steve robb
f029468598
Fixed up a lot of bool-taking container resize functions to take EAllowShrinking instead.
...
[CL 30729174 by steve robb in ue5-main branch]
2024-01-19 16:41:35 -05:00
rob krajcarski
3426e607f7
Fix EVVM implementations of external register handlers to properly handle when a register is unused
...
#rb Stu.McKenna
[CL 30296969 by rob krajcarski in ue5-main branch]
2023-12-13 12:51:07 -05:00
rob krajcarski
37c29f7116
Fix for stomp when optimizing a div + mad combo with the EVVM. We don't support the result of the div being the last operand of the mad, but we would end up allocating space for the registers, which would then be iterated over.
...
#rb Michael.Galetzka
[CL 29975506 by rob krajcarski in ue5-main branch]
2023-11-28 18:55:14 -05:00
rob krajcarski
ffc60bf148
More fixes for EVVM with compressed (fp16) channels. In particular this addresses issues with handling offsetting buffers when an InstanceOffset is in effect
...
#rb Stu.McKenna
[CL 29779140 by rob krajcarski in ue5-main branch]
2023-11-16 11:15:11 -05:00
stu mckenna
a2f11d1a5f
- Fix buffer underrun from EVVM
...
- If we output < 4 instance we are writing to memory before the allocation
- Data buffers are allocated 16 byte aligned and each component is aligned to 16 bytes so we can therefore just loop until all components are written
#rb rob.krajcarski
[CL 29443739 by stu mckenna in ue5-main branch]
2023-11-03 22:30:45 -04:00
marc audy
e3a65bdf57
Fix warning V501: There are identical sub-expressions
...
[CL 29377146 by marc audy in ue5-main branch]
2023-11-02 16:10:45 -04:00
rob krajcarski
5f52564b3c
Fix for incorrect output buffer type being assigned when converting a float to a half. Corrects crash/assert that was happening with the destruction VFX
...
#rb stu.mckenna
[CL 29221752 by rob krajcarski in ue5-main branch]
2023-10-28 20:14:47 -04:00
rob krajcarski
ae2c615999
Adding some more checks to investigate apparent crashes with the EVVM
...
#rb stu.mckenna
[CL 29189692 by rob krajcarski in ue5-main branch]
2023-10-27 14:00:04 -04:00
rob krajcarski
5adae429ec
Fix for merged gt/ge + select ops not being properly mirrored during VM optimization pass
...
#rb stu.mckenna
[CL 29142472 by rob krajcarski in ue5-main branch]
2023-10-26 13:46:05 -04:00
rob krajcarski
89efbc3c0f
Fix divi for the EVVM to treat the operands as signed integers
...
#rb stu.mckenna
[CL 29024830 by rob krajcarski in ue5-main branch]
2023-10-23 19:25:08 -04:00
shawn mcgrath
90c67d3d53
- EVVM ARM half float to match X64
...
#rb none
#trivial
[CL 28738538 by shawn mcgrath in ue5-main branch]
2023-10-12 21:03:13 -04:00
shawn mcgrath
9f4ee2f6db
- new VectorVM bug fixes for half float
...
#rb rob.krajcarski
[CL 28717827 by shawn mcgrath in ue5-main branch]
2023-10-12 13:39:49 -04:00
shawn mcgrath
4e139a6169
- EVVM codesize improvement. (just removed a single FORCEINLINE)
...
#rb none
#trivial
[CL 28630652 by shawn mcgrath in ue5-main branch]
2023-10-10 14:34:37 -04:00
shawn mcgrath
19d9a303fc
New VectorVM half float and cast instruction fixes.
...
#rb rob.krajcarski
[CL 28540332 by shawn mcgrath in ue5-main branch]
2023-10-06 10:49:41 -04:00
ionut matasaru
7f8829bd4c
Fixed usage of TRACE_CPUPROFILER_EVENT_SCOPE macros in various engine files.
...
* TRACE_CPUPROFILER_EVENT_SCOPE(ScopeName) --> to be used with a plain text as scope name (not a string!). If the parameter is a string ("abc" or TEXT("abc")), the quotes (and the TEXT prefix) will be included in the actual timer name.
* TRACE_CPUPROFILER_EVENT_SCOPE_STR("scope name") --> to be used with a static string (const ANSICHAR* or const TCHAR*); ex: when the scope name needs to include a space char
* TRACE_CPUPROFILER_EVENT_SCOPE_TEXT(*MyScopeName) --> to be used with a dynamic string (const ANSICHAR* or const TCHAR*)
See also the comments for these macros in ProfilingDebugging\CpuProfilerTrace.h.
#rb Catalin.Dragoiu
[CL 28337510 by ionut matasaru in ue5-main branch]
2023-09-29 04:05:39 -04:00
rob krajcarski
34a47d29bc
Fix deterministic output for EVVM byte code
...
#rb stu.mckenna
[CL 27535251 by rob krajcarski in ue5-main branch]
2023-08-31 16:00:24 -04:00
shawn mcgrath
d6be895e0f
#jira UE-193982
...
#rb none
VectorVM aligned store fix for Linux
[CL 27484467 by shawn mcgrath in ue5-main branch]
2023-08-30 10:40:47 -04:00
rob krajcarski
55c4555cb3
Fix for EVVM scripts not being tagged properly as including randoms if the script only contains merged random functions (random_add or random_2x)
...
#rb stu.mckenna
[CL 27353508 by rob krajcarski in ue5-main branch]
2023-08-24 15:20:21 -04:00
rob krajcarski
79b5d6242b
Collection of EVVM fixes
...
-Cleans up the checks of the constant buffers at runtime
-Removes the search for conflicts that was exclusive to exec_index & i2f and makes sure merging exec_index only considers temp registers
-Adds an EVVM assembly string representation in the Generated Code view
-fixes random_addi implementation that was sharing the buffer increment value between the random seed and the addition operand
-fixes a OOB access/stomp when dealing with newly identified unused ext function output operands
#jira UE-193006, UE-193112
#rb stu.mckenna
[CL 27182518 by rob krajcarski in ue5-main branch]
2023-08-17 15:34:02 -04:00
bryan sefcik
39a0a9b5bd
[Backout] - CL27078945
...
[FYI] bryan.sefcik
Original CL Desc
-----------------------------------------------------------------
Win ARM64 compile fix.
[FYI] shawn.mcgrath
#jira
#rnx
[CL 27079579 by bryan sefcik in ue5-main branch]
2023-08-14 14:40:36 -04:00
bryan sefcik
c375ac2461
Win ARM64 compile fix.
...
[FYI] shawn.mcgrath
#jira
#rnx
[CL 27079045 by bryan sefcik in ue5-main branch]
2023-08-14 14:29:03 -04:00
shawn mcgrath
d700e06855
compiler fix
...
[CL 26809460 by shawn mcgrath in ue5-main branch]
2023-08-03 10:05:39 -04:00
shawn mcgrath
fa57ee7b7f
fixing compiler error
...
[CL 26808955 by shawn mcgrath in ue5-main branch]
2023-08-03 09:40:37 -04:00
shawn mcgrath
8d361f2fcf
Experimental Vector VM improvements: faster execution speed, faster optimization speed, uses less runtime memory, added debugging features
...
[CL 26804446 by shawn mcgrath in ue5-main branch]
2023-08-03 07:16:11 -04:00
stu mckenna
fd6cae6b42
- Add VVM support for bit casting between floats and ints
...
#rb rob.krajcarski
#preflight 6466532e743f7c995bc67a0f
[CL 25548196 by stu mckenna in ue5-main branch]
2023-05-19 13:51:29 -04:00
