|
acetone vapor is nature's antialiasing
|
#
?
Oct 19, 2017 02:17
|
|
|
|
| # ? Jun 9, 2024 22:39 |
|
|
Sagebrush posted:please post lots more about this because it's extraordinarily cool to me and also relevant to what i'm teaching the kids it's definitely true that current fdm tooling and machines are focused on doing polygons and line segments and have a lot of higher order behavior based on that. the thing you have to realize is the limiting factor of resolution is legitimately not hardware aside from the rigidity thing. especially for what you mention with interior holes, a lot of that is driven by the assumption that line segments are the controlling primitive. the thing about fdm cam is that everything is controllable. the reason that interior holes decay into polygons is a) if you assume that your machine draws straight segments, paths between segments are.... straight b) the vertices are what you control, and you do a good job of making sure that the vertices are where the STL says they are So you draw lines between vertices and you do your best to make sure the vertices stay where they are by making two or more outlines of an interior hole to distribute the load created by the shrinking of the plastic. and they turn into polygons anyway because of a couple reasons, against based on the fdm machines i've worked on, which are 1. computation time and execution speed is a tradeoff, from a system design perspective, if you want a machine that is self hosting 2. computation time is driven by, shockingly, segments/sec 3. if you are a company that makes both machine and pather, because you are able to control both computation and execution time, and (very importantly) because your base assumption is that you work with a triangular 3d model, you're making conscious choices in your pather about your quantization level based on your already-quantized input mesh So you set a minimum sustained segment sized based on your understanding, usually based on experimentation, of your throughput. And interior holes are the worst case for that because they're small radius circles. So you quantize them down because that's better than having motion control buffer exhaustion based gantry halts that you didn't plan for. And since your toolpather controls vertices (and does a good job, also based on experimentation, of anchoring those vertices against the so-called "inner shell") the beads undergo thermal contraction between the well-anchored vertices and... turn into a polygon. And t hat's all setting aside the mechanical accuracy of the system, which to be honest is strictly on a you get what you pay for basis. To be fair the pather is too, less in terms of the algorithms and more in terms of a) how well characterized the machine is (which is also very dependent on 1) the quality of your process and manufacturing and 2) your volume) and b) the generality of the pather; both with how much time and brainpower the pather team has, and how closely defined their product targets are in terms of what they support.
|
|
#
?
Oct 19, 2017 03:21
|
|
|
Sapozhnik posted:well, the 1000 fps figure on its own means nothing. it could be 1000 fps while drawing 300 amps of current. scale that down to 60 fps and that's still 18 amps. I'm the 300 Amp gpu
|
|
#
?
Oct 19, 2017 06:03
|
|
|
Bloody posted:I'm the 300 Amp gpu um its called a mining rig
|
|
#
?
Oct 19, 2017 06:38
|
|
|
Bloody posted:I'm the 300 Amp gpu i thought this was crazy the first time i heard it but turns out: nope modern gpus run at ~1v logic, and the newest gtx titan z has a tdp of 375 watts 340 fuckin' amps running through that processor even my little 1060 is 120w = 110 amps. as much as i use to tig-weld 1/8" steel
|
|
#
?
Oct 19, 2017 06:51
|
|
|
Sapozhnik posted:well, the 1000 fps figure on its own means nothing. it could be 1000 fps while drawing 300 amps of current. scale that down to 60 fps and that's still 18 amps.  in general raw perf to ppw is reasonable to do assuming even sublinear scaling with fps -- just race to idle on vsync
|
|
#
?
Oct 19, 2017 12:54
|
|
|
I want to figure out how to upload code to the AMD29000 on my Apple Display Card 8•24GC that would be a fun way to display a rotating cube just need an AMD29000 disassembler and assembler you can already treat it as a blitter by creating an on-card GWorld containing your sprites and then using CopyBits() to blit them to the screen; once everything is on the card, none of the work will go through the CPU
|
|
#
?
Oct 19, 2017 21:21
|
|
|
Sagebrush posted:i thought this was crazy the first time i heard it but turns out: nope im also the use of amps when watts would be more meaningful
|
|
#
?
