line masking using vector instructions

examples/demo/main/CMakeLists.txt

@@ -1,3 +1,4 @@
 set(app_sources "main.c")
 
+set(CMAKE_EXPORT_COMPILE_COMMANDS ON)
 idf_component_register(SRCS ${app_sources} REQUIRES epdiy)

src/diff.S

@@ -118,3 +118,37 @@ epd_interlace_4bpp_line_VE:
     //l8ui a10, a10, 0
 
     retw.n
+
+
+.global epd_apply_line_mask_VE
+.type   epd_apply_line_mask_VE,@function
+
+// void epd_apply_line_mask_VE(
+//     uint8_t *line,
+//     const uint8_t *mask,
+//     int mask_len
+// )
+epd_apply_line_mask_VE: 
+// line   - a2
+// mask   - a3
+// mask_len   - a4
+
+    entry	a1, 32
+
+    // divide by 16 for loop count
+    srli a4, a4, 4
+
+    // Instructions sometimes are in an unexpected order
+    // for best pipeline utilization
+    loopnez a4, .loop_end_mask
+
+        EE.VLD.128.IP q0, a2, 0
+        EE.VLD.128.IP q1, a3, 16
+
+        EE.ANDQ q0, q0, q1
+
+        EE.VST.128.IP q0, a2, 16
+
+.loop_end_mask:
+
+    retw.n

src/output_common/line_queue.c

@@ -5,6 +5,7 @@
 #include <esp_heap_caps.h>
 
 #include "line_queue.h"
+#include "render_method.h"
 
 static inline int ceil_div(int x, int y) { return x / y + (x % y != 0); }
 
@@ -65,9 +66,14 @@ void IRAM_ATTR lq_commit(LineQueue_t* queue) {
         atomic_fetch_add(&queue->current, 1);
     }
 
+#ifdef RENDER_METHOD_LCD
+    void epd_apply_line_mask_VE(uint8_t *line, const uint8_t *mask, int mask_len);
+    epd_apply_line_mask_VE(queue->bufs[current], queue->mask_buffer, queue->mask_buffer_len);
+#else
     for (int i=0; i < queue->mask_buffer_len / 4; i++) {
         ((uint32_t*)(queue->bufs[current]))[i] &= ((uint32_t*)(queue->mask_buffer))[i];
     }
+#endif
 }
 
 int IRAM_ATTR lq_read(LineQueue_t* queue, uint8_t* dst) {