diff --git a/src/add-ons/accelerants/nvidia/engine/nv_acc.c b/src/add-ons/accelerants/nvidia/engine/nv_acc.c index 9ecda311eb..f5eabe1443 100644 --- a/src/add-ons/accelerants/nvidia/engine/nv_acc.c +++ b/src/add-ons/accelerants/nvidia/engine/nv_acc.c @@ -26,8 +26,7 @@ blit nVidia hardware info: We should be able to do FIFO assignment setup changes on-the-fly now, using all the engine-command-handles that are pre-defined on any FIFO channel. - (check channel to be emptied, then re-assign to new command, then use: etc.) - + Maybe we can even setup new additional handles to previously unused engine commands now, and there might even be a chance DMA can be setup(?). */ @@ -1060,6 +1059,13 @@ static void nv_init_for_3D(void) } } +/* fixme? (check this out..) + * Looks like this stuff can be very much simplified and speed-up, as it seems it's not + * nessesary to wait for the engine to become idle before re-assigning channels. + * Because the cmd handles are actually programmed _inside_ the fifo channels, it might + * well be that the assignment is buffered along with the commands that still have to + * be executed! + * (sounds very plausible to me :) */ void nv_acc_assert_fifo(void) { /* does every engine cmd this accelerant needs have a FIFO channel? */ diff --git a/src/add-ons/accelerants/nvidia/engine/nv_info.c b/src/add-ons/accelerants/nvidia/engine/nv_info.c index f65c921929..6c4392ce8a 100644 --- a/src/add-ons/accelerants/nvidia/engine/nv_info.c +++ b/src/add-ons/accelerants/nvidia/engine/nv_info.c @@ -1,7 +1,7 @@ /* Read initialisation information from card */ /* some bits are hacks, where PINS is not known */ /* Author: - Rudolf Cornelissen 7/2003-10/2004 + Rudolf Cornelissen 7/2003-12/2004 */ #define MODULE_BIT 0x00002000 @@ -2807,7 +2807,9 @@ static void pinsnv30_arch_fake(void) { case NV31: case NV36: + /* fixme? could be all >= NV40 cards have extended PLL's... (these 2 are confirmed) */ case NV40: + case NV43: /* we have a extended PLL */ si->ps.ext_pll = true; break;