Re: [Mesa-dev] [PATCH 14/14] radeonsi: increase the maximum UBO size to 2 GB

2018-08-09 Thread Marek Olšák
piglit testing the max uniform block size: Before: FRAG PROPERTY FS_COLOR0_WRITES_ALL_CBUFS 1 DCL OUT[0], COLOR DCL CONST[0][0] DCL CONST[1][0..4095] DCL TEMP[0], LOCAL IMM[0] UINT32 {0, 16, 0, 0} 0: UMUL TEMP[0].x, CONST[0][0]., IMM[0]. 1: LOAD OUT[0], CONSTBUF[1], TEMP[0]. 2:

Re: [Mesa-dev] [PATCH 14/14] radeonsi: increase the maximum UBO size to 2 GB

2018-08-09 Thread Roland Scheidegger
Am 09.08.2018 um 17:57 schrieb Marek Olšák: > On Thu, Aug 9, 2018 at 1:35 AM, Roland Scheidegger wrote: >> I'm not quite convinced you can really use huge ubos safely? At least >> direct addressing in tgsi can't work (you've only got a 16bit register >> index, and it's signed too). > > UBOs use t

Re: [Mesa-dev] [PATCH 14/14] radeonsi: increase the maximum UBO size to 2 GB

2018-08-09 Thread Marek Olšák
On Thu, Aug 9, 2018 at 1:35 AM, Roland Scheidegger wrote: > I'm not quite convinced you can really use huge ubos safely? At least > direct addressing in tgsi can't work (you've only got a 16bit register > index, and it's signed too). UBOs use the LOAD instruction if PIPE_CAP_LOAD_CONSTBUF is supp

Re: [Mesa-dev] [PATCH 14/14] radeonsi: increase the maximum UBO size to 2 GB

2018-08-08 Thread Roland Scheidegger
I'm not quite convinced you can really use huge ubos safely? At least direct addressing in tgsi can't work (you've only got a 16bit register index, and it's signed too). Roland Am 09.08.2018 um 01:55 schrieb Marek Olšák: > From: Marek Olšák > > Same as the closed driver. > > This causes a fail

[Mesa-dev] [PATCH 14/14] radeonsi: increase the maximum UBO size to 2 GB

2018-08-08 Thread Marek Olšák
From: Marek Olšák Same as the closed driver. This causes a failure in GL45-CTS.compute_shader.max, which has a trivial bug. --- src/gallium/drivers/radeonsi/si_get.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/gallium/drivers/radeonsi/si_get.c b/src/gallium/drivers/