diff --git a/src/secp256k1/.travis.yml b/src/secp256k1/.travis.yml
--- a/src/secp256k1/.travis.yml
+++ b/src/secp256k1/.travis.yml
@@ -41,6 +41,8 @@
     - OPENSSL_TESTS=auto
     - MULTISET=no
     - CTIMETEST=yes
+    - BENCH=yes
+    - SECP256K1_BENCH_ITERS=2
   jobs:
     - SCALAR=32bit    RECOVERY=yes
     - SCALAR=32bit    FIELD=32bit       ECDH=yes  EXPERIMENTAL=yes  MULTISET=yes
@@ -54,10 +56,10 @@
     - BIGNUM=no
     - BIGNUM=no       ENDOMORPHISM=yes  RECOVERY=yes  EXPERIMENTAL=yes  MULTISET=yes
     - BIGNUM=no       STATICPRECOMPUTATION=no
-    - AUTOTOOLS_TARGET=distcheck       CMAKE_TARGET=install    CTIMETEST=
+    - AUTOTOOLS_TARGET=distcheck       CMAKE_TARGET=install    CTIMETEST= BENCH=
     - AUTOTOOLS_EXTRA_FLAGS=CPPFLAGS=-DDETERMINISTIC CMAKE_EXTRA_FLAGS=-DCMAKE_C_FLAGS=-DDETERMINISTIC
     - AUTOTOOLS_EXTRA_FLAGS=CFLAGS=-O0               CMAKE_EXTRA_FLAGS=-DCMAKE_BUILD_TYPE=Debug
-    - AUTOTOOLS_TARGET=check-java      CMAKE_TARGET=check-secp256k1-java JNI=yes ECDH=yes EXPERIMENTAL=yes  CTIMETEST=
+    - AUTOTOOLS_TARGET=check-java      CMAKE_TARGET=check-secp256k1-java JNI=yes ECDH=yes EXPERIMENTAL=yes CTIMETEST= BENCH=
     - ECMULTGENPRECISION=2
     - ECMULTGENPRECISION=8
     - SCHNORR=no
diff --git a/src/secp256k1/src/bench.h b/src/secp256k1/src/bench.h
--- a/src/secp256k1/src/bench.h
+++ b/src/secp256k1/src/bench.h
@@ -73,7 +73,7 @@
     printf("%s", &buffer[ptr]);
 }
 
-void run_benchmark(char *name, void (*benchmark)(void*), void (*setup)(void*), void (*teardown)(void*), void* data, int count, int iter) {
+void run_benchmark(char *name, void (*benchmark)(void*, int), void (*setup)(void*), void (*teardown)(void*, int), void* data, int count, int iter) {
     int i;
     int64_t min = INT64_MAX;
     int64_t sum = 0;
@@ -84,10 +84,10 @@
             setup(data);
         }
         begin = gettime_i64();
-        benchmark(data);
+        benchmark(data, iter);
         total = gettime_i64() - begin;
         if (teardown != NULL) {
-            teardown(data);
+            teardown(data, iter);
         }
         if (total < min) {
             min = total;
@@ -121,4 +121,13 @@
     return 0;
 }
 
+int get_iters(int default_iters) {
+    char* env = getenv("SECP256K1_BENCH_ITERS");
+    if (env) {
+        return strtol(env, NULL, 0);
+    } else {
+        return default_iters;
+    }
+}
+
 #endif /* SECP256K1_BENCH_H */
diff --git a/src/secp256k1/src/bench_ecdh.c b/src/secp256k1/src/bench_ecdh.c
--- a/src/secp256k1/src/bench_ecdh.c
+++ b/src/secp256k1/src/bench_ecdh.c
@@ -28,20 +28,18 @@
         0xa2, 0xba, 0xd1, 0x84, 0xf8, 0x83, 0xc6, 0x9f
     };
 
-    /* create a context with no capabilities */
-    data->ctx = secp256k1_context_create(SECP256K1_FLAGS_TYPE_CONTEXT);
     for (i = 0; i < 32; i++) {
         data->scalar[i] = i + 1;
     }
     CHECK(secp256k1_ec_pubkey_parse(data->ctx, &data->point, point, sizeof(point)) == 1);
 }
 
