aappleby

31 ms·

451.
by aappleby 5y ago
And incredibly annoying as the developer - so much work translating and debugging gates, and it's still not "correct" for some definition of correct ;)
452.
GateBoy – A gate-level simulation of the original Game Boy hardware(github.com)
416 points by aappleby 5y ago|66 comments
453.
by aappleby 5y ago
I love my tungsten cube <3
454.
by aappleby 5y ago
can confirm. -austin (murmur author)
455.
by aappleby 5y ago
biden was president for how many of those 20 years?
456.
by aappleby 5y ago
Very nice work. Try putting byteswap into the set of possible instructions and see what happens. -Austin, murmur3 author
457.
by aappleby 5y ago
in mice
458.
by aappleby 5y ago
This was a good article until I got to this bullshit - "There are only two genders, male/female, masculine/feminine, and only two pronouns, he/she, him/her, and the whole transgender, gender-fluid and other crappy “
459.
by aappleby 5y ago
FWIW, filling a screen with monospaced pixel-font text (plus colors and stuff) takes single-digit microseconds on a modern GPU once everything is in on-device memory. Fancier fonts or distance-field-based antialiasing is a few times that. A
460.
by aappleby 5y ago
"If there are enough rabbits, the people eat till their stomachs are distended; but no matter how much they eat they feel unsatisfied."
461.
by aappleby 5y ago
Getting it wrong is part of the job.
462.
by aappleby 5y ago
Ah, I missed that there was a data-dependent rotation applied in the output. So it's a lcg plus xorshift plus ddr. Still weaker than I'd like, but if it passes DieHard etc then it passes.
463.
by aappleby 5y ago
I'll read the paper.
464.
by aappleby 5y ago
I could very well be wrong, I haven't checked yet. That said, I have done a bit of work with these sort of things.
465.
by aappleby 5y ago
Yes, I predict it will fail most of the DieHard tests.
466.
by aappleby 5y ago
I predict that prng will totally fail the DieHard tests.
467.
by aappleby 5y ago
The _advance function in this prng is also way too weak, purely linear with a tiny bit of output mixing. Replacing self.state with the (64-bit) result of the xorshifts would help a lot.
468.
by aappleby 5y ago
You can get the same result by hashing seed+level_id+some_counter, you don't really need a prng.
469.
by aappleby 5y ago
Oh and yeah the remaining glBufferData would def be on my fix-next list. Unless the tilemap changes dynamically, all those tile triangles should be grouped into chunks in one "vertex atlas" and their screen position controlled by
470.
by aappleby 5y ago
The bit about optimizing his blit shader surprised me (as a 20+ year graphics dev) - the few extra multiply-adds should be virtually free on most architectures as that small a shader would generally be ROP-limited.
471.
by aappleby 6y ago
no mention of https://en.wikichip.org/wiki/mtr-mm%C2%B2 ?
472.
by aappleby 6y ago
If you can lift the device it's stored on with your bare hands, it's not big data.
473.
by aappleby 6y ago
You set up a timer to fire a "calculate next sample" interrupt at ~4khz, then your main thread just walks through your MIDI file or whatever and tweaks the synthesizer settings.
474.
by aappleby 6y ago
I had 8-ish channel FM synthesis (using a bunch of lookup tables) on a stock 8-bit ATMega-based Arduino back in the day. Any modern Cortex-based microcontroller has more than enough oomph to handle 4x as many voices plus filters, etc.
475.
by aappleby 6y ago
Paywall.
476.
by aappleby 6y ago
Incredibly misleading article. I'm reminded of the old joke about "It works in practice, but how does it hold up in theory?"
477.
by aappleby 6y ago
Nice. Wish I'd seen this ~15 years ago when I was doing a font renderer.
478.
by aappleby 6y ago
All renders, no reality.
479.
by aappleby 6y ago
Because it's like staring into a mirror all day.
480.
by aappleby 6y ago
Oh and it does zero dynamic allocations except during atlas resize, which was very helpful in older builds of V8.

More ›