Forum Discussion
ASIC modules, WiGig, microwave cavities
I keep running into reoccurring hard limits regarding today's consumer-grade HPC as I ponder next-gen apps. Eventually it got me thinking, what about an architecture where the main intercom fabric was a microwave cavity? You could have ports on its faces and address both in frequency and space inside the manifold cavity.
Modules would be specialized ASICs, such as the shady ones currently minting cryptocoin, perhaps in addition to FPGAs. For style you could put an FPGA in one end for input and one in the other for output. When you load the device with a bad computation it achieves a very high voltage and possibly explodes! And the modules would probably become rugged enough to survive ninja theft...
I mean, it's perfect!
The power-performance ratio of the 7970 I bought just when they hit the web with their launch prices of somewhere over 500€ appeared to be ideal, but since then I have concluded that the entire form-factor of air-cooled computers is somehow wrong for a big resistor outputting 300 Watts of heat. It would suffice for a demo of each one of my ideas but not for much more than two in parallel. 900 Watts would not be bearable in the summer.
There isn't a single component on a PCB which is designed to be liquid cooled, much less immersed and left in water while operating. - I think HPC circuit boards should have a pressure tolerance written on the housing.
Pretty soon we will start adding 60GHz WiGig tech to consumer grade HPC. Looking at the trend of 2.4GHz I begin to imagine serious voltage differentials. How convenient if we could just submerse the entire kit in radiation-absorbent H2O and at high load we were simply heating water? No information leaks, unless you were about to run out of steam.
The central transmission fabric is an encapsulated void with antennas at known intervals. Those antennas may or may not be connected to a transmitter, receiver, or both. Each manifold could have characteristic defects to make it unique thanks to modern manufacturing accuracy and any antenna that you populated with a module could increase this defect space. You could achieve a very high bit rate at 60GHz with say eightfold symmetry in your microwave cavity. More so if each module could command more than one antenna and then use beam-forming.
I figure, it might be enough for running Luxrender at 60Hz, 4K wide. Maybe if the modules were like object-oriented programming objects, standard libraries which could be etched in silicon and plugged into the manifold. Or perhaps Luxrender is a single module and in two years you can get 8K by paying the same price.
For sensors distributed around a room wireless access to computation ought to be popular. These sensors would however almost certainly be used for security. If they transmit when something moves, an observer of the transmission can detect movements which may be private. These sensors have to be resistant to spoofing.
60GHz is severely limited by O2 in the atmosphere so the radiation should never travel very far. This is a convenient physical property of the oxygen molecule. Your neighbours might still spy on you but at least drones can't. :P
Modules would be specialized ASICs, such as the shady ones currently minting cryptocoin, perhaps in addition to FPGAs. For style you could put an FPGA in one end for input and one in the other for output. When you load the device with a bad computation it achieves a very high voltage and possibly explodes! And the modules would probably become rugged enough to survive ninja theft...
I mean, it's perfect!
The power-performance ratio of the 7970 I bought just when they hit the web with their launch prices of somewhere over 500€ appeared to be ideal, but since then I have concluded that the entire form-factor of air-cooled computers is somehow wrong for a big resistor outputting 300 Watts of heat. It would suffice for a demo of each one of my ideas but not for much more than two in parallel. 900 Watts would not be bearable in the summer.
There isn't a single component on a PCB which is designed to be liquid cooled, much less immersed and left in water while operating. - I think HPC circuit boards should have a pressure tolerance written on the housing.
Pretty soon we will start adding 60GHz WiGig tech to consumer grade HPC. Looking at the trend of 2.4GHz I begin to imagine serious voltage differentials. How convenient if we could just submerse the entire kit in radiation-absorbent H2O and at high load we were simply heating water? No information leaks, unless you were about to run out of steam.
The central transmission fabric is an encapsulated void with antennas at known intervals. Those antennas may or may not be connected to a transmitter, receiver, or both. Each manifold could have characteristic defects to make it unique thanks to modern manufacturing accuracy and any antenna that you populated with a module could increase this defect space. You could achieve a very high bit rate at 60GHz with say eightfold symmetry in your microwave cavity. More so if each module could command more than one antenna and then use beam-forming.
I figure, it might be enough for running Luxrender at 60Hz, 4K wide. Maybe if the modules were like object-oriented programming objects, standard libraries which could be etched in silicon and plugged into the manifold. Or perhaps Luxrender is a single module and in two years you can get 8K by paying the same price.
For sensors distributed around a room wireless access to computation ought to be popular. These sensors would however almost certainly be used for security. If they transmit when something moves, an observer of the transmission can detect movements which may be private. These sensors have to be resistant to spoofing.
60GHz is severely limited by O2 in the atmosphere so the radiation should never travel very far. This is a convenient physical property of the oxygen molecule. Your neighbours might still spy on you but at least drones can't. :P
12 Replies
- @dsp4: All geniuses are a little crazy, and all madmen have a bit of genius too. They have more in common than you may expect, with the main difference being how they channel their creativity.
- do what now?
"geekmaster" wrote:
@dsp4: All geniuses are a little crazy, and all madmen have a bit of genius too. They have more in common that you may expect, with the main difference being how they channel their creativity.