-static void bench_ecdh(void* arg) {
+static void bench_ecdh(void* arg, int iters) {
     int i;
     unsigned char res[32];
     bench_ecdh_data *data = (bench_ecdh_data*)arg;
 
-    for (i = 0; i < 20000; i++) {
+    for (i = 0; i < iters; i++) {
         CHECK(secp256k1_ecdh(data->ctx, res, &data->point, data->scalar, NULL, NULL) == 1);
     }
 }
@@ -49,6 +47,13 @@
 int main(void) {
     bench_ecdh_data data;
 
-    run_benchmark("ecdh", bench_ecdh, bench_ecdh_setup, NULL, &data, 10, 20000);
+    int iters = get_iters(20000);
+
+    /* create a context with no capabilities */
+    data.ctx = secp256k1_context_create(SECP256K1_FLAGS_TYPE_CONTEXT);
+
+    run_benchmark("ecdh", bench_ecdh, bench_ecdh_setup, NULL, &data, 10, iters);
+
+    secp256k1_context_destroy(data.ctx);
     return 0;
 }
diff --git a/src/secp256k1/src/bench_ecmult.c b/src/secp256k1/src/bench_ecmult.c
--- a/src/secp256k1/src/bench_ecmult.c
+++ b/src/secp256k1/src/bench_ecmult.c
@@ -18,7 +18,6 @@
 #include "secp256k1.c"
 
 #define POINTS 32768
-#define ITERS 10000
 
 typedef struct {
     /* Setup once in advance */
@@ -55,13 +54,13 @@
     return 1;
 }
 
-static void bench_ecmult(void* arg) {
+static void bench_ecmult(void* arg, int iters) {
     bench_data* data = (bench_data*)arg;
 
-    size_t count = data->count;
     int includes_g = data->includes_g;
-    size_t iters = 1 + ITERS / count;
-    size_t iter;
+    int iter;
+    int count = data->count;
+    iters = iters / data->count;
 
     for (iter = 0; iter < iters; ++iter) {
         data->ecmult_multi(&data->ctx->error_callback, &data->ctx->ecmult_ctx, data->scratch, &data->output[iter], data->includes_g ? &data->scalars[data->offset1] : NULL, bench_callback, arg, count - includes_g);
@@ -76,10 +75,10 @@
     data->offset2 = (data->count * 0x7f6f537b + 0x6a1a8f49) % POINTS;
 }
 
-static void bench_ecmult_teardown(void* arg) {
+static void bench_ecmult_teardown(void* arg, int iters) {
     bench_data* data = (bench_data*)arg;
-    size_t iters = 1 + ITERS / data->count;
-    size_t iter;
+    int iter;
+    iters = iters / data->count;
     /* Verify the results in teardown, to avoid doing comparisons while benchmarking. */
     for (iter = 0; iter < iters; ++iter) {
         secp256k1_gej tmp;
@@ -104,10 +103,10 @@
     CHECK(!overflow);
 }
 
-static void run_test(bench_data* data, size_t count, int includes_g) {
+static void run_test(bench_data* data, size_t count, int includes_g, int num_iters) {
     char str[32];
     static const secp256k1_scalar zero = SECP256K1_SCALAR_CONST(0, 0, 0, 0, 0, 0, 0, 0);
-    size_t iters = 1 + ITERS / count;
+    size_t iters = 1 + num_iters / count;
     size_t iter;
 
     data->count = count;
@@ -130,7 +129,7 @@
 
     /* Run the benchmark. */
     sprintf(str, includes_g ? "ecmult_%ig" : "ecmult_%i", (int)count);
-    run_benchmark(str, bench_ecmult, bench_ecmult_setup, bench_ecmult_teardown, data, 10, count * (1 + ITERS / count));
+    run_benchmark(str, bench_ecmult, bench_ecmult_setup, bench_ecmult_teardown, data, 10, count * iters);
 }
 
 int main(int argc, char **argv) {
@@ -139,6 +138,8 @@
     secp256k1_gej* pubkeys_gej;
     size_t scratch_size;
 
