Framapiaf

0 message0 participant0 message aujourd’hui

Kevin Karhan :verified:I've yet to see any <a href="https://infosec.space/tags/GPU" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#GPU</a> at <a href="https://infosec.space/tags/MSRP" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#MSRP</a> in <a href="https://infosec.space/tags/Germany" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#Germany</a>!<ul><li>They all are <a href="https://infosec.space/tags/overpriced" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#overpriced</a> for no valid reason whatsoever and it <a href="https://www.youtube.com/watch?v=SPE95_RnL_Q&t=35m14s" rel="nofollow noopener noreferrer" target="_blank">pisses me off</a>, because I didn't buy a <a href="https://infosec.space/tags/GTX1050" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#GTX1050</a> or <a href="https://infosec.space/tags/Vega64" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#Vega64</a> as an <a href="https://infosec.space/tags/invsstment" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#invsstment</a>, but right now both cards.that I was able to snipe at €100 each are actually rising in value because <a href="https://infosec.space/tags/AMD" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#AMD</a>, <a href="https://infosec.space/tags/nvidia" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#nvidia</a> & <a href="https://infosec.space/tags/intel" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#intel</a> only shit out expensive stuff and don't even have MSRP-priced merch available.</li></ul>Worse there ain't even <a href="https://infosec.space/tags/LowEnd" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#LowEnd</a> [< €100] or actual <a href="https://infosec.space/tags/MidRange" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#MidRange</a> [€250 - €500] cards available at MSRP!

Aaro KoskinenThis <a href="https://bitwoods.duckdns.org/tags/ePaper" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#ePaper</a> seems to have 7 internal LUT sets for different temperatures. At 20'C refresh takes 4 secs. By fixing temp sensor to > 50'C, refresh takes 3.5 secs and there's no difference in picture.I copied this "fast" table and started reducing iterations and frame counts so that picture quality still remains intact.I've now reached < 1 s refresh, even for the full 648x480 and it still looks very nice. And now it's fast enough for simple <a href="https://bitwoods.duckdns.org/tags/Linux" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#Linux</a> terminal use!<a href="https://bitwoods.duckdns.org/tags/DIY" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#DIY</a> <a href="https://bitwoods.duckdns.org/tags/Computing" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#Computing</a> <a href="https://bitwoods.duckdns.org/tags/LowEnd" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#LowEnd</a>

Aaro KoskinenLittle progress today but still: Dumped the OTP table data from e-paper which helped a lot. In read direction, these things lack MISO pin so <a href="https://bitwoods.duckdns.org/tags/Linux" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#Linux</a> spidev can't read anything but I was able to do it with GPIO bit-bang from userspace.So I have a copy of LUT data and I'm now able to modify and re-load it back to device. There's a baseline and method for the optimization.A crucial document is UC8179 doc that documents the OTP data.<a href="https://bitwoods.duckdns.org/tags/ReverseEngineering" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#ReverseEngineering</a> <a href="https://bitwoods.duckdns.org/tags/SaturdayNight" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#SaturdayNight</a> <a href="https://bitwoods.duckdns.org/tags/Hobby" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#Hobby</a> <a href="https://bitwoods.duckdns.org/tags/LowEnd" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#LowEnd</a> <a href="https://bitwoods.duckdns.org/tags/Computing" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#Computing</a>

Aaro KoskinenThis week's <a href="https://bitwoods.duckdns.org/tags/ePaper" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#ePaper</a> <a href="https://bitwoods.duckdns.org/tags/hacking" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#hacking</a> status: Got partial updates working nicely (pixel-wise) on 15€ unbranded 5.83" WFT0583CZ61. It's just as slow as full refresh, but limits flickering to the updated area, like a single 8x16 char, so the other part of the screen remains readable during refresh.Now I try to make custom LUTs to make it faster. Nice; all registers are documented but no mention about operation principle whatsoever. Typical trial-and-error <a href="https://bitwoods.duckdns.org/tags/Linux" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#Linux</a> driver programming.<a href="https://bitwoods.duckdns.org/tags/LowEnd" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#LowEnd</a> <a href="https://bitwoods.duckdns.org/tags/Computing" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#Computing</a>

