Grafik Komit

3983 Melakukan

Penulis SHA1 Pesan Tanggal
Chauncey Hui ebda86f961 SWDEV-2 - Change OpenCL version number from 3264 to 3265 2021-02-26 03:00:10 -05:00
German Andryeyev a9b0e20d26 SWDEV-272496 - Fix a regression in PAL
Skip notification for markers with direct dispatch only,
since they are blocking always

Change-Id: I6bb17650f73371dae6e29c59fd6bb2012cc062fd
2021-02-25 11:11:42 -05:00
Chauncey Hui 9c2f9b208a SWDEV-2 - Change OpenCL version number from 3263 to 3264 2021-02-25 03:00:07 -05:00
Vladislav Sytchenko 184b2631d5 SWDEV-271964 - Revert "SWDEV-264244 Fix StreamSync"
This reverts commit a148a71075.

Change-Id: I870c8b71edeb31f587fffe2447762acba61a7938
2021-02-24 11:43:08 -05:00
Vladislav Sytchenko 9e8a2f3266 SWDEV-86035 - [PAL] Correct allocation logging
Since the allocation can be a suballocation, we should print both the
VA range for the allocation and the underlying memory object.

Change-Id: Ic9c707bbb78113b366d1b2c688e6fd33bdc8fd94
2021-02-24 11:35:49 -05:00
kjayapra-amd 46a50965c4 SWDEV-274058 - Porting HIP, ROCclr gfx90a changes to mainline.
Change-Id: I4f4220df77e57f749a00c1dbb66743ac5af4959a
2021-02-24 09:55:54 -05:00
Chauncey Hui d5e6769959 SWDEV-2 - Change OpenCL version number from 3262 to 3263 2021-02-24 03:00:06 -05:00
German Andryeyev 0587fb7450 SWDEV-272496 - Disable notification for the previous notify
Direct disaptch doesn't insert extra barriers for Markers if
AQL barrier was the last issued command already.

Change-Id: I00fbc658547d83dd3ee64ec391ed50e5f8a08e30
2021-02-23 17:04:59 -05:00
Jason Tang 54a7170e40 SWDEV-198364 - Only enable clgl sharing in ROCm path when building LinuxPro
Change-Id: Ie4d87e252519d090a62b930f7ebb315d3477b690
2021-02-23 14:15:04 -05:00
Vladislav Sytchenko c93f66a5ec SWDEV-270013 - Fix uninitialized variable warning
Change-Id: Ic4c8090e94e82798c294ac4532ca20efbc5a3a54
2021-02-23 11:49:28 -05:00
Chauncey Hui 708d3d6621 SWDEV-2 - Change OpenCL version number from 3261 to 3262 2021-02-23 03:00:06 -05:00
Vladislav Sytchenko dbcf023cb1 SWDEV-86035 - [PAL] Propagate gl2Uncached setting to the backend
Change-Id: I594f304a20d448fe198ec27901e6c97b265c1183
2021-02-21 13:52:38 -05:00
Vladislav Sytchenko 263173914f SWDEV-86035 - [PAL] Fix runtime init
The settings need to be populated ASAP, otherwise the dummy context is
not created properly.

Change-Id: Iede0066308bb601dc68164e894775a646a0372f1
2021-02-20 16:39:13 -05:00
Chauncey Hui b625a46db1 SWDEV-2 - Change OpenCL version number from 3260 to 3261 2021-02-19 03:00:07 -05:00
Jason Tang 0d47b06928 SWDEV-1 - Fix warnings when building with clang++
Change-Id: Idebd2b618b9a3360147984a0e33852dbe2e65818
2021-02-18 10:02:46 -05:00
Chauncey Hui ae7cb73fe5 SWDEV-2 - Change OpenCL version number from 3259 to 3260 2021-02-18 03:00:06 -05:00
German Andryeyev ac387f9b03 SWDEV-272496 - Wait on CPU before switching to GPU wait
GPU waits have noticeable overheads on compute with extra
AQL barrier packet and on SDMA with power saving features. This
change introduces a wait on CPU for 30 us in case the app has tiny
operations.

