Computing a Cure For HIV

← Back to Stories (view on slashdot.org)

Posted by Soulskill on Saturday June 21, 2014 @05:45AM from the petaflops-for-science dept.

aarondubrow writes: The tendency of HIV to mutate and resist drugs has made it particularly difficult to eradicate. But in the last decade scientists have begun using a new weapon in the fight against HIV: supercomputers. Using some of the nation's most powerful supercomputers, teams of researchers are pushing the limits of what we know about HIV and how we can treat it. The Huffington Post describes how supercomputers are helping scientists understand and treat the disease.

48 of 89 comments (clear)

Min score:

Reason:

Sort:

Bitcoin mining? by Anonymous Coward · 2014-06-21 05:53 · Score: 3, Interesting

Imagen if all that computer power was put to use such as finding the cure of HIV.... We would be done by lunch time.
1. Re:Bitcoin mining? by Anonymous Coward · 2014-06-21 06:07 · Score: 5, Insightful
  
  It isn't lack of computing power that is holding back most theoretical biophysics research.. it's the lack of people who have the rare combination of skills of a programmer, mathematician, chemist, biologist and drug engineer coming up with novel and unique ideas to combat disease who will sacrifice industry paychecks to work in academic fields.
2. Re:Bitcoin mining? by cashman73 · 2014-06-21 06:36 · Score: 2
  
  who will sacrifice industry paychecks to work in academic fields.
  Why do researchers have to sacrifice an industry paycheck to do it? In other words, why won't industrial pharma hire more talented scientists. They seem instead to be more interested in hiring salespeople, lawyers and MBAs, then contracting with academia so they can take advantage of "cheap labor" due to the overabundant supply of low-paid graduate students and post-docs. But then they wonder why the amount of NDAs (New Drug Applications) has been declining.
3. Re:Bitcoin mining? by gnupun · 2014-06-21 06:54 · Score: 4, Insightful
  
  Why do researchers have to sacrifice an industry paycheck to do it? In other words, why won't industrial pharma hire more talented scientists. They seem instead to be more interested in hiring salespeople, lawyers and MBAs...
  Perhaps, it has something to do with the high failure rate of such research. Would you pay a salary to 1000 employees, of which only one employee gives you solid results and the remaining fail? That's not very business friendly. This type of research is more feasible under govt. grants.
4. Re:Bitcoin mining? by bmo · 2014-06-21 07:56 · Score: 5, Insightful
  
  Perhaps, it has something to do with the high failure rate of such research. Would you pay a salary to 1000 employees, of which only one employee gives you solid results and the remaining fail?
  >implying that this is bad
  Typical bean-counter/MBA attitude.
  That's not very business friendly.
  Companies like HP, Xerox, etc, built empires on that kind of research.
  They declined when they spun off or closed their research divisions because management failed to see the value/use the output of the research labs. The HP example is particularly striking - they went from an advanced technology company to a schlock printer seller, one that is sneered at and loathed, in a handful of years. Xerox is also striking in that PARC laid the foundation for a lot of modern computing but management only saw money to be made in copiers and filing paper and thusly ignored most of PARC's output, ceding the computer revolution to other companies.
  It is also part of a larger problem. Because of the emphasis on short-term profits (quarters are too long!) at the expense of everything else, we in the West are so enthusiastic at shoving all our production to the Chinese and others saying "We can't be arsed to get our hands dirty; we want to just do the high-level stuff like design and company management" totally ignoring the fact where the production goes, so does the engineering development, science research, and eventually even upper-management. This was learned by Samuel Slater, Francis Cabot Lowell, and others who founded the "silicon valley" (Blackstone Valley) of the Industrial Revolution. A lesson forgotten through complacency, greed, and snobbery.
  Alexander Graham Bell is shouting at you from his grave calling you a moron.
  --
  BMO
5. Re:Bitcoin mining? by SuricouRaven · 2014-06-21 08:07 · Score: 1
  
  Industry does do a lot of research. There's money to be made in curing HIV, after all. They do shy away from the purely theoretical though, because there's no money to be made if you don't get a patentable drug at the end of it.
6. Re:Bitcoin mining? by Anonymous Coward · 2014-06-21 08:24 · Score: 1
  
  And yet, many people do just that - when given a choice between doing good for humanity and making money, they choose to do good.
  It's almost weird.
7. Re:Bitcoin mining? by Anonymous Coward · 2014-06-21 08:41 · Score: 1
  
