Re: [Mesa-dev] TGSI 16-bit support

2017-08-26 Thread Nicolai Hähnle
On 26.08.2017 00:46, Timothy Arceri wrote: On 26/08/17 00:50, Nicolai Hähnle wrote: On 25.08.2017 13:58, Marek Olšák wrote: Nicolai, Have you thought about switching to NIR for radeonsi completely to get 16-bit support? We need NIR support anyway for spirv, right? Would be it be easier tha

Re: [Mesa-dev] TGSI 16-bit support

2017-08-25 Thread Timothy Arceri
On 26/08/17 00:50, Nicolai Hähnle wrote: On 25.08.2017 13:58, Marek Olšák wrote: Nicolai, Have you thought about switching to NIR for radeonsi completely to get 16-bit support? We need NIR support anyway for spirv, right? Would be it be easier than adding 16-bit support into TGSI, glsl2tgsi

Re: [Mesa-dev] TGSI 16-bit support

2017-08-25 Thread Connor Abbott
On Aug 25, 2017 4:10 PM, "Matt Turner" wrote: On Fri, Aug 25, 2017 at 10:50 AM, Nicolai Hähnle wrote: > On 25.08.2017 13:58, Marek Olšák wrote: >> >> Nicolai, >> >> Have you thought about switching to NIR for radeonsi completely to get >> 16-bit support? We need NIR support anyway for spirv, rig

Re: [Mesa-dev] TGSI 16-bit support

2017-08-25 Thread Matt Turner
On Fri, Aug 25, 2017 at 10:50 AM, Nicolai Hähnle wrote: > On 25.08.2017 13:58, Marek Olšák wrote: >> >> Nicolai, >> >> Have you thought about switching to NIR for radeonsi completely to get >> 16-bit support? We need NIR support anyway for spirv, right? Would be it be >> easier than adding 16-bit

Re: [Mesa-dev] TGSI 16-bit support

2017-08-25 Thread Nicolai Hähnle
On 25.08.2017 13:58, Marek Olšák wrote: Nicolai, Have you thought about switching to NIR for radeonsi completely to get 16-bit support? We need NIR support anyway for spirv, right? Would be it be easier than adding 16-bit support into TGSI, glsl2tgsi, and tgsi2llvm? Well. What's missing from

Re: [Mesa-dev] TGSI 16-bit support

2017-08-25 Thread Marek Olšák
Nicolai, Have you thought about switching to NIR for radeonsi completely to get 16-bit support? We need NIR support anyway for spirv, right? Would be it be easier than adding 16-bit support into TGSI, glsl2tgsi, and tgsi2llvm? Marek ___ mesa-dev mailing

Re: [Mesa-dev] TGSI 16-bit support

2017-08-24 Thread Marek Olšák
On Thu, Aug 24, 2017 at 3:39 PM, Nicolai Hähnle wrote: > On 24.08.2017 14:19, Erik Faye-Lund wrote: >> >> On Wed, Aug 23, 2017 at 6:19 PM, Marek Olšák wrote: >>> >>> On Wed, Aug 23, 2017 at 3:08 PM, Nicolai Hähnle >>> wrote: Here's another question: What does "low precision" mean

Re: [Mesa-dev] TGSI 16-bit support

2017-08-24 Thread Nicolai Hähnle
On 24.08.2017 14:19, Erik Faye-Lund wrote: On Wed, Aug 23, 2017 at 6:19 PM, Marek Olšák wrote: On Wed, Aug 23, 2017 at 3:08 PM, Nicolai Hähnle wrote: Here's another question: What does "low precision" mean on a texture instruction? Are the offsets low precision or is it the output? Maybe we

Re: [Mesa-dev] TGSI 16-bit support

2017-08-24 Thread Erik Faye-Lund
On Wed, Aug 23, 2017 at 6:19 PM, Marek Olšák wrote: > On Wed, Aug 23, 2017 at 3:08 PM, Nicolai Hähnle wrote: >> >> Here's another question: What does "low precision" mean on a texture >> instruction? Are the offsets low precision or is it the output? Maybe we can >> punt on this for now -- at lea

Re: [Mesa-dev] TGSI 16-bit support

2017-08-23 Thread Marek Olšák
On Wed, Aug 23, 2017 at 3:08 PM, Nicolai Hähnle wrote: > On 22.08.2017 22:39, Roland Scheidegger wrote: >> >> Am 22.08.2017 um 19:10 schrieb Marek Olšák: >>> >>> Hi, >>> >>> I'd like to discuss 16-bit float and integer support in TGSI. I'm >>> proposing this: >>> >>> struct tgsi_instruction >>>

Re: [Mesa-dev] TGSI 16-bit support

2017-08-23 Thread Ilia Mirkin
On Wed, Aug 23, 2017 at 10:48 AM, Nicolai Hähnle wrote: > On 23.08.2017 16:36, Ilia Mirkin wrote: >> >> On Wed, Aug 23, 2017 at 10:30 AM, Nicolai Hähnle >> wrote: >>> >>> On 23.08.2017 15:15, Nicolai Hähnle wrote: On 22.08.2017 19:32, Marek Olšák wrote: > > > On Tue, Au

Re: [Mesa-dev] TGSI 16-bit support

2017-08-23 Thread Nicolai Hähnle
On 23.08.2017 16:36, Ilia Mirkin wrote: On Wed, Aug 23, 2017 at 10:30 AM, Nicolai Hähnle wrote: On 23.08.2017 15:15, Nicolai Hähnle wrote: On 22.08.2017 19:32, Marek Olšák wrote: On Tue, Aug 22, 2017 at 7:28 PM, Ilia Mirkin wrote: How do you propose defining the semantics for e.g. loadin