+    int iters = get_iters(10000);
+
     data.ctx = secp256k1_context_create(SECP256K1_CONTEXT_SIGN | SECP256K1_CONTEXT_VERIFY);
     scratch_size = secp256k1_strauss_scratch_size(POINTS) + STRAUSS_SCRATCH_OBJECTS*16;
     data.scratch = secp256k1_scratch_space_create(data.ctx, scratch_size);
@@ -167,8 +168,8 @@
     data.scalars = malloc(sizeof(secp256k1_scalar) * POINTS);
     data.seckeys = malloc(sizeof(secp256k1_scalar) * POINTS);
     data.pubkeys = malloc(sizeof(secp256k1_ge) * POINTS);
-    data.expected_output = malloc(sizeof(secp256k1_gej) * (ITERS + 1));
-    data.output = malloc(sizeof(secp256k1_gej) * (ITERS + 1));
+    data.expected_output = malloc(sizeof(secp256k1_gej) * (iters + 1));
+    data.output = malloc(sizeof(secp256k1_gej) * (iters + 1));
 
     /* Generate a set of scalars, and private/public keypairs. */
     pubkeys_gej = malloc(sizeof(secp256k1_gej) * POINTS);
@@ -185,14 +186,20 @@
     free(pubkeys_gej);
 
     for (i = 1; i <= 8; ++i) {
-        run_test(&data, i, 1);
+        run_test(&data, i, 1, iters);
     }
 
-    for (p = 0; p <= 11; ++p) {
-        for (i = 9; i <= 16; ++i) {
-            run_test(&data, i << p, 1);
+    /* This is disabled with low count of iterations because the loop runs 77 times even with iters=1
+    * and the higher it goes the longer the computation takes(more points)
+    * So we don't run this benchmark with low iterations to prevent slow down */
+     if (iters > 2) {
+        for (p = 0; p <= 11; ++p) {
+            for (i = 9; i <= 16; ++i) {
+                run_test(&data, i << p, 1, iters);
+            }
         }
     }
+
     if (data.scratch != NULL) {
         secp256k1_scratch_space_destroy(data.ctx, data.scratch);
     }
diff --git a/src/secp256k1/src/bench_internal.c b/src/secp256k1/src/bench_internal.c
--- a/src/secp256k1/src/bench_internal.c
+++ b/src/secp256k1/src/bench_internal.c
@@ -56,271 +56,271 @@
     memcpy(data->data + 32, init_y, 32);
 }
 
-void bench_scalar_add(void* arg) {
+void bench_scalar_add(void* arg, int iters) {
     int i, j = 0;
     bench_inv *data = (bench_inv*)arg;
 
-    for (i = 0; i < 2000000; i++) {
+    for (i = 0; i < iters; i++) {
         j += secp256k1_scalar_add(&data->scalar_x, &data->scalar_x, &data->scalar_y);
     }
-    CHECK(j <= 2000000);
+    CHECK(j <= iters);
 }
 
-void bench_scalar_negate(void* arg) {
+void bench_scalar_negate(void* arg, int iters) {
     int i;
     bench_inv *data = (bench_inv*)arg;
 
-    for (i = 0; i < 2000000; i++) {
+    for (i = 0; i < iters; i++) {
         secp256k1_scalar_negate(&data->scalar_x, &data->scalar_x);
     }
 }
 
-void bench_scalar_sqr(void* arg) {
+void bench_scalar_sqr(void* arg, int iters) {
     int i;
     bench_inv *data = (bench_inv*)arg;
 
-    for (i = 0; i < 200000; i++) {
+    for (i = 0; i < iters; i++) {
         secp256k1_scalar_sqr(&data->scalar_x, &data->scalar_x);
     }
 }
 
-void bench_scalar_mul(void* arg) {
+void bench_scalar_mul(void* arg, int iters) {
     int i;
     bench_inv *data = (bench_inv*)arg;
 
-    for (i = 0; i < 200000; i++) {
+    for (i = 0; i < iters; i++) {
         secp256k1_scalar_mul(&data->scalar_x, &data->scalar_x, &data->scalar_y);
     }
 }
 
 #ifdef USE_ENDOMORPHISM
-void bench_scalar_split(void* arg) {
+void bench_scalar_split(void* arg, int iters) {
     int i, j = 0;
     bench_inv *data = (bench_inv*)arg;
 