  Typical bean-counter/MBA attitude.
  And who do you think makes those decisions?
  Research only gives a pay off for the next CEO. Fuck that guy. It's much better to downsize everything, cut down costs and jump with a golden parachute and repeat the formula elsewhere.
  Industry doesn't wants theoretical results that might be useful ten years from now. They want results for this quarter. So while they do invest in crazy ideas, small improvements is more their thing. It goes the other way around for academicals: even if it isn't a viable solution right now, you can publish it and in ten years or a century either it will be forgotten or someone will do something incredibly clever with it.
8. Re:Bitcoin mining? by fuzzyfuzzyfungus · 2014-06-21 09:36 · Score: 1
  
  Is computational biology really looking to compute a lot of SHA hashes?
9. Re:Bitcoin mining? by currently_awake · 2014-06-21 12:41 · Score: 1
  
  This is the reason that most major medical research is done by government grant (historically). Big pharma companies like to tell everyone they are the solution but really they are only the endpoint in the process.
10. Re:Bitcoin mining? by pepty · 2014-06-21 13:08 · Score: 1
  
  Companies like HP, Xerox, etc, built empires on that kind of research.
  Tech research is a lot less risky and has a much shorter wait til revenue then Pharma/Biotech (these days).
11. Re:Bitcoin mining? by pepty · 2014-06-21 13:14 · Score: 1
  
  Sort of. Basically we need someone to invent a force field (mathematical model) that can simulate proteins, protein-substrate interactions, and protein-protein interactions so well that it can make accurate and useful de novo predictions about drug candidates and disease models. After that we've got plenty of med chemists, structural biologists, etc., ready to make use of it.
12. Re:Bitcoin mining? by pepty · 2014-06-21 13:20 · Score: 1
  
  Year in, year out, 75-80% of new drugs are invented privately in biotechs/pharma. The remainder are invented by academia. In terms of money and sheer manpower required to get from "what causes this disease?" to "new drug approval" the pharma/biotech portion often isn't the short end of the stick either.
13. Re:Bitcoin mining? by pepty · 2014-06-21 13:23 · Score: 1
  
  You don't always need a patentable drug, just FDA exclusivity. If you find a new indication for an old generic drug and do the research to back it up you can get market exclusivity for a while and charge out the wazoo for it.
14. Re:Bitcoin mining? by dcollins117 · 2014-06-21 13:33 · Score: 1
  
  Why do researchers have to sacrifice an industry paycheck to do it? In other words, why won't industrial pharma hire more talented scientists.
  There is more money in treating a medical condition than in curing it. Once a disease is cured, there is no need to take expensive medications anymore. The financial incentives for both doctors and pharmaceutical companies is to keep a patient in treatment for as long as possible.
15. Re:Bitcoin mining? by Marxist+Hacker+42 · 2014-06-21 13:46 · Score: 1
  
  Jewishness isn't contagious. But Leprosy is, and this is how we practically wiped it out of the Western Hemisphere.
  
  --
  SJW: a person who perceives an injustice, and while correcting it, commits a greater injustice.
16. Re:Bitcoin mining? by pepty · 2014-06-21 13:55 · Score: 1
  
  Research only gives a pay off for the next CEO. Fuck that guy.
  But the stock price moves based on the latest clinical trial results and letters from the FDA. What's a CEO to do?
  ILOS: in license, out source. It's much faster to buy a promising drug candidate (in license) then to herd your own to phase II. It may also be less risky. Hence Valeant trying to buy Allergan. For your remaining pipeline outsourcing R &D to contract labs can cut your R&D budget nicely, and it's the investors looking at R&D spending as a black hole into which their money disappears that makes the CEO nervous. Though you will also lose all of the insight and acumen accumulated by the specialists you laid off when you closed that department.
  You can also try lots of management tricks: give everyone metrics to measure up to so that they always look busy, the pipeline looks full, the competition creatively destructed. Just don't tell anyone the truth: the candidates you are advancing are actually shit, but you have to advance some to keep your bonus. Hey, that sort of thing worked for the Veterans Administration, should work fine for pharma.
17. Re:Bitcoin mining? by the+gnat · 2014-06-21 16:04 · Score: 1
  
  it's the lack of people who have the rare combination of skills of a programmer, mathematician, chemist, biologist and drug engineer coming up with novel and unique ideas to combat disease who will sacrifice industry paychecks to work in academic fields.
  Guess what: there are vastly more jobs like this available in academic groups than in industry. (I know this firsthand, because I work in a related field and am basically stuck in academia unless I can change careers somehow.) The bigger problem is that given the limitations of the simulations and our knowledge of human biology, there is still a huge leap from docking results or MD simulations to working drugs. Effective in silico is light years away from effective in vivo.
18. Re:Bitcoin mining? by the+gnat · 2014-06-21 16:10 · Score: 1
  