Change-Id: I761ba3af595f3f48544980058a9077dda15aa5f9
2021-02-17 15:19:17 -05:00
Chauncey Hui 9900b46c21 SWDEV-2 - Change OpenCL version number from 3258 to 3259 2021-02-17 03:00:07 -05:00
Jason Tang 82c6aabb0a SWDEV-268345 - Remove ADDRESS_SANITIZER option from cmake
DevOps is going to add this through build script.

Change-Id: I536d691ea8c05d8289bad1448cac532a8ef93cc4
2021-02-16 17:37:48 -05:00
Vladislav Sytchenko 671778bdd3 SWDEV-232428 - Push hostcall implementation to the device layer
This change unifies the hostcall implementation for all the backends,
by pushing the common logic to the device layer. This is done by
replacing the use of hsa_signal_t with device::Signal (a light wrapper
around it).

Change-Id: I7b6fca7930b5a0b199da5d85e2e048354cc04e7b
2021-02-16 17:19:57 -05:00
Jason Tang c65be06c4c SWDEV-265304 - Disable GFX8 in LinuxPro ROCr build
Change-Id: Ia26bb025a133a4acb255a539a45668975019cc6e
2021-02-16 13:32:47 -05:00
Vladislav Sytchenko ac237ac094 SWDEV-271486 - [PAL] Temporarily disable reporting sramecc support
Change-Id: I089af8f8a4645ce5353f351dbf3a018cd31df216
2021-02-16 10:39:18 -05:00
Vladislav Sytchenko 660e543f49 SWDEV-270013 - Fix Windows build
Change-Id: I9573ef8b9d94367b8fa1dd90d40a93ddc4992f18
2021-02-16 10:03:51 -05:00
Chauncey Hui a1b21ae7d2 SWDEV-2 - Change OpenCL version number from 3257 to 3258 2021-02-16 03:00:08 -05:00
Ravi C Akkenapally 0a5f9a3b10 SWDEV-179105 - Stream Operations: Add support for Wait and Write
Change-Id: Ibffa1d6d573826b64763da280074a77271d66808
2021-02-15 17:02:38 -08:00
kjayapra-amd 2df099df9e SWDEV-270013 - Allocate kernel_arguments from kern_arg & finegrain pool instead of coarse grain.
Change-Id: Id4c6977934fdd6ef2311f6e75593801f1e51983c
2021-02-15 18:20:08 -05:00
Saleel Kudchadker 0f14c54c04 SWDEV-272673 - Add changes to dump log to a file
Env var AMD_LOG_LEVEL_FILE would dump the log to file.
Change-Id: I6add4a1ae6788f376ce116797cc0573007502e73
2021-02-15 10:28:06 -08:00
Chauncey Hui 196fada9c8 SWDEV-2 - Change OpenCL version number from 3256 to 3257 2021-02-13 03:00:07 -05:00
German Andryeyev 70f06c79fa SWDEV-86035 - switch PAL interface to 656 version
Change-Id: I162ba6e49aa866c013f37a114df5bb7ea3e08874
2021-02-12 13:08:59 -05:00
Chauncey Hui 272e29a1cd SWDEV-2 - Change OpenCL version number from 3255 to 3256 2021-02-12 03:00:07 -05:00
German Andryeyev e7c636c5e7 SWDEV-86035 - Enable PAL for gfx8
Change-Id: Ia6623993e44aeb4bdf317628ee8a84af6c4cacc7
2021-02-11 14:25:43 -05:00
Payam a2e0b0495c SWDEV-257937 - Updated fix for ROC_BARRIER_SYNC=0
Change-Id: I7e28e541b654db57fb0890d7dbb7519cfb2d93db
2021-02-11 14:01:45 -05:00
German Andryeyev fb142e7b2f SWDEV-271806 - Disable pinning for views.
Only parent object should pin system memory

