cheap ISP for low-bandwidth use?
Tom Roche
Tom_Roche at pobox.com
Wed Oct 28 20:17:10 MST 2015
Tom Roche http://lists.phxlinux.org/lurker/message/20151027.194425.556c2517.en.html
>>> how cheap can one rent a reliable, Linux-compatible internet connection in the Valley?
Stephen Partington http://lists.phxlinux.org/lurker/message/20151027.202526.8af0ce8c.en.html
>> [CenturyLink] is offering 20/mo for 10 down / something up (im guessing 2mb or so)
Thanks for the pointer. Of course, what I'd really like to know is what their actual performance is :-) When I was last with TWC, they promoted {10 MB down, 2 MB up} but I never got close to that: IIRC my best TWC home download ever was 1.92 MB/s, briefly, while on a direct wire connection to the cablemodem. And that was on a laptop that regularly did 20 MB/s downloads on my academic network.
>> but i have no idea what they filter and block.
I'll check. FWIW I definitely prefer to post via SMTP--it's one of the reasons why I use pobox.com, though the primary reason is the stable alias.
Michael Butash http://lists.phxlinux.org/lurker/message/20151028.061401.686176e5.en.html
> who really uses [GB] speeds aside from a service provider or having 10 kids all using netflix at the same time watching at 4k res?
One legit usecase for environmental modeling and similar subfields of scientific computing is testing via visualization.
While a CS student I drank the TDD koolaid, and spent much time as a corporate coder automating tests. (At least, when management would permit, but that's another issue.) Then I took the vow of poverty and went into modeling, where all-too-often the only feasible way to test output from a reasonably-successful run is to eyeball a visualization. (One writes statistical code to screen out less-successful runs, e.g.: concentrations=NAN in all voxels ? abend : continue) The problem is, instead of acceptance-testing an MB-scale webpage or remote-desktop UI (the assessment of which one can of course automate, though I'll admit to occasional laziness and one-offs), visualization of a large/complex spatiotemporal field usually requires downloading one or more GB-scale (or 1e8-scale) PNGs or PDFs. (Then asking, "does that look reasonable?")
That being said, right now I'm between contracts, so I'm trying to keep costs down, so ...
your assistance is appreciated, Tom Roche <Tom_Roche at pobox.com>
More information about the PLUG-discuss
mailing list