725f46aa94
Merge pull request #237 from bensander/use_ctxptr_for_p2p
...
Use ctxptr for p2p
[ROCm/clr commit: fe32685fbc
2017-11-01 18:55:25 +01:00
4df6b371ab
Add ns-level timer for HIP API routines
...
Refactor some miuses of ihipLogStatus, these should only be in top-level
HIP APIs and should be paired with HIP_API_INIT calls.
[ROCm/clr commit: dc7d993a02
2017-10-30 20:20:51 +00:00
fd5b9cb650
Merge pull request #222 from bensander/fix_device_prop
...
Fix device prop
[ROCm/clr commit: 7e8b39fc96
2017-10-30 17:58:48 +01:00
ec29721b16
Check for null copyEngine before looking at peers.
...
[ROCm/clr commit: 4c7b2be1c2
2017-10-30 16:58:03 +00:00
fb169b4de6
Fix bug with peer-to-peer combined with context API
...
- Store context inside the tracker rather than using int deviceID that
was always mapped to primary context
- IsPeerWatcher now based on device IDs rather than specific peers.
[ROCm/clr commit: a417241507
2017-10-26 19:44:22 +00:00
a4af82519a
Enhance debug for copy pointers
...
- show more pointer tracking fields
- show pointer info before and after "tailoring'
[ROCm/clr commit: 698721be34
2017-10-26 19:44:22 +00:00
5f24981c61
replace __hcc_workweek__ with HC_FEATURE_PRINTF flag
...
[ROCm/clr commit: da7c37947c
2017-10-23 18:30:08 -04:00
b5e3de043c
Remove printf
...
[ROCm/clr commit: b4c7876244
2017-10-20 13:24:04 -07:00
2774542487
Update device properties.
...
- clear properties to defined initial state.
- enable some property flags which are now supported.
[ROCm/clr commit: ed85b15c3e
2017-10-20 15:52:13 +00:00
339cf92137
Modify device properties to use pool API.
...
- Also better error code checking
[ROCm/clr commit: e738fa66b5
2017-10-20 14:49:29 +00:00
5ccf3d4d9b
hipDeviceReset(): make sure to reinitialize the printf buffer in hcc RT
...
[ROCm/clr commit: a1956f64e6
2017-10-18 16:26:13 -04:00
a6159384a1
Bump device major version from 2 to 3
...
This would significantly improve performance for certain apps in kernel
selection logic.
[ROCm/clr commit: 1efc99e69f
2017-09-15 15:47:39 +00:00
49b1276e3c
Merge branch 'master' into hip_init_alloc
...
[ROCm/clr commit: 1ea468e279
2017-09-14 11:53:33 -05:00
2e98e3c4a0
Add HIP_INIT_ALLOC to init allocated memory.
...
[ROCm/clr commit: fff42fd591
2017-09-13 23:31:48 +00:00
5134d6a2e9
hipStreamQuery uses av::is_empty. Add HIP_FORCE_NULL_STREAM.
...
[ROCm/clr commit: fff74eee21
2017-08-31 03:00:14 +00:00
b45217fef0
Refactor hipStreamWaitEvent
...
- Null streams use same flow as non-null.
- Add HIP_SYNC_STREAM_WAIT
- Resolve null stream.
[ROCm/clr commit: ed8c3ba7e7
2017-08-31 03:00:14 +00:00
070bdb3736
Lock streams when waiting on event completion or querying event safety.
...
[ROCm/clr commit: bc9ba7cd81
2017-08-28 18:40:16 -05:00
d74896f873
[texture] guard new HCC APIs under workweek
...
Change-Id: I4f60a64fb0b0496ca1eb01ffe6ddda121c25d976
[ROCm/clr commit: 172a568aa6
2017-08-15 15:51:38 +05:30
a7f95bb8ee
[HIP Texture] The GPU virtual address for texture memory needs to be
...
aligned.
In hcc_am, a bigger buffer will be allocated for alignment purpose
and _unalignedDevicePointer is added in struct AmPointerInfo for
original allocated address.