  Year in, year out, 75-80% of new drugs are invented privately in biotechs/pharma. The remainder are invented by academia.
  Correct; what government grants pay for is the majority of the basic research that informs efforts to find a cure. Naturally, private companies are (mostly) free to use this information when searching for new drugs - this is part of the point of federal funding for basic research. The vast majority of that research won't directly lead to a cure, of course, but it does contribute to our overall knowledge of biomedicine. In contrast, I've heard the drug development process at some companies compared to "piling up stacks of money and setting it on fire", which is why I'm really, really glad the universities and governments don't try to get deeply into the drug development business.
19. Re:Bitcoin mining? by the+gnat · 2014-06-21 16:27 · Score: 1
  
  There is more money in treating a medical condition than in curing it.
  Not for the insurance companies or government, there isn't. And given the immense cost of long-term treatment for many conditions, pharma companies would be able to charge much more for a drug that completely stopped a disease.* In reality, the reason most medications merely treat rather than cure diseases is that actually eliminating the root cause of a disease without debilitating side effects, for instance death of the patient, is usually really fucking hard.
  (* For instance, the common cold is estimated to be a $40 billion per year drain on the US economy. This suggests that if a drug company could come up with a cure, they would be fabulously rich. Every time I get a cold, my employer loses hundreds of dollars in lost productivity; a $100 pill that returned me to work after a day would be a huge savings, far more effective than $10 of Nyquil.)
20. Re:Bitcoin mining? by the+gnat · 2014-06-21 16:41 · Score: 1
  
  It's because the Republicans won't let them work on that research.
  The National Institutes of Health - the single largest government sponsor of biomedical research in the world - spends
  approximately $3 billion per year on AIDS research. That's about 10% of their entire budget. In comparison, they currently spend about $5.5 billion per year on cancer, which affects vastly more Americans than AIDS, and also kills more in wealthy countries (because AIDS patients - or their insurers - can afford the treatments that enable long-term survival with low viral load). Due to federal budget issues, both funding pools have declined since 2010, but AIDS research only slightly - cancer funding is significantly lower.
  As for treating the cure versus the symptoms, it is extraordinarily difficult to "cure" viral infections with drugs, and HIV has proven to be incredibly difficult to vaccinate against.
21. Re:Bitcoin mining? by Bite+The+Pillow · 2014-06-21 18:18 · Score: 1
  
  You rolled all of R&D into one pile, so horseshit.
  Drug R&D is way different from tech R&D. Microsoft is sitting on billions of unmonetized discoveries. They could, if they converted 10% of the non-operational Microsoft Research into marketable assets, push Apple off the map.
  In comparison, look at the price for a cure for HCV. $100 per pill, or $90,000+ in total. It seems to be a solved problem, and not one of those things where big pharma hides a cure because long-term treatment is financially more beneficial.
  If you want to attack big pharma, the number to argue is the advertising to cost, or advertising to research dollar amount. Not arguing unrelated numbers pulled out of your ass.
  You accused me of being a troll my first post, because you apparently disagreed with me. After pointing out multiple flaws in your reasoning, you are officially a troll. Maybe unintentionally, out of either ignorance or stupidity, but nevertheless you have a hook upon which you post some faith, and with which you intend to catch respondents. Troll under the bridge, troll on a boat, same same bmo.
22. Re:Bitcoin mining? by X0563511 · 2014-06-21 19:23 · Score: 1
  
  They might try not publicly trading. There's no shareholders to yank your leash around. Don't jump into the pool if you don't want to get wet.
  
