Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
TheFaxMisher, thanks. The Odroid XU4 heat sink shim idea was something on the back burner for a while, and that needed the 'details' for making sure the shim would clear all them surface mount capacitor solder ends, other wise the size of the stuff would be fine as a displacement map. Down side of using displacement maps is the stuff is not exactly visible in the view field, so it can make doing some stuff incredibly difficult without the actual geometry there to see (and in mesh editing programs).
I let myself get distracted with making a simple generic SSD and a memory dimm (sort of a custom creation). No, the 8-CPU Power8(tm) MCM is not included, for many reasons, lol. That stuff shouldn't need any smoothing trickery unlike the Odroid power connector, and I have no idea if it will even hold up to sub-d at all (again, why overcomplicate it, lol).
BTW, the only parts that have backs on the Odroid are the cube place-holders for plugs/buttons I've yet to make for the thing. I'm not sure why the chip sides vanished in Kettu's example, it may be a trick of the lighting possibly (shrugs shoulders).
Ah, that explains some of it, hex is moving stuff on me. Hex has a three digit precision for placing stuff, and some times it moves stuff the forth digit out when your messing with small stuff. 0.125 became 0.1247 with only the three-digit rounded up number showing in hex at the original 10x scale used to make the stuff.
Especially when you start selecting multiple faces to move them. I was working at ten times size in hex as it was barely enough for most of the stuff. I am now redoing some of the problematic parts at one-hundred times size (as shown in these screen-caps).
I am currently UV mapping this plug. With all the corners back at 90-degree angles, that odd shading appears to be addressed in Daz Studio.
Vat Da...
, question.
When I hit 'R' to do a rectangular flatten of a rectangle, why is Headus spontaneously 'Pinning corners' with them green dots, and how do I get rid of the pin-dots without saving it and reopening the thing? Looks like Hex double connected some dots, grrrr.
Getting there bit by bit.
(bottom of the page for the latest zip with heat sinks as well)
Chip heights under the big heat sink are based on posts from the Odroid forum and guessing on the smaller bits, I've yet to verify them my self. The other stuff is a WIP.
If it's not a perfect grid of lines it'll do weird things like that. To get rid of pins hover pointer over green dot and press "p"
(if the pins circle the whole thing select the island, press shift-p and then "u" (for unpin))
OK, thank you, and thank you. I really need to make a key-code card like what Word Perfect for dos use to include. It sometimes takes longer to find the key-code to undo a mistake then to just start over from scratch, lol. Especially on complex shapes like that 30-pin connector, I accidently Dropped the thing twice when trying to Cut up the pins on the thing (C, D, W are just a tad to close on this keyboard, and the FJ nubs are almost not there at all), lol.
I need to make coffee, and clear white stuff off the deck, again
.
Team O.S.W.A.T.T. question of the day, can a 8mm or 10mm heatpipe fit under the Odroid XU4 heatsink and get all five chips
without getting stranded pockets of hot vapor or cool liquid.
food for thought while I attempt to not clog up the shopping cart plumbing, lol.
Normally they do not get hot, yet when they start doing what they are there for things get a bit dicey, lol. There is a reason the NCP372 chip has a massive SMT thermal lug on the bottom of the thing, it's just best to keep it cool and happy.
It's not the lighting, it's just that the bottom loop of a 3D rectangle that has no bottom face needs more subdiv weight. When the edges are all set to 7, the sides fall out... when the bottom loop is set to 10, it stays solid.
I'm wondering actually if there are any products on the market that use subdivision weighting on edges and/or vertices.
Here's a view with part of the board hidden to see the chip better.
Don't know. I am aware of the 'Lingo' yet never fussed with it, so stuff like rigidity on clothing buttons and that sub-d weight stuff is at best "That magic stuff PA's do" that I've never had the time nor pressing need to figure out how to do it, lol.
I am curious if adding a base mash loop at the bottom will help at all with that, and how close do the loops need to be to the edge/corner without making it impossible to dice up in Headus. And considering how many parts are on the board, just how many loops would be 'reasonable' before it enters the Giga-Byte range of never fitting in a GPU for Iray. I admit I have not done much for the chips under the heat sink, because I'm out of thermal-goop and the shipment is stuck somewhere out there in the blizzard. Not in a major rush on that either, I have a million little things I can get the digital calipers on that I need to fine tune around the massive heat sink. There is three more smaller SMD caps around the NIC that I did not initially think I needed to know where they are, for example.
Back in zip revision 007 (I think it was) I took Fisty's advice and removed all the bottom faces of the stuff that still had it (I didn't know the little transistors next to the uSD plug was just cubes, I'll fix that when I get in there to measure them). So now it shouldn't crash Iray, yet it will be far worse with that Sub-D stuff
I can't win, lol.
Then again, I hit some wrong key somewhere, and now the arrows for moving stuff in hex are gone and will not come back.
I have the move thing selected top left and bottom center, and yet nothing, it is broken. I even tried restarting hex to no avail, the move arrows are gone. I can select surfaces and that's it. So much for that idea.
(EDIT) probably the most difficult way to fix it, and I still have no clue what broke in hex. Use the bridge to send the object from Hex to Daz Studio, and make sure the item is in Studio. Then go back to hex and clear out the scene in Hex (File - New), then go to Daz Studio and send that object copy back to Hex using the bridge, then save it in hax as a new scene before fussing with any thing else. I still have no clue why the arrows went away.
