Timeline for Should I connect 2-3 laptops to form a supercomputer?
Current License: CC BY-SA 4.0
19 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jun 17, 2021 at 17:10 | comment | added | lupe | As an HPC geek, yeah, this isn't going to work well - though, if you have access to a lot of consumer grade machines, HTCondor is pretty good at getting consumer grade hardware into a useful HPC like resource - be warned, I've fried at least 6 machines with it - making an auto restart script is a bad idea | |
| Jun 17, 2021 at 13:53 | comment | added | Buttonwood | Stackoverlow has a dedicated tag «cluster-computing», with entries like this and this. Some build their own cluster (example) as a cheaper home / DIY procject. | |
| Jun 17, 2021 at 9:58 | comment | added | schoon | Actually I remember now there were lots of issues around networking which took me awhile because I had no networking experience. | |
| Jun 17, 2021 at 9:56 | comment | added | schoon | Took me a long time to get it right! But a lot of problems were due to the machines' limitations. If you want hadoop Cloudera's auto-install got better over the years but there were always problems. If all goes well just a few hours, but nothing ever goes well! | |
| Jun 17, 2021 at 8:16 | comment | added | Parmeet Singh EP 066 | @schoon How much time would it take to assemble considering I have no knowledge about HPC and other computing stuff. I am also pretty new to computational chemistry. | |
| Jun 17, 2021 at 7:55 | comment | added | schoon | I created a cluster about 4 years ago using crappy laptops to run Hadoop. Was a bit slow but not so bad. Just used an ethernet switch and Ubuntu (can get replace Win with that). It allowed me to work on much bigger data sets. | |
| Jun 16, 2021 at 22:45 | answer | added | Jimmio92 | timeline score: 4 | |
| Jun 16, 2021 at 18:30 | comment | added | wzkchem5 | Can't you buy some CPU-hours from a supercomputer service? If you are part of a research group in a university or a research institute, it should always be possible to do so, typically by registering an account at a supercomputer center and paying for your CPU-hours at regular intervals | |
| Jun 16, 2021 at 17:07 | comment | added | Nike Dattani - No Free Time | @Greg Sounds like that could be written as an answer! | |
| Jun 16, 2021 at 16:59 | answer | added | ciamej | timeline score: 10 | |
| Jun 16, 2021 at 16:57 | comment | added | A rural reader | Might be worthwhile to check out aws.amazon.com/hpc/parallelcluster. | |
| Jun 16, 2021 at 16:50 | comment | added | Greg | No. 0) "clusters" with 2-3 nodes make no sense. one node runs all kinds of admin software, the remaining 1 or 2 nodes doing the actual computation... zero speedups compare to one node. 1) admin of any cluster is a lot of work (= cost), 2) laptop hardware is not the most reliable compared to the price, generally has bad cooling, small memory, etc, so your project would be frying after a day or two 3) DFT in 3D scales purely, so you will need 10-20X more computational power to see any difference. | |
| Jun 16, 2021 at 16:38 | history | became hot network question | |||
| Jun 16, 2021 at 16:34 | history | edited | Nike Dattani - No Free Time | CC BY-SA 4.0 |
edited title
|
| Jun 16, 2021 at 10:39 | vote | accept | Parmeet Singh EP 066 | ||
| Jun 16, 2021 at 1:22 | comment | added | Camps♦ | My strong recommendation is: if you don't have access to huge computational resources, go with SIESTA that it is much less resource consuming. | |
| Jun 15, 2021 at 20:41 | answer | added | Nike Dattani - No Free Time | timeline score: 18 | |
| Jun 15, 2021 at 20:24 | history | edited | Nike Dattani - No Free Time | CC BY-SA 4.0 |
added 18 characters in body; edited tags; edited title
|
| Jun 15, 2021 at 20:18 | history | asked | Parmeet Singh EP 066 | CC BY-SA 4.0 |