw someone on the makerbot mailing list who made this drupal site for.. uh.. i guess selling ads or something.. about 3d printing.. yeah, that must be it | 17:57 |
| @kanzure | and he's saying it will be like a homebrew computer club | 17:57 |
| @kanzure | fact: your crappy drupal site is completely unlike the homebrew computer club. | 17:57 |
| @kanzure | it's even worse than calling diybio that | 17:58 |
| brownies | haha | 18:04 |
| brownies | we can just compare random bullshit projects to the homebrew computer club now? | 18:04 |
| @kanzure | yes | 18:06 |
| @kanzure | you know what, fuck that | 18:06 |
| @kanzure | brownies: from now on, compare everything to the manhatten project | 18:06 |
| brownies | no, need something less offensive | 18:06 |
| @kanzure | no, to the space race | 18:06 |
| brownies | "basically, our web app devleopment team is just like CERN" | 18:06 |
| @kanzure | hahah | 18:06 |
| @kanzure | "Our particles are high energy" | 18:07 |
| brownies | "if you think about it, CERN is basically just a series of tubes" | 18:07 |
| brownies | ...i could get a good blog post about this | 18:07 |
| @kanzure | oh god what have i done | 18:07 |
| brownies | "Run your startup like a high-energy particle accelerator" | 18:07 |
| @kanzure | well, to be fair, it's probably better than the other crap they're doing | 18:08 |
| brownies | hahah | 18:08 |
| @kanzure | i don't understand how they all gravitate to the same comparison though | 18:08 |
| @kanzure | like, you guys can't all be the reincarnation of the homebrew computer club | 18:08 |
| streety | maybe it's catching | 18:09 |
| brownies | it's viral! | 18:09 |
| @kanzure | "Many of the original members of the Homebrew Computer Club continue to meet (as of 2009), having formed the 6800 Club, named after the Motorola (now Freescale) 6800 microprocessor." | 18:10 |
| @kanzure | oh look, they even still meet | 18:10 |
| @kanzure | i bet these douchebags calling themselves reincarnations of the homebrew computer club have never gone to a 6800 meeting | 18:10 |
| streety | "like the homebrew computer club" doesn't seem like much of a claim, from the way wikipedia reads every computing/programming meetup I've been to is "like the homebrew computer club" | 18:15 |
| streety | granted so far they haven't had quite the same impact | 18:16 |
| @kanzure | the difference is that people don't go around saying their twitter clone is like "the moomoo mobile mondays of boston" | 18:16 |
| -!- klafka [~textual@c-67-174-253-229.hsd1.ca.comcast.net] has joined ##hplusroadmap | 18:51 | |
| -!- klafka [~textual@c-67-174-253-229.hsd1.ca.comcast.net] has quit [Excess Flood] | 18:51 | |
| -!- klafka [~textual@c-67-174-253-229.hsd1.ca.comcast.net] has joined ##hplusroadmap | 18:52 | |
| -!- klafka [~textual@c-67-174-253-229.hsd1.ca.comcast.net] has quit [Excess Flood] | 18:52 | |
| -!- klafka [~textual@c-67-174-253-229.hsd1.ca.comcast.net] has joined ##hplusroadmap | 18:53 | |
| -!- klafka [~textual@c-67-174-253-229.hsd1.ca.comcast.net] has quit [Excess Flood] | 18:53 | |
| -!- mode/##hplusroadmap [+b klafka!*@*] by kanzure | 18:53 | |
| -!- augur [~augur@c-69-250-16-111.hsd1.md.comcast.net] has quit [Read error: Connection reset by peer] | 18:57 | |
| -!- augur_ [~augur@c-69-250-16-111.hsd1.md.comcast.net] has joined ##hplusroadmap | 18:57 | |
| -!- soylentbomb [~k@unaffiliated/soylentbomb] has joined ##hplusroadmap | 19:21 | |
| -!- augur_ [~augur@c-69-250-16-111.hsd1.md.comcast.net] has quit [Read error: Connection reset by peer] | 19:42 | |
| -!- augur [~augur@c-69-250-16-111.hsd1.md.comcast.net] has joined ##hplusroadmap | 19:43 | |
| -!- augur [~augur@c-69-250-16-111.hsd1.md.comcast.net] has quit [Remote host closed the connection] | 19:52 | |
| -!- augur [~augur@c-69-250-16-111.hsd1.md.comcast.net] has joined ##hplusroadmap | 19:57 | |
| -!- soylentbomb [~k@unaffiliated/soylentbomb] has quit [Quit: Leaving] | 19:59 | |
| -!- delinquentme [~asdfasdf@rrcs-74-219-159-104.central.biz.rr.com] has joined ##hplusroadmap | 20:06 | |
| @kanzure | ugh. now people are spamming diybio with their surveys. today is a survey from "desktop genetics". | 20:23 |
| yashgaroth | "How many plasmids are you making at the moment" man who even knows these days how many plasmids they're making | 20:24 |
| @kanzure | "2mg" | 20:25 |
| yashgaroth | Q: How many people are in your lab and who also make plasmids? | 20:27 |
| yashgaroth | o Less than 2 | 20:27 |
| yashgaroth | o 2-5 | 20:27 |
| @kanzure | hmmm | 20:27 |
| yashgaroth | if only there were a shorter way to say '1', oh wait | 20:27 |