Alright, now I need to figure out where the contacts are inside this thing. It looks like five USB 3.0 contacts up front, and four more contacts behind that. Time to hunt down a PDF with dimensions on that stuff.
OK, a quick visual tour of the Odroid XU4 OBJ in zip revision 009 (second post on this page).
The contacts in the one USB port are kind of visible from a few angles, however the ones in the dual 3.0 jack are kind of on the underside, so I'm some what not sure that was necessary considering how it made the geometry count shoot threw the roof.
I am happy with how it looks tho.
I am still trying to figure out what to do with these
So far, turning OFF smoothing in the surface tab appears to be the simplest and best solution for non-right-angles.
Morning update from zip v010.
I just added the small SMD caps around the NIC, and double checked some of the measurements with some minor tweaks of component placement over there. The large cap next to the NIC is 1.0mm tall, and the chip is 0.9mm in height. The smaller caps (excluding the one on the fan circuit) are 0.75mm tall. So that large SMD cap is something to look out for when putting a heat sink on that chip, and it makes another idea a bit dicey for using the NIC plug housing as a heat sink with a strip of copper folded up between them
Won't clicking that little ghost figure button (fifth from the left) bring the manips back? There might also be a keyboard shortcut. I think. Honestly, I think it's a kind of magic that you (and a number of other people) manage to be productive using Hexagon, what with how quirky it is.
And if you just bevel edges so that they don't need subdivision for closeups, will it crash Iray or do other weird things?
It wont, not for that whatever that was. There is a key on the keyboard
'Somewhere' 
(hit by the digital calipers accidently when I was measuring the Odroid) that makes the object un-manipulatable even if all the boxes are checked properly. The only fix is to export the thing and re-import it, or start from scratch with a fresh cube (shrugs shoulders). Yea, hex is very flaky, yet it has an understandable interface that is easy for n00b's like me, and it is affordable for individuals unlike rent-wear targeted at the likes of Pixar or ILM, lol.
I think I got something for the crystals, and it is based on an actual PDF part for the actual NIC 25MHz crystal on my Odroid. I'm dicing it up now in Headus.
So far no more anomalous shading artifacts, so far.
Crash I don't think so.
yet anything not 90 degree corners clearly has shading anomalies, like the odd angle edges on the original version of the crystals I was fussing with yesterday.
The new one is much Much MUCH better in that regard...
, I just need to figure out what part number is the one next to the CPU to make that one. The two 25MHz crystals appear to be normal 3.2x2.5 mm SMD devices, that smaller 24MHz one appears to be an odd ball size, hmm.
This is 9 times out of 10 what makes Hex non responsive. (click and drag on them to move the view around of course, but actually having them selected means you can't do anything)
(sorry, that's the best my handwriting gets with my right hand and a mouse, I'm left handed)
It was worth a try and often that is all it is, that move thing get's deselected when I'm panning around something. Whatever it was, the digital calipers found the one out of ten key that defies that simple make sure the arrows are selected thing, lol. Even going back to that hex save just now, I can not extrude that top tab up for the dual USB plug. It's of minor thing as I figured out a way around it, yet I am curious what key the calipers found that did that, I just don't dare openly share the hex file as some may use it for ill will. And that calligraphy is far superior to my inability to spell
, I'm sure 'Dave' would approve of your take on 'DaveCAD' (tm EEVblog), lol
As for other things, I was looking at chip PDF files and fine tuning the horizontal sizes of some things.
I got to fix that cat5 plug, it dose not look right without them contacts in it, ugh.
At least the HDMI plug looks better now.
Morning, not much for today, just looking at random things like TCAS for cars (V2X) and other what's new things. It looks like electric cars are getting better, however I have a few doubts for my neck of the woods. TCAS works because pilots are incredibly focused when in the cockpit and dedicated to all things related to flying planes, not the same level of dedication is given by many drivers of cars. Then there is that watt thing vs range. "Range" in a few ways, I'll give you that, lol. How far out can V2X detect possible collision paths given fixed RF power from other devices, and after all the other gadgets sucking watts what is left for moving the car down the road.
Will that electric thing be able to dethaw me after I've finished digging it out of the snow, or is that electric car craze best suited for more temperate climates that don't regularly see sub-zero temps that give pitot tube heaters a run for there money, lol. Speaking of, looks out door and grabs camera to take cool pics from this morning....
Looks like I need to fire up the 2-meter radio and melt some of that impedance mismatch stuff off the 'harpoon' antenna, lol. You will note the lack of care about the 440MHz javelin or 2-meter Lance antennas, they will be fine.
I wonder how Electric cars will bolster the ability or lack there of to keep the occupants from being cryogenicily frozen like some stasis tube from a few Sci-Fi movies. Now honestly that melting the snow of the antenna is just a joke, and often involves brute force and a sledge hammer to nock the stuff off the tower, lol. Actual heating elements dissipate energy in a very different way from antennas (and 2-meter is just the wrong band to be 'microwaving' snow, lol). Every watt used to heat the occupants or keep the windows and mirrors clear is watts not there for mileage.