[ROCm/clr commit: e4de2d1138
2017-08-08 11:18:00 -04:00
602c3466f9
Merge pull request #135 from bensander/fix_tracing
...
Some fixes to tracing.
[ROCm/clr commit: 06b51109c6
2017-07-31 10:24:41 +05:30
6079b4b4f6
Some fixes to tracing.
...
[ROCm/clr commit: 9e9f384899
2017-07-28 22:13:43 -05:00
8b7ce77fc0
Merge pull request #122 from bensander/enable_async_null_stream
...
Set HIP_SYNC_NULL_STREAM=0.
[ROCm/clr commit: 8330fb3fe0
2017-07-28 09:15:56 +05:30
1aa0656d34
Set HIP_SYNC_NULL_STREAM=0.
...
Optimizes null stream synchronization so it uses GPU-side dependency
resolution. Requires HCC __hcc_workweek__ > 17300.
[ROCm/clr commit: cd42711134
2017-07-27 11:11:54 -05:00
235b60165f
Make host memory allocations coherent by default.
...
Associated change is to optimize event recording so it uses
agent-scope releaes (since it was only using system-scope release
to support non-coherent host mem).
Flags and environment variables exist to obtain previous behavior
if desired. Options are documented in new performance guide.
[ROCm/clr commit: 6576201ec2
2017-07-26 19:20:34 -05:00
0e9272cfd7
Enable HCC_OPT_FLUSH=1 (if HCC compiler new enough)
...
[ROCm/clr commit: cdc4291431
2017-07-24 18:57:19 -05:00
c824a3ea0c
Temporarily disables HCC_OPT_FLUSH
...
Change-Id: I290791e58dd52ab3823f6c3315e42b0d386e9d64
[ROCm/clr commit: 32909c4a92
2017-07-12 16:08:36 +00:00
f1f98a0d2f
Set default HIP_SYNC_NULL_STREAM=1.
...
[ROCm/clr commit: 61fafdceb1
2017-06-30 19:01:14 -05:00
bd427d5b47
automate gcnarch detection
...
Change-Id: Ibbad22db136f7f5e2be84c82e9169298a144cc77
[ROCm/clr commit: 0fe0381608
2017-06-29 12:01:40 -05:00
1ff995a92a
Fixed hipDeviceGetPCIBusId for HIP/HCC
...
Change-Id: I3688fa2476e1baada2d3c5fc3735cec3f15a1e21
[ROCm/clr commit: a7c727f352
2017-06-28 23:48:27 +05:30
639a6d5e46
Set default HIP_HIDDEN_FREE_MEM
...
[ROCm/clr commit: eb2c5e166c
2017-06-27 12:17:12 -05:00
2582679911
Add support of HIP_HIDDEN_FREE_MEM, to deduct the returned available
...
memory from hipMemGetInfo API, measured in MB.
Change-Id: I7a8260c12e032e04e26611db4c38c893a29f2653
[ROCm/clr commit: e5ce585307
2017-06-26 15:29:38 -05:00
f1bed37a34
Clean up old USE_* and RELEASE.md notes.
...
[ROCm/clr commit: 42882ddf9c
2017-06-23 18:05:30 -05:00
daae691cdb
Enable HCC_OPT_FLUSH=1.
...
Requires appropriate HCC with this support :
commit 38e392b517a46a09a3b1c8f388e6a0db3741c510
[ROCm/clr commit: c2baa4f6e6
2017-06-07 00:15:05 -05:00
d1fe7f1683
Enable HIP_SYNC_NULL_STREAM=0 optimization.
...
[ROCm/clr commit: 344b6cb0c0
2017-06-05 08:50:41 -05:00
7237ed04f3
Fix HIP_SYNC_NULL_STREAM=0 mode.
...
- Fix null-stream sync
- hipStreamDestroy of null stream returns hipErrorInvalidResourceHandle
- Update documentation.
- Add tests for null stream sync, hipEventElapsedTime.
- Rename internal enum hipEventStatusRecorded to hipEventStatusComplete
- refactor hipStreamWaitEvent to streamline control-flow
[ROCm/clr commit: 823281dcba
2017-06-05 08:50:22 -05:00
6aaeed821d
Update tests, add p2p coherency test.