I couldn't agree more.- >_>
I think hacker insurance against exploding mainframes would be cool. You could make a wave cavity switch grid with interesting geometry. There would have to be repeaters and fat pipes. It's the Intertubes, man! "ganzuul" wrote:
>_>
I think hacker insurance against exploding mainframes would be cool. You could make a wave cavity switch grid with interesting geometry. There would have to be repeaters and fat pipes. It's the Intertubes, man!
That post deserves some relevant quotes from great thinkers:Genius is the recovery of childhood at will.
― Arthur RimbaudThe public is wonderfully tolerant. It forgives everything except genius.
― Oscar WildeNo great mind has ever existed without a touch of madness.
― AristotleThere's a fine line between genius and insanity. I have erased this line.
― Oscar Levant
Your post seems to have crossed the line rather than erasing it. Tin-foil hat time! ;)- Positive I didn't follow all of that but some of it sounded like what they do with Radar on private yachts. The problem with it being closer than 4 ft of Humans is cancer, no kids, brain scramble, etc. Talk to a xray tech or a radiologist, they might have some insight you can use.
"jrspradlin" wrote:
Positive I didn't follow all of that but some of it sounded like what they do with Radar on private yachts. The problem with it being closer than 4 ft of Humans is cancer, no kids, brain scramble, etc. Talk to a xray tech or a radiologist, they might have some insight you can use.
Claims such as yours that make broad general claims but do not include "picky details" (such as transmit power level and exposure periods in this case) are almost always false. Such absolute claims are rarely true, and require significant supporting evidence to be believed. However, in this case, the evidence is overwhelmingly against it (regarding non-ionizing radiation).
We had this debate before on another forum. I won. Your claims fall on the losing side in that debate:
http://www.mobileread.com/forums/showthread.php?t=189286
Radar (including 60 GHz radio emissions) causes tissue heating that can be damaging to eyes and some other heat-sensitive organs in doses that exceed the body's cooling capacity. It uses non-ionizing radiation, which is below the frequency of short-wave ultraviolet light. Such energy causes heating, but does not cause direct cellular damage or cancer. Only ionizing radiation photons contain enough energy to
What most people think of as "radiation" is not radio. It is ionizing radiation, which can modify chemicals in the body by removing electrons and creating harmful ions:" We take advantage of its properties to generate electric power, to kill cancer cells, and in many manufacturing processes....
Non-ionizing Radiation
Radiation that has enough energy to move atoms in a molecule around or cause them to vibrate, but not enough to remove electrons, is referred to as "non-ionizing radiation." Examples of this kind of radiation are sound waves, visible light, and microwaves.
Scare-mongers spread their FUD by confusing harmful ionizing radiation (what most people call "radiation") and non-ionizing radiation (what most people call "radio"). This FUD is so prevalent that some government agencies such as WHO have issued statements to placate the fear-mongers, saying that "radiofrequency electromagnetic fields as possibly carcinogenic to humans based on an increased risk for glioma, a malignant type of brain cancer, associated with wireless phone use." The problem with that is that the original "wireless phone" study was found to be flawed, and (except for one biases study in 2004) studies done since then (including much larger studies) did not find such a link. [I would normally go find links to back that up, but I have better things to do at the moment, so please Google it yourself]. However, as I mentioned, heating itself can cause some damage (such as heat-induced cataracts in the eyes).
Many xray techs and radiologists use ionizing radiation are are justified for their safety concerns for their equipment. However, that does not make them aware of the properties of non-ionizing radiation.
Some people need to either study more, or restrict their opinions to stuff they actually know. Luckily, there are tools such as Google and WikiPedia (an so much more) that can help expand your field of knowledge in almost any area of interest.
Just about anything in daily life is harmful in excessive quantity. You can even die by drinking too much water (which dilutes electrolytes and depolarizes your nerves). In fact, larger than "normal' amounts of natural ionizing radiation have been found to increase lifespan, as determined by human regional longevity studies. It is thought that such radiation "tunes" the immune system to teach the body to better identify and eliminate small harmful tumors. This is similar to studies showing that children who grow up in an overly clean environment tend to have more allergies (which cause the poorly-trained immune system to attack itself). Like everything else, even the immune system needs practice. This seems to include small amounts of exposure to ionizing radiation.
P.S. Some people who carry their cellphones in their pants pocket still have kids.- Click for full-size image:radcomp.jpg
Click for full-size image:radkiller.jpg
60 GHz may sound scary. But it really isn't. You cannot drown in 60 GHz. You can drown in water. Water is scary. The OP mentioned water too. And drones. Oh noes! scary drones!
:lol:
No, drones are not scary. Except for the military kind that carry weapons. Those are scary, especially if used to police civilian populations. But let's not venture into tin-foil hat territory, okay?
;)
P.S. I like the OP. That post has potential. And great ideas. Food for thought. Expanding the borders of possibility. I like that...
:D - Maintaining strict access control between the inside and outside of the cavity would be important. These 60GHz WiGig transceivers have 7GHz of bandwidth. At the edges of the range there is less attenuation from O2, so you could get maybe 100Mbps at Wi-Fi ranges.
You should easily be able to detect an O2 atmosphere given a pair of antennas some distance apart. Inside the manifold there is a pristine vacuum. If you're in space, outside there is a vacuum too but international treaties about transmission strength may still apply. Not that you'd care cause you're in spaaaace and don't care about bogus laws!
