February 9, 2017 
< | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | >
February 9, 2017 [00:03:28] *** pa <pa!~pa@unaffiliated/pa> has joined ##vulkan[00:09:14] *** Jackneill_ <Jackneill_!~Jackneill@unaffiliated/jackneill> has quit IRC (Read error: Connection reset by peer)[00:10:35] *** Girafferson <Girafferson!~giraffers@c-50-152-51-82.hsd1.co.comcast.net> has quit IRC ()[00:11:41] *** GMpow2srV <GMpow2srV!~goedelmac@55d41599.access.ecotel.net> has joined ##vulkan[00:22:33] *** Lypho <Lypho!~Lypho@lypho.tv> has quit IRC (Quit: :D)[00:25:48] *** BearishMushroom <BearishMushroom!~BearishMu@90-231-174-194-no159.tbcn.telia.com> has quit IRC (Read error: Connection reset by peer)[00:26:10] *** derhass <derhass!~derhass@ipservice-092-216-126-046.092.216.pools.vodafone-ip.de> has quit IRC (Quit: leaving)[00:36:13] *** ector <ector!~asdf@37.247.11.157> has quit IRC ()[00:38:48] *** kdub <kdub!~kdub@d28-23-138-68.dim.wideopenwest.com> has quit IRC (Quit: goodnight)[00:42:40] *** tarceri <tarceri!~tarceri@CPE-120-144-40-53.lnse5.win.bigpond.net.au> has quit IRC (Quit: Leaving)[00:43:09] *** tarceri <tarceri!~tarceri@CPE-120-144-40-53.lnse5.win.bigpond.net.au> has joined ##vulkan[00:55:46] *** doomlord <doomlord!~textual@host86-148-102-241.range86-148.btcentralplus.com> has quit IRC (Quit: My MacBook Pro has gone to sleep. ZZZzzz…)[00:57:53] *** slidercrank <slidercrank!~slidercra@unaffiliated/slidercrank> has quit IRC (Ping timeout: 252 seconds)[01:02:36] *** Feedz <Feedz!~Feedz@unaffiliated/feedz> has quit IRC (Remote host closed the connection)[01:27:48] *** gamester <gamester!~gamester@31-209-216-75.dsl.dynamic.simnet.is> has joined ##vulkan[01:30:25] <gamester> Reading about Input Attachments, the spec says "Loads from input attachments are unfiltered". What does this mean?[01:31:11] <|3b|> "no interpolation"?[01:32:43] <|3b|> as in you get either value of pixel A or value of adjacent pixel B, but can't get an intermediate value by asking for pixel A.5 (like you could with a linear filtered texture)[01:33:10] *** Karethoth <Karethoth!~karethoth@89.33.6.195> has quit IRC (Ping timeout: 240 seconds)[01:34:36] *** john1 <john1!~john@host86-143-93-17.range86-143.btcentralplus.com> has joined ##vulkan[01:34:37] <gamester> ahh ok[01:35:36] *** Karethoth <Karethoth!~karethoth@89.33.6.195> has joined ##vulkan[01:59:08] *** w-t-h_ <w-t-h_!~chatzilla@72.67.52.110> has quit IRC (Ping timeout: 240 seconds)[02:07:08] *** antilogic_ <antilogic_!~antilogic@85.206.109.17> has quit IRC (Ping timeout: 260 seconds)[02:13:18] *** drasich <drasich!~indefini@124x35x243x164.ap124.ftth.ucom.ne.jp> has joined ##vulkan[02:20:19] *** ratchetfreak <ratchetfreak!~ratchetfr@ptr-82s3g7lne6wzjp8iec8.18120a2.ip6.access.telenet.be> has quit IRC (Ping timeout: 255 seconds)[02:28:41] *** matriarchetype <matriarchetype!~matriarch@d149-67-254-201.try.wideopenwest.com> has quit IRC (Quit: later nerds)[02:31:27] *** ZexaronS <ZexaronS!~zexpc@cpe-194-152-22-45.static.triera.net> has quit IRC (Read error: Connection reset by peer)[02:32:42] *** ZexaronS <ZexaronS!~zexpc@cpe-194-152-22-45.static.triera.net> has joined ##vulkan[02:33:33] *** w-t-h_ <w-t-h_!~chatzilla@72.67.52.110> has joined ##vulkan[03:06:29] *** Kingsquee <Kingsquee!~kingsley@d206-116-125-16.bchsia.telus.net> has joined ##vulkan[03:13:36] *** drasich <drasich!~indefini@124x35x243x164.ap124.ftth.ucom.ne.jp> has quit IRC (Read error: Connection reset by peer)[03:16:01] *** ZexaronS <ZexaronS!~zexpc@cpe-194-152-22-45.static.triera.net> has quit IRC (Quit: Leaving)[03:17:57] *** _28_ria <_28_ria!~igoryonya@opfr028.ru> has quit IRC (Read error: Connection reset by peer)[03:20:10] *** _28_ria <_28_ria!~igoryonya@opfr028.ru> has joined ##vulkan[03:46:20] *** GMpow3srV <GMpow3srV!~goedelmac@55d41599.access.ecotel.net> has joined ##vulkan[03:49:29] *** GMpow2srV <GMpow2srV!~goedelmac@55d41599.access.ecotel.net> has quit IRC (Ping timeout: 252 seconds)[03:57:24] *** GMpow5srV <GMpow5srV!~goedelmac@55d41599.access.ecotel.net> has joined ##vulkan[03:57:31] *** tacchinotacchi_ <tacchinotacchi_!~tacchinot@net-93-144-206-17.cust.dsl.teletu.it> has joined ##vulkan[03:59:56] *** GMpow3srV <GMpow3srV!~goedelmac@55d41599.access.ecotel.net> has quit IRC (Ping timeout: 252 seconds)[03:59:56] *** tacchinotacchi <tacchinotacchi!~tacchinot@net-93-144-206-17.cust.vodafonedsl.it> has quit IRC (Ping timeout: 252 seconds)[03:59:56] *** Biolunar_ <Biolunar_!Biolunar@dslb-092-073-188-133.092.073.pools.vodafone-ip.de> has quit IRC (Ping timeout: 252 seconds)[03:59:57] *** Guest74768 <Guest74768!~jeremy@50-253-122-174-static.hfc.comcastbusiness.net> has quit IRC (Ping timeout: 252 seconds)[04:00:21] *** Guest74768 <Guest74768!~jeremy@50-253-122-174-static.hfc.comcastbusiness.net> has joined ##vulkan[04:06:21] *** Biolunar <Biolunar!Biolunar@dslb-092-073-188-133.092.073.pools.vodafone-ip.de> has joined ##vulkan[04:16:20] *** TechnoCrunch <TechnoCrunch!