diff --git a/src/ds/gx.c b/src/ds/gx.c
index 24984f69e..4430f1d13 100644
--- a/src/ds/gx.c
+++ b/src/ds/gx.c
@@ -137,7 +137,7 @@ static void _pullPipe(struct DSGX* gx) {
 }
 
 static void _updateClipMatrix(struct DSGX* gx) {
-	DSGXMtxMultiply(&gx->clipMatrix, &gx->projMatrix, &gx->posMatrix);
+	DSGXMtxMultiply(&gx->clipMatrix, &gx->posMatrix, &gx->projMatrix);
 }
 
 static int32_t _dotViewport(struct DSGXVertex* vertex, int32_t* col) {
diff --git a/src/ds/matrix.c b/src/ds/matrix.c
index 3b4b0de6b..d36b14fb9 100644
--- a/src/ds/matrix.c
+++ b/src/ds/matrix.c
@@ -35,22 +35,22 @@ void DSGXMtxIdentity(struct DSGXMatrix* mtx) {
 void DSGXMtxMultiply(struct DSGXMatrix* out, const struct DSGXMatrix* a, const struct DSGXMatrix* b) {
 	struct DSGXMatrix o;
 	// XXX: This is transposed because DS matrices are transposed
-	o.m[0] = _dot(&a->m[0], &b->m[0]);
-	o.m[1] = _dot(&a->m[1], &b->m[0]);
-	o.m[2] = _dot(&a->m[2], &b->m[0]);
-	o.m[3] = _dot(&a->m[3], &b->m[0]);
-	o.m[4] = _dot(&a->m[0], &b->m[4]);
-	o.m[5] = _dot(&a->m[1], &b->m[4]);
-	o.m[6] = _dot(&a->m[2], &b->m[4]);
-	o.m[7] = _dot(&a->m[3], &b->m[4]);
-	o.m[8] = _dot(&a->m[0], &b->m[8]);
-	o.m[9] = _dot(&a->m[1], &b->m[8]);
-	o.m[10] = _dot(&a->m[2], &b->m[8]);
-	o.m[11] = _dot(&a->m[3], &b->m[8]);
-	o.m[12] = _dot(&a->m[0], &b->m[12]);
-	o.m[13] = _dot(&a->m[1], &b->m[12]);
-	o.m[14] = _dot(&a->m[2], &b->m[12]);
-	o.m[15] = _dot(&a->m[3], &b->m[12]);
+	o.m[0] = _dot(&b->m[0], &a->m[0]);
+	o.m[1] = _dot(&b->m[1], &a->m[0]);
+	o.m[2] = _dot(&b->m[2], &a->m[0]);
+	o.m[3] = _dot(&b->m[3], &a->m[0]);
+	o.m[4] = _dot(&b->m[0], &a->m[4]);
+	o.m[5] = _dot(&b->m[1], &a->m[4]);
+	o.m[6] = _dot(&b->m[2], &a->m[4]);
+	o.m[7] = _dot(&b->m[3], &a->m[4]);
+	o.m[8] = _dot(&b->m[0], &a->m[8]);
+	o.m[9] = _dot(&b->m[1], &a->m[8]);
+	o.m[10] = _dot(&b->m[2], &a->m[8]);
+	o.m[11] = _dot(&b->m[3], &a->m[8]);
+	o.m[12] = _dot(&b->m[0], &a->m[12]);
+	o.m[13] = _dot(&b->m[1], &a->m[12]);
+	o.m[14] = _dot(&b->m[2], &a->m[12]);
+	o.m[15] = _dot(&b->m[3], &a->m[12]);
 	*out = o;
 }