What would be the mileage of them electric cars be, in say, Alaska or Halifax during the winter? And like wise, how would the electric cars fare around Moscow (during the winter), and would the charging circuits even work on the Russian grid voltage (is that 380V 50Hz, 4-phase
or 3-phase + neutral, I don't know?) ?
In separate news, yea I did play a bit with RF shielding for some of the stuff around here. The incapable PI2 got the resistor-art of a shield (clear case to have fun with). If only it had more of a FSB then a single USB 1.0 channel (2 wires, 1 RX and 1 TX
) connecting the CPU to the CAT5 and sound ports along with other stuff.
I am much more an Odroid XU4 fan, because it is far more capable of connecting it's video output to a streaming website (It's FSB is far more robust then the PI2, and it dose not look like the PI3 is any better then the PI2). You just can't get enough data to the CPU of the fruit computers to do any thing useful with them, unlike the Odroid XU4.
Your eyes are not deceiving you, them are dedicated differential signaling (4-wires, 2 RX and 2 TX) links to the USB hub chip and a separate link to the NIC chip. Even with the stock heat sink and always thermal throttling to keep it's cool, the XU4 blows the fruit computers out of the water. And it has more CPU cores then most i7 processors ten times the cost of the XU4. This is a SBC, that can
.
LOL my spelling isn't much better than yours. Interesting story.. when I was in middle school (about 13 years old) I had a student teacher in art class for a few weeks, it was her job to teach us how to write calligraphy. So, no big deal right.. unless you're left handed but don't write with your wrist bent over. (I write just like a right hander, I just turn the paper the other way) She INSISTED that left handers had to use the left handed nibs.. for people who write like me that means the angle on the tip is the wrong dirrection. She insisted to the point of failing me if I didn't... so I did it with my right hand. I got a C but I passed. Six years later I was winning awards in the SCA for my calligraphy and illumination - with my left hand and a right hand nib, dammit.
Not sure about cars but I have a friend in Alaska that vapes and he found the lowest temperature a rechargable battery (an 18650) would function.. I don't remember exactly but it was really cold, like -40 F.
Stop saving my last post as a draft Forum, grrrrr.
The jury is out on if, IF, I got the correct stuff for the Odroid. The MX4 is for the GPU project, and is not available in bulk large enough for Odroid experimentation. I need a tad more then a grain of rice per application for the Odroid, and MX4 is not available in larger quantities at less the the cost of gold per unit-weight lots. And don't start sending out the hate mail just yet... the stuff from the land of Bang-Good LED lights (tm, bigclivedotcom), just may be the proper counterfeit stuff for cooling transistors and other TO-220 parts
, If I can figure out what tooth paste tube of stuff is the correct one for that. Nothing any where indicated what one was "Electrically Non-Conductive", just thermal properties.
The funny thing is, the photo on the store page was of a jar similar to what GamersNexus has in a few vids of GPU teardowns, not tooth paste tubes.
Wild question, as gold may be a bit on the cheep side of things. Is MX4 more expensive per unit-weight then titanium or lithium these days, hmmm. Twenty USD for a grain of rice quantity of MX4,
YGTBCM. Twenty USD, That better move the hair from the chest to the shoulders and back of that GTX960, Expletive!
The HY510 may just be the correct gray stuff, mark that as the first time in over ten years that a product with the CE mark actually worked and was not something less then advertised (if not outright defective parts sold as new good items). HY510 is "Electrically Non-Conductive" and dose NOT have any sizable capacitive influence beyond air. (chin on floor) Time to mash that buy-now button for the rest of that stuff. The 610 is giving me some doubts, I need to re-test it.
The HY410 looks like the generic white goop used for many simple things. "Electrically Non-Conductive" and dose NOT have any sizable capacitive influence beyond air. The consistency is a bit thinner then the white goop from the local Radio-not outlet. So it will be better for getting a thinner gap with better thermal conductivity then the thick paste sold locally.
Just some pics from yesterday as I wake up a bit. How much dose the XU4 flex under the weight of such a massive cooler, lots.
Only the lower half of that heat sink is worm most of the time, so I have some doubts about the effectiveness of the excessively vertical fins. At least I had a good seat between the chips-shims and heat spreader. can't say as much for the heat sink.
Some of the chips under the heat sink are a tad taller then what was posted on the forum, so I may need to check in to that some more.
I guess I'll need to look into getting another one to measure one I did not put nail polish on the caps.
One downside of that clear nail polish, is it tends to absorb any contaminants floating around the air.
Yep, I had the pore thing in pieces. Oh, that led-Y-cable thing, no that is not for a silly illuminated case, it is an indicator light to let me know when the Odroid is asking for more cooling. That AFB0405MA is much quieter then other things around here (like the furnace), so the light is to let me know when the fan is on. I like that Delta fan much better then the CFM-lacking fan suggested by others for the XU4, I honestly think a better heat sink fin design would do much better, I just need to consider a few angles for that (good fin design, easy to make, cheep to produce, proper MX440 mount layout, etc).
No idea off the top of my head, but I could look it up. I have heard about consumer e-cars and a few charging stations for them, but it's still very much a novelty, so I'm not sure I could find any reliable specs. But what we've had for years is electric trams and trolleybuses, would these be of interest to you?
