Back
[00:04:21] <rue_shop4> a mouse had the balls to walk into my shop while I was sitting here
[00:04:28] <rue_shop4> *zing*
[00:04:30] <rue_shop4> anyhow
[00:05:00] <rue_shop4> its hard to design a flipflop like that around the negitive region of a 2222 cause its to small
[00:19:15] <rue_house> on a neon, its a few 100 volts wide, on the 2222, its 0.5v
[00:20:43] * katsmeow nods
[00:21:58] <rue_house> 25 ohms or so over 20mA of an led
[23:16:28] <rue_shop4> the only thing
[23:16:38] <rue_shop4> not handling the test of my half bridge driver
[23:16:46] <rue_shop4> is the breadboard and the load resistor
[23:16:52] <rue_shop4> and the power supply
[23:17:04] <rue_shop4> so far, the driver is just laughing it up
[23:17:11] <rue_shop4> katsmeow, :)
[23:17:21] <katsmeow> so you are smoking the test jig, that's a good sign
[23:17:50] <rue_shop4> continious 2A, and I cant even detect any part of the driver getting warm
[23:18:24] <katsmeow> what happens when you power the motor with it?
[23:18:44] <rue_shop4> I thought the votage drop had gone way up since I tested the circuit, but its all breadboard and power supply
[23:18:55] <rue_shop4> its good, I even stuck in an inductor for a while
[23:19:28] <katsmeow> kool, congrats
[23:20:19] <rue_shop4> at 2A, I'm losing 2.5V between the power supply and the breadboard connectors
[23:20:54] <rue_shop4> 2V is across the 1R resistor, the rest is actual driver drop
[23:21:29] <rue_shop4> kat, what are you workin on latley
[23:21:39] <rue_shop4> according to a map I saw, its still pretty hot up there
[23:21:56] <katsmeow> very hot, and i am mostly up on the roof :-/
[23:22:17] <katsmeow> patching the leaks , now that they finally let me worrk on th e house again
[23:24:54] <rue_shop4> hmm, I want a 4V signal at 2A
[23:25:25] <rue_shop4> at 2A, the 0.5R sense has 1V
[23:25:33] <rue_shop4> so, I want a gain of 4
[23:25:52] <rue_shop4> its a non-inverting amp
[23:26:07] <rue_shop4> so if the input resistance is 100k
[23:26:14] <rue_shop4> I want the feedback to be
[23:26:20] <rue_shop4> 300k?
[23:27:23] <katsmeow> that's a dogawefully high resistor value in a high current motor h-bridge driver
[23:27:40] <rue_shop4> its just a 2A driver tho
[23:27:46] <rue_shop4> actually, you have a point
[23:27:56] <rue_shop4> its that high because I was gonna put it direct into the adc
[23:28:02] <rue_shop4> but I need a peak detector
[23:28:09] <rue_shop4> so I'm throwing in gain too
[23:28:16] <katsmeow> 12v across 300k is 40 microamps
[23:28:37] <rue_shop4> oh, yea, the Rf bleeds off the peak holding cap
[23:28:47] <katsmeow> i guess you have not checked email in a few days agin
[23:28:49] <rue_shop4> amp output -> diode -> capacitor
[23:29:00] <rue_shop4> I asnwered this afternoon
[23:29:03] <rue_shop4> I think
[23:29:08] <katsmeow> so i just paind $28 for some sodimm 144p sockets
[23:29:28] <katsmeow> you replied only about books
[23:29:32] <rue_shop4> oh yea, I didn't sit down on that one, needed to ask more detail
[23:29:44] <rue_shop4> I sit down with that stuff last thing at night
[23:29:59] <rue_shop4> almost anytime I'm trying to do a good hunt
[23:30:16] <katsmeow> well, i got as many sockets as i have sodimms to fit into them
[23:31:22] <katsmeow> winter project this yr might be replacing the 64k of ram in a C64 wth 64M of ram
[23:31:29] <rue_shop4> wholy #!@$% china wants $31 for 5pcs!
[23:31:41] <katsmeow> yeas, they are very pricey, dunno why
[23:31:48] <rue_shop4> I dont even know if thats the right one
[23:32:32] <katsmeow> afaik, the p144 socks fit only sdr sdrams, and i'd be surprised if they still make them for laptops anymore
[23:32:44] <rue_shop4> at that rate I should strip the ones out of the printer control boards and sell them
[23:33:32] <rue_shop4> were you looking for just one then?
[23:33:42] <katsmeow> no
[23:33:55] <rue_shop4> how do you do the refresh if the C64 can only access up to 640k
[23:34:09] <katsmeow> the C64 can acces only 64k
[23:34:11] <rue_shop4> does the old refresh even work on new ram?