  --
  For large sets, this will be our guide even unto death, for the LORD will work for each type of data it is applied to...
23. Re:Bitcoin mining? by sillybilly · 2014-06-22 02:01 · Score: 1
  
  The true cure of HIV is a genetic mutation in a human, with HIV resistant genes. The highest likelihood of such a mutation emerging is in an area ravaged by HIV deaths, such as South Africa. If HIV doesn't really cause death (i.e. does not really increase mortality) then it's not really a disease worth mutating against. There are probably a gazillion kinds of bacteria and viruses that live in you, that do nothing special, and these HIV FUD (fear, uncertainty, doubt) spreading people will use one of these standard viruses, and call it the AIDS virus. By the way it's extrememly, extremely difficult, if not impossible, to properly select and filter for a virus - a sodium bloodtest is easy, so is low molecular weight stuff like sugars, etc, but fishing out a virus, staining it with a dye under an electron microscope (because it's too small for optical microscopes, and the electric microscope requires a gold or osmium tetroxide gas layer coating and even then it vaporizes the metal as the electron beam scans, so you can't really dye a metal coating like you can dye the naked virus), then calling it HIV after you stained in with 8 different dies giving you a yes no answer (i.e. 256 bits, as in there are only 256 different viruses out there.) (Notice my recent love of grammatically incomplete/incorrect sentences lacking a verb.) Going for DNA fragment special short chain combination abundance (such as ACTGCTCAACTGACG, a couple of these fingerprint markers specific to each virus) might work, but it's hard to accurately chop DNA to sticks of say 22 units each, and not fully disassemble it to just A, C, T, G, and even then there is a lot of noise, because all life on the planet is based on DNA, and all DNA is made of 2 bit(i.e. 4 variations, 8 bit having 256 variations, and 1 bit 2 variations) A, C, T, G units.
24. Re:Bitcoin mining? by pepty · 2014-06-22 03:41 · Score: 1
  
  Shareholders are hard to avoid. A new drug approval might cost just $200M (Optimer's Dificid), but on average companies are racking up ~$4B in R&D per new drug. (by new drug I mean new molecule, not new dose/formulation/indication for an old one). If you take venture capital they will be looking for a payout well before a drug is approved, probably by either selling the company or its ideas to a big pharma. Keeping it closely held has worked in the past - but you may need vasty deep (10 figure deep) pockets willing to stay the course for a decade.
25. Re:Bitcoin mining? by pepty · 2014-06-22 04:00 · Score: 1
  
  In contrast, I've heard the drug development process at some companies compared to "piling up stacks of money and setting it on fire", which is why I'm really, really glad the universities and governments don't try to get deeply into the drug development business.
  Lots of folks express that exact feeling about the NIH's efforts to get into translational medicine, which is establishing centers to do preclinical and clinical development. I'm pretty sanguine about that too. I could see some payouts happening though even if not many new drugs get made: 1, having publicly funded (and disclosed) research into how to conduct pharma R&D might improve the whole industry; 2, a stable funding source could allow really, really long clinical trials, which could be critical for areas like dementia; 3, an extra carrot for orphan disease drugs in addition to the ones supplied by the FDA.
26. Re:Bitcoin mining? by SuricouRaven · 2014-06-22 06:27 · Score: 1
  
  I'm trying to figure out what you are saying, but I find your post rambling and self-contradictory.
  The test for HIV doesn't use an electron microscope. It's a form of the ELISA test, using specific antibodies that attach only to HIV. Are you trying to claim that HIV is actually harmless, or accusing scientists of fabricating the tests where no virus exists? Because either way, there are about 35 million corpses that disagree with you.
27. Re:Bitcoin mining? by ToddInSF · 2014-06-22 06:50 · Score: 1
  
  Given the extraordinary amounts of money dumped into HIV/AIDS research every year, for decades, the huge numbers of researchers involved globally, and the simply massive amount of published research, when you take it all in and look at it with a degree of curiosity and attempt to compare all this with other historically similar phenomenon...
  
  Well, that's the problem. It's not really like anything else.
  
  Sometimes what is needed isn't more people or money or research. Sometimes one is compelled to look at the reality and ask what is wrong with this picture, and if there needs to be an entirely new paradigm developed, because the results one would expect given the resources dumped into this one just are not there.
28. Re:Bitcoin mining? by ToddInSF · 2014-06-22 07:02 · Score: 1
  
  You know, I keep hearing that HIV has "ravaged" all these places in Africa, based on WHO estimates of infection, yet when I look at the population growth numbers it just does not add-up.
  
  I remember back in the 90's Oprah saying how HIV would impact straight America profoundly, yet none of that came true either.
  
  If HIV is half the threat we've always been told it is, we should have seen some pretty serious widespread population impact by now. Obviously that never happened, is not going to ever happen, and what we've been told is not true.
  