Domestic electricity is 220 V 3-phase here afaik.
Morning, the entire Electric car bit was just something to think about thing. As for trams and trolleybuses, I'm sure they have enclosures for some of them and a large number of people to maintain them and the tracks unlike what many individuals can afford for there cars. Plus, I'm sure some even swap out drivers without turning off the heaters and engines for many of them, so the car you step into at a station has had a few hours to worm up (if not days) unlike the "I miss my Blazer" in my drive way that I drive possibly five times a month. I can imagine getting into a car to go somewhere, and need to operate buttons and stuff with nothing to worm them up (yes, the Jeep is that bad).
Voltages, that was a genuine curiosity when the vid I had seen mentioned 220-volt chargers for the cars, I guess the charger would be fine over yonder given how popular 50Hz is. So that just leaves traction on white stuff, and the ability to thaw-out the occupant compartment without significant reduction in mileage (not to forget about AC for the summer).
Cool CGI vid (oscilloscope add), somewhat in line with what I was doing for that SBC project. I have no idea what render engine was used, tho it dose look kind of cool how they did up the circuit boards. I can imagine similar kind of stuff done for other stuff (amplifiers, supercomputers, etc).
http://www.youtube.com/watch?v=C4W20HdVjA0
There is some pics I ned to get off the camera from yesterday I think, I need coffee and to look.
Drinking coffee and looking at pics from yesterday. I took the opportunity to plane the Heat sink a bit more to get it a tad flatter, tho the mount is just as much a limitation for the thing.
The CPU sits mostly under a lip that holds the mount-thing, so the majority of the heat sink base is off to the side of the CPU without rotating the thing 90 degrees (the mount-thing won't let you do that). So I modified the plug a bit so the heat sink could get a tad closer to the plug on the CPU.
The odd stuff across the fins in some pics was residue from an experiment with a fan air guide of a sorts.
Also I was looking at how to get it into a nice looking box. Oh, that goop on the back plate, that was just a random Why not try it kind of thing and resulted in a 2c to 3c improvement in temps (not much, the plastic spacers are not exactly the best thermal conductive stuff, lol).
I'm a bit more worried about how rapidly and much the temps shoot up when the CPU goes under load (keep in mind, this is a 20watt computer not a Killa-watt-hog GPU). The MX440, ATI7000, 36GB raptors, and other 20watt stuff don't ever see 50c with smaller heatsinks
, so I'm still looking at what is up with that. I'm going to run to the store to get a short grounded cord to put the kill-a-watt on the desk so I can get better numbers then just thermals (needed to do that for a while now).... back from that, looks like somewhere in the ballpark of 3c/w for that cooling set up with it sitting under the monitor in the path of the desk fan that's keeping the monitors cool (possibly a light breeze, nothing significant or noisy). And 73c is not unheard of for the thing, It's just much wormer then I'm comfortable with for only 720p video playback.
(EDIT) watt disclaimer, there is a generic Non illuminated USB keyboard (V7, KC0A1-4N6P) and track ball (Kensington K64325) plugged into the XU4, and a Powered USB hub with a USB sound-thing. I have no idea how much power they use from the XU4, at the very least it is a measure of how much power is going threw both NCP372 chips.
Probably. My grandfather worked for the Metro, fixing the trains, and there are huge places dedicated to just that; the over-ground tram depots might be set up similarly.
Going without being turned off for days seems unlikely, though - no public transportation here works 24/7, it's always off for the night. They only started using cars with heaters maybe ten years ago (and not on every route at once), and I remember that it used to be damn cold inside in winter all the time, barely a few degrees warmer than in the street.
Yea, that was a bit of (there may be exceptions) to what I typed about trams. I was on a passenger train a few times when I was younger, and there was a crew switchover that happened half way threw the trip, I just was not sure how common that kind of thing was with other transit (whatever the word is). The train would pull into a station, the new crew would get on the train with it still running (engine running, train stopped) and then the old crew would leave after a chat (paperwork, etc), the train engines were kept running the entire time. Also I had seen a few cold starts of locomotive engines (around the net), and that is a far more involved process then just getting in a car and "turning they key", followed by letting the engine worm up as you scrape frost off the windshield, lol. Then there was a few vids I had watch recently that looked like there was an indoor Turntable station kind of thing (I'm guessing that was essentially an indoor garage kind of place).
It looks very different from where my Jeep sits when I'm not driving, lol. Then again that may be just an End-of-the-line U-turn kind of place for that track, regardless it is not exactly outside in the elements, lol.
That's a tad large for just an end-of-the-line turnaround building, it's at least a shop or something as well.
Odroid XU4 status. I decided that I'm going to stick with the chip heights posted on the Hardkernel forum for the time being. I'm not sure how much variation there is with the heights of some of the chips, and the short-out protection I applied to some SMD caps (nail polish) is making it difficult to measure some chips at multiple locations to get an accrete height. The shims I put on the chips was not all that 'scientific' as well. I found that regular "Copper step flashing" I had scraps of on hand, just happened to be "close enough" eyeballing it for two chips, and it was easy to sand it down for the third chip. Now that I have a more accrete tool for measuring, the step flashing ranges in thickness from 0.40 mm to 0.51mm in thickness (depending on the individual sheet from the scrap bin). After some measurements of them on the board with ultra-thin thermal tape, I could have done much better.