-    for (i = 0; i < 20000; i++) {
+    for (i = 0; i < iters; i++) {
         secp256k1_scalar_split_lambda(&data->scalar_x, &data->scalar_y, &data->scalar_x);
         j += secp256k1_scalar_add(&data->scalar_x, &data->scalar_x, &data->scalar_y);
     }
-    CHECK(j <= 20000);
+    CHECK(j <= iters);
 }
 #endif
 
-void bench_scalar_inverse(void* arg) {
+void bench_scalar_inverse(void* arg, int iters) {
     int i, j = 0;
     bench_inv *data = (bench_inv*)arg;
 
-    for (i = 0; i < 2000; i++) {
+    for (i = 0; i < iters; i++) {
         secp256k1_scalar_inverse(&data->scalar_x, &data->scalar_x);
         j += secp256k1_scalar_add(&data->scalar_x, &data->scalar_x, &data->scalar_y);
     }
-    CHECK(j <= 2000);
+    CHECK(j <= iters);
 }
 
-void bench_scalar_inverse_var(void* arg) {
+void bench_scalar_inverse_var(void* arg, int iters) {
     int i, j = 0;
     bench_inv *data = (bench_inv*)arg;
 
-    for (i = 0; i < 2000; i++) {
+    for (i = 0; i < iters; i++) {
         secp256k1_scalar_inverse_var(&data->scalar_x, &data->scalar_x);
         j += secp256k1_scalar_add(&data->scalar_x, &data->scalar_x, &data->scalar_y);
     }
-    CHECK(j <= 2000);
+    CHECK(j <= iters);
 }
 
-void bench_field_normalize(void* arg) {
+void bench_field_normalize(void* arg, int iters) {
     int i;
     bench_inv *data = (bench_inv*)arg;
 
-    for (i = 0; i < 2000000; i++) {
+    for (i = 0; i < iters; i++) {
         secp256k1_fe_normalize(&data->fe_x);
     }
 }
 
-void bench_field_normalize_weak(void* arg) {
+void bench_field_normalize_weak(void* arg, int iters) {
     int i;
     bench_inv *data = (bench_inv*)arg;
 
-    for (i = 0; i < 2000000; i++) {
+    for (i = 0; i < iters; i++) {
         secp256k1_fe_normalize_weak(&data->fe_x);
     }
 }
 
-void bench_field_mul(void* arg) {
+void bench_field_mul(void* arg, int iters) {
     int i;
     bench_inv *data = (bench_inv*)arg;
 
-    for (i = 0; i < 200000; i++) {
+    for (i = 0; i < iters; i++) {
         secp256k1_fe_mul(&data->fe_x, &data->fe_x, &data->fe_y);
     }
 }
 
-void bench_field_sqr(void* arg) {
+void bench_field_sqr(void* arg, int iters) {
     int i;
     bench_inv *data = (bench_inv*)arg;
 
-    for (i = 0; i < 200000; i++) {
+    for (i = 0; i < iters; i++) {
         secp256k1_fe_sqr(&data->fe_x, &data->fe_x);
     }
 }
 
-void bench_field_inverse(void* arg) {
+void bench_field_inverse(void* arg, int iters) {
     int i;
     bench_inv *data = (bench_inv*)arg;
 
-    for (i = 0; i < 20000; i++) {
+    for (i = 0; i < iters; i++) {
         secp256k1_fe_inv(&data->fe_x, &data->fe_x);
         secp256k1_fe_add(&data->fe_x, &data->fe_y);
     }
 }
 
-void bench_field_inverse_var(void* arg) {
+void bench_field_inverse_var(void* arg, int iters) {
     int i;
     bench_inv *data = (bench_inv*)arg;
 
-    for (i = 0; i < 20000; i++) {
+    for (i = 0; i < iters; i++) {
         secp256k1_fe_inv_var(&data->fe_x, &data->fe_x);
         secp256k1_fe_add(&data->fe_x, &data->fe_y);
     }
 }
 
-void bench_field_sqrt(void* arg) {
+void bench_field_sqrt(void* arg, int iters) {
     int i, j = 0;
     bench_inv *data = (bench_inv*)arg;
     secp256k1_fe t;
 
