log☇︎
1800+ entries in 0.195s
mod6: http://www.mod6.net/fg/fg-test/20170602_150946.jpg
mod6: http://www.mod6.net/fg/fg-test/20170602_142155.jpg http://www.mod6.net/fg/fg-test/20170602_142450.jpg
mod6: just throwing a matrix together of all the test data.
mod6: http://www.mod6.net/fg/fg-test/fg-test.html
mod6: but the battery cell is somehow better for this test?
asciilifeform does not recommend the histogram thing as a firstline test, it was written as diagnostic to ferret out broken os
asciilifeform: the perhaps worst 'test' ~of a working rng in particular~ is to look at the hex with naked eye
mod6: makes sense, i think it was worth covering this though -- wanted to be sure that I have one solid test-bed.
asciilifeform: ( if there were, you would fail dieharder's 'birthdays' test at the very least. )
asciilifeform: first test however is always 'ent', you can see in pete_dushenski's sample that ent immediately produces barf
asciilifeform: and wanted the thing to be repeatable ('yoke test' per the docs)
asciilifeform: so we have him, who won't test (phree!111) sample of FUCKGOATS...
asciilifeform: they don't test for it
a111: Logged on 2017-05-30 05:11 pete_dushenski: asciilifeform: http://contravex.com/fg_test_alf.bin
pete_dushenski: asciilifeform: http://contravex.com/fg_test_alf.bin ☟︎
asciilifeform: how much this even changes when it ~is~ put to test ? e.g. the folx during the plague who carried around paper talisman, word 'arsenicum' (because obtaining actual arsenic for talisman costs a little more..)
mircea_popescu: but the point is that their temple will never be put to the test.
mod6: So that should conclude the collections for now. I'll be putting some effort into some fg-anal sometime soon -- maybe nothing more than some basic facts extrapolated from the test results per collection and links to the raw results.
mod6: OK here are the results from a recollection of entropy (with 'fullblock') from FG #2: http://www.mod6.net/fg/fg-test/fg2.ent_run1.txt http://www.mod6.net/fg/fg-test/fg2.dieharder_run1.txt
mod6: thanks! last dieharder test is running now, actually.
mod6: Now, I just need to re-run one more test from FG #2, then the collections will be complete, all using 'fullblock'.
mod6: http://www.mod6.net/fg/fg-test/fg1.dieharder_run3.txt
mod6: http://www.mod6.net/fg/fg-test/fg1.ent_run3.txt
mod6: I have re-run all three entropy collections from FG #1 with 'fullblock', here are the results: http://www.mod6.net/fg/fg-test/fg1.ent_run1.txt http://www.mod6.net/fg/fg-test/fg1.dieharder_run1.txt http://www.mod6.net/fg/fg-test/fg1.ent_run2.txt http://www.mod6.net/fg/fg-test/fg1.dieharder_run2.txt
asciilifeform: at some point i intend to try to convince mircea_popescu to read it and it'll be ultimate test
IamAGirlsoInever: Oh, a test from way back when, I implemented an XML reader and negelected to cap off recursion, so I tried a file full of <_> commands, just open a short tag.
asciilifeform: post. example. of something you tested. and the test result.
IamAGirlsoInever: test like science, double blind, use a premise, try to prove a point?
asciilifeform: word 'test' implies a result..
a111: Logged on 2017-05-24 22:16 asciilifeform: previously i thought that you could not have an auditable single-crystal rng, but possibly this is not so : if you make it so that only co-incident (on 2 detectors) gamma is picked up, you can 'yoke'-audit 2 units by placing back to back around test source.
asciilifeform: all them folx who skimped on 'test p-1 and q-1'
a111: Logged on 2017-05-25 03:08 mod6: !~later tell gabriel_laddel_p Let me know how much entropy you'd like, I'll run ent test & dieharder against it, let you decide if you want it. I'll ask 0.1 BTC per Gb. What is your bid?
a111: Logged on 2017-05-25 03:08 mod6: !~later tell gabriel_laddel_p Let me know how much entropy you'd like, I'll run ent test & dieharder against it, let you decide if you want it. I'll ask 0.1 BTC per Gb. What is your bid?
mod6: !~later tell gabriel_laddel_p Let me know how much entropy you'd like, I'll run ent test & dieharder against it, let you decide if you want it. I'll ask 0.1 BTC per Gb. What is your bid? ☟︎☟︎
asciilifeform: theoretically you dun need a physical unit, for such a test
asciilifeform: previously i thought that you could not have an auditable single-crystal rng, but possibly this is not so : if you make it so that only co-incident (on 2 detectors) gamma is picked up, you can 'yoke'-audit 2 units by placing back to back around test source. ☟︎
asciilifeform: compared performance of 'kindergarten' multiplier ( halfword at a time, uses machine's MUL instr. which apparently IS constant-time on x86 (but NOT on arm! would need test routine at warmup to establish if can use MUL ) )
asciilifeform: and yes, this is still same old tests, nothing magically different about them, there will ~never~ be a mechanical test for 'good rng'
asciilifeform: so next step is: once we have a sadness equation, you can take it and test its guess rate vs expected value, with new output sample.
