#garfield Logs
Feb 19 2015
#garfield Calendar
02:26 katsmeow-afk 20F here, low teens in N Alaforkingbama
04:13 katsmeow-afk fwiw, i just discovered the door to my "garage space" is of the same design, and roughly the same size, as the front door to Bletchly Park
04:28 katsmeow-afk 4:06 AM CST on February 19, 2015
04:28 katsmeow-afk Clear
04:28 katsmeow-afk 18.3 °F
04:48 katsmeow-afk hmm, Brits say "cemetery" and "symmetry" the same way
05:50 katsmeow-afk Of course a machine cannot think like a human, it's not a human. But since it is thinking differently, how can you say it is not thinking?
05:52 katsmeow-afk The smarter question is: when do you know it is thinking, if it is capable out being smarter than you, and doesn't want you to know it is thinking?
06:30 katsmeow-afk [05:55] <RifRaf> http://i.imgur.com/vFq7Igp.jpg
06:31 katsmeow-afk 10pm Rif's time
06:50 katsmeow-afk rue, i figured out how to make a state machine chose if-then breaks and jumps: align the end of the if- with an odd memory location, plug in the output of the comparator or status register bit to the lsb of the next sequential address, the contents of the address so chosen will be the actual addy to jmp to
06:51 katsmeow-afk the if- ends on the odd, the next addy will end (LSB) in 0 , the one after ends in 1 , that's the bit in the status register, the overflow bit, or the sign bit, or the greaterthan bit, or etc
06:52 katsmeow-afk the side effect is you need a wide ram that holds all the addresses, or you are limited to byte-wide jmps
06:53 katsmeow-afk if you use a ram that holds only addresses, then you don't need an address counter, only a latch, because the next program addy in output from the addy ram
06:55 tsmeow-afk listens to Herman's Her
06:56 katsmeow-afk i am thinking of this mostly because i have those 1200 rams from the 100's of cable tv boxes
06:57 katsmeow-afk even as slow as they are, if i use them all, the combined clk rate is 10Ghz
06:58 katsmeow-afk if i change the basic design to use 20ns rams, the clk rate is 60Ghz
07:00 katsmeow-afk if there was just some way to hybrid the descrete address ram and a traditional address counter scheme
07:01 katsmeow-afk the descrete ram would eliminate the addy miss penalty of 50% of jump instructions, the counter would free up 2x as many ram chips for real instructions
07:04 katsmeow-afk an exciting speedup of that scheme of indexing using the status register tothe LSB of the next address is you can combine status reg bits if you align the end of the if- right, so you can do "is negative and greater than" , etc
07:04 katsmeow-afk you land on 1 of the next 4 instructions, that's the next instruction addy
07:05 katsmeow-afk err
07:05 katsmeow-afk you land on 1 of the next 4 *addresses*, that's the next instruction addy
07:07 katsmeow-afk you simply load it normally into the latch for the next instruction
07:08 katsmeow-afk but if you don't use a separate ram for the addresses, how do you do that?
07:09 katsmeow-afk i think it gets real pipeline-y, and that means speed hits
07:09 katsmeow-afk or you haveto set up a jmp table somewhere, and that gets squirrelly
07:10 katsmeow-afk if you don't pipelinethe addy counter, you haveto include it's time with the ram accesstime, that alone could halve the cpu clock rate
07:10 katsmeow-afk from 20ns to 40ns
07:12 katsmeow-afk every pipeline where the next addy is precomputer will have ahigh failure rate with any jump (brk, if-then, goto, gosub, etc)
08:01 katsmeow-afk twas 14F here at sunrise today
08:33 katsmeow-afk Lenovo has been selling laptops that come loaded with Superfish 'malware'
08:47 katsmeow-afk US gluten imports - mostly from Australia, Canada, China, and Europe - more than doubled between 1997 and 2007, reaching 386 million pounds
08:52 katsmeow-afk i wonder why the usa is importing 386 million pounds, or 193,000 tons, of ANY food ingredient
09:19 rue_more because it cant produce for itself anymore
09:19 katsmeow-afk going to get worse with the drought
09:20 katsmeow-afk woo, temperature is up now to 19 F / -7 C !
09:21 katsmeow-afk so the trick to fast indexing into a jmp table is to just align the table address to end in 0, or 00 , or 000, depending on how may conditions you wanna test for at the same time
09:22 katsmeow-afk then apply the status register bits to the LSB of the address
09:27 katsmeow-afk you could test for <, =, and > in one swoop
09:28 katsmeow-afk the jump would land on one of 3 addresses in the 8-addy table, which would be a jmp address to whatever code handles that condition
09:28 rue_more maybe we should do away with addressing and bits, just makea matrix that stores data , without bounding how its actaully stored
09:29 rue_more or where
09:29 katsmeow-afk hmm?
09:29 rue_more something thats not designed, but works because of the way its built
09:29 katsmeow-afk so don't design the way it's built?
