-
Enhancement
-
Resolution: Fixed
-
P4
-
20
-
b25
-
riscv
-
linux
| Issue | Fix Version | Assignee | Priority | Status | Resolution | Resolved In Build |
|---|---|---|---|---|---|---|
| JDK-8297599 | 19.0.2 | Vladimir Kempik | P4 | Resolved | Fixed | b07 |
| JDK-8311736 | 17.0.9 | Fei Yang | P4 | Resolved | Fixed | b01 |
Currently copy_memory stub on risc-v is doing 8-byte copy per loop at best
We can improve that to 32-bytes per loop ( doing it similar to memcpy in newlib) - four ld and four sd per loop
Also copy_memory produces some dead ( never executed) code when is_aligned is true. Lets eliminate that as well
We can improve that to 32-bytes per loop ( doing it similar to memcpy in newlib) - four ld and four sd per loop
Also copy_memory produces some dead ( never executed) code when is_aligned is true. Lets eliminate that as well
- backported by
-
JDK-8297599 RISC-V: improve performance of copy_memory stub
-
- Resolved
-
-
-
- Resolved
-
- links to
-
Commit
openjdk/jdk17u-dev/966fc82d
-
Commit
openjdk/jdk19u/7e340b3c
-
Commit
openjdk/jdk/bd57e213
-
Review
openjdk/jdk17u-dev/1427
-
Review
openjdk/jdk19u/77
-
Review
openjdk/jdk/11058
-
Review
openjdk/riscv-port-jdk17u/9
(4 links to)