Change-Id: Ic9a0a34e3aff2263501c564c3cebd33f4f7e2455
2021-02-11 13:05:58 -05:00
Vladislav Sytchenko ee1ec3258d SWDEV-252745 - Fix uninitialized variable
Change-Id: I9167dc3c85779771a44626f2f88eec4134c5a5ad
2021-02-11 10:26:38 -05:00
Chauncey Hui 91ba74f314 SWDEV-2 - Change OpenCL version number from 3254 to 3255 2021-02-11 03:00:07 -05:00
Rahul Garg df0b14ff47 SWDEV-271182 - Fix HdpMemFlushCntl and HdpRegFlushCntl
Change-Id: Ib690b475c2c4514c862ded5fc2bdd1049e27bd46
2021-02-10 11:40:35 -05:00
Chauncey Hui de52aa7e7c SWDEV-2 - Change OpenCL version number from 3253 to 3254 2021-02-10 03:00:06 -05:00
cjatin 5cfea2981b SWDEV-252745 - Adding precompiled headers to build option
Change-Id: Iab03e8a75215d80c91817ee0227d034fefa9dfad
2021-02-09 23:33:55 -05:00
German Andryeyev 6966d8098e SWDEV-269654 - Fix HIP stream busy query
- Avoid GPU wait on the marker submission and update the command
batch after HSA signal callback upon HSA barrier completion.

Change-Id: I5c1c97212aefc2ae4b99aa9e2a81627ee9a38c1c
2021-02-09 12:57:12 -05:00
Chauncey Hui 12e92b603e SWDEV-2 - Change OpenCL version number from 3252 to 3253 2021-02-09 03:00:06 -05:00
Vladislav Sytchenko 738fd1d24b SWDEV-232428 - Use std atomics
Change-Id: Ieca85738621d056e612faf4de1e2011a817b8965
2021-02-08 16:50:41 -05:00
German Andryeyev 47f8d55313 SWDEV-271888 - Remove an explanation link
Change-Id: I157adac7a248d50ee8db677c19be20c00fcd5ecc
2021-02-05 17:28:28 -05:00
Chauncey Hui c82cda6a7c SWDEV-2 - Change OpenCL version number from 3251 to 3252 2021-02-04 03:00:06 -05:00
Julia Jiang 699b61b5c1 SWDEV-271125 - update ROCclr readme
Change-Id: Icc3df1eca16c685d5ab1c74bebf37bbf169841c0
2021-02-02 17:20:46 -05:00
Chauncey Hui 20ab999164 SWDEV-2 - Change OpenCL version number from 3250 to 3251 2021-02-02 03:00:06 -05:00
Jason Tang 4791447370 SWDEV-269880 - Fix luxmark 3.1 segfaults
To workaround a bug in std::regex() implementation.

Change-Id: I4ea2b40e0b74ee1ee79d31283937e43d361d37c2
2021-02-01 16:12:36 -05:00
Alex Xie 5330679473 SWDEV-268914 - Linux, AMF, tests fail to scale images down
In pitch workaround, we need to copy the image to copy buffer
when application wants to read image to buffer. After this
patch, we unconditionally copy the image data to the copy buffer.

Change-Id: I71b0d19459542dfbb3ca51a2c8a3a81367fa2fb5
2021-02-01 15:22:15 -05:00
Joseph Greathouse 54d1d69c0a SWDEV-270773 - Fix workgroup calculation logic for GWS initialization
The existing workgroup calculation logic for GWS initialization is
incorrect. It tries to add together workgroups across dimensions,
leading to major under-count in 2D and 3D kernels. An (x,y,z) kernel
uses x * y * z blocks, not x + y + z.

In addition, the previous logic was incorrect for the case of launching
a single-threaded kernel. It calculated 0 workgroups, leading to
initializing GWS to -1.

Change-Id: I1bb20a0d5b6e0cc10ac55901c28d8f93aac61c09
2021-01-31 01:16:48 -06:00
Chauncey Hui b7f0fd0354 SWDEV-2 - Change OpenCL version number from 3249 to 3250 2021-01-29 03:00:06 -05:00
Saleel Kudchadker 629a2d8ef3 SWDEV-257787 - Add log for tracking copy signals
Change-Id: I713e8463916a85a634a1ec2309bbd46a11c461a8
2021-01-28 13:25:49 -05:00