...
[ROCm/clr commit: 15f54fb943
2017-06-03 17:11:34 -05:00
0ca3262f0a
Add event controls for release fences.
...
Env var : HIP_EVENT_SYS_RELEASE
Event allocation flags : hipEventReleaseToDevice, hipEventReleaseToSystem
(remove hipEventDisableSystemRelease)
Update test for new functionality.
[ROCm/clr commit: 942ec0eff8
2017-05-27 16:02:34 -05:00
d6e8f5bbdc
Cleanup hipEvent. (Intermediate checkpoint)
...
Support hipEventDisableSystemRelease flag.
Update test.
Remove stray printf
[ROCm/clr commit: c8178c6838
2017-05-27 16:02:34 -05:00
b4363ffcba
Remove HIP_NUM_KERNELS_INFLIGHT. (redundant with HCC controls)
...
[ROCm/clr commit: 35212632e7
2017-05-24 01:03:28 -05:00
d302498787
Add hipHostMallocCoherent, hipHostMallocNonCoherent
...
Provide per-allocation control over coherent/non-coherent mem.
These overrid the default HIP_COHERENT_HOST_ALLOC setting.
[ROCm/clr commit: dda70ae514
2017-05-24 00:48:10 -05:00
ae983e1b09
Remove HIP_MAX_QUEUES (replaced with HCC_MAX_QUEUES)
...
[ROCm/clr commit: d43d57d39c
2017-05-23 23:48:01 -05:00
2e8625a208
Use accelerator_scope for create_marker and create_blocking_marker.
...
As optimization when system-scope is not needed.
[ROCm/clr commit: 2d5b3359c6
2017-05-23 23:15:45 -05:00
4ac6ac9d1d
Add initial HIP_SYNC_NULL_STREAM=0 mode.
...
This eliminates host-synchronization for null stream. Instead, the
null-stream uses GPU-side events to wait for other streams.
Default is OFF pending additional testing.
Add enhanced null-stream test.
Also refine HIP_TRACE_API.
[ROCm/clr commit: 8bc6ee5932
2017-05-16 19:04:25 -05:00
ee05975efa
Add HIP_TRACE_API=4. Only display memory allocation/free apis.
...
[ROCm/clr commit: 7e7ba5027f
2017-05-16 19:04:25 -05:00
feca2ed0dc
added gfx900 to hipDeviceProp_t
...
Change-Id: I49e7a32f218926fd55f1c94c5dc2366d6c8ac4ca
[ROCm/clr commit: a6dc00f167
2017-05-12 21:43:34 -05:00
0d3c99eb6e
Remove old USE_ switches no longer needed.
...
[ROCm/clr commit: c7c62dd022
2017-05-12 16:06:03 -05:00
9b222c0ae4
Add hipEventDisableSystemRelease flag.
...
[ROCm/clr commit: 2c2625cb9e
2017-05-12 16:06:03 -05:00
90e5930923
hipHostMalloc allocation are mapped to all devices by default.
...
Support hipHostMallocPortable flag.
Default flags are hipHostMallocPortable | hipHostMallocMapped.
Also:
-refactor tests to move addCount and addCountReverse into HipTest
namespace.
-test multi-GPU host memory.
[ROCm/clr commit: ff9bed6535
2017-05-10 17:34:36 -05:00
0e042a1529
Tailor pointer info for src/dst before calling HCC copy routines.
...
HCC sometimes uses the srcPtrInfo or dstPtrInfo to determine the
pointer. Make sure these use the actual pointer and not the base
of the allocation.
[ROCm/clr commit: 3da8e94cbf
2017-04-24 22:30:54 -05:00
eef8282c96
Fix hip debug for case where copyAgent is null (host-to-host)
...
[ROCm/clr commit: dfacfbb641
2017-04-24 22:30:54 -05:00
Ben Sander
Aditya Atluri
Siu Chi Chan
Wen-Heng (Jack) Chung
Maneesh Gupta
Weixing Zhang
Rahul Garg
Sun, Peng