Re: [Mesa-dev] TGSI 16-bit support

2017-08-23 Thread Ilia Mirkin
On Wed, Aug 23, 2017 at 10:30 AM, Nicolai Hähnle wrote: > On 23.08.2017 15:15, Nicolai Hähnle wrote: >> >> On 22.08.2017 19:32, Marek Olšák wrote: >>> >>> On Tue, Aug 22, 2017 at 7:28 PM, Ilia Mirkin >>> wrote: How do you propose defining the semantics for e.g. loading a 16-bit val

Re: [Mesa-dev] TGSI 16-bit support

2017-08-23 Thread Nicolai Hähnle
On 23.08.2017 15:15, Nicolai Hähnle wrote: On 22.08.2017 19:32, Marek Olšák wrote: On Tue, Aug 22, 2017 at 7:28 PM, Ilia Mirkin wrote: How do you propose defining the semantics for e.g. loading a 16-bit value from a constbuf/ssbo? Would those get separate instructions? st/mesa should use UP2

Re: [Mesa-dev] TGSI 16-bit support

2017-08-23 Thread Nicolai Hähnle
On 23.08.2017 16:00, Roland Scheidegger wrote: Am 23.08.2017 um 15:08 schrieb Nicolai Hähnle: On 22.08.2017 22:39, Roland Scheidegger wrote: Am 22.08.2017 um 19:10 schrieb Marek Olšák: Hi, I'd like to discuss 16-bit float and integer support in TGSI. I'm proposing this: struct tgsi_instru

Re: [Mesa-dev] TGSI 16-bit support

2017-08-23 Thread Roland Scheidegger
Am 23.08.2017 um 15:08 schrieb Nicolai Hähnle: > On 22.08.2017 22:39, Roland Scheidegger wrote: >> Am 22.08.2017 um 19:10 schrieb Marek Olšák: >>> Hi, >>> >>> I'd like to discuss 16-bit float and integer support in TGSI. I'm >>> proposing this: >>> >>>   struct tgsi_instruction >>>   { >>> uns

Re: [Mesa-dev] TGSI 16-bit support

2017-08-23 Thread Nicolai Hähnle
On 22.08.2017 19:32, Marek Olšák wrote: On Tue, Aug 22, 2017 at 7:28 PM, Ilia Mirkin wrote: How do you propose defining the semantics for e.g. loading a 16-bit value from a constbuf/ssbo? Would those get separate instructions? st/mesa should use UP2H, PK2H and similar opcodes for I16 and U16,

Re: [Mesa-dev] TGSI 16-bit support

2017-08-23 Thread Nicolai Hähnle
On 22.08.2017 22:39, Roland Scheidegger wrote: Am 22.08.2017 um 19:10 schrieb Marek Olšák: Hi, I'd like to discuss 16-bit float and integer support in TGSI. I'm proposing this: struct tgsi_instruction { unsigned Type : 4; /* TGSI_TOKEN_TYPE_INSTRUCTION */ unsigned NrTokens

Re: [Mesa-dev] TGSI 16-bit support

2017-08-22 Thread Roland Scheidegger
Am 22.08.2017 um 19:10 schrieb Marek Olšák: > Hi, > > I'd like to discuss 16-bit float and integer support in TGSI. I'm > proposing this: > > struct tgsi_instruction > { > unsigned Type : 4; /* TGSI_TOKEN_TYPE_INSTRUCTION */ > unsigned NrTokens : 8; /* UINT */ > unsigned O

Re: [Mesa-dev] TGSI 16-bit support

2017-08-22 Thread Marek Olšák
On Tue, Aug 22, 2017 at 7:55 PM, Ilia Mirkin wrote: > On Tue, Aug 22, 2017 at 1:32 PM, Marek Olšák wrote: >> On Tue, Aug 22, 2017 at 7:28 PM, Ilia Mirkin wrote: >>> How do you propose defining the semantics for e.g. loading a 16-bit >>> value from a constbuf/ssbo? Would those get separate instru

Re: [Mesa-dev] TGSI 16-bit support

2017-08-22 Thread Ilia Mirkin
On Tue, Aug 22, 2017 at 1:32 PM, Marek Olšák wrote: > On Tue, Aug 22, 2017 at 7:28 PM, Ilia Mirkin wrote: >> How do you propose defining the semantics for e.g. loading a 16-bit >> value from a constbuf/ssbo? Would those get separate instructions? > > st/mesa should use UP2H, PK2H and similar opco

Re: [Mesa-dev] TGSI 16-bit support

2017-08-22 Thread Marek Olšák
On Tue, Aug 22, 2017 at 7:28 PM, Ilia Mirkin wrote: > How do you propose defining the semantics for e.g. loading a 16-bit > value from a constbuf/ssbo? Would those get separate instructions? st/mesa should use UP2H, PK2H and similar opcodes for I16 and U16, and drivers can replace them with MOV i

Re: [Mesa-dev] TGSI 16-bit support

2017-08-22 Thread Ilia Mirkin
How do you propose defining the semantics for e.g. loading a 16-bit value from a constbuf/ssbo? Would those get separate instructions? On Tue, Aug 22, 2017 at 1:10 PM, Marek Olšák wrote: > Hi, > > I'd like to discuss 16-bit float and integer support in TGSI. I'm > proposing this: > > struct tgsi

[Mesa-dev] TGSI 16-bit support

2017-08-22 Thread Marek Olšák
Hi, I'd like to discuss 16-bit float and integer support in TGSI. I'm proposing this: struct tgsi_instruction { unsigned Type : 4; /* TGSI_TOKEN_TYPE_INSTRUCTION */ unsigned NrTokens : 8; /* UINT */ unsigned Opcode : 8; /* TGSI_OPCODE_ */ unsigned Saturate : 1;