That embarrassment out of the way, on to the rather impressive looking heat sink that could have been a lot better after a closer look.
The incredibly heavy outer fins don't have a good thermal path to the base and are less then effective at higher heat throughput. That leaves the inner rows of fins to dissipate a majority of heat, and the air flow paths are not all that impressive to be honest. (Remember, them outer corners will be mostly blocked by the screws that hold on the fan.)
In fact two rows of fins actually block air flow from a fan, doing nothing other then producing noise from a fan. So I decided to order a backup heat sink from ameridroid and have at this one with a dremel. It's a tad difficult to get in there to some of the fins, so it is very ugly looking.
I decided to attempt to cut a grove in the center of the fin tops, just to increase the surface area a bit, and to see how thin such a cut could be made reasonably. It's ugly looking as all hell, tho I like the concept.
Yea, that notch cut at an angle at the base of the center row of fins was an attempt to open that up for air flow, and I attempted to round the face of the fins with a file to be less of a wall for air from the desk fan. I can see what's going on here between fan-size vs air path cross area on the sides, tho I'll measure it for some numbers. The fan is about 38mm (1134.11 mm2) with a motor roughly 20mm (314.16 mm2) in the center, for a total air path cross area of 819.95 square mm. There are eight fin gaps about 1.8mm wide and about 27mm at the tallest, 20mm tall (36 mm2 each gap) with the tape there to force air threw the fins rather then bypassing them at the tops (total of 288 square mm on each side). That's 576 square mm to exhaust air from a 819.95 square mm fan intake, I don't think that will work all that well, lol. So I decided to attempt to open things up a tad, and it needs more work still, I just need to figure out how to proceed (what cut with what tool, where and how).
To be honest, the heatsink on the right with a 40mm fan on top of it has outperformed everything else I've tried on the XU4 so far, it's mounting is just a tad iffy at best. It never had MX440 mounts and had to be diced up to fit with pressure from the fan holding it down with essentially bailing wire, lol.
Here, the Metro train drivers do switch at certain stations, but I'm not sure about tram/trolleybus drivers.
Those buildings might well host some sort of a workshop, but I can't say for sure... you imagine they don't let general public in.
Morning, working on my first cup of coffee, and it looks like some rather important info is not available. The question was just how close to the impossible am I with this heat sink excursion. It dose not look like the simple c/w numbers (from glass to solder balls, and from glass to 'top' of package) are available for the Exynos 5422 chip, so I will have to guess. It may just be that my lack of being awake yet combined with the occasional flaky search results of google may just be working against me this morning, lol. I do have some numbers on another chip that can give me a guess, keeping in mind that the black shell on the Exynos 5422 chip is thinner then the BUF634 chips.
Just looking at that, it looks like the range of 3 to 4 c/w that I'm getting (glass to air threw a heatsink) is rather chin dropping, yet it can be a lot better (IF, if the Exynos chip had a "thermal slug" on the top of it instead of just plastic). In any case, I did do a bit of a mod to get some more air to the center of that heat sink, then the dremel died half way threw working on the other side (Refraining from another BOLTR episode, it's done, time for a new one).
So, in any case. I added a bit of a slot to the inner fins to the sides, to open that up for air flow a bit more, on both sides this time.
And keeping in mind that the ultra thick outer fins do nothing other then block air flow with minimal heat dissipation, this is what I had in mind.
Now, I intended to trim up the top a bit more where the outer fins are blocking the fan, unfortunately the dremel stopped letting the pixies dance around just before I got to that, so the fan still has a bit of blockage on the outer edges. Them spring-wire-clips just won't keep a good grip on the wires in the dremel and is not worth fussing with for less the five minuets of usefulness for each dremel dissection to put the wires back in the clips, grr. So I cleaned up what I had and let the XU4 temps stabilize over night. Hopefully today is a bit better.
Some are perfectly happy with thermal cycling shock and temps over 70c for there stuff that is not all that critical and easy for them to replace, I don't live in that world, lol. I have stuff I need to do that dose not involve repeatedly fixing a dremel, so I will be reduced to living off of Ramen noodles for a few weeks to scrape up enough to replace that
. I will see if I can make a facsimile of what I've done to this heat sink in hex, that may or may not happen today (I have other stuff I must do sooner then later. I'm delinquent with another project), those outer fins look kind of wild almost like an "Impossible trident", lol.
Kettu, it's understandably, as many don't know the safe and unsafe places/activities in such places. Even simple car repair shops will be a tad uncomfortable with people just wandering around the shop. As for that incredible looking ceiling, I suspected it had to do with weight rather then other factors (kind of like the Pantheon), and to let in something resembling day light.
There are a lot of award-winning and otherwise groundbreaking architectural designs among Russian "mundane" industrial kinda buildings. So I don't know if in the video we see one of those and hence different from all others, or if the roof is a common element in tram depots (?). I'm trying to recall if I've seen many of those IRL, but honestly I'm not paying that much attention to non-Metro stuff, non-bridges and non-towers =D I can only say the place where Granddad worked looks different.