  But go ahead and keep ignoring the facts and keep repeating the bullshit, people that don't have HIV that draw a paycheck by inflating numbers and misleading the public and diverting resources are relying on it.
29. Re:Bitcoin mining? by godel_56 · 2014-06-22 10:02 · Score: 1
  
  Imagen if all that computer power was put to use such as finding the cure of HIV.... We would be done by lunch time.
  If anyone wants to contribute to computer research on HIV with their own systems then there is a World Community Grid project called Fight Aids At home (FAAH) that uses your computer's spare cycles to work on AIDS research, using the BOINC platform.
  There are versions for Windows, Apple, Linux, and Android software.
30. Re:Bitcoin mining? by the+gnat · 2014-06-22 12:29 · Score: 1
  
  the results one would expect given the resources dumped into this one just are not there.
  I don't know, what kind of results do you expect? HIV is a really tough bug to fight - it's almost the opposite of smallpox where a universal and exceptionally effective vaccine was found early on. Tricking the immune system into killing a virus that is evolved to prey on the immune cells was never going to be easy. But the leading antiviral therapies allow most infected patients to live almost indefinitely while maintaining relatively high quality-of-life, whereas 30 years ago they would nearly all have been doomed (and some of the earlier therapies were debilitating). I consider that a pretty impressive achievement of medical technology.
  Now, the fact that millions of Africans (and others) still have AIDS is less impressive, but the reasons for that are entirely social and political, not technological.
31. Re:Bitcoin mining? by ToddInSF · 2014-06-22 13:33 · Score: 1
  
  "But the leading antiviral therapies allow most infected patients to live almost indefinitely while maintaining relatively high quality-of-life"
  
  That's a pretty neat trick right there, I'm not aware of any drug which makes that possible !
  
  Nothing about HIV makes any sense, outside of the political and financial aspect of it. That's the point.
32. Re:Bitcoin mining? by Admiral_Grinder · 2014-06-23 04:21 · Score: 1
  
  So do you have a control group to present to us that didn't buy into the HIV bullshit?
  One of the earliest public cases of HIV was a elementary school kid in Indiana that got it through a blood transfusion. That helped push for higher standards in blood donations. It also pushed for higher standards in dealing with spilled bodily fluids. Before all this though the general public thought was only a STD that the gay community passed around.
  Remember that real bullshit is pretty green grass before being consumed. Without the widespread focus on preventing these diseases it may have been twice as a bad as predicted. It only looks like bullshit today because of the hidden effort to reduce the damage doesn't get presented to (or ignored by) the public which only remembers the hysteria. Same goes for Y2K bug, global warming, and general pollution
33. Re:Bitcoin mining? by ClioCJS · 2014-06-23 04:38 · Score: 1
  
  No, we would not.
  
  --
  -Clio
  Karma: Bad (mostly from not giving a fuck)
  Blog: http://clintjcl.wordpress.com
34. Re:Bitcoin mining? by ToddInSF · 2014-06-23 20:24 · Score: 1
  
  I was very specific in stating that none of the estimates and projections are ever accurate.
  
  If you have a problem with that, take it up with the organizations making the inflated estimates and predictions.
  
  I'm not playing the bullshit game of accepting demands to disprove something that's already well documented.
  
  Pull your own head out of your ass, anon.
Computing also prevents HIV by Rashdot · 2014-06-21 08:28 · Score: 2

Because we don't get out much.

--
This is not the sig you're looking for.
Re:It may be wiser... by fuzzyfuzzyfungus · 2014-06-21 09:45 · Score: 1

The 'meh, it's just homos and drug addicts' theory of epidemiology often leads to some...fascinating...discoveries concerning unexpected transmission paths between the filthy expendables and the good, decent, people who we assumed were safe.

As a social experiment, it's actually pretty interesting. not so much with the 'advisable'; but interesting.
/Very/ different hardware by gentryx · 2014-06-21 09:50 · Score: 2

Computational drug design and bitcoin miners have in common that both run best on custom hardware. The crux is, that both require very different types of hardware. As an example, please refer to Anton, designed by DE Shaw Research exactly for molecular dynamics (MD) codes.
Bitcoin mining is classified as a so called embarrassingly parallel algorithm, while MD is a tightly coupled problem. Hence an efficient parallelization for MD codes is much harder to speed up: communication gets in the way, and communication is essentially always bound by the speed of light.
ps: fun fact: bitcoin mining and MD can be carried out (at least somewhat) efficiently on GPUs.