-    for (i = 0; i < 20000; i++) {
+    for (i = 0; i < iters; i++) {
         t = data->fe_x;
         j += secp256k1_fe_sqrt(&data->fe_x, &t);
         secp256k1_fe_add(&data->fe_x, &data->fe_y);
     }
-    CHECK(j <= 20000);
+    CHECK(j <= iters);
 }
 
-void bench_group_double_var(void* arg) {
+void bench_group_double_var(void* arg, int iters) {
     int i;
     bench_inv *data = (bench_inv*)arg;
 
-    for (i = 0; i < 200000; i++) {
+    for (i = 0; i < iters; i++) {
         secp256k1_gej_double_var(&data->gej_x, &data->gej_x, NULL);
     }
 }
 
-void bench_group_add_var(void* arg) {
+void bench_group_add_var(void* arg, int iters) {
     int i;
     bench_inv *data = (bench_inv*)arg;
 
-    for (i = 0; i < 200000; i++) {
+    for (i = 0; i < iters; i++) {
         secp256k1_gej_add_var(&data->gej_x, &data->gej_x, &data->gej_y, NULL);
     }
 }
 
-void bench_group_add_affine(void* arg) {
+void bench_group_add_affine(void* arg, int iters) {
     int i;
     bench_inv *data = (bench_inv*)arg;
 
-    for (i = 0; i < 200000; i++) {
+    for (i = 0; i < iters; i++) {
         secp256k1_gej_add_ge(&data->gej_x, &data->gej_x, &data->ge_y);
     }
 }
 
-void bench_group_add_affine_var(void* arg) {
+void bench_group_add_affine_var(void* arg, int iters) {
     int i;
     bench_inv *data = (bench_inv*)arg;
 
-    for (i = 0; i < 200000; i++) {
+    for (i = 0; i < iters; i++) {
         secp256k1_gej_add_ge_var(&data->gej_x, &data->gej_x, &data->ge_y, NULL);
     }
 }
 
-void bench_group_jacobi_var(void* arg) {
+void bench_group_jacobi_var(void* arg, int iters) {
     int i, j = 0;
     bench_inv *data = (bench_inv*)arg;
 
-    for (i = 0; i < 20000; i++) {
+    for (i = 0; i < iters; i++) {
         j += secp256k1_gej_has_quad_y_var(&data->gej_x);
     }
-    CHECK(j == 20000);
+    CHECK(j == iters);
 }
 
-void bench_ecmult_wnaf(void* arg) {
+void bench_ecmult_wnaf(void* arg, int iters) {
     int i, bits = 0, overflow = 0;
     bench_inv *data = (bench_inv*)arg;
 
-    for (i = 0; i < 20000; i++) {
+    for (i = 0; i < iters; i++) {
         bits += secp256k1_ecmult_wnaf(data->wnaf, 256, &data->scalar_x, WINDOW_A);
         overflow += secp256k1_scalar_add(&data->scalar_x, &data->scalar_x, &data->scalar_y);
     }
     CHECK(overflow >= 0);
-    CHECK(bits <= 256*20000);
+    CHECK(bits <= 256*iters);
 }
 
-void bench_wnaf_const(void* arg) {
+void bench_wnaf_const(void* arg, int iters) {
     int i, bits = 0, overflow = 0;
     bench_inv *data = (bench_inv*)arg;
 
-    for (i = 0; i < 20000; i++) {
+    for (i = 0; i < iters; i++) {
         bits += secp256k1_wnaf_const(data->wnaf, &data->scalar_x, WINDOW_A, 256);
         overflow += secp256k1_scalar_add(&data->scalar_x, &data->scalar_x, &data->scalar_y);
     }
     CHECK(overflow >= 0);
-    CHECK(bits <= 256*20000);
+    CHECK(bits <= 256*iters);
 }
 
 
-void bench_sha256(void* arg) {
+void bench_sha256(void* arg, int iters) {
     int i;
     bench_inv *data = (bench_inv*)arg;
     secp256k1_sha256 sha;
 
-    for (i = 0; i < 20000; i++) {
+    for (i = 0; i < iters; i++) {
         secp256k1_sha256_initialize(&sha);
         secp256k1_sha256_write(&sha, data->data, 32);
         secp256k1_sha256_finalize(&sha, data->data);
     }
 }
 