Oct 20, 2017 01:53
|
|
|
Bloody posted:im also the use of amps when watts would be more meaningful yeah well im as powerful as a 120 volt battery 
|
|
#
?
Oct 20, 2017 02:10
|
|
|
i take 3ohm dumps
|
|
#
?
Oct 20, 2017 02:42
|
|
|
Bloody posted:im also the use of amps when watts would be more meaningful hw people always quote amps and not watts, i'm a prick
|
|
#
?
Oct 20, 2017 03:10
|
|
|
power dissipation in an individual section of a circuit is I2R. quoting wattage is only really useful for telling you the overall power consumption.
|
|
#
?
Oct 20, 2017 03:44
|
|
|
overall power consumption is a much better measure of how hard the card is working than the current through any given point, though
|
|
#
?
Oct 20, 2017 03:46
|
|
|
as someone who doesn't do mobile i don't really care as long as the card isn't on fire actually i wouldn't care about that either if it kept working anyway
|
|
#
?
Oct 20, 2017 04:08
|
|
|
Jabor posted:power dissipation in an individual section of a circuit is I2R. quoting wattage is only really useful for telling you the overall power consumption. that's just Joule heating there's also other frequency and voltage dependent CMOS switching losses there's a company that uses servers to provide space heating in homes with fiber connections since computers are basically rocks that get hot and can occasionally add numbers together but principally just get hot
|
|
#
?
Oct 20, 2017 09:52
|
|
|
Bloody posted:I'm the 300 Amp gpu my 980Ti draws north of 300 amps (~450W@1.38V) when overclocked as far as it'll go 
|
|
#
?
Oct 20, 2017 11:14
|
|
|
Malcolm XML posted:that's just Joule heating there's also other frequency and voltage dependent CMOS switching losses  is that like in apartment complexes or something because that seems like it would get outrageously expensive real quick
|
|
#
?
Oct 20, 2017 16:08
|
|
|
Farmer Crack-rear end posted:
i assume the server company would subsidize at least part of the electricity since otherwise it's just an even more inefficient electric furnace that "wastes" some of the power it could be turning into heat by shuffling bits around
|
|
#
?
Oct 20, 2017 16:16
|
|
|
ate all the Oreos posted:i assume the server company would subsidize at least part of the electricity since otherwise it's just an even more inefficient electric furnace that "wastes" some of the power it could be turning into heat by shuffling bits around the shuffling bits around still manifests as heat tho
|
|
#
?
Oct 20, 2017 16:18
|
|
|
yeah but weve been working towards intentionally reducing TDP on servers. now if you got a early-mid 2000s era netburst xeon..........
|
|
#
?
Oct 20, 2017 16:30
|
|
|
Cocoa Crispies posted:the shuffling bits around still manifests as heat tho actually yeah for some reason i thought information would consume energy in some kind of magical other way
|
|
#
?
Oct 20, 2017 16:38
|
|
|
I suppose technically some of those electrons get sent up the wire to your monitor where they are fractionally converted into light instead The light turns into heat when it hits your body or the walls though. Can't get away from it E: no, the monitor backlight is actually not going to be electrically connected to the data input. hmm
|
|
#
?
Oct 20, 2017 16:48
|
|
|
yo its real easy. if mechanical work is done (objects physically move) then it wont be 100% heat. the only things that apply here are: hard drives spin optical disk drives spin fans spin speaker cones move if you're 100% solid state and dont have speakers then ever bit of energy that comes in except fans is turning into heat
|
|
#
?
Oct 20, 2017 16:53
|
|
|
ate all the Oreos posted:i assume the server company would subsidize at least part of the electricity since otherwise it's just an even more inefficient electric furnace that "wastes" some of the power it could be turning into heat by shuffling bits around oh hell it didn't even occur to me the occupant would be paying for the electricity, i just meant in terms of the fiber connections. wait, these aren't fuckin' rackmount servers are they? god jesus who would want that much noise in their home just for some piddling heating
|
|
#
?
Oct 20, 2017 17:44
|
|
|
NEED MORE MILK posted:yo its real easy. if mechanical work is done (objects physically move) then it wont be 100% heat. doesn't it eventually turn into heat anyway since the energy use after the fans/spindles get to speed is to keep them spinning against frictional losses?
|
|
#
?
