Randy Simons

@Sh41 @aperezdc @electron_greg The floating point representation and math differs quite a lot between these old machines, including in the precision. MSX uses doubles by default, 14 digits precision. That's quite some overhead for these simple FOR loops used here.

However, dropping to single precision (DEFSNG A-Z) the time goes down from 53.4s to 50.8s.

Might be fun to compare the "integer version" performance of this code on these machines. The MSX version thus is 26.4s.

@aperezdc @electron_greg Had to check on MSX turbo R :) Integer math again. Runtime: 4.6 seconds. And, with KUN BASIC compiler: 0.2 seconds :D

@aperezdc @electron_greg Using integer math makes it twice as fast: from 53.4s to 26.4s. Screenshot shows both versions, on an emulated FS-A1WSX #MSX for good measure. First the original, then the version with integer math. Note that ¥ is actually a \.

@ilpianista Chum GUI does lack a "what's new" feature, which Open Repos / Storeman does have. That hurts discoverability of new apps.

@leeg@fosstodon.org @jani@floss.social Sure, and there are others, like Olmo, but those models are currently not really feasible for software development. Of course, I do hope those orgs will keep up, and eventually release models better suited for development as well.

@jani@floss.social @leeg@fosstodon.org I don't know if we differ on that. I use Claude/GPT/Gemini models for work. They *are* way better than qwen3-coder-next, which I've used on my recent hobby project. But it's (already) useful. My bigger concern is that these free models are still too expensive to train by free/libre orgs. We just got some freebies/appetizers from big commercial orgs. If (when?) those dry up, we indeed get to a point were only subscription to proprietary services exist as option.

@jani@floss.social @leeg@fosstodon.org But that qwen3-coder-next model is more free (as in libre) than that VC++ license ever was. Runs locally on my machine. Never requires new hardware. So present, affordable hardware (minus DRAM...) can already run free, capable AI. It will get better still. And newer (free) models might require beefier hardware, but how is that different from free software?

@leeg@fosstodon.org @jani@floss.social None. It's impractically slow for actual work-use, but I run qwen3-coder-next on a little ~€750 AMD machine with regular 64GiB DDR5 DIMMs. Granted, I got that RAM 2.5 years ago before the craze. But with Strix Halo hardware I'd say its feasible. Like you said, how accessible was hardware, let alone software, in the early days of (personal) computing? I paid ~€130 for a STUDENT license Visual C++ in ~1998! The original premisse feels almost entitled.

@mathowie But even e-mail is decentralized, whereas Signal is fully centralized. Yes that might be convenient and all, but centralization is a hard pass for me.

@marleenstikker@waag.social Ik mis nog een punt over het digitale contact met de burger, en de afhankelijk daarvan met Big Tech. Veel overheidsapps worden enkel voor iOS en Android gemaakt. Gebruikers van een besturingssysteem dat niet van Apple, Microsoft of Google is hebben het nakijken. Mijn pet peeve in deze is DigiD. Inloggen met extra ID-check is voor mij onmogelijk daardoor.