It's all good, I often don't go off the deep end on building architecture often. It's a that looks nice with random thoughts more on the technical side then artistic rambling. When I went back to get the screen-caps I was thinking about the caveats of climate control for such a structure and weather or not double/triple pane glass on that ceiling would be worth having at all. It looks like there is at least two huge doors for the trains going outside, and heating such a chamber with doors like that probably is not 'cost effective' from a few point's of view. The roof over that Turntable was probably to keep ice and snow out of the cogs for year round operation more then comfort of people under it.
Been busy looking at something that jest gets more pathetic looking the more I dive into it. The some what anemic HSF combo that came with the XU4 vs the first thing I tossed at it out of wild curiosity.
Some of them spindly little things are no thicker then the graphite in a mechanical pencil (0.7mm), giving me the impression there is not enough meat to them to get heat to the ends of the fins. And the heftier fins are completely blocked by the screws that hold on the fan. There was just so much to chose for improving that thing, I've yet to decide where to begin other then outright replacing it.
The Pin Fin Array heat sink scavenged from an outdated dead motherboard is just so far beyond that stock cooler that weighs less then a triple-A battery with the fan on it (the stock cooler can not possibly be aluminum, aluminum is a lot heavier then that). At the very least lifting the fins up so the thing can breath without the connectors choking the thing was one direction I was looking into.
Yet after seeing the thermals of that Pin Fin Array vs the Stock cooler I have abandoned any efforts to improve the stock one. It honestly looks like the HSF maker took an extruded heat sink capable of dissipating possibly fifteen or more watts with active cooling (a fan on top), and gutted the fins out of it to cram in the largest low profile fan they could find, with an end result of a cooler that can't breath and probably is not rated for more then five watts with the fan sounding like a clogged up vacuum cleaner
, lol. (The XU4 stock HSF is actually much quieter than most graphics cards I've seen the past ten years, so even the fan in it is rather gutless, lol.) Given that the fan dose not turn on at all till the thing gets over 65c, I'm surprised the XU4 did leave a hole melted in my desk like an Alien(tm) Xenomorph blood spill. The stock HSF is not completely useless, it just leaves a lot to be desired for such a capable SBC. Just to drive the point home, the Pin Fin Array I found is not all that tall (I wanted a 16mm to 20mm tall one) and is not anything more exotic then regular aluminum, yet it packs a lot more surface area to cool things down to reasonable levels (I was actually looking for a beefier Pin Fin Array then what I had in the scrap bin and made a OBJ model of).
So, that is the status of things. I'm sort of in a holding patron as I wait for another unit to arrive to test and make a model of. I'm sort of using the XU4 at the moment, so can not get the calipers in there to measure the other heat sink at the moment. I will probably play a bit more in hex with an Idea that's been on the back burner for a while now.
Oh, by the way. I added both these coolers to the zip so you can look at them yourself. Odroid XU4 OBJ and miscellaneous parts by ZGD, Revision 019 (with my prliminary heatsink idea as well).
https://www.dropbox.com/s/jabaum7r2uthvks/OdroidXu4ObjByZdg_v019.zip?dl=0
P.S. To be fair, I've spent the past year and change looking for a MX440 mount style heat sink with a 40x40mm base (not smaller or excessively larger), and there is practically nothing available with just that criteria. If you add in the requirement for beefy cooling capacity with passive mode (fan off), there is Nothing out there. So in all honesty, the shame falls more on heat sink makers then the Hardkernel crew. So I'm looking into a simple heat sink Idea that can be extruded and cross-cut as a concept for SBC cooling. We need something that actually works, instead of the "All Looks and Nonfunctional" garbage that is on the market.
Just a still alive post, I've been deep in maths and technical stuff the past few days just to see if a few ideas was viable at all or not. Originally I wanted to stuff the heat sink under the Odroid lid and put the fan on top with rubber bushings, however it dose not look like that will work at all for a few reasons.
There is only about 16mm clearance under the odroid lid for a heat sink, and it looks like the heat sink fins will need to be taller then that (not including copper heat spreader and heat sink base). After some quick algebra and google circle calculations, it looks like the minimum height of an adequate 40x40mm heat sink will need to be much taller then 16mm for airflow, and that is just the beginning of the restrictions with the original idea, hmmm.
Looks like a 'Transition Duct' will be required to allow air to enter the top of the heat sink directly under the fan motor for minimum air cross area intake exhaust balance stuff. So I've mostly been looking into that the past few days and have a crude draft in the works.
I may need to make that 40mm 'Transition Duct' a tad taller then only 4mm, I just don't know any specifics beyond that just yet. That 4mm is just for air cross area under the fan motor, and dose not consider how air will flow around that sharp corner (or more accurately how that sharp corner will restrict air flow to be honest).
Oh, and what I mentioned months ago about two gentleman that understood cooling and air flow dynamics far better then me, here is a short vid showing what a "Transition Duct" is in a jet engine. In this case, replace the combustor cans with a 40mm fan (with no air flow in the center of the fan exhaust) and the turbine with the top of the 40x40mm heat sink (with intake area in the center of the heat sink top). Just look at it, you'll get the idea of what a "Transition Duct" is for, and how complex they can be to achieve best performance.
http://www.youtube.com/watch?v=p2xQzwzvgZs
So the lesson for today. Is slapping a fan directly up against a heat sink the best thing to do for optimum cooling and air flow performance (on most heat sinks except the stock Intel round and spiral shaped things), NO. That Pin Fin Array heat sink off a dead motherboard had room for improvement for far better cooling performance then what I had done, even tho it's fins are to short for adequate exhaust cross area on the sides of the thing.