~TechnoCru@101.100.137.146> has joined ##vulkan[04:21:44] *** GMpow2_2 <GMpow2_2!~5@55d41add.access.ecotel.net> has joined ##vulkan[04:21:58] *** GMpow3srV <GMpow3srV!~goedelmac@55d41add.access.ecotel.net> has joined ##vulkan[04:25:11] *** GMpow2 <GMpow2!~5@55d41599.access.ecotel.net> has quit IRC (Ping timeout: 240 seconds)[04:25:31] *** GMpow5srV <GMpow5srV!~goedelmac@55d41599.access.ecotel.net> has quit IRC (Ping timeout: 240 seconds)[04:29:05] *** GMpow3srV <GMpow3srV!~goedelmac@55d41add.access.ecotel.net> has quit IRC (Ping timeout: 252 seconds)[04:29:50] *** GMpow2_2 <GMpow2_2!~5@55d41add.access.ecotel.net> has quit IRC (Ping timeout: 256 seconds)[04:29:55] *** lpghatguy <lpghatguy!~LPGhatguy@63.153.99.104> has joined ##vulkan[04:49:20] *** TechCrunch <TechCrunch!~TechnoCru@101.100.137.146> has joined ##vulkan[04:52:05] *** TechnoCrunch <TechnoCrunch!~TechnoCru@101.100.137.146> has quit IRC (Ping timeout: 240 seconds)[04:53:10] *** TechCrunch is now known as TechnoCrunch[05:38:39] *** _28_ria <_28_ria!~igoryonya@opfr028.ru> has quit IRC (Read error: Connection reset by peer)[05:39:48] *** doras <doras!~doras@77.125.128.185> has joined ##vulkan[05:39:56] *** _28_ria <_28_ria!~igoryonya@opfr028.ru> has joined ##vulkan[05:45:27] <doras> A bit late, but thanks for the information, Ralith and fazias.[05:51:02] <doras> AMD's host-coherent device-local heap is definitely an interesting one. I guess it's best used for data that is frequently-changed on the CPU side.[05:55:50] <doras> And is obviously light in size, so no images or models in bulk. Perhaps instance data and special buffers/images.[06:07:49] *** tambre <tambre!~tambre@650e-f29b-eeed-bc53-0e80-8a31-07d0-2001.dyn.estpak.ee> has joined ##vulkan[06:09:50] <doras> I'm still trying to figure out what HOST_CACHED_BIT means, and why NVIDIA doesn't offer it.[06:12:07] <Ralith> it means it's slower to access, presumably because all accesses go over the PCIe bus and the gpu's MMU[06:12:27] <doras> Sorry, I misread. They do offer it. My bad.[06:13:06] <doras> Yeah, which surprised me, because it was on a system memory heap.[06:13:13] <doras> But, again, I misread :)[06:52:44] *** echelog-3 <echelog-3!~echelog-3@108.61.103.42> has joined ##vulkan[07:02:17] *** LanDi <LanDi!~landi@191.33.22.177> has joined ##vulkan[07:18:55] *** echelog-3 <echelog-3!~echelog-3@108.61.103.42> has joined ##vulkan[07:20:06] *** nitroxis <nitroxis!n@6.nxs.re> has joined ##vulkan[07:22:07] *** minsky <minsky!minsky@231-242-183-107.Static.PanicBNC.us> has quit IRC (Ping timeout: 255 seconds)[07:27:56] *** nsf <nsf!~nsf@jiss.convex.ru> has joined ##vulkan[07:28:04] *** salek__ <salek__!~salek@91-159-91-43.elisa-laajakaista.fi> has joined ##vulkan[07:30:01] *** KaetemiC__ <KaetemiC__!sid109773@gateway/web/irccloud.com/x-txnylusyyivuqtff> has joined ##vulkan[07:32:32] *** ville_ <ville_!~ville@37-136-69-165.rev.dnainternet.fi> has joined ##vulkan[07:32:45] *** Karethoth_ <Karethoth_!~karethoth@89.33.6.195> has joined ##vulkan[07:33:08] *** KaetemiC_ <KaetemiC_!sid109773@gateway/web/irccloud.com/x-gpxukctbawnlrdyu> has quit IRC (Ping timeout: 240 seconds)[07:33:14] *** Karethoth <Karethoth!~karethoth@89.33.6.195> has quit IRC (Ping timeout: 240 seconds)[07:33:15] *** ville <ville!~ville@37-136-69-165.rev.dnainternet.fi> has quit IRC (Ping timeout: 240 seconds)[07:33:18] *** salek_ <salek_!~salek@91-159-91-43.elisa-laajakaista.fi> has quit IRC (Ping timeout: 240 seconds)[07:33:22] *** KaetemiC__ is now known as KaetemiC_[07:34:34] *** w-t-h_ <w-t-h_!~chatzilla@72.67.52.110> has quit IRC (Ping timeout: 264 seconds)[07:51:13] *** ShaneC1 <ShaneC1!~shane@cpe-104-175-221-156.socal.res.rr.com> has joined ##vulkan[07:52:26] *** ShaneC <ShaneC!~shane@cpe-104-175-221-156.socal.res.rr.com> has quit IRC (Ping timeout: 258 seconds)[07:53:09] *** nitroxis <nitroxis!n@6.nxs.re> has quit IRC (Read error: Connection reset by peer)[07:54:15] *** nitroxis <nitroxis!n@6.nxs.re> has joined ##vulkan[07:55:49] *** LanDi <LanDi!~landi@191.33.22.177> has quit IRC (Remote host closed the connection)[07:58:39] *** john1 <john1!~john@host86-143-93-17.range86-143.btcentralplus.com> has quit IRC (Ping timeout: 245 seconds)[08:17:53] *** ShaneC1 <ShaneC1!~shane@cpe-104-175-221-156.socal.res.rr.com> has quit IRC (Ping timeout: 252 seconds)[08:19:01] *** ShaneC <ShaneC!~shane@cpe-104-175-221-156.socal.res.rr.com> has joined ##vulkan[08:22:17] *** ShaneC1 <ShaneC1!~shane@cpe-104-175-221-156.socal.res.rr.com> has joined ##vulkan[08:23:29] *** ShaneC <ShaneC!~shane@cpe-104-175-221-156.socal.res.rr.com> has quit IRC (Ping timeout: 258 seconds)[08:24:55] *** elect_ <elect_!~elect@business-092-079-135-041.static.arcor-ip.net> has joined ##vulkan[08:33:53] *** ville_ is now known as ville[08:37:35] *** funfunctor <funfunctor!~edward@1.129.96.139> has quit IRC (Ping timeout: 240 seconds)[08:45:40] *** doomlord <doomlord!