Felix Kuehling 590c8e522c Fix node 0 system memory allocation for dGPU ...

This is a hack to allow the Runtime to allocate system memory with
PreferredNode=0 on a dGPU system. We allocated it from Node 1
instead so that the node 1 GPU can map the memory. A proper fix
will be implemented together with multi-GPU support.

Change-Id: Ieb52599e5275781c04ee34405ea850bf782c523a

2015-10-21 20:00:01 -04:00

..

debug.c

Modify HW dbg thunk,to adapt to new KFD-thunk interface

2015-02-08 21:44:47 +02:00

events.c

Fix hard-coded usage of Node 0

2015-10-14 17:27:38 -04:00

fmm.c

Reserve more SVM process address space

2015-10-19 17:44:23 -04:00

fmm.h

Revert "Topology, memory allocation, cleanup issue for gGPU"

2015-09-21 10:47:30 -04:00

globals.c

Support gfx802 dGPU

2015-08-30 14:13:53 +03:00

libhsakmt.c

Move from direct call to kfd_ioctl to wrapping kmtIoctl

2014-12-29 15:37:42 +02:00

libhsakmt.h

Fix init process apertures

2015-10-15 15:38:14 -04:00

libhsakmt.ver

Add interface to set CU mask

2015-07-23 15:44:01 +08:00

Makefile

make: add 'deb' target for creating deb packages

2015-10-16 19:13:51 -04:00

memory.c

Fix node 0 system memory allocation for dGPU

2015-10-21 20:00:01 -04:00

openclose.c

Revert "Topology, memory allocation, cleanup issue for gGPU"

2015-09-21 10:47:30 -04:00

perfctr.c

Add exclusive trace access

2014-12-29 10:13:32 +02:00

pmc_table.c

Add Performance Counter Support on gfx801.

2015-04-09 00:00:42 +08:00

pmc_table.h

Add pmc table module

2014-07-29 13:58:01 +03:00

queues.c

Fix hard-coded usage of Node 0

2015-10-14 17:27:38 -04:00

time.c

Move from direct call to kfd_ioctl to wrapping kmtIoctl

2014-12-29 15:37:42 +02:00

topology.c

Fix init process apertures

2015-10-15 15:38:14 -04:00

version.c

Move from direct call to kfd_ioctl to wrapping kmtIoctl

2014-12-29 15:37:42 +02:00