asciilifeform: my original contention is that any test which, upon 'failure', cannot produce a sadness equation, is ill-conceived.
asciilifeform: say an rng fails the 'birthdays' test.
mircea_popescu: but this is not what you're asking, for a test that sometimes finds something, is it.
mircea_popescu: necessarily no test can be formulated.
asciilifeform: and any test that cannot be formulated in this way -- is rubbish and oughta go away
asciilifeform: but to display ~existing~ test failures in predictor-algorithm form ☟︎
asciilifeform: http://btcbase.org/log/2017-05-23#1660248 << this ain't the problem -- already right off the bat , sane operator knows that it is physically impossible to test for ~good~ rng, only for ~bad rng~ ☝︎
mod6: re-doing the tests for FG1 and the first test of FG2. Basically these were run prior to the 'fullblock' thread.
mod6: I was about to embark on some fg-anal, but realized that I needed to re-run all of the test from FG #1 and one test from FG #2 -- as those were run before the thread about dd 'fullblock'.
mod6: Here's the third entropy collection from FG #5: http://www.mod6.net/fg/fg-test/fg5.ent_run3.txt http://www.mod6.net/fg/fg-test/fg5.dieharder_run3.txt
Framedragger: "We intended to test the hypothesis that flattery of the academic Left’s moral architecture in general, and of the moral orthodoxy in gender studies in particular, is the overwhelming determiner of publication in an academic journal in the field." noshit.jpg but also applause.jpg
asciilifeform: when i'm satisfied with the numerics, incl. primality test, i'ma genesis it..
mod6: Here's the results from the second run of FG #5: http://www.mod6.net/fg/fg-test/fg5.ent_run2.txt http://www.mod6.net/fg/fg-test/fg5.dieharder_run2.txt
a111: Logged on 2016-08-19 15:26 asciilifeform: (interest in the subj is slim, on account of miller-rabin probabilistic test provably converging - i.e. probability of catastrophic failure can be made smaller than probability of machine flattened by asteroid the same night)
asciilifeform: http://trilema.com/2012/o-hai-i-was-justing-doing-a-penetration-test-of-your-site/#comment-122056 << ftr these are bots
mod6: Here's the results from the first entropy collection of FG #5: http://www.mod6.net/fg/fg-test/fg5.ent_run1.txt http://www.mod6.net/fg/fg-test/fg5.dieharder_run1.txt
Framedragger: test
a111: Logged on 2017-05-14 05:12 asciilifeform: if it works on arbitrary addrs, seems like it'd be wuite asy to test
ben_vulpes: http://btcbase.org/log/2017-05-14#1655564 << easy to test when complete, easy to show that it does not work, not easy to test in the sense of "i have changed implementation details, and would like to see how the system behaves now" ☝︎
ben_vulpes: relatedly, i have a patch in abeyance that fixes the test target. i'll bring that out of the refrigerator and start wiring this new IsMine implementation into it.
ben_vulpes: i'm going to step away from testing the indexer /in toto/, and plug this IsMine overload into the test target
trinque: will test ben_vulpes
mircea_popescu: "very confused salesman" test. i gotta remember this one.
asciilifeform: why not do same test by attempting the original ( now - public ) remote diddle ?
asciilifeform: if it works on arbitrary addrs, seems like it'd be wuite asy to test ☟︎
ben_vulpes: i am considering testing this in conjunction with a solipsistic miner, but may test it against the extant blockchain instead. input on this also welcome
ben_vulpes: now i have to decide how to test it, for the odds are that it does not work as-written. but i do invite commentary on the design!
jhvh1: asciilifeform: Can You Really Cook Salmon in a Dishwasher ? | Kitchn: <http://www.thekitchn.com/can-you-really-cook-salmon-in-a-dishwasher-putting-tips-to-the-test-in-the-kitchn-218048>; 5 Foods You Can Cook in Your Dishwasher ... if You Dare | HuffPost: <http://www.huffingtonpost.com/the-stir/dishwasher-cooking_b_3605859.html>; Dishwasher Salmon with a Piquant Dill Sauce Recipe | Bob Blumer ...: (1 more message)
mod6: Here are the test results on third run of FG #4: http://www.mod6.net/fg/fg-test/fg4.ent_run3.txt http://www.mod6.net/fg/fg-test/fg4.dieharder_run3.txt
BingoBoingo: This is why test/experiment!
asciilifeform: BingoBoingo: what kind of test was this ?