09:30 rue_more you make it not do what you dont want it to do, meaning what it does it some form of what you want
09:31 tsmeow-afk offers an instruction to NOT do anything but add 1
09:31 rue_more :)
09:31 katsmeow-afk but it looks exactly like an instructon to add 1 + 1
09:32 katsmeow-afk how would you build your matrix with 8kx8 static rams?
09:32 rue_more we use numbers to represent letters that represent sounds that represent thoughts
09:33 rue_more how many layers do you think might be in the brain?
09:33 katsmeow-afk umm,,,, 12?
09:33 rue_more is there 1 neuron that represents a kettle?
09:33 rue_more what if that neuron dies?
09:34 katsmeow-afk then you forget what a kettle is, exactly,, it maybecome a "water carrier", or "bowl with a handle"
09:34 rue_more how many neurons have to die before I cant remember the word I'm looking for
09:35 tsmeow-afk tries to remember that
09:35 katsmeow-afk you can rememberit to a new neuron as soon asyou remember a expereince where you did name it properly
09:36 rue_more what if the origional kettly neuron starts responding again?
09:37 katsmeow-afk problem : if you arenot using actualword tags, then how you store the data hastobe far more rigid
09:37 katsmeow-afk then it's easier to remember
09:37 katsmeow-afk knowing theitemname, and knowing anexpereince when you knew the item's name, isn't exclusive
09:38 katsmeow-afk if you decideto number from one to 2^32 items, and use the binary number as the name, you can't just store that name anywhere, because it's unrecogniseable as an item name
09:39 katsmeow-afk you'll need to store it in a definite location where you expect item names
09:39 rue_more if two of those form, and the word I want isn't in the one I'm looking in, I might be stuck not able to remmember the word
09:40 rue_more untill I stop looking, back off, and let the room behind the door change
09:40 rue_more then look again
09:41 katsmeow-afk but NAME memory isn't the same as EXPERIENCE memory, if you make a pass thrut he one that hasn't forgotten, then you can re-write the one that did
09:42 rue_more this might happen if a conciousness varries it occupation of several memory spaces, each of which has different details or contexts
09:42 rue_more if the conciousness is in the wrong memory space, it cant recall what it wants
09:43 rue_more but like you said
09:43 rue_more some details are always available to it
09:43 katsmeow-afk but recalling is program, what it's looking up is in data
09:43 rue_more hmm
09:43 katsmeow-afk you keep program and data separate
09:45 rue_more where are memories stored?
09:45 katsmeow-afk data store
09:45 rue_more I dont recall an area of the brain that can be poked to invoke childhood memories
09:46 katsmeow-afk https://duckduckgo.com/?q=location+of+memories+in+human+brain
09:47 katsmeow-afk during brain surgery, poking around caneven make poeple taste or smell things that aren't there
09:48 katsmeow-afk or hear songs
09:51 katsmeow-afk i suspect the best way to make memories is a linear "tape recording" log, and then for each item that is being expereinced, it's data isupgraded, like "the cat had 4 kittens at 4:12 am"
09:51 katsmeow-afk that way tou can remember just the cat stuff from onelocation, or the istory of the morning from the other,,a ndyou canrebuild either from the other
09:52 katsmeow-afk otherwise you spend alot of time searching, brute-forcing, testing
11:06 katsmeow-afk https://duckduckgo.com/?q=robot-feeds-marathon-runners-tomatoes&ia=images
13:45 Tupolev pasta time
13:47 Tupolev hey
13:47 le_noob_haxor what is this
13:47 Tupolev trash
13:47 Tupolev :(
20:17 katsmeow-afk so i come back tothe puter, it's off, i fire it up, get a dozen google updater and google other stuff popping up, so i doa search and destroy in exploder foor google and chrome, then in regedit, come to find out, there's also gupdater and meebo and otehr stuff
20:34 katsmeow-afk it also looks like google pushes updates to chrome, even several times per day
20:34 rue_more wow
20:36 katsmeow-afk it tried to "sync" me to stuff i had never signed up for
20:38 rue_more almost have to make an outbound network filter
20:38 e_more closes ch
20:42 rue_shop3 there was a real house fire today
20:42 rue_shop3 kitchen fire
20:42 rue_shop3 set sprinklers off
20:42 rue_shop3 building next to work office
20:42 rue_shop3 I was there
20:42 katsmeow-afk sprinklers in a house?
20:42 rue_shop3 rattled me, didn't expect that
20:42 rue_shop3 condo
20:43 rue_shop3 dealing with a fire is one thing
20:43 rue_shop3 doing it while there is a 65db bell riing in your ear is another
20:43 rue_shop3 65db from 10' I think
20:44 tsmeow-afk looks at all the iron pipe she bought to do sprinklers here, 100's of ft of it, and ended up cutting it all into fence stakes because of
20:44 katsmeow-afk 65db isn'tthat loud, many peole normally talk that loud
20:44 rue_shop3 oh must have forgotten my numbers
20:46 rue_shop3 ah, this page says 86
20:46 rue_shop3 db
20:46 katsmeow-afk yeas, 90db gets annoying really fast
20:49 katsmeow-afk <fireman> why are all these people inhere?? <tenant> i don't know, the alarm sounded, and people began running in from all over the place, i have never seen this many people around here!