Oct 20, 2017 17:45
|
|
|
Farmer Crack-rear end posted:oh hell it didn't even occur to me the occupant would be paying for the electricity, i just meant in terms of the fiber connections. people with lots of space had coworkers in colorado with testing clusters in their basement, and they just didn't notice it over household noise in the rest of the house
|
|
#
?
Oct 20, 2017 18:59
|
|
|
as a person with a rack in their house: you cant hear anything unless youre right outside the door. just put it in a utility room or something
|
|
#
?
Oct 20, 2017 19:04
|
|
|
Farmer Crack-rear end posted:oh hell it didn't even occur to me the occupant would be paying for the electricity, i just meant in terms of the fiber connections. I CANT EBELIEVE YOU HAVE ANEW AVATAR
|
|
#
?
Oct 20, 2017 19:05
|
|
|
NEED MORE MILK posted:yeah but weve been working towards intentionally reducing TDP on servers. now if you got a early-mid 2000s era netburst xeon.......... eh, bullshit. we've been working towards reducing power per teraflop or whatever metric, yes. but modern server cpus with tens of threads are all 200-300w space heaters, and you can cheaply jam like 2-4 per 1u these days if you have the need. hairdryers in 1u format
|
|
#
?
Oct 20, 2017 19:08
|
|
|
Truga posted:eh, bullshit. if you're not at a kw / u ur not even trying
|
|
#
?
Oct 20, 2017 19:17
|
|
|
NEED MORE MILK posted:as a person with a rack in their house: you cant hear anything unless youre right outside the door. just put it in a utility room or something i have a closet that conveniently is where the fiber comes into the apartment, where the patch panel for the ethernet in the walls is, and it even has its own little A/C vent for cooling. I put my lil' half-height rack in there and just close the door and don't even hear it, though all my servers are 4-u ones with fancy quiet fans in them that aren't loud normally
|
|
#
?
Oct 20, 2017 20:18
|
|
|
yep too bad fios doesnt make a racked ONT, that would own
|
|
#
?
Oct 20, 2017 20:53
|
|
|
echinopsis posted:I CANT EBELIEVE YOU HAVE ANEW AVATAR yeah. i'm definitely gonna go back to space moose at some point but i felt this was too perfect an av not to rock for a bit.
|
|
#
?
Oct 20, 2017 23:22
|
|
|
Sagebrush posted:I suppose technically some of those electrons get sent up the wire to your monitor where they are fractionally converted into light instead in SEDs (OLEDs) it is
|
|
#
?
Oct 21, 2017 00:45
|
|
|
NEED MORE MILK posted:yo its real easy. if mechanical work is done (objects physically move) then it wont be 100% heat. lol how do you think actual work gets done? I mean you're like 99% right but the actual computation consumes some energy, just a very tiny bit of it space heater computer company: https://www.qarnot.com actually on topic: need more amps???? https://www.evga.com/articles/01140/evga-epower-v/ EVGA's got u, fam
|
|
#
?
Oct 21, 2017 00:49
|
|
|
most of a graphics card is power delivery fyi and maybe ram, but on HBM cards that's on the interposer so like 80% of the area is just for delivering power
|
|
#
?
Oct 21, 2017 00:50
|
|
|
Can't wait for FIVRs on GPUs
|
|
#
?
Oct 21, 2017 00:51
|
|
|
Malcolm XML posted:lol how do you think actual work gets done? I mean you're like 99% right but the actual computation consumes some energy, just a very tiny bit of it like the actual barriers for making MOSFETs better are thermal limits, same with solar cells
|
|
#
?
Oct 21, 2017 00:52
|
|
|
Malcolm XML posted:like the actual barriers for making MOSFETs better are thermal limits, same with solar cells on the electron scale ofc
|
|
#
?
Oct 21, 2017 00:54
|
|
|
|
| # ? Jun 9, 2024 22:39 |
|
|
Malcolm XML posted:lol how do you think actual work gets done? I mean you're like 99% right but the actual computation consumes some energy, just a very tiny bit of it how does computation itself consume energy? where does that energy go if it doesn't just become heat?
|
|
#
?
Oct 21, 2017 01:18
|
|