~textual@host86-148-102-241.range86-148.btcentralplus.com> has joined ##vulkan[08:49:51] *** funfunctor <funfunctor!~edward@hs1.gbsw.oxynet.net.au> has joined ##vulkan[08:50:19] *** _28_ria <_28_ria!~igoryonya@opfr028.ru> has quit IRC (Ping timeout: 258 seconds)[08:52:14] *** doras <doras!~doras@77.125.128.185> has quit IRC (Remote host closed the connection)[08:53:31] *** minsky <minsky!minsky@gateway/shell/panicbnc/x-tdelylhssjtfslmd> has joined ##vulkan[09:08:34] *** lpghatguy <lpghatguy!~LPGhatguy@63.153.99.104> has quit IRC (Read error: Connection reset by peer)[09:14:07] *** _28_ria <_28_ria!~igoryonya@opfr028.ru> has joined ##vulkan[09:16:19] *** funfunctor1 <funfunctor1!~edward@1.129.96.139> has joined ##vulkan[09:19:50] *** funfunctor <funfunctor!~edward@hs1.gbsw.oxynet.net.au> has quit IRC (Ping timeout: 258 seconds)[09:25:03] *** funfunctor1 is now known as funfunctor[09:32:35] *** salek__ <salek__!~salek@91-159-91-43.elisa-laajakaista.fi> has quit IRC (Ping timeout: 240 seconds)[09:36:13] *** Kingsquee <Kingsquee!~kingsley@d206-116-125-16.bchsia.telus.net> has quit IRC (Ping timeout: 255 seconds)[09:50:40] *** realitix <realitix!~realitix@92.103.166.6> has joined ##vulkan[09:57:35] *** Kingsquee <Kingsquee!~kingsley@d206-116-125-16.bchsia.telus.net> has joined ##vulkan[10:02:53] *** ratchetfreak <ratchetfreak!c18021f8@gateway/web/freenode/ip.193.128.33.248> has joined ##vulkan[10:06:57] *** Salek <Salek!~salek@91-159-91-43.elisa-laajakaista.fi> has joined ##vulkan[10:14:14] *** ravior <ravior!~crapitea@89.121.200.106> has joined ##vulkan[10:14:48] *** n4yru <n4yru!~n4yru@181.red-88-26-129.staticip.rima-tde.net> has joined ##vulkan[10:21:44] *** funfunctor <funfunctor!~edward@1.129.96.139> has quit IRC (Ping timeout: 256 seconds)[10:48:22] *** n4yru <n4yru!~n4yru@181.red-88-26-129.staticip.rima-tde.net> has quit IRC (Remote host closed the connection)[10:51:31] *** n4yru <n4yru!~n4yru@181.red-88-26-129.staticip.rima-tde.net> has joined ##vulkan[10:53:00] *** bjz <bjz!~bjz@pa49-184-163-9.pa.vic.optusnet.com.au> has joined ##vulkan[11:04:42] *** funfunctor <funfunctor!~edward@1.129.96.139> has joined ##vulkan[11:08:42] *** bjz <bjz!~bjz@pa49-184-163-9.pa.vic.optusnet.com.au> has quit IRC (Quit: My MacBook Pro has gone to sleep. ZZZzzz…)[11:13:43] *** TechnoCrunch <TechnoCrunch!~TechnoCru@101.100.137.146> has quit IRC (Read error: Connection reset by peer)[11:14:16] *** Matthijs <Matthijs!~quassel@unaffiliated/matthijs> has joined ##vulkan[11:16:12] *** slidercrank <slidercrank!~slidercra@unaffiliated/slidercrank> has joined ##vulkan[11:22:49] *** Fats <Fats!~Fats@109.201.138.226> has quit IRC (Remote host closed the connection)[11:23:56] *** Fats <Fats!~Fats@46.166.137.194> has joined ##vulkan[11:27:42] *** bjz <bjz!~bjz@14-201-215-168.tpgi.com.au> has joined ##vulkan[11:35:10] *** Matthijs <Matthijs!~quassel@unaffiliated/matthijs> has quit IRC (Ping timeout: 240 seconds)[11:41:40] *** ville <ville!~ville@37-136-69-165.rev.dnainternet.fi> has quit IRC (Ping timeout: 258 seconds)[11:42:13] *** kuldeep <kuldeep!~kuldeepdh@unaffiliated/kuldeepdhaka> has quit IRC (Remote host closed the connection)[11:45:08] *** Kingsqueee <Kingsqueee!~kingsley@d206-116-125-16.bchsia.telus.net> has joined ##vulkan[11:46:42] *** SHC <SHC!~quassel@45.251.245.117> has joined ##vulkan[11:46:52] *** ravior <ravior!~crapitea@89.121.200.106> has quit IRC (Ping timeout: 240 seconds)[11:46:52] *** Kingsquee <Kingsquee!~kingsley@d206-116-125-16.bchsia.telus.net> has quit IRC (Ping timeout: 240 seconds)[11:50:10] *** Matthijs <Matthijs!~quassel@unaffiliated/matthijs> has joined ##vulkan[11:53:35] *** ville <ville!~ville@87-95-24-10.bb.dnainternet.fi> has joined ##vulkan[11:59:28] *** kuldeep <kuldeep!~kuldeepdh@unaffiliated/kuldeepdhaka> has joined ##vulkan[12:00:56] *** ravior <ravior!~crapitea@89.121.200.106> has joined ##vulkan[12:13:12] *** slidercrank <slidercrank!~slidercra@unaffiliated/slidercrank> has quit IRC (Ping timeout: 240 seconds)[12:15:37] *** n4yru <n4yru!~n4yru@181.red-88-26-129.staticip.rima-tde.net> has quit IRC (Remote host closed the connection)[12:18:06] *** n4yru <n4yru!~n4yru@181.red-88-26-129.staticip.rima-tde.net> has joined ##vulkan[12:21:40] *** Rangar <Rangar!~Dave@101.98.215.224> has quit IRC (Ping timeout: 240 seconds)[12:25:30] *** n4yru <n4yru!~n4yru@181.red-88-26-129.staticip.rima-tde.net> has quit IRC (Remote host closed the connection)[12:26:05] *** n4yru <n4yru!~n4yru@38.88.216.101> has joined ##vulkan[12:27:05] *** robjperez <robjperez!~n4yru@181.red-88-26-129.staticip.rima-tde.net> has joined ##vulkan[12:27:47] *** z3r0_ <z3r0_!~z3r0@197.234.219.78> has joined ##vulkan[12:27:59] *** z3r0_ <z3r0_!~z3r0@197.234.219.78> has quit IRC (Remote host closed the connection)[12:30:53] *** n4yru <n4yru!~n4yru@38.88.216.101> has quit IRC (Ping timeout: 252 seconds)[12:31:40] <neure> how does gles lowp / mediump / highp map to bit width?