-void bench_hmac_sha256(void* arg) {
+void bench_hmac_sha256(void* arg, int iters) {
     int i;
     bench_inv *data = (bench_inv*)arg;
     secp256k1_hmac_sha256 hmac;
 
-    for (i = 0; i < 20000; i++) {
+    for (i = 0; i < iters; i++) {
         secp256k1_hmac_sha256_initialize(&hmac, data->data, 32);
         secp256k1_hmac_sha256_write(&hmac, data->data, 32);
         secp256k1_hmac_sha256_finalize(&hmac, data->data);
     }
 }
 
-void bench_rfc6979_hmac_sha256(void* arg) {
+void bench_rfc6979_hmac_sha256(void* arg, int iters) {
     int i;
     bench_inv *data = (bench_inv*)arg;
     secp256k1_rfc6979_hmac_sha256 rng;
 
-    for (i = 0; i < 20000; i++) {
+    for (i = 0; i < iters; i++) {
         secp256k1_rfc6979_hmac_sha256_initialize(&rng, data->data, 64);
         secp256k1_rfc6979_hmac_sha256_generate(&rng, data->data, 32);
     }
 }
 
-void bench_context_verify(void* arg) {
+void bench_context_verify(void* arg, int iters) {
     int i;
     (void)arg;
-    for (i = 0; i < 20; i++) {
+    for (i = 0; i < iters; i++) {
         secp256k1_context_destroy(secp256k1_context_create(SECP256K1_CONTEXT_VERIFY));
     }
 }
 
-void bench_context_sign(void* arg) {
+void bench_context_sign(void* arg, int iters) {
     int i;
     (void)arg;
-    for (i = 0; i < 200; i++) {
+    for (i = 0; i < iters; i++) {
         secp256k1_context_destroy(secp256k1_context_create(SECP256K1_CONTEXT_SIGN));
     }
 }
 
 #ifndef USE_NUM_NONE
-void bench_num_jacobi(void* arg) {
+void bench_num_jacobi(void* arg, int iters) {
     int i, j = 0;
     bench_inv *data = (bench_inv*)arg;
     secp256k1_num nx, norder;
@@ -329,51 +329,53 @@
     secp256k1_scalar_order_get_num(&norder);
     secp256k1_scalar_get_num(&norder, &data->scalar_y);
 
-    for (i = 0; i < 200000; i++) {
+    for (i = 0; i < iters; i++) {
         j += secp256k1_num_jacobi(&nx, &norder);
     }
-    CHECK(j <= 200000);
+    CHECK(j <= iters);
 }
 #endif
 
 int main(int argc, char **argv) {
     bench_inv data;
-    if (have_flag(argc, argv, "scalar") || have_flag(argc, argv, "add")) run_benchmark("scalar_add", bench_scalar_add, bench_setup, NULL, &data, 10, 2000000);
-    if (have_flag(argc, argv, "scalar") || have_flag(argc, argv, "negate")) run_benchmark("scalar_negate", bench_scalar_negate, bench_setup, NULL, &data, 10, 2000000);
-    if (have_flag(argc, argv, "scalar") || have_flag(argc, argv, "sqr")) run_benchmark("scalar_sqr", bench_scalar_sqr, bench_setup, NULL, &data, 10, 200000);
-    if (have_flag(argc, argv, "scalar") || have_flag(argc, argv, "mul")) run_benchmark("scalar_mul", bench_scalar_mul, bench_setup, NULL, &data, 10, 200000);
+    int iters = get_iters(20000);
+
+    if (have_flag(argc, argv, "scalar") || have_flag(argc, argv, "add")) run_benchmark("scalar_add", bench_scalar_add, bench_setup, NULL, &data, 10, iters*100);
+    if (have_flag(argc, argv, "scalar") || have_flag(argc, argv, "negate")) run_benchmark("scalar_negate", bench_scalar_negate, bench_setup, NULL, &data, 10, iters*100);
+    if (have_flag(argc, argv, "scalar") || have_flag(argc, argv, "sqr")) run_benchmark("scalar_sqr", bench_scalar_sqr, bench_setup, NULL, &data, 10, iters*10);
+    if (have_flag(argc, argv, "scalar") || have_flag(argc, argv, "mul")) run_benchmark("scalar_mul", bench_scalar_mul, bench_setup, NULL, &data, 10, iters*10);
 #ifdef USE_ENDOMORPHISM
-    if (have_flag(argc, argv, "scalar") || have_flag(argc, argv, "split")) run_benchmark("scalar_split", bench_scalar_split, bench_setup, NULL, &data, 10, 20000);
+    if (have_flag(argc, argv, "scalar") || have_flag(argc, argv, "split")) run_benchmark("scalar_split", bench_scalar_split, bench_setup, NULL, &data, 10, iters);
 #endif
     if (have_flag(argc, argv, "scalar") || have_flag(argc, argv, "inverse")) run_benchmark("scalar_inverse", bench_scalar_inverse, bench_setup, NULL, &data, 10, 2000);
     if (have_flag(argc, argv, "scalar") || have_flag(argc, argv, "inverse")) run_benchmark("scalar_inverse_var", bench_scalar_inverse_var, bench_setup, NULL, &data, 10, 2000);
 
