Search logs:

channel logs for 2004 - 2010 are archived at ·· can't be searched

#osdev2 = #osdev @ Libera from 23may2021 to present

#osdev @ OPN/FreeNode from 3apr2001 to 23may2021

all other channels are on OPN/FreeNode from 2004 to present

Sunday, 1 May 2022

00:02:00 <nomagno> mrvn: 900 kmh is the kind of speed at which most physics engines blow up...
00:02:00 <nomagno> But indeed the hectograms part is actually a bad idea
00:03:00 <nomagno> Because a modern f1 weighs 8000 hectograms, which would be enough to make the calculations go poof at small speeds
00:11:00 <geist> doesn't GBA have an arm7tdmi?
00:11:00 <geist> that has a perfectly capable multiplier
00:11:00 <mrvn> The Game Boy Advance (GBA) is a 32-bit handheld game console.
00:11:00 <mrvn> Why not go with 16.16 fixed point even?
00:12:00 <mrvn> CPU: ARM7TDMI @ 16.78 MHz
00:12:00 <geist> yah agreed, 16.16 would at least be more regular if you have to implement it in software
00:15:00 <mrvn> On my Amiga I used 16.16 too because that fit nicely with 16*16=32 bit multipleis.
00:16:00 <mrvn> geist: does your 68010 have a bigger MUL or was that added at 040 level?
00:17:00 <mrvn> or maybe 030
00:22:00 <mrvn> nomagno: don't overlook that {uint8_t; uint16_t} on a 32bit cpu will have 1 byte padding so you save nothing over the full 32bit.
00:23:00 <mrvn> imho 24bit numbers would only make sense as 3 byte.
00:23:00 <mrvn> for 8/16 bit cpus
00:34:00 <geist> mrvn: i think it des the full mul from day 1, its just the ALU is narrow until 020
00:35:00 <geist> but there maqy have been a 64 mul or something added later
00:39:00 <geist> yah i dont see any div or mul that wasn't defined on the 68000
01:00:00 <mrvn> geist: There is a mul for d0 * d1 = d0:d1, 32 * 32 = 64
01:01:00 <mrvn> Some of the 64bit stuff go removed on the 060.
03:54:00 <jjuran> mrvn: 68020 added long MUL and DIV
03:55:00 <jjuran> geist: ^
03:56:00 <jjuran> 68000 and 68010 have 16*16 -> 32 MUL and 32/16 -> 16r|16q DIV
06:04:00 <clever> jjuran: that reminds me, the vector core on the rpi also has a 16*16 -> 32 MUL opcode, that works on 16 lanes at once, ive tried to implement a 32bit mult with that, but ran into trouble with the addition step
06:06:00 <jjuran> clever: Sure, you can FOIL the multiplies, but you still have shifts and adds to do
06:09:00 <clever> FOIL?
06:09:00 <clever> and yeah, it was the 64bit add that i had trouble with
06:59:00 <jjuran> clever: First, Outer, Inner, Last
07:01:00 <jjuran> (a + b)(c + d) —> ac + ad + bc + bd
07:03:00 <jjuran> and recognizing that a longword is the sum of the low word and the shifted high word
07:09:00 <clever> yeah, i did brute force my way into basically doing ah*bh, al*bl, ah*bl, and al*bh
07:09:00 <clever> roughly stealing x86's AH + AL = AX naming scheme
07:09:00 <clever> so i get 4 32bit products
07:09:00 <clever> but then i need to shift them by 0, 16, 16, and 32, and add them together
07:28:00 <jjuran> and carry any overflowed bits
07:29:00 <geist> jjuran: aah must have been a variant of the same instruction. i didn' see it called out seprately in the 68k programmer manual
07:29:00 <geist> but it probably just listed as a sub variant
07:33:00 <geist> ah yeah i see it now. MULU.L and MULS.L
07:34:00 <geist> 020+
10:24:00 <nomagno> Ended up adding uint32_t requirement to the engine's compatibility diagnostics header. Life's too short
10:27:00 <nomagno> I do have enough space for simulating an F1 crash in a flat-out corner now :P
12:36:00 <yasar11732> Looking for a bootable HDD test program, something like memtest
12:36:00 <yasar11732> any suggestions?
13:09:00 <zid> gparted?
13:18:00 <yasar11732> zid, does gparted have disc scanning feature?
13:31:00 <kazinsal> it's been a decade or so since I've used it but Hiren's BootCD used to have some solid HDD utils
14:32:00 <GeDaMo> ?
14:32:00 <bslsk05> ​ SystemRescue - System Rescue Homepage
19:35:00 <geist> so might be time to start doing something about my crashy server
19:35:00 <geist> it's crashing so much now (about once a day or two) that it's unusuable
19:35:00 <geist> i'm guessing i should start by reseating everything and then seeing if it explodes again
19:37:00 <GeDaMo> I used to have an intermittent crashing problem that was caused by the power connector to the motherboard not having a latch :/
19:39:00 <geist> yah. this is a hard lockup. the sytem just freezes cold, so dunno
19:40:00 <geist> have to reset it
19:44:00 <GeDaMo> Sometimes mine would freeze, other times it would try to reboot but fail somewhere in the BIOS checks making me think it was connected to drives or USB
19:44:00 <GeDaMo> I jiggled a lot of cables before spotting it :P
20:03:00 <mrvn> removing everything and cleaning all contacts could help
20:44:00 <geist> yah, i can't quite tell when it started, either when i put the new cpu in it (2700x -> 3950x) or when i added the 10gbe card
20:44:00 <geist> so probably first thing after cleaning everything and if it continues i'll pull out the card next