[23:34:24] <rue_shop4> 64kbytes
[23:34:37] <rue_shop4> I assumed you meant your C64 only had 64kbits
[23:34:43] <katsmeow> the oem refresh is controlled by the video chip, the sdrams have build in refresh
[23:34:58] <rue_shop4> but you have to tell them to do it?
[23:35:03] <katsmeow> 65536 bytes
[23:35:11] <rue_shop4> yea
[23:35:30] <katsmeow> who cares? the sdram is so fast the cpu won't notice it
[23:35:46] <rue_shop4> its the system that does it that I'm wondering about
[23:36:00] <rue_shop4> cause, something would need to i presume
[23:36:33] <rue_shop4> so I can sell ram sockets at like $4 to $5 ea
[23:36:54] <katsmeow> i am less concerned about that than simply access the 26bit address buss with a cpu that's got only 16bit address buss
[23:37:23] <rue_shop4> well you make a protected space (the first 16k?) and page the rest
[23:37:26] <katsmeow> usng pointers, 26bits if an odd register size
[23:37:51] <rue_shop4> nothng could be odder than x86 memory bits
[23:38:02] <katsmeow> i have another idea: decentralise the pu
[23:38:38] <katsmeow> pas ponter addresses to a math unit, for instance, dma moves to a dma unit
[23:39:09] <rue_shop4> you need two machines, a paged C64 and the KAT machine
[23:40:20] <rue_shop4> or start programming graphics processors (126 vector processors on one chip)
[23:40:30] <katsmeow> a paged machine in 16k of *useable* ram (16k for two variables = 32k of ram, out of 64k total) will still be very slow processing a webpage
[23:41:10] <rue_shop4> more bits didn't help out much, more processors did
[23:41:26] <rue_shop4> 126 4Ghz vector processors
[23:41:38] <rue_shop4> I think my guess of 126 is low too
[23:45:14] <katsmeow> intel sells a cpu with 128 cores(?) , $3000 ballpark
[23:47:21] <rue_shop4> "Stacked with 5760 cores and 12 GB of memory, this dual GPU gives you the power to drive even the most insane multi-monitor displays and 4K hyper P"
[23:47:47] <rue_shop4> and each core, as I under stand, has multiple vector machines in it
[23:48:24] <rue_shop4> "A Quadro K5000 has 1536 CUDA cores and a GTX 780 has 2304"
[23:48:58] <katsmeow> the Propeller 2 is supposed to be coming out this yr with insane specs, but it's got horrible bandwidth issues, too few pins on the chip
[23:49:09] <rue_shop4> I suspect running an sentient AI realtime on a GPU would not be a problem
[23:50:13] <rue_shop4> GTX980ti- 2816 cores
[23:50:13] <rue_shop4> GTX780ti- 2880 cores
[23:50:13] <rue_shop4> Titan X- 3072 cores
[23:50:13] <rue_shop4> Titan Z- ("Dual GPU card") 2880*2= 5760 cores)
[23:50:13] <rue_shop4> Titan Black- 2880 cores
[23:51:00] <katsmeow> good grief
[23:51:11] <rue_shop4> "The news came out today that the first Kepler GPU, the GeForce GTX 680, has been launched. A single GPU has 1,536 CUDA Cores. This means that those high-end workstations with 8 PCIe slots will be able to pack 12,288 CUDA cores into a single computer. That’s some serious computational power."
[23:51:36] <rue_shop4> I think I should learn more about CUDA cores
[23:51:51] <katsmeow> with one pipe to the hdd, and one to the ethernet
[23:52:20] <rue_shop4> yea, thats the problem I had with my optical processor
[23:52:23] <rue_shop4> its io bound
[23:52:38] <rue_shop4> processor running at 4Thz, always waiting
[23:52:48] <katsmeow> heh, yeas
[23:53:01] <rue_shop4> even for ram
[23:53:15] <rue_shop4> which would be just a legth of optical fiber
[23:53:29] <rue_shop4> it would have to wait for the data to come around agian
[23:54:13] <rue_shop4> the good news it that everything is going serial anyhow
[23:54:33] <rue_shop4> so aligning parallel optical paths is less of an issue
[23:54:56] <rue_shop4> which is good cause the laser trimming would get beyond insane
[23:55:11] <rue_shop4> esp cause you need to account for lightwave phase
[23:55:36] <rue_shop4> didn't work out thermal control either
[23:55:53] <rue_shop4> I suppose if you run hot, its more controllable