-    if (have_flag(argc, argv, "field") || have_flag(argc, argv, "normalize")) run_benchmark("field_normalize", bench_field_normalize, bench_setup, NULL, &data, 10, 2000000);
-    if (have_flag(argc, argv, "field") || have_flag(argc, argv, "normalize")) run_benchmark("field_normalize_weak", bench_field_normalize_weak, bench_setup, NULL, &data, 10, 2000000);
-    if (have_flag(argc, argv, "field") || have_flag(argc, argv, "sqr")) run_benchmark("field_sqr", bench_field_sqr, bench_setup, NULL, &data, 10, 200000);
-    if (have_flag(argc, argv, "field") || have_flag(argc, argv, "mul")) run_benchmark("field_mul", bench_field_mul, bench_setup, NULL, &data, 10, 200000);
-    if (have_flag(argc, argv, "field") || have_flag(argc, argv, "inverse")) run_benchmark("field_inverse", bench_field_inverse, bench_setup, NULL, &data, 10, 20000);
-    if (have_flag(argc, argv, "field") || have_flag(argc, argv, "inverse")) run_benchmark("field_inverse_var", bench_field_inverse_var, bench_setup, NULL, &data, 10, 20000);
-    if (have_flag(argc, argv, "field") || have_flag(argc, argv, "sqrt")) run_benchmark("field_sqrt", bench_field_sqrt, bench_setup, NULL, &data, 10, 20000);
+    if (have_flag(argc, argv, "field") || have_flag(argc, argv, "normalize")) run_benchmark("field_normalize", bench_field_normalize, bench_setup, NULL, &data, 10, iters*100);
+    if (have_flag(argc, argv, "field") || have_flag(argc, argv, "normalize")) run_benchmark("field_normalize_weak", bench_field_normalize_weak, bench_setup, NULL, &data, 10, iters*100);
+    if (have_flag(argc, argv, "field") || have_flag(argc, argv, "sqr")) run_benchmark("field_sqr", bench_field_sqr, bench_setup, NULL, &data, 10, iters*10);
+    if (have_flag(argc, argv, "field") || have_flag(argc, argv, "mul")) run_benchmark("field_mul", bench_field_mul, bench_setup, NULL, &data, 10, iters*10);
+    if (have_flag(argc, argv, "field") || have_flag(argc, argv, "inverse")) run_benchmark("field_inverse", bench_field_inverse, bench_setup, NULL, &data, 10, iters);
+    if (have_flag(argc, argv, "field") || have_flag(argc, argv, "inverse")) run_benchmark("field_inverse_var", bench_field_inverse_var, bench_setup, NULL, &data, 10, iters);
+    if (have_flag(argc, argv, "field") || have_flag(argc, argv, "sqrt")) run_benchmark("field_sqrt", bench_field_sqrt, bench_setup, NULL, &data, 10, iters);
 