BingoBoingo: In other things, sync test is now within 1500 blox
jpt4: Give fbar a test.
asciilifeform: http://asciilifeform.com << asciilifeform's mp-wp test install. comments dun work...
mod6: Here's the results from the second entropy collection from FG #4: http://www.mod6.net/fg/fg-test/fg4.ent_run2.txt http://www.mod6.net/fg/fg-test/fg4.dieharder_run2.txt
mod6: Here's the results from the first entropy collection from FG #4: http://www.mod6.net/fg/fg-test/fg4.ent_run1.txt http://www.mod6.net/fg/fg-test/fg4.dieharder_run1.txt
asciilifeform: asciilifeform was recently buying faraday ~bags~ to test, after noticing that they have become cheap and ubiquitous, and then went to find WHY
asciilifeform: 3 ) trivially boosted to almost any bit rate you might want, by simple expedient of putting 'test capsule' from geiger kit in the box...
asciilifeform: exactly not like. 'test socket' is what folx used to call a 'zif', gizmo for use in factory to test ~chips~ on conveyor, rated for many insertion cycles and easy to open/close
mircea_popescu: so not like eg http://www.arieselec.com/products/data/23026-bga-high-temp-burn-in-and-test-socket.htm ?
mod6: Hi all: Ok 3rd run of entropy collection & `ent` & `dieharder` tests are done on FG #3: http://www.mod6.net/fg/fg-test/fg3.ent_run3.txt http://www.mod6.net/fg/fg-test/fg3.dieharder_run3.txt
a111: Logged on 2017-05-04 03:12 mod6: here's the results from FG #3, second run of entropy collection1.1Gb worth. `ent` && `dieharder` results: http://www.mod6.net/fg/fg-test/fg3.ent_run2.txt http://www.mod6.net/fg/fg-test/fg3.dieharder_run2.txt
a111: Logged on 2017-05-04 03:12 mod6: here's the results from FG #3, second run of entropy collection1.1Gb worth. `ent` && `dieharder` results: http://www.mod6.net/fg/fg-test/fg3.ent_run2.txt http://www.mod6.net/fg/fg-test/fg3.dieharder_run2.txt
mod6: here's the results from FG #3, second run of entropy collection1.1Gb worth. `ent` && `dieharder` results: http://www.mod6.net/fg/fg-test/fg3.ent_run2.txt http://www.mod6.net/fg/fg-test/fg3.dieharder_run2.txt ☟︎☟︎
Framedragger: re. per minute, i guess i should reach out to consult first eh - this was borne from a solipsistic "me as a customer" consideration: e.g. i want to test out some trb feature across multiple different instances, i need good i/o and memory; monthly costs would not be trivial (for the purpose at hand), so i wish to be charged on smaller timescales.
mod6: mircea_popescu, asciilifeform, et. al: Third FG, first run of entropy collection & testing of `ent` && `dieharder` complete: http://www.mod6.net/fg/fg-test/fg3.ent_run1.txt http://www.mod6.net/fg/fg-test/fg3.dieharder_run1.txt
asciilifeform: this test only works in one direction
asciilifeform: seems like the only guaranteed solutions are either to trial-divide every time (eat 2x the space and 2x the time!) or to keep around 'behind the scenes' 2x the nominal register, so as to have where to overflow into, and then test it for zero (2x the space, and 2x the time ! just like the other 'solution')
mats: the provided answer was - first test for Intel CPU with CPUID (so you don't trigger f00f) and then for anything GenuineIntel, run it and check for undefined opcode exception
mats: relatedly - a trivia q i didn't know the answer to, until it was shared - what is the proper way to test for support for the cmpxchg8b instruction?
ben_vulpes: current thinking in re testing dumper is to solo mine and test but that's a not-small pile of test harness to write
ben_vulpes: gotta back out the c++11 work, implement a predicate and test the shit out of it
mircea_popescu: http://trilema.com/2011/test-de-cultura-botanica/ << these are artichoke babies. they grew up great.
mircea_popescu: http://trilema.com/2012/test-de-cultura-botanica-iii/
asciilifeform: Framedragger: test with/without numlock. betcha it worx as-is.
mod6: doesn't probably do us all justice if I run some of these test collections on some shitty hardware i may or may not have.
mod6: once all 5 are done, i'll pull together something more comprehensive for comparison; currently, i just have the raw output test data.
mod6: Test run #3 of `ent` && `dieharder` against 2nd FG finished: http://www.mod6.net/fg/fg-test/fg2.ent_run3.txt http://www.mod6.net/fg/fg-test/fg2.dieharder_run3.txt
mircea_popescu: i figure we can afford a test unit for some dude you read.
mod6: if you're looking at this chart: http://www.mod6.net/eatblock-test/trb_offline_eatblock.png
mod6: Ladies and Gentlemen of The Most Serene Republic: My Second Offline Eatblock Sync Test is complete (with DB Read Wait Stats): http://www.mod6.net/eatblock-test/