[12:31:54] <neure> in spirv[12:32:28] <neure> hmm[12:32:29] <neure> "RelaxedPrecision"[12:34:49] *** robjperez is now known as n4yru[12:35:02] <lieff[m]> there no direct mapping, it's hardware dependent[12:35:42] <neure> but does compiler produce anything else than OpTypeFloat 32 ?[12:36:16] <neure> oh[12:36:35] <neure> OpTypeFloat 16 only exists to access buffers that have 16-bit floats[12:36:37] <ratchetfreak> it may support AMD's halffloat extension[12:36:45] *** realitix <realitix!~realitix@92.103.166.6> has quit IRC (Excess Flood)[12:36:54] <neure> then there is 64 bit float[12:37:07] *** realitix <realitix!~realitix@92.103.166.6> has joined ##vulkan[12:37:20] <neure> lowp simply is not supported?[12:37:46] <neure> "The RelaxedPrecision Decoration allows 32-bit integer and 32-bit floating-point operations to execute with a relaxed precision of somewhere between 16 and 32 bits."[12:41:39] <sharpneli> Doesn't seem to be supported. And even on GLSL it was always just a storage format[12:41:50] <sharpneli> I don't know if there has been any HW that has lopw units internally[12:42:03] <sharpneli> In last 10 years that is[12:42:45] <neure> i thought img had lowp for fragment shaders[12:44:19] <sharpneli> For storage at least. But hard to say if they actually had lowp ALU units[12:44:46] *** jkilpatr <jkilpatr!~jkilpatr@2602:30a:c7da:b600:1202:b5ff:fe3b:d41d> has quit IRC (Ping timeout: 255 seconds)[12:44:54] <sharpneli> You can already get most bang for the buck by having lowp registers and then just stuffing that to half float ALU[12:45:10] *** john1 <john1!~john@host86-143-93-17.range86-143.btcentralplus.com> has joined ##vulkan[12:50:06] <neure> sure[12:50:21] <neure> but spirv doesnt seem to support that[12:50:49] *** Kingsqueee <Kingsqueee!~kingsley@d206-116-125-16.bchsia.telus.net> has quit IRC (Quit: https://i.imgur.com/qicT3GK.gif)[12:51:09] <neure> just an observation, no, i dont actually care about lowp[12:51:15] <neure> even mediump is horrible[12:51:20] <sharpneli> True. Maybe it was decided that it's unnecessary feature[12:51:28] <sharpneli> Because you can use lowp as storage format already[12:51:59] <neure> how?[12:53:02] <sharpneli> Usual RGBA8 image is lowp storage format basically :D[12:53:12] <neure> well lowp is a bit more than that[12:53:21] <neure> http://www.codersnotes.com/notes/wrangling-halfs/ look at that table[12:53:25] <neure> shows how horrible mediump is[12:53:39] <neure> assuming mediump is anything near half float[12:53:42] <sharpneli> That's mediump[12:53:55] <sharpneli> Mediump basically is half float[12:53:57] *** slidercrank <slidercrank!~slidercra@unaffiliated/slidercrank> has joined ##vulkan[12:54:00] <neure> yeah, and even mediump sucks for a lot[12:54:12] <sharpneli> Sure. Especially the normalize function breaks with it as it can overflow internally[12:54:19] <neure> value greatet than 10 have not much precision left[12:54:23] <sharpneli> But as a storage format for things like vertices and as render targets for HDR it's ok[12:55:41] <exDM69> fp16 gives 1mm accuracy at 1 meter range[12:55:52] <exDM69> it's quite alright for vertex data[12:56:07] <exDM69> and color buffer[12:56:15] <neure> for those it is fine[12:56:25] <sharpneli> neure: Lowp actually is exactly RGBA8. Or pretty much it. As it requires 2^-8 absolute precision[12:56:32] <sharpneli> Instead of relative precision required by mediump and highp[12:56:38] <sharpneli> Just looking at glsl spec[12:57:09] <neure> the other day i checked out someone using a float time in seconds uniform in fragment shader[12:57:18] <sharpneli> Nooo don't do that[12:57:18] <neure> that didnt work too well :P[12:57:21] <neure> indeed[12:57:40] <neure> i heard unity has time in seconds[12:57:43] <neure> how does that work[12:58:13] <neure> _Time float4 Time (t/20, t, t*2, t*3), use to animate things inside the shaders.[12:58:17] <neure> ah cool[12:58:19] <ratchetfreak> microsecondTime/1000000.0f[12:58:30] <neure> they put t/20, t, t*2 and t*3 into vec4[12:58:38] <neure> works.. a bit longer[12:58:44] <neure> still probably sucks in fragment shader[12:58:51] <sharpneli> Depends on when they overflow[12:59:11] <ratchetfreak> you'll want to rebase to 0 very often[12:59:42] *** jkilpatr <jkilpatr!jkilpatr@nat/redhat/x-idatajtwaqxtdpzy> has joined ##vulkan[12:59:43] <sharpneli> Not too often is required with 32bit float. Minute or so was enough iirc[13:00:04] <neure> i thought 32 bit float (highp in vertex shader for example) is good for 24 hours[13:00:31] <neure> 7.81 milliseconds[13:00:40] <neure> precision at ~1 day[13:00:54] <neure> according to https://randomascii.wordpress.com/2012/02/13/dont-store-that-in-a-float/[13:01:03] <neure> maybe that's not enough precision for something[13:01:06] <sharpneli> Depends on what your precision requirement is[13:05:16] *** bjz_ <bjz_!~bjz@104.222.140.17> has joined ##vulkan[13:07:00] *** bjz <bjz!~bjz@14-201-215-168.tpgi.com.au> has quit IRC (Ping timeout: 240 seconds)[13:07:07] *** kdub <kdub!