| @kanzure | nobody roles their own plasmids | 20:27 |
| @kanzure | there's not going to be multiple people makin' new plasmids | 20:27 |
| yashgaroth | is it too passe to say that making 20 different gene compiler programs doesn't make things easier for biologists? | 20:29 |
| @kanzure | nope, go right ahead and say it | 20:30 |
| @kanzure | you should tell them to just work on a single open source project instead of making proprietary crap that doesn't even work yet | 20:31 |
| yashgaroth | exactly, it's not like there's megabux to be made in this software | 20:31 |
| @kanzure | i don't even know what they are expecting to do.. make visualization widgets? that's what that "herf derf i'm with mozilla" guy keeps blabbering about. | 20:31 |
| @kanzure | connor. | 20:31 |
| -!- augur [~augur@c-69-250-16-111.hsd1.md.comcast.net] has quit [Remote host closed the connection] | 20:32 | |
| yashgaroth | trying to apply web 3.0, or whatever version we're on, to biology somehow just speaks to their ignorance of biology | 20:32 |
| @kanzure | eh he might be going for desktop dna synthesis ("desktop genetics") | 20:33 |
| yashgaroth | nah | 20:33 |
| * delinquentme is running biology 6.04 Natucket nanner | 20:40 | |
| * delinquentme nods | 20:40 | |
| delinquentme | alsoooo if anyone is in ohio feel free to come hang out :D | 20:40 |
| yashgaroth | oh apparently they do have bio backgrounds, which explains why their linkedin links on "gust" are broken | 20:41 |
| yashgaroth | also http://m4.licdn.com/media/p/4/000/17d/3fd/2847097.jpg "hundread" sounds legit | 20:42 |
| @kanzure | $7500 for a survey? | 20:42 |
| -!- SDr [SDr@unaffiliated/sdr] has quit [Remote host closed the connection] | 20:58 | |
| @kanzure | wow there's an attendance fee to go to genspace meetings? | 21:15 |
| -!- cpopell [ad42e490@gateway/web/freenode/ip.173.66.228.144] has joined ##hplusroadmap | 21:20 | |
| cpopell | What's up all? | 21:20 |
| -!- roksprok [~Zac@adsl-64-169-176-150.dsl.snfc21.pacbell.net] has joined ##hplusroadmap | 21:40 | |
| -!- mode/##hplusroadmap [-b klafka!*@*] by kanzure | 21:40 | |
| -!- augur [~augur@208.58.5.87] has joined ##hplusroadmap | 21:42 | |
| -!- Cat4D [182bc9d2@gateway/web/freenode/ip.24.43.201.210] has joined ##hplusroadmap | 21:56 | |
| -!- skorket [~skorket@cpe-24-58-232-122.twcny.res.rr.com] has joined ##hplusroadmap | 22:05 | |
| skorket | just got back from maker faire | 22:05 |
| brownies | which one? | 22:10 |
| skorket | nyc | 22:11 |
| -!- roksprok_ [~Zac@adsl-75-36-187-206.dsl.pltn13.sbcglobal.net] has joined ##hplusroadmap | 22:27 | |
| -!- roksprok [~Zac@adsl-64-169-176-150.dsl.snfc21.pacbell.net] has quit [Read error: Connection reset by peer] | 22:28 | |
| -!- roksprok [~Zac@adsl-64-169-176-150.dsl.snfc21.pacbell.net] has joined ##hplusroadmap | 22:30 | |
| -!- roksprok [~Zac@adsl-64-169-176-150.dsl.snfc21.pacbell.net] has quit [Client Quit] | 22:31 | |
| -!- roksprok_ [~Zac@adsl-75-36-187-206.dsl.pltn13.sbcglobal.net] has quit [Ping timeout: 255 seconds] | 22:31 | |
| -!- cpopell [ad42e490@gateway/web/freenode/ip.173.66.228.144] has quit [Ping timeout: 245 seconds] | 22:32 | |
| brownies | at 1am on a sunday? | 22:32 |
| yashgaroth | sounds about right | 22:33 |
| skorket | I'm in Ithaca | 22:34 |
| skorket | left at around 5pm, hit bad traffic, just got home | 22:34 |
| -!- Mokstar [~Nate@c-76-115-136-13.hsd1.or.comcast.net] has joined ##hplusroadmap | 22:46 | |
| -!- Mokstar1 [~Nate@c-76-115-136-13.hsd1.or.comcast.net] has quit [Ping timeout: 255 seconds] | 22:46 | |
| -!- marainein [~net@2001:388:608c:6cb5:613f:d749:c60c:a0db] has quit [Quit: Ex-Chat] | 22:53 | |
| -!- Mokstar1 [~Nate@c-76-115-136-13.hsd1.or.comcast.net] has joined ##hplusroadmap | 22:55 | |
| -!- Mokstar [~Nate@c-76-115-136-13.hsd1.or.comcast.net] has quit [Ping timeout: 245 seconds] | 22:56 | |
| -!- yashgaroth [~f@cpe-66-27-117-179.san.res.rr.com] has quit [Quit: Leaving] | 23:05 | |
| Mariu | Ithaca - New York ? | 23:07 |
| Mariu | nice | 23:07 |
| Mariu | got a friend there | 23:07 |
| Mariu | what're you doing in Ithaca? Are you with Cornell? | 23:19 |
| -!- ThomasEgi [~thomas@pd95bdd66.dip0.t-ipconnect.de] has joined ##hplusroadmap | 23:32 | |
| -!- ThomasEgi [~thomas@pd95bdd66.dip0.t-ipconnect.de] has quit [Changing host] | 23:32 | |
| -!- ThomasEgi [~thomas@panda3d/ThomasEgi] has joined ##hplusroadmap | 23:32 | |
| -!- Cat4D [182bc9d2@gateway/web/freenode/ip.24.43.201.210] has quit [Ping timeout: 245 seconds] | 23:54 | |
| --- Log closed Mon Oct 01 00:00:42 2012 | ||
Generated by irclog2html.py 2.15.0.dev0 by Marius Gedminas - find it at mg.pov.lt!
