diff --git a/workspace/all/common/api.c b/workspace/all/common/api.c
index 19f3f5439..35be15bc4 100644
--- a/workspace/all/common/api.c
+++ b/workspace/all/common/api.c
@@ -3858,6 +3858,22 @@ int PAD_tappedSelect(uint32_t now)
 	return PAD_tappedBtn(BTN_SELECT, now);
 }
 
+///////////////////////////////
+
+// the RLIMIT_STACK default (8MB per thread) wastes address space
+// workaround for libretro/gpsp#248 and NextUI#814
+#define THREAD_STACK_SIZE (1024 * 1024)
+
+static int spawn_thread(pthread_t *pt, void *(*fn)(void *), void *arg)
+{
+	pthread_attr_t attr;
+	pthread_attr_init(&attr);
+	pthread_attr_setstacksize(&attr, THREAD_STACK_SIZE);
+	int rc = pthread_create(pt, &attr, fn, arg);
+	pthread_attr_destroy(&attr);
+	return rc;
+}
+
 ///////////////////////////////
 static struct VIB_Context
 {
@@ -3891,7 +3907,7 @@ static void *VIB_thread(void *arg)
 void VIB_init(void)
 {
 	vib.queued_strength = vib.strength = 0;
-	pthread_create(&vib.pt, NULL, &VIB_thread, NULL);
+	spawn_thread(&vib.pt, &VIB_thread, NULL);
 	vib.initialized = 1;
 }
 void VIB_quit(void)
@@ -4026,7 +4042,7 @@ void PWR_init(void)
 
 	PWR_updateBatteryStatus();
 
-	pthread_create(&pwr.battery_pt, NULL, &PWR_monitorBattery, &pwr);
+	spawn_thread(&pwr.battery_pt, &PWR_monitorBattery, &pwr);
 	LOG_info("PWR_init complete\n");
 }
 void PWR_quit(void)
diff --git a/workspace/all/minarch/minarch.c b/workspace/all/minarch/minarch.c
index b3e156eda..fd6afb977 100644
--- a/workspace/all/minarch/minarch.c
+++ b/workspace/all/minarch/minarch.c
@@ -1,5 +1,8 @@
 #include <stdlib.h>
 #include <msettings.h>
+#ifdef __GLIBC__
+#include <malloc.h>
+#endif
 
 #include <SDL2/SDL_image.h>
 
@@ -139,6 +142,14 @@ int main(int argc , char* argv[]) {
 	if(argc < 2)
 		return EXIT_FAILURE;
 
+	// keep malloc arenas and thread stacks out of the address range dynarec cores
+	// need for their JIT cache (must happen before any thread is created)
+	// workaround for libretro/gpsp#248 and NextUI#814
+#ifdef __GLIBC__
+	mallopt(M_ARENA_MAX, 2); // glibc extension
+#endif
+	SDL_SetHint(SDL_HINT_THREAD_STACK_SIZE, "1048576");
+
 	PWR_setCPUSpeed(CPU_SPEED_PERFORMANCE); // start in performance mode for fast loading
 	PWR_pinToCores(CPU_CORE_PERFORMANCE); // thread affinity