~kdub@d28-23-138-68.dim.wideopenwest.com> has joined ##vulkan[13:21:47] *** n4yru <n4yru!~n4yru@181.red-88-26-129.staticip.rima-tde.net> has quit IRC (Remote host closed the connection)[13:57:18] *** mithsv <mithsv!~mithsv@217.149.135.12> has joined ##vulkan[13:57:25] *** mithsv <mithsv!~mithsv@217.149.135.12> has quit IRC (Client Quit)[13:59:09] <neure> does glslangvalidator support preprocessor defines?[13:59:26] <neure> i cant find any support for defines[14:01:15] *** MatthijsH <MatthijsH!~quassel@unaffiliated/matthijs> has joined ##vulkan[14:04:52] *** Matthijs <Matthijs!~quassel@unaffiliated/matthijs> has quit IRC (Ping timeout: 240 seconds)[14:11:24] <sharpneli> If it doesn't the run it trough c preprocessor[14:13:51] <baldurk> it does support defines, but you have to define them yourself there's no top-level interface to add them from outside the source IIRC[14:15:53] <sharpneli> We do it so that we just pass it trough a normal C preprocessor. Allows includes etc too.[14:19:23] *** Ipe <Ipe!~I@a91-156-244-1.elisa-laajakaista.fi> has joined ##vulkan[14:21:46] *** n4yru <n4yru!~n4yru@181.red-88-26-129.staticip.rima-tde.net> has joined ##vulkan[14:30:51] *** nitroxis <nitroxis!n@6.nxs.re> has quit IRC (Read error: Connection reset by peer)[14:31:53] *** nitroxis <nitroxis!n@6.nxs.re> has joined ##vulkan[14:38:25] *** bjz <bjz!~bjz@14-201-215-168.tpgi.com.au> has joined ##vulkan[14:39:07] *** n4yru <n4yru!~n4yru@181.red-88-26-129.staticip.rima-tde.net> has quit IRC (Remote host closed the connection)[14:39:41] *** n4yru <n4yru!~n4yru@38.88.216.101> has joined ##vulkan[14:40:07] *** bjz_ <bjz_!~bjz@104.222.140.17> has quit IRC (Ping timeout: 255 seconds)[14:58:36] *** MatthijsH is now known as Matthijs[15:07:16] *** bjz <bjz!~bjz@14-201-215-168.tpgi.com.au> has quit IRC (Quit: My MacBook Pro has gone to sleep. ZZZzzz…)[15:08:16] *** nsf <nsf!~nsf@jiss.convex.ru> has quit IRC (Quit: WeeChat 1.7)[15:09:16] *** funfunctor <funfunctor!~edward@1.129.96.139> has quit IRC (Quit: Leaving.)[15:24:01] *** robjperez <robjperez!~n4yru@181.red-88-26-129.staticip.rima-tde.net> has joined ##vulkan[15:27:04] *** n4yru <n4yru!~n4yru@38.88.216.101> has quit IRC (Ping timeout: 258 seconds)[15:38:59] *** Feedz <Feedz!~Feedz@unaffiliated/feedz> has joined ##vulkan[15:53:47] *** zeduckmaster <zeduckmaster!~zeduckmas@85.106.1.181> has quit IRC (Quit: Leaving)[15:57:24] *** tacchinotacchi_ <tacchinotacchi_!~tacchinot@net-93-144-206-17.cust.dsl.teletu.it> has quit IRC (Remote host closed the connection)[16:01:12] *** robjperez is now known as n4yru[16:24:22] *** ville <ville!~ville@87-95-24-10.bb.dnainternet.fi> has quit IRC (Ping timeout: 264 seconds)[16:26:37] *** nsf <nsf!~nsf@jiss.convex.ru> has joined ##vulkan[16:33:09] <gamester> For texel buffers, is texelFetch used?[16:40:34] *** ImQ009 <ImQ009!~ImQ009@unaffiliated/imq009> has joined ##vulkan[16:42:05] <gamester> or imageLoad?[17:10:18] *** ravior <ravior!~crapitea@89.121.200.106> has quit IRC (Quit: Konversation terminated!)[17:11:24] *** n4yru <n4yru!~n4yru@181.red-88-26-129.staticip.rima-tde.net> has quit IRC (Remote host closed the connection)[17:19:26] *** haagch <haagch!~haagch@mx.frickel.club> has quit IRC (Ping timeout: 255 seconds)[17:20:30] *** notadeveloper <notadeveloper!~notadevel@2602:306:bd2a:a160:424a:a59c:b306:ff3e> has quit IRC (Quit: Leaving)[17:21:46] *** notadeveloper <notadeveloper!~notadevel@2602:306:bd2a:a160:424a:a59c:b306:ff3e> has joined ##vulkan[17:23:31] *** haagch <haagch!~haagch@mx.frickel.club> has joined ##vulkan[17:24:48] <tomaka> ONE_TIME_SUBMIT command buffers are the norm rather than the exception, right?[17:25:17] <tomaka> I imagine most programs have a few non-ONE-TIME-SUBMIT CBs, but most of the logic is in command buffers only submitted once?[17:26:27] <ratchetfreak> one time submit implies that you'll rerecord every frame[17:26:55] <ratchetfreak> I believe that most people start that way[17:28:15] <ratchetfreak> it takes a bit of finagling to be able to reuse command buffers frame to frame[17:30:24] <tomaka> I was wondering if "experts" indeed reuse command buffers, or if it's impossible anyway[17:30:44] <ratchetfreak> it's possible[17:30:50] <tomaka> because simply streaming data from the CPU to the GPU most likely requires rerecording a command buffer every frame for example[17:31:11] <ratchetfreak> for example if you have a round robin staging buffer and a set UBO location for each swapchain image[17:31:26] <ratchetfreak> then you can use a separate copy buffer to get the uniforms in place[17:31:42] <tomaka> hmm[17:31:45] <ratchetfreak> but the big render command buffer can stay the same[17:32:00] <sharpneli> It depends on what you do[17:32:12] <baldurk> AFAIK reusing command buffers is uncommon in shipping programs, because it's typically going scene traversal, culling, etc on CPU. There's not much of the frame that you know is fixed up front[17:32:14] <sharpneli> Things