21:26 katsmeow-afk Wedding ends in brawl after bride marries guest instead of groom
21:26 katsmeow-afk Fox News Staff
21:28 katsmeow-afk www.pcworld.com/article/2886357/lenovo-preinstalls-man-in-the-middle-adware-that-hijacks-https-traffic-on-new-pcs.html
21:42 katsmeow-afk Backblazes storage infrastructure, which it refers to as "Storage Pods". The company has almost 40,000 drives in its data center, and a study over the past several years shows that SMART software stats often don't relate to the health of those drives.
21:44 katsmeow-afk For example, SMART stat 187 reports the number of reads that could not be corrected using hardware error correction code (ECC). Drives with 0 uncorrectable errors hardly ever fail, Budman said, "but once SMART 187 goes above 0, we schedule the drive for replacement."
21:48 katsmeow-afk ok, i go away then
22:18 katsmeow-afk wtf is "purpiss" ??
22:19 rue_more 12 Power_Cycle_Count 0x0032 100 100 000 Old_age Always - 615
22:19 rue_more 190 Airflow_Temperature_Cel 0x0022 071 057 045 Old_age Always - 29
22:19 rue_more dont have a 187
22:19 rue_more 12 Power_Cycle_Count 0x0032 100 100 000 Old_age Always - 75
22:19 rue_more 192 Power-Off_Retract_Count 0x0032 200 200 000 Old_age Always - 50
22:19 rue_more or that one
22:19 rue_more 12 Power_Cycle_Count 0x0032 100 100 000 Old_age Always - 71
22:19 rue_more 192 Power-Off_Retract_Count 0x0032 200 200 000 Old_age Always - 50
22:20 rue_more or that one
22:20 katsmeow-afk i'm sorry
22:20 rue_more hmm
22:20 rue_more huh
22:20 rue_more Power_On_Hours 0x0032 055 055 000 Old_age Always - 32886
22:20 rue_more I'v gone 32886hrs with only about 75 power cycles
22:21 rue_more 435hrs average
22:21 rue_more :/ 18 days?
22:21 rue_more well that sucks
22:21 rue_more its an average tho, it happens in bursts
22:22 rue_more the drive with 615 power cycles is a used drive I have in there right now
22:22 rue_more not one I bought new
22:25 tsmeow-afk notes rule of thumb is a drive is prone to failure after being on 24-7 for 5 yrs, they sell 15 yr old drives on ebay, and most of my drives are 15 yrs old
22:26 rue_more mine run 24/7
22:27 rue_more so my drive are 3.75yrs old
22:27 rue_more +- loooong power outages
22:27 rue_more ^s
22:28 rue_more 1 Raw_Read_Error_Rate 0x002f 200 200 051 Pre-fail Always - 0
22:28 rue_more I watch...
22:28 rue_more 196 Reallocated_Event_Count 0x0032 200 200 000 Old_age Always - 0
22:28 rue_more if it starts relocating sectors, its going down
22:29 rue_more the last number is the count
22:29 rue_more first number is the param num
22:29 rue_more 198 Offline_Uncorrectable 0x0030 200 200 000 Old_age Offline - 0
22:30 rue_more can you get smart data in windows?
22:31 katsmeow-afk https://en.wikipedia.org/wiki/Comparison_of_S.M.A.R.T._tools
22:32 rue_more can YOU get smart data in windows
22:32 katsmeow-afk not yet
22:32 katsmeow-afk my next 45 minutes or so is prettry booked
23:13 katsmeow-afk Power On Count 208 times
23:13 katsmeow-afk Power On Time 2462.8 days
23:18 katsmeow-afk the Seagate and two WD don't have the same SMART parms reported
23:24 katsmeow-afk 11:00 PM CST on February 19, 2015
23:24 katsmeow-afk Clear
23:24 katsmeow-afk 16.5 °F
23:25 katsmeow-afk that must be wrong
23:28 katsmeow-afk ok, see, this bites, and no one mentions it (granted this is an old puter, but the same applies tonew puters too) : 166Mhz sdram, which is 6ns clock , BUT cycle time is 7 clocks, which is 42ns or **_23_MHZ_**
23:49 katsmeow-afk hmm, this 15 yr old drive has only 115 days powered on ?
23:49 rue_more yes,a nd no, latency vs bandwidth
23:49 rue_more each cycle I think you start a new request
23:49 rue_more new computing LOVES lots and lots of latency
23:50 katsmeow-afk well, burst more is one addy per cycle, but if you cross a row , you lose burst mode
23:50 katsmeow-afk and not all memory support long bursts
23:50 rue_more imagine being able to say 166Mhz ram but not having to say 42ns latency
23:51 rue_more must be intel, AMD always had faster ram than that
23:51 rue_more intel has always been about good core speed and poor io speed
23:52 rue_more amd has good io speed for compromised core speed
23:52 rue_more as far as I'm concerned, 99% of everything a computer does is prettymuch just moving data around, so I always use amd
23:52 tsmeow-afk