-
Alongside `VFMA.F16`/`VFMS.F16`, AArch32 offers `VMLA.F16`/`VMLS.F16` instructions which performs multiply-add operation **with** intermediate rounding. Importantly, the vector-by-vector lane form (e.…
-
Because ARM devices are not yet part of the CI testing farm for Eclipse OMR (but we are working on it!), it is useful to at least cross-compile build the components without actually running any tests.…
-
### Description
When compiling cpuinfo for arm 64 platforms, conan requires the target `arch` to be set to `armv8`. Unfortunately, when this is passed to `cpuinfo` it treats `armv8` as an arm 32bit…
-
hello i have a modem binary that is aarch32 and I can load it in ghidra with your script but I cannot use it with Firmwire - I get an error message about not finding a loader for it. On the wiki it me…
-
I found both GCC12 and Clang11 call __aeabi_uldivmod when dividing a uint64_t by a constant on AArch32, and I found it's possible to emulate the AArch64 umulh instruction which enables division by a c…
-
I need to request a new machine:
- New machine operating system (e.g. linux/windows/macos/solaris/aix): Linux
- New machine architecture (e.g. x64/aarch32/arm32/ppc64/ppc64le/sparc): aarch64
- Pr…
-
Most notably for AArch32 but AArch64 also has brainfarts/spillage:
https://godbolt.org/z/GY1sG56qd
For comparison GCC's codegen has zero spillage even for AArch32 (with half as many registers)...
-
This architecture has been deprecated and obsolete for over a decade, and nobody should invest *any* time into maintaining packaging for it. The maintenance burden has become too large, and most Linux…
-
llvm 17.0.1 /llvm/lib/ExecutionEngine/JITLink/aarch32.cpp
```
168 template
169 bool checkRegister(const ThumbRelocation &R, HalfWords Reg) {
170 uint16_t Hi = R.Hi & FixupInfo::…
-
I target set by user is: rk3566
D adb path: /home/friva/.local/lib/python3.10/site-packages/rknn/3rdparty/platform-tools/adb/linux-x86_64/adb
I Get hardware info: target_platform = rk3566, os = Linu…