Re: [Mesa-dev] is_tex bit for TGSI sampling instructions

2011-08-25 Thread Christoph Bumiller
On 25.08.2011 23:15, Zack Rusin wrote: > On Thursday, August 25, 2011 05:06:05 PM Christoph Bumiller wrote: >> What ? No. >> It would be: >> SAMPLE DST[0], SRC[0], RES[ADDR[0] - 4], SAMP[ADDR[0] - 4], >> OpenGL does not decouple resources and samplers at GLSL level. > Exactly, meaning that a group

Re: [Mesa-dev] is_tex bit for TGSI sampling instructions

2011-08-25 Thread Zack Rusin
On Thursday, August 25, 2011 05:06:05 PM Christoph Bumiller wrote: > What ? No. > It would be: > SAMPLE DST[0], SRC[0], RES[ADDR[0] - 4], SAMP[ADDR[0] - 4], > OpenGL does not decouple resources and samplers at GLSL level. Exactly, meaning that a group of samplers will have exactly the same resourc

Re: [Mesa-dev] is_tex bit for TGSI sampling instructions

2011-08-25 Thread Christoph Bumiller
On 25.08.2011 22:46, Zack Rusin wrote: > On Thursday, August 25, 2011 04:42:21 PM Christoph Bumiller wrote: >> Of course -4 isn't a valid resource, I just meant to emphasize that I >> cannot tell at all the resource to be accessed. >> ADDR[0].x - 4 should of course be >= 0 and valid. >> >> And, may

Re: [Mesa-dev] is_tex bit for TGSI sampling instructions

2011-08-25 Thread Zack Rusin
On Thursday, August 25, 2011 04:42:21 PM Christoph Bumiller wrote: > Of course -4 isn't a valid resource, I just meant to emphasize that I > cannot tell at all the resource to be accessed. > ADDR[0].x - 4 should of course be >= 0 and valid. > > And, maybe it seems silly to you, but OpenGL allows i

Re: [Mesa-dev] is_tex bit for TGSI sampling instructions

2011-08-25 Thread Christoph Bumiller
On 25.08.2011 22:42, Christoph Bumiller wrote: > On 25.08.2011 22:40, Zack Rusin wrote: >> On Thursday, August 25, 2011 04:30:26 PM Christoph Bumiller wrote: >>> Putting the resource type in the resource declaration instead of the >>> instruction isn't necessarily smart considering indirect resourc

Re: [Mesa-dev] is_tex bit for TGSI sampling instructions

2011-08-25 Thread Christoph Bumiller
On 25.08.2011 22:40, Zack Rusin wrote: > On Thursday, August 25, 2011 04:30:26 PM Christoph Bumiller wrote: >> Putting the resource type in the resource declaration instead of the >> instruction isn't necessarily smart considering indirect resource access >> ... I just hope the person implementing

Re: [Mesa-dev] is_tex bit for TGSI sampling instructions

2011-08-25 Thread Zack Rusin
On Thursday, August 25, 2011 04:30:26 PM Christoph Bumiller wrote: > Putting the resource type in the resource declaration instead of the > instruction isn't necessarily smart considering indirect resource access > ... I just hope the person implementing it remembers to provide a base > array addre

Re: [Mesa-dev] is_tex bit for TGSI sampling instructions

2011-08-25 Thread Christoph Bumiller
On 25.08.2011 06:42, Zack Rusin wrote: > On Wednesday, August 24, 2011 10:14:48 PM Bryan Cain wrote: >> Like Dave said, the GLSL->TGSI translator needs to account for this. > Probably not, at least yet. All of those instructions are DX10.1 level > instructions which support splitting of samplers a

Re: [Mesa-dev] is_tex bit for TGSI sampling instructions

2011-08-25 Thread Dave Airlie
On Thu, Aug 25, 2011 at 5:42 AM, Zack Rusin wrote: > On Wednesday, August 24, 2011 10:14:48 PM Bryan Cain wrote: >> Like Dave said, the GLSL->TGSI translator needs to account for this. > > Probably not, at least yet. All of those instructions are DX10.1 level > instructions which support splitting

Re: [Mesa-dev] is_tex bit for TGSI sampling instructions

2011-08-24 Thread Bryan Cain
On 08/24/2011 11:42 PM, Zack Rusin wrote: > On Wednesday, August 24, 2011 10:14:48 PM Bryan Cain wrote: >> Like Dave said, the GLSL->TGSI translator needs to account for this. > Probably not, at least yet. All of those instructions are DX10.1 level > instructions which support splitting of sampler

Re: [Mesa-dev] is_tex bit for TGSI sampling instructions

2011-08-24 Thread Zack Rusin
On Wednesday, August 24, 2011 10:14:48 PM Bryan Cain wrote: > Like Dave said, the GLSL->TGSI translator needs to account for this. Probably not, at least yet. All of those instructions are DX10.1 level instructions which support splitting of samplers and the underlaying resource, we don't suppor

Re: [Mesa-dev] is_tex bit for TGSI sampling instructions

2011-08-24 Thread Bryan Cain
On 08/24/2011 07:26 AM, Michal Krol wrote: > - Original Message - >> Any reasons >> >>{ 1, 2, 0, 0, 0, 0, "LOAD",TGSI_OPCODE_LOAD }, >>{ 1, 2, 0, 0, 0, 0, "LOAD_MS", TGSI_OPCODE_LOAD_MS }, >>{ 1, 3, 0, 0, 0, 0, "SAMPLE", TGSI_OPCODE_SAMPLE }, >>{ 1, 4, 0, 0,

Re: [Mesa-dev] is_tex bit for TGSI sampling instructions

2011-08-24 Thread Dave Airlie
On Wed, Aug 24, 2011 at 1:26 PM, Michal Krol wrote: > - Original Message - >> Any reasons >> >>    { 1, 2, 0, 0, 0, 0, "LOAD",        TGSI_OPCODE_LOAD }, >>    { 1, 2, 0, 0, 0, 0, "LOAD_MS",     TGSI_OPCODE_LOAD_MS }, >>    { 1, 3, 0, 0, 0, 0, "SAMPLE",      TGSI_OPCODE_SAMPLE }, >>    { 1

Re: [Mesa-dev] is_tex bit for TGSI sampling instructions

2011-08-24 Thread Michal Krol
- Original Message - > Any reasons > >{ 1, 2, 0, 0, 0, 0, "LOAD",TGSI_OPCODE_LOAD }, >{ 1, 2, 0, 0, 0, 0, "LOAD_MS", TGSI_OPCODE_LOAD_MS }, >{ 1, 3, 0, 0, 0, 0, "SAMPLE", TGSI_OPCODE_SAMPLE }, >{ 1, 4, 0, 0, 0, 0, "SAMPLE_B",TGSI_OPCODE_SAMPLE_B }, >

[Mesa-dev] is_tex bit for TGSI sampling instructions

2011-08-24 Thread Dave Airlie
Any reasons { 1, 2, 0, 0, 0, 0, "LOAD",TGSI_OPCODE_LOAD }, { 1, 2, 0, 0, 0, 0, "LOAD_MS", TGSI_OPCODE_LOAD_MS }, { 1, 3, 0, 0, 0, 0, "SAMPLE", TGSI_OPCODE_SAMPLE }, { 1, 4, 0, 0, 0, 0, "SAMPLE_B",TGSI_OPCODE_SAMPLE_B }, { 1, 4, 0, 0, 0, 0, "SAMPLE_C",TGSI_OP