Must make coffee moment, lol. And a how is it going as I try to gather my thoughts into something a tad more coherent about this heat sink 'Idea' in the works. Here is some pics of the transition duct that I settled on a minimum of 8mm for a 40mm fan just to let air get around the fan motor to the heat sink, and a revised fan that better matches the duct interface shape of modern (higher quality) fans.
I could have gone off the deep end and made the fan perfectly match the Delta(tm) that is on my XU4, however I wanted to keep the geometry count at a reasonable level, and this is good enough to get an idea of what is going in.
The same was also true for the Transition duct, and I was considering overhang for mold injection process as well as 3d-printing limitations as well.
I did think about making a true match up to the fan duct, however that is going to be a geometry nightmare to be honest, that I may look at again later on (if I ever get a desire for self torment, lol).
Oh, and that heat sink. It's a WIP with most of the concept there. Given limitations for machining ultra thin tall bits of metal and getting the most amount of heat to the tops of the pin-fins (
things), I opted for a 2x2 mm dimension rather then something thinner for most of the pins. The top 10mm of the inner pins do taper in a bit in the extrude direction, however the cross cuts are just simple straight cuts of various depths. The outer perimeter of pins are intentionally thicker and heftier, just for mechanical strength while handling and to protect the more fragile inner pins. The total size of the heat sink is 40x40x20mm (excluding mounting wings) that should fit nicely on the Odroid XU4. Combined with the 8mm transition duct and a "AFB0405MA-A" Delta(tm) 40mm fan, the entire HSF is around 38.4mm tall. Add a 2mm thick copper heat spreader under that and it is a nice 40x40x40 mm cube of coolness 
for the XU4, lol.
As for air flow, I'm still looking at all them 2mm spaces between the pins and considering if that is to narrow for natural convection or not. I do have the cross cuts intentionally 2mm deeper in the inner rows for two reasons. First it leaves more metal laterally to conduct more heat from the CPU over to the far side of the heat sink, And it also allows more air to exit the heat sink on the sides that have less obstructions and allow the rest of the XU4 to get some extra cooling as well.
Oh, and one other random note. Both the fan mount holes, and the heat sink mount holes are 3mm diameter. I just didn't feel like adding unnecessary mesh complexity to make them perfectly round (and they don't need to be).
https://www.dropbox.com/s/jabaum7r2uthvks/OdroidXu4ObjByZdg_v019.zip?dl=0
As much as I would love to make this heat sink and use it on my XU4, that is honestly beyond my financial capabilities and skills with a hand held hack-saw.
.
(edit) After I catch up with my willfully overdo inventory of what I purchased the past few months, and get coffee going. I'm going to take all three heat sinks back in to hex to remove all the surfaces blocked by fans and mounting heard wear. Then when I have Headus pack the UV maps, it should give a much better visualization of cooling surface area then the mess of colored-areas I have at the moment.
I do understand why modern stuff is designed to within an inch of it's life. It's a drive for the least amount of material in a product as possible to barely get the job done (more often then not resulting in it not being good enough to do what it was portrayed to be able to do). Less material results in cheaper cost of material to make something, and the lighter weight makes it cheaper to ship to the other side of the planet. (some time later...)
The blue-filled square is the 40x40mm bottom of each heat sink that conducts heat off of the XU4 chips, I left that in each map to give me something to scale them all to about the same surface area size roughly. The green is the surfaces that are in contact with fresh air to dissipate the heat, after removing everything else that was blocked by mounting hardware. The number in the little red box, is the amount of the map in percent that is actual surface on the object. Given that the stock XU4 map on the right is the least efficiently filled map of the three, it's remaining cooling surface area is a tad exaggerated in this comparison. I had to give it at least some benefit of a doubt
, there has got to be a reason it is the 'Only' HSF for this mount setup on the market (other then it never sold out years ago when the MX440 was popular), lol.
The Aavid "375124B00032G" is the only thing I've been able to find (any where) that is remotely close to having the correct 40x40 mm base with cooling pins tall enough for a 40mm fan to be able to breath. However it dose NOT have the correct mounting to bolt nicely to the XU4 without bailing wire holding it down by a fan. They do have heat sinks that would bolt up nicely to the XU4, however every single one I've located so far is to short for a 40mm fan to breath and incapable of dissipating over 16 watts from an Odroid XU4, Grrr. I may have an idea for the Transition duct to get around the lack of mounts on the heat sink, it just wont be as solid as actual mounting wings on the base of the heat sink.
I was holding off for a bit while playing in hex to make a post, yet I just came across something that just made my day (actually it's the second thing, tho the first I can't discuss just yet). I have not been oblivious to the technology hype that As Steve put so accurately, the hype had reach speeds capable of achieving low earth orbit, lol. I was just hesitant about two thing aside from a retired fixed income budget, I already have eight calculators in this FX8350 cruising along at 4.0GHz, and I read somewhere that the cell phone interface version of windows was required to use "the new thing".