--
Computer simulation made easy -- LibGeoDecomp
1. Re:/Very/ different hardware by loufoque · 2014-06-23 07:56 · Score: 1
  
  It's not that specialized. It's just plenty of DSPs strapped together on a torus.
  Unlike what wikipedia claims, you could probably achieve comparable performance using a more classical and general-purpose supercomputer setup with GPU or Xeon Phi accelerators, provided the network topology is well tuned to address this sort of communication scheme (most recent supercomputers don't use tori)
Re:Misleading headline by pepty · 2014-06-21 13:26 · Score: 2

Humans have never found a "cure" for a virus. We've been able to find cures for bacterial infections, and we've been able to find immunizations for SOME viruses (NOT a cure, just a stimulation of the body's own immune capabilities) but NEVER a cure for a virus.

Hepatitis C.
It was easy after all by manu0601 · 2014-06-21 13:45 · Score: 1

We bothered about HIV for decades while we just had to run supercomputers to get rid of it. I wonder why nobody thought about it before.
OR... by peter.kowalchuk.reid · 2014-06-21 15:12 · Score: 2

Condoms. Cheap, plentiful, and 99% effective.
1. Re:OR... by As_I_Please · 2014-06-21 19:16 · Score: 1
  
  Those aren't much help to the people already infected.
2. Re:OR... by DigiShaman · 2014-06-22 04:20 · Score: 1
  
  Or don't have casual sex with people you know little about. AKA don't go fucking around.
  
  --
  Life is not for the lazy.
Re:Misleading headline by KingOfBLASH · 2014-06-21 19:58 · Score: 1

As I understand it, the key problem is rate of mutation. We can easily develop a cure for the common cold, for instance, but any cure we develop will be strain based, and the virus just mutates too fast for any "cure" to be effective for more than a few percentage of the population.
The same holds for HIV. If you give the right person a blood sample, he can come back with a cure for your particular virus. However by the time he comes back, your virus has mutated to something different.
Re:It may be wiser... by Admiral_Grinder · 2014-06-23 04:33 · Score: 1

"non-promiscuous, turned out to be pretty much safe"
Some how I bet this kid wouldn't agree with you:
http://en.wikipedia.org/wiki/R...
Yesno? by gentryx · 2014-06-23 08:33 · Score: 1
It's not that specialized. It's just plenty of DSPs strapped together on a torus.
Actually Anton uses ASICS, their cores are specially geared at MD codes. This goes way beyond just "strapping together DSPs". They have IIRC ~70 hardware engineers on site. (Source: I've been to DE Shaw Research last year).

Unlike what wikipedia claims, you could probably achieve comparable performance using a more classical and general-purpose supercomputer setup with GPU or Xeon Phi accelerators, provided the network topology is well tuned to address this sort of communication scheme
No, you can't, and here is why: Anton is built for strong scaling of smallish, long running simulations. If you ran the same simulations on a "x86 + accelerator" system (think ORNL's Titan) then you'd observe two effects:
- The GPU itself might idle a lot as each timestep only involves few computations, leaving many shaders idle or waiting for the DRAM.
- Anton's network is insanely efficient for this use case. IIRC it's got a mechanism equivalent to Active Messages, so when data arrives, the CPU can immediately forward it to the computation which is waiting for it. That leads to a very low latency compared to a mainstream "InfiniBand + GPU" setup.
(most recent supercomputers don't use tori)
Let's take a look at the current Top 500:
- #1 Tianhe-2: Fat Tree
- #2 Titan: 3D Torus
- #3 Sequoia: 5D Torus
- #4 K Computer: 6D Torus
- #5 Mira: 5D Torus
- #6 Piz Daint: 3D Torus
- #7 Stampede: Fat Tree
- #8 JUQUEEN: 5D Torus
- #9 Vulcan: 5D Torus
- #10 nn: 3D Torus
So, torus networks are the predominant topology for current supercomputers.
--
Computer simulation made easy -- LibGeoDecomp
1. Re:Yesno? by loufoque · 2014-06-23 08:40 · Score: 1
  
  Indeed, the GPU constrained programming model is ill-suited for this, but a Xeon Phi isn't. And the next-gen Xeon Phi will have very low latency networking on board, too.
  My bad about tori, the supercomputers I have had access to lately were all fat tree.