like transparent objects where order changes need re-recording anyway[17:32:38] <tomaka> I'm not doing anything in fact, still trying to wrap my head around a design that handles command buffers and inter-command buffers synchronization with as few overhead as possible[17:32:56] <ratchetfreak> unless you sort the indices of the transparent tris and copy the sorted index buffer around[17:33:00] <sharpneli> baldurk: A neat trick for culling is to use drawindirect with instance count 0 or 1 depending if it passed culling or not with the data in GPU accessible memory[17:34:37] <tomaka> Rust has this nice division between closures that can only be executed once, closures that can be executed multiple times but not in parallel, and closures that can be executed in parallel[17:34:56] <tomaka> vulkan's command buffers have exactly the same properties, so I'm trying to draw a parallel[17:35:06] <baldurk> sharpneli: that's good for fine-grained culling but would you want to submit your entire scene all the time and then do that?[17:35:39] <sharpneli> baldurk: You can easily put things into chunks and use secondary command buffers.[17:35:51] <tomaka> just closures on the CPU have access to the variables of their environment but there are restrictions depending on how they are called, command buffers could access resources with the same kind of restrictions[17:35:54] <tomaka> just like*[17:36:38] <ratchetfreak> sharpneli: I'd be worried about the latency of the query though[17:36:42] <baldurk> sure, I just wonder if that's worthwhile since recording is designed to be cheap and parallelisable[17:37:28] <sharpneli> ratchetfreak: What latency?[17:37:41] <baldurk> unless you're doing full-GPU rendering redlynx style, but then your command buffers are probably tiny and who cares if you re-record :)[17:38:00] <sharpneli> Recording is not necessarily cheap, it's just parallelisable[17:38:23] *** ravior <ravior!~crapitea@5-13-152-135.residential.rdsnet.ro> has joined ##vulkan[17:38:25] <ratchetfreak> occlusion query latency[17:38:27] <baldurk> fair point, not cheap in an absolute sense, but it should be cheaper than previous APIs at least[17:38:35] <sharpneli> ratchetfreak: I'm talking about normal CPU culling[17:39:41] <sharpneli> ratchetfreak: Instead of re-recording the command buffer just use vkCmdDrawIndexedIndirect with buffer that contains instance counts.[17:40:12] <sharpneli> So for something that got culled you put instance count as 0. The draw command goes in but does nothing. Saves the vertex workload which is the whole point of culling[17:41:33] <ratchetfreak> I was thinking about an occlusion pass with a separate query per chunk and then the results get copied into the indirect draw buffer using a compute shader[17:42:03] <sharpneli> That's even more advanced, but it's not really needed when considering re-recording command buffer or not[17:42:56] <sharpneli> Actually we use that method right now. For the boring opaque objects the only thing the engine has to do is update the uniforms like camera and model matrices in memory, update the culling buffer and voila. Just resubmit the old buffer.[17:43:15] <ratchetfreak> and you are then drawing the same scene 3 times: once for depth, once for occlusion and once for actual rendering[17:43:42] <sharpneli> Doom does that kind of thin. They use the occlusion query for the next frame for more precise culling[17:44:34] <sharpneli> http://www.adriancourreges.com/blog/2016/09/09/doom-2016-graphics-study/ Check this. They render depth, then occlusion query pass and then forward+[17:45:19] <ratchetfreak> for transparent objects you can sort the indirect draw buffer.[17:45:36] <ratchetfreak> if they are all in the same set of vbos at least[17:46:27] *** sla_ro|master <sla_ro|master!~sla.ro@95.76.45.217> has joined ##vulkan[17:46:36] <sharpneli> Doesn't work like that. Because you actually have to issue the draw commands back to front in api order[17:46:41] <sharpneli> You'd have to sort the index buffer[17:46:48] <sharpneli> And that will get so slow[17:48:12] *** haagch_ <haagch_!~haagch@mx.frickel.club> has joined ##vulkan[17:48:13] *** haagch <haagch!~haagch@mx.frickel.club> has quit IRC (Ping timeout: 255 seconds)[17:48:34] <gamester> I'm wondering: I know Vulkan is more efficient for the CPU and mobile GPUs, but what about desktop GPUs? Is Vulkan sometimes slightly slower? I don't know why that would be, but sometimes benchmarks have shown that? Idk.[17:49:07] <ratchetfreak> but api order for indirect draw is the order of the indirect draw commands[17:49:46] <ratchetfreak> split the object up in non overlapping convex patches and it should work[17:50:02] *** haagch_ <haagch_!~haagch@mx.frickel.club> has quit IRC (Client Quit)[17:50:54] <sharpneli> ratchetfreak: You cannot change the draw order by moving just ints around[17:51:13] <sharpneli> ratchetfreak: Just like if you swap the model matrices of 2 objects in memory the object will not flip, their positions will flip.