Well, the new processor is not a 5GHz 12-core 48-thread per-chip Power8(tm)
, it looks like Ryzen can possibly deliver reasonable performance for Non-Gamers on a budget from the looks of things (I would only need around eight AM4 sockets instead of sixteen for reasonable spot-renders of AoA subsurface pre-compute on HD meshes). So, on to the blatant fail of mimicking the Android cell phone interface OS (OK, it was not a completely bad copy of a cell phone interface, it just fails at being useful as a workstation interface)...
Honestly, I don't have the funds to get a MoBo and Memory and 1800x at this time, so I can not confirm that the windows 7 drivers are usable. And I suspect that simply installing windows 7 on a new Ryzen computer without another working computer available to download drivers and stuff may be a tad difficult if not impossible. If the win7 drivers do work along with Daz Studio on Ryzen, that is a possible upgrade path for me, even tho what I really need and can't afford is a IBM Power9(tm) supercomputer.
On that note, I will do something a tad rare. Thank you Paul (with fishy-cat Blender render test), and Steve (with Monkey-heads Blender render tests), and the crew at Linus Tech Tips. This is the first time in a long while that I've been able to look at benchmarks and see how what I have compares to products from the other manufacturer. Thank you for not just testing Ryzen vs i7 processors. When it comes to multi-threaded CPU rendering (like 3delight), the FX-8350 and FX-8370 are exactly the same thing (8 Piledriver cores at 4.0GHz, because the boost clock thing is for single thread stuff only).
Morning y'all. I've had a long somber weekend to mill over some numbers that have not been available to me prior to last week (again, thanks to all the reviewers that posted r15/blender/3ds scores on something other then i7 processors in the Ryzen reviews). Just to be sure of where I stand, I decided it was prudent to look at what I spent for the core components in my work station. Way back in 2012 I got the M5A97 because the old motherboard died after the chipset fan failed. Two years later I had managed to scrape up for a FX8350 and 32GB of memory. So honestly such things are quite painful for me financially, and genuinely involve weeks of living off Romen noodles to save up for (the top end i7 is just to expensive for my blood). I stopped looking at intell processors when they started the game of each processor would only work with a specific chipset and nothing else even if the motherboard had the correct CPU socket for it (there is just no upgrade path with that at all for people with my kind of finances). I'm sure that (I hope) things have improved some since then, however given that motherboard makers cant always provide bios updates for newer processors for older motherboards, I will not say that the processors in the following chart will actually work with the motherboard listed with it (I was only looking at socket type to get an idea of the entry cost for the possible systems).
The FX8350/8370 is at the end of the line, and the only way to upgrade is to get a completely new system at this point, something I really am not looking forward to (it is going to be quite painful). I am not aware of an affordable processor for the x99 platform that would allow me to get the motherboard and ram now and then upgrade the processor later on, so the entry cost is just way to far out of my budget range. So, as of this morning, after making a quick spread sheet to compare cost of entry vs the improvement in render performances (humble thanks to Steve and crew and fellow reviewers), here is a quick cost (MoBo + CPU + 32GB memory) vs R15 and Blender render times. The "R15/$" is R15 score points per dollar (higher is better), and the Blender "L/km" (Liters per kilometer)
is a humorous comparison of how many USD spent per extra minute it takes to render (lower is better), lol.
These numbers scare me and my ledger. I think I will need to dig a bit to see if there is an upgrade path for the i7-7700K socket LGA1151, however I suspect like the FX8350 it is the end of the line for more CPU performance without replacing the entire core with a x99 system (Yep, there is Nothing beyond 4-core for the LGA1151 socket, for the moment it appears to be a dead-end platform.). The R1700 (non-X) looks a bit better (even tho I don't know any scores for it yet), however I do have some reservations with going with something with a lower clock then what I currently have. I have much to consider in the meantime. Also, at this time, there is just to many question marks regarding memory kits larger then 16GB for the R7 motherboards, so I would hold off a bit more till further info becomes available regarding Ryzen 32 and 64 GB memory kits. I have 32GB in my computer right now, any thing less is a downgrade for me.
It looks like the R7 processor is a good production processor with what is known about it. It's just that the play was rushed and the pass was fumbled leaving product partners running around the field unable to grab the foot ball as it bounces around chaotically, lol. Memory kit verification, BIOS updates, etc, This will take some time.
The i7-5960 R15 score came from "Hardware Unboxed", it looked close enough to the other i7s to consider it a valid value for my chart. The FX8350 Blender score is a guess based on the almost twin-like similarities with the FX8370 (I don't think there is any difference at all between them, and have not been able to locate anything to indicate that there is other then the irrelevant 100MHz higher single-thread-only boost clock.). Also HerdwareCanucks had a nice review with some other CG render tests (The i7-6950X R15 score is from them).
(edit) Looks like the retailer suggested the most expensive x99 motherboard on the planet for a combo for the i7-6900K when I looked up the price on the processor
. I'm redoing the charts, replacing the $579.99 "MSI X99A GodLike Gaming Carbon" with a more reasonable $309.99 "MSI X99A GAMING PRO CARBON".
And as I watch the latest vid from Steve, boxes from Aavid and Ameridroid just arrived
.