2022-04-21 00:00:12 aha! you're benchmarking! 2022-04-21 00:00:46 Oh yeah - I got set up for that lats week. Wrote 'uet' which gets me universal time to the microsecond. 2022-04-21 00:44:07 Ok - I THINK I've just measured a rough "average time per primitive" and gotton 1.5 ns. 2022-04-21 00:44:40 I modified the tick counter so it just starts from zero and increments upward (temporary - just for tests). And I had my uet word for the other day. 2022-04-21 00:45:28 So I did this: uet uet perod . which gave me 130 us. Then I did tick tick swap - and got 88201. 2022-04-21 00:45:51 88201 trips through next between those tick samples. 2022-04-21 00:46:17 I'm reasoning that that measures some sort of "typical code" scattered about throughout my system. 2022-04-21 00:46:38 130 us is 130000000 ps, and that divided by the ticks is 1473. 2022-04-21 00:46:48 So 1473 picoseconds -> 1.5 ns. 2022-04-21 00:46:53 Does that seem sane? 2022-04-21 00:48:18 So that comes out to about 679 million primitives per second. 2022-04-21 00:50:05 "sane" ?! computers these days are crazy :-) 2022-04-21 00:50:40 loop a billion times takes seconds 2022-04-21 00:50:49 Oh, I meant my calculation - not the result. 2022-04-21 00:50:50 even less 2022-04-21 00:51:08 Also, I measured NEXT the other day and got right at 1 ns. 2022-04-21 00:51:32 yeah single digit nanoseconds sounds right 2022-04-21 00:51:51 So if "next" is a nanosecond and "average primitive time" is 1.5, then that says 2/3 of my time is spent in NEXT. That matches up fairl well with the roughly 3x C vs. Forth comparison I got on one of my previous systems. 2022-04-21 00:52:29 Most of my primitives are pretty short, which makes NEXT a larger percentage. 2022-04-21 00:55:32 And THAT means that the whole business I was talking about a few days ago, of profiling to find the burdensome code, and potentially writing it in assembly, has a lot of payoff to offer. 2022-04-21 00:55:47 Eliminating NEXT from deep inner loops stands to pay off 3:1. 2022-04-21 04:15:48 I keep coming back to Forth, but I feel like it's only after retirement I'll have the time to use it 2022-04-21 11:36:10 Morning, folks. 2022-04-21 12:29:16 good morning KipIngram 2022-04-21 16:32:51 Agh it annoys me when a website that doesn't have anything fancy going on spins my fans up to an audible level. Once the information is on my screen it's just a static image - it shouldn't be taking any serious amount of power. 2022-04-21 16:34:03 I'd prefer they not be running at all, but the definitely shouldn't be distinctly audible. 2022-04-21 16:43:44 What's frustrating is that there's no good way to tell which tab is doing it, so I wind up closing tabs I'd rather have left open in my fever to find it and kill it. 2022-04-21 16:47:21 what browser? 2022-04-21 16:47:36 Chrome. 2022-04-21 16:48:19 It's like some sites must sit there and chit-chat with your browser the whole time you're there or something. 2022-04-21 16:48:53 I like the original model: please give me this file - thank you, now go away until I ask for something else. 2022-04-21 16:48:54 chrome should have a task manager that may help identify which tabs are being bad (I think under the three dot menu, then 'more tools') 2022-04-21 16:49:12 Oh, ok - that would be cool. I'll take a look. 2022-04-21 16:51:43 I just discovered the lm_sensors package earlier. 2022-04-21 16:51:52 Lets me find out my fan speed. 2022-04-21 16:52:13 Before I closed the tabs it was up at 4636 rpm. 2022-04-21 16:54:47 Now it's a big happy goose egg again. 2022-04-21 17:25:27 Oh, yes - the Task Manager is under "More tools..." 2022-04-21 17:28:28 That makes more sense - if it had been directly in the "three dot" menu I'd think I would have noticed it. 2022-04-21 21:33:13 Man - APL is just not the easiest thing in the world to learn. 2022-04-21 21:33:31 I'm still picking up bits, but it's a slow haul. 2022-04-21 21:46:52 It's fun, though. 2022-04-21 22:32:58 Oh, I'm quite fascinated with it, and I definitely sense that there's a lot of power there. And power of that sort is almost always fun. I'm an engineer by training, but I've always been sort of heavily "science / math / formality" oriented. It seemed apparent right away that APL has a *really* strong connection to math formalism - and I love "crafting with rigorous rules" like that. 2022-04-21 22:32:58 It looks like the games you can play with "variable shaped data structures" are quite extensive.