[17:52:18] <sharpneli> Ah now I get it what you mean[17:52:23] <sharpneli> By using the optional multidrawindirect feature[17:52:45] *** haagch <haagch!~haagch@mx.frickel.club> has joined ##vulkan[17:53:00] <tomaka> gamester, a game well-optimized for OpenGL may perform more poorly if it writes crappy Vulkan code[17:53:09] <ratchetfreak> the VkDrawIndexedIndirectCommand struct includes the first index and the count[17:53:24] <ratchetfreak> swap 2 of them and you swap the order they are executed in[17:53:43] <sharpneli> Ah true. So yeah you can do that if they indeed are statically baked into the same vbo[17:54:10] <sharpneli> Could be useful for things like transparent particles[17:54:15] <tomaka> gamester, "crappy" is exaggerated, but as an example OpenGL would perform some operations in a background thread, and if your game does all its rendering in a single thread, when switching to Vulkan it'll lose parallelism[17:54:44] <ratchetfreak> also some of the optimizations the opengl driver does behind the application's back could be non obvious[17:55:02] <ratchetfreak> like sorting the vertex and index buffer for better cache locality[17:55:45] <gamester> tomaka: yeah I'm aware of crappy ports, like The Talos Principle (at least when they initially ported it).[17:58:03] <gamester> ratchetfreak: ahh, interesting[18:01:11] <gamester> A lot of doom players complained about higher frame time variance on Vulkan. You'd think it'd be the opposite, and probably will in the future. What could the problem be?[18:01:29] <gamester> Here is one such thread: http://www.neogaf.com/forum/showthread.php?t=1249353[18:01:43] <ratchetfreak> there is no real way to wait on vsync with vulkan[18:02:09] <ratchetfreak> unless you use the fifo swapchain but then you are lagging behind by a few frames[18:03:15] *** nsf <nsf!~nsf@jiss.convex.ru> has quit IRC (Quit: WeeChat 1.7)[18:04:40] <gamester> ratchetfreak: will this limitation be addressed in the future?[18:05:46] <ratchetfreak> most likely[18:07:58] <sharpneli> Shouldn't fifo with a short buffer pretty much address it perfectly?[18:08:59] *** n4yru <n4yru!~n4yru@181.red-88-26-129.staticip.rima-tde.net> has joined ##vulkan[18:10:54] *** n4yru <n4yru!~n4yru@181.red-88-26-129.staticip.rima-tde.net> has quit IRC (Client Quit)[18:19:27] *** clintar <clintar!~clintar@68.69.164.130> has quit IRC (Remote host closed the connection)[18:19:49] *** realitix <realitix!~realitix@92.103.166.6> has quit IRC (Quit: Leaving)[18:21:35] *** GMpow2 <GMpow2!~5@55d41add.access.ecotel.net> has joined ##vulkan[18:21:49] *** ratchetfreak <ratchetfreak!c18021f8@gateway/web/freenode/ip.193.128.33.248> has quit IRC (Ping timeout: 260 seconds)[18:25:42] *** ZexaronS <ZexaronS!~zexpc@cpe-194-152-22-45.static.triera.net> has joined ##vulkan[18:32:46] *** Matthijs <Matthijs!~quassel@unaffiliated/matthijs> has quit IRC (Ping timeout: 264 seconds)[18:33:35] *** clintar <clintar!~clintar@68.69.164.130> has joined ##vulkan[18:36:46] *** slidercrank <slidercrank!~slidercra@unaffiliated/slidercrank> has quit IRC (Quit: Leaving)[18:40:14] *** BearishMushroom <BearishMushroom!~BearishMu@90-231-174-194-no159.tbcn.telia.com> has joined ##vulkan[18:58:46] *** ratchetfreak <ratchetfreak!~ratchetfr@ptr-82s3g7lne6wzjp8iec8.18120a2.ip6.access.telenet.be> has joined ##vulkan[19:00:26] <fazias> sad part about drawcalls that don't render anything, still cost all the cacheflushes it needs to operate and the cost of launching it... and do nothing and move on to next command.[19:01:05] *** ShadowIce <ShadowIce!~pyoro@unaffiliated/shadowice-x841044> has joined ##vulkan[19:02:57] *** SHC <SHC!~quassel@45.251.245.117> has quit IRC (Remote host closed the connection)[19:07:35] *** nsf <nsf!~nsf@jiss.convex.ru> has joined ##vulkan[19:13:08] *** Ipe <Ipe!~I@a91-156-244-1.elisa-laajakaista.fi> has quit IRC (Ping timeout: 240 seconds)[19:13:43] *** Ipe <Ipe!~I@a91-156-244-1.elisa-laajakaista.fi> has joined ##vulkan[19:15:37] *** ShadowIce <ShadowIce!~pyoro@unaffiliated/shadowice-x841044> has quit IRC (Quit: Leaving)[19:28:10] *** jkilpatr_ <jkilpatr_!jkilpatr@nat/redhat/x-isyejbudkvnyknxy> has joined ##vulkan[19:28:52] *** ShadowIce <ShadowIce!~pyoro@unaffiliated/shadowice-x841044> has joined ##vulkan[19:31:11] *** jkilpatr <jkilpatr!jkilpatr@nat/redhat/x-idatajtwaqxtdpzy> has quit IRC (Ping timeout: 276 seconds)[19:31:24] *** w-t-h_ <w-t-h_!~chatzilla@72.67.52.110> has joined ##vulkan[19:37:21] *** ector <ector!~asdf@37-247-11-157.customers.ownit.se> has joined ##vulkan[19:49:44] *** derhass <derhass!~derhass@ipservice-092-216-126-046.092.216.pools.vodafone-ip.de> has joined ##vulkan[19:56:22] *** slidercrank <slidercrank!~slidercra@unaffiliated/slidercrank> has joined ##vulkan[20:18:18] *** CaptainLex <CaptainLex!~CaptainLe@67-220-31-207.fttp.usinternet.com> has joined ##vulkan[20:44:47] *** CaptainLex <CaptainLex!