Aaro KoskinenHeh, I've never been an artsy-fartsy graphics guy but I just now recalled how typewriters and matrix printers implemented bold text with 1 color back in the day - just rewind back and type harder!So I tried the similar with my e-ink display. Just go back and redraw a bold character with 1 pixel offset - and you get a nice readable bold text!I consider this a major personal achievement of the week.<a href="https://bitwoods.duckdns.org/tags/Computing" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#Computing</a> <a href="https://bitwoods.duckdns.org/tags/LowEnd" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#LowEnd</a> <a href="https://bitwoods.duckdns.org/tags/TextMode" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#TextMode</a> <a href="https://bitwoods.duckdns.org/tags/Monochrome" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#Monochrome</a> <a href="https://bitwoods.duckdns.org/tags/ASCII" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#ASCII</a> <a href="https://bitwoods.duckdns.org/tags/Guthenberg" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#Guthenberg</a>

Aaro KoskinenToday I'm testing my #8086 emulator running <a href="https://bitwoods.duckdns.org/tags/DOS" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#DOS</a> on a 10€ <a href="https://bitwoods.duckdns.org/tags/RISC" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#RISC</a>-V computer running <a href="https://bitwoods.duckdns.org/tags/Linux" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#Linux</a>.Plenty of speed with negligible power consumption! Pretty impressive and way better than any RPi model. This board would be ideal for running a headless <a href="https://bitwoods.duckdns.org/tags/retrocomputing" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#retrocomputing</a> telnet BBS.<a href="https://bitwoods.duckdns.org/tags/Linux" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#Linux</a> <a href="https://bitwoods.duckdns.org/tags/LowEnd" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#LowEnd</a> <a href="https://bitwoods.duckdns.org/tags/Tinkering" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#Tinkering</a>

Aaro KoskinenGot my first <a href="https://bitwoods.duckdns.org/tags/RISCV" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#RISCV</a> computer - LicheeRV Nano w/64-bit CPU, 256 MB RAM, Ethernet, UART, SD card slot etc at the 10 euro price point.The vendor <a href="https://bitwoods.duckdns.org/tags/Linux" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#Linux</a> image is booting fine, but mainline experience is less smooth - I'm currently stuck at U-boot as the SD card fails to respond?! The first in-kernel DTS will be available in v6.13; maybe I'm just expecting too much at this point... though on U-boot side docs already say the SD should be functional.<a href="https://bitwoods.duckdns.org/tags/LowEnd" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#LowEnd</a> <a href="https://bitwoods.duckdns.org/tags/Computing" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#Computing</a> <a href="https://bitwoods.duckdns.org/tags/Hobbyist" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#Hobbyist</a>

Aaro KoskinenTesting my touchscreen fixes for mainline kernel regressions on <a href="https://bitwoods.duckdns.org/tags/Nokia" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#Nokia</a> 770 using <a href="https://bitwoods.duckdns.org/tags/busybox" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#busybox</a> shell and a virtual keyboard on framebuffer. The keyboard is very small but accurate enough for typing shell commands reliably with a pen.The userspace binary size of my touch GUI is around 20 KB including bitmaps for the "keys" which I generated using the UNIX banner command and sed.<a href="https://bitwoods.duckdns.org/tags/Linux" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#Linux</a> <a href="https://bitwoods.duckdns.org/tags/OMAP" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#OMAP</a> <a href="https://bitwoods.duckdns.org/tags/Hobbies" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#Hobbies</a> <a href="https://bitwoods.duckdns.org/tags/Permacomputing" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#Permacomputing</a> <a href="https://bitwoods.duckdns.org/tags/Minimalism" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#Minimalism</a> <a href="https://bitwoods.duckdns.org/tags/LowEnd" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#LowEnd</a>

Aaro KoskinenBuilding Perl 5 natively has been running for 2 days on OMAP1 w/32MB RAM, so I decided to look if there's something to optimize... It used to finish within a day. The compilation seems to proceed well, it's just slow because of swapping.GCC 12.4 cc1 binary alone is around 22 megs. If I compile it with -Oz it shrinks almost 4 megs! On other binaries (like binutils) difference is not that big but still measurable. Going to check if this helps.<a href="https://bitwoods.duckdns.org/tags/Linux" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#Linux</a> <a href="https://bitwoods.duckdns.org/tags/ARM" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#ARM</a> <a href="https://bitwoods.duckdns.org/tags/SoftwareBloat" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#SoftwareBloat</a> <a href="https://bitwoods.duckdns.org/tags/LowEnd" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#LowEnd</a> <a href="https://bitwoods.duckdns.org/tags/Hobbyist" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#Hobbyist</a>

Recherches récentes

Options de recherche

Administré par :

Statistiques du serveur :

#lowend