~CaptainLe@67-220-31-207.fttp.usinternet.com> has quit IRC (Ping timeout: 252 seconds)[20:49:59] *** CaptainLex <CaptainLex!~CaptainLe@67-220-31-207.fttp.usinternet.com> has joined ##vulkan[21:02:21] *** Guest74768 <Guest74768!~jeremy@50-253-122-174-static.hfc.comcastbusiness.net> has quit IRC (Remote host closed the connection)[21:03:38] *** tambre <tambre!~tambre@650e-f29b-eeed-bc53-0e80-8a31-07d0-2001.dyn.estpak.ee> has quit IRC (Ping timeout: 258 seconds)[21:05:44] *** jeremy <jeremy!~jeremy@50-253-122-174-static.hfc.comcastbusiness.net> has joined ##vulkan[21:06:07] *** jeremy is now known as Guest65426[21:12:17] *** CaptainLex <CaptainLex!~CaptainLe@67-220-31-207.fttp.usinternet.com> has quit IRC (Ping timeout: 252 seconds)[21:18:25] *** TechnoCrunch <TechnoCrunch!~TechnoCru@101.100.137.146> has joined ##vulkan[21:18:31] *** jkilpatr_ <jkilpatr_!jkilpatr@nat/redhat/x-isyejbudkvnyknxy> has quit IRC (Quit: Leaving)[21:19:12] *** jkilpatr <jkilpatr!jkilpatr@nat/redhat/x-bahrbuywjydywynr> has joined ##vulkan[21:20:50] *** Rangar <Rangar!~Dave@101.98.215.224> has joined ##vulkan[21:25:08] *** toor_ <toor_!~toor@76.ip-51-254-204.eu> has quit IRC (Ping timeout: 240 seconds)[21:32:14] *** lpghatguy <lpghatguy!~LPGhatguy@63.153.99.104> has joined ##vulkan[21:51:58] *** [xyzzy] <[xyzzy]!~xyzzy@2600:8803:e400:d600:fcbd:e14b:d2aa:5daa> has quit IRC (Ping timeout: 255 seconds)[21:52:02] *** _28_ria <_28_ria!~igoryonya@opfr028.ru> has quit IRC (Read error: Connection reset by peer)[21:54:36] *** _28_ria <_28_ria!~igoryonya@opfr028.ru> has joined ##vulkan[21:57:40] *** ShaneC1 <ShaneC1!~shane@cpe-104-175-221-156.socal.res.rr.com> has quit IRC (Ping timeout: 240 seconds)[22:00:16] *** ShaneC <ShaneC!~shane@cpe-104-175-221-156.socal.res.rr.com> has joined ##vulkan[22:00:29] *** [[xyzzy]] <[[xyzzy]]!~xyzzy@ip68-229-227-253.ok.ok.cox.net> has joined ##vulkan[22:01:23] *** [xyzzy] <[xyzzy]!~xyzzy@ip68-229-227-253.ok.ok.cox.net> has joined ##vulkan[22:01:59] *** [xyzzy] <[xyzzy]!~xyzzy@ip68-229-227-253.ok.ok.cox.net> has quit IRC (Client Quit)[22:01:59] *** [[xyzzy]] is now known as [xyzzy][22:03:13] *** bjz <bjz!~bjz@pa49-185-190-131.pa.vic.optusnet.com.au> has joined ##vulkan[22:07:47] *** ImQ009 <ImQ009!~ImQ009@unaffiliated/imq009> has quit IRC (Quit: Leaving)[22:17:10] *** bjz <bjz!~bjz@pa49-185-190-131.pa.vic.optusnet.com.au> has quit IRC (Quit: My MacBook Pro has gone to sleep. ZZZzzz…)[22:24:38] *** toor_ <toor_!~toor@76.ip-51-254-204.eu> has joined ##vulkan[22:31:32] *** Ipe <Ipe!~I@a91-156-244-1.elisa-laajakaista.fi> has quit IRC (Quit: Konversation terminated!)[22:34:06] *** jkilpatr <jkilpatr!jkilpatr@nat/redhat/x-bahrbuywjydywynr> has quit IRC (Ping timeout: 258 seconds)[22:39:37] *** _28_ria <_28_ria!~igoryonya@opfr028.ru> has quit IRC (Read error: Connection reset by peer)[22:40:13] *** robot-beethoven <robot-beethoven!~robot-bee@c-73-65-29-166.hsd1.mn.comcast.net> has joined ##vulkan[22:43:00] *** _28_ria <_28_ria!~igoryonya@opfr028.ru> has joined ##vulkan[22:43:53] *** _28_ria <_28_ria!~igoryonya@opfr028.ru> has quit IRC (Read error: Connection reset by peer)[22:46:11] *** balrog <balrog!~balrog@unaffiliated/balrog> has quit IRC (Ping timeout: 276 seconds)[22:47:20] *** Thinkofdname_ <Thinkofdname_!~Think@178.62.1.224> has joined ##vulkan[22:52:27] *** jkilpatr <jkilpatr!~jkilpatr@2602:30a:c7da:b600:1202:b5ff:fe3b:d41d> has joined ##vulkan[22:53:14] *** Thinkofname <Thinkofname!Think@oops.i.broke.thinkofdeath.uk> has quit IRC (Ping timeout: 245 seconds)[22:53:14] *** Thinkofdname_ is now known as Thinkofname[22:57:40] *** balrog <balrog!~balrog@unaffiliated/balrog> has joined ##vulkan[23:03:00] *** abax is now known as Apaksi[23:03:10] *** Apaksi is now known as abax[23:03:10] *** sla_ro|master <sla_ro|master!~sla.ro@95.76.45.217> has quit IRC ()[23:19:47] *** plasticle <plasticle!~plasticle@cpe-107-15-169-8.nc.res.rr.com> has joined ##vulkan[23:37:40] *** balrog <balrog!~balrog@unaffiliated/balrog> has quit IRC (Ping timeout: 240 seconds)[23:42:41] *** doomlord <doomlord!~textual@host86-148-102-241.range86-148.btcentralplus.com> has quit IRC (Quit: My MacBook Pro has gone to sleep. ZZZzzz…)[23:43:24] *** ShadowIce <ShadowIce!~pyoro@unaffiliated/shadowice-x841044> has quit IRC (Quit: Leaving)[23:44:45] *** balrog <balrog!~balrog@unaffiliated/balrog> has joined ##vulkan[23:45:45] *** nsf <nsf!~nsf@jiss.convex.ru> has quit IRC (Quit: WeeChat 1.7)[23:46:10] *** ravior <ravior!~crapitea@5-13-152-135.residential.rdsnet.ro> has quit IRC (Quit: Konversation terminated!)[23:46:23] *** ravior <ravior!~crapitea@5-13-152-135.residential.rdsnet.ro> has joined ##vulkan[23:50:10] *** w-t-h_ <w-t-h_!~chatzilla@72.67.52.110> has quit IRC (Ping timeout: 264 seconds)[23:53:48] *** [xyzzy] <[xyzzy]!~xyzzy@ip68-229-227-253.ok.ok.cox.net> has quit IRC (Ping timeout: 240 seconds)[23:59:17] *** [xyzzy] <[xyzzy]!~xyzzy@2600:8803:e400:d600:21e:6ff:fecc:bedc> has joined ##vulkan