-    if (have_flag(argc, argv, "group") || have_flag(argc, argv, "double")) run_benchmark("group_double_var", bench_group_double_var, bench_setup, NULL, &data, 10, 200000);
-    if (have_flag(argc, argv, "group") || have_flag(argc, argv, "add")) run_benchmark("group_add_var", bench_group_add_var, bench_setup, NULL, &data, 10, 200000);
-    if (have_flag(argc, argv, "group") || have_flag(argc, argv, "add")) run_benchmark("group_add_affine", bench_group_add_affine, bench_setup, NULL, &data, 10, 200000);
-    if (have_flag(argc, argv, "group") || have_flag(argc, argv, "add")) run_benchmark("group_add_affine_var", bench_group_add_affine_var, bench_setup, NULL, &data, 10, 200000);
-    if (have_flag(argc, argv, "group") || have_flag(argc, argv, "jacobi")) run_benchmark("group_jacobi_var", bench_group_jacobi_var, bench_setup, NULL, &data, 10, 20000);
+    if (have_flag(argc, argv, "group") || have_flag(argc, argv, "double")) run_benchmark("group_double_var", bench_group_double_var, bench_setup, NULL, &data, 10, iters*10);
+    if (have_flag(argc, argv, "group") || have_flag(argc, argv, "add")) run_benchmark("group_add_var", bench_group_add_var, bench_setup, NULL, &data, 10, iters*10);
+    if (have_flag(argc, argv, "group") || have_flag(argc, argv, "add")) run_benchmark("group_add_affine", bench_group_add_affine, bench_setup, NULL, &data, 10, iters*10);
+    if (have_flag(argc, argv, "group") || have_flag(argc, argv, "add")) run_benchmark("group_add_affine_var", bench_group_add_affine_var, bench_setup, NULL, &data, 10, iters*10);
+    if (have_flag(argc, argv, "group") || have_flag(argc, argv, "jacobi")) run_benchmark("group_jacobi_var", bench_group_jacobi_var, bench_setup, NULL, &data, 10, iters);
 
-    if (have_flag(argc, argv, "ecmult") || have_flag(argc, argv, "wnaf")) run_benchmark("wnaf_const", bench_wnaf_const, bench_setup, NULL, &data, 10, 20000);
-    if (have_flag(argc, argv, "ecmult") || have_flag(argc, argv, "wnaf")) run_benchmark("ecmult_wnaf", bench_ecmult_wnaf, bench_setup, NULL, &data, 10, 20000);
+    if (have_flag(argc, argv, "ecmult") || have_flag(argc, argv, "wnaf")) run_benchmark("wnaf_const", bench_wnaf_const, bench_setup, NULL, &data, 10, iters);
+    if (have_flag(argc, argv, "ecmult") || have_flag(argc, argv, "wnaf")) run_benchmark("ecmult_wnaf", bench_ecmult_wnaf, bench_setup, NULL, &data, 10, iters);
 
-    if (have_flag(argc, argv, "hash") || have_flag(argc, argv, "sha256")) run_benchmark("hash_sha256", bench_sha256, bench_setup, NULL, &data, 10, 20000);
-    if (have_flag(argc, argv, "hash") || have_flag(argc, argv, "hmac")) run_benchmark("hash_hmac_sha256", bench_hmac_sha256, bench_setup, NULL, &data, 10, 20000);
-    if (have_flag(argc, argv, "hash") || have_flag(argc, argv, "rng6979")) run_benchmark("hash_rfc6979_hmac_sha256", bench_rfc6979_hmac_sha256, bench_setup, NULL, &data, 10, 20000);
+    if (have_flag(argc, argv, "hash") || have_flag(argc, argv, "sha256")) run_benchmark("hash_sha256", bench_sha256, bench_setup, NULL, &data, 10, iters);
+    if (have_flag(argc, argv, "hash") || have_flag(argc, argv, "hmac")) run_benchmark("hash_hmac_sha256", bench_hmac_sha256, bench_setup, NULL, &data, 10, iters);
+    if (have_flag(argc, argv, "hash") || have_flag(argc, argv, "rng6979")) run_benchmark("hash_rfc6979_hmac_sha256", bench_rfc6979_hmac_sha256, bench_setup, NULL, &data, 10, iters);
 
-    if (have_flag(argc, argv, "context") || have_flag(argc, argv, "verify")) run_benchmark("context_verify", bench_context_verify, bench_setup, NULL, &data, 10, 20);
-    if (have_flag(argc, argv, "context") || have_flag(argc, argv, "sign")) run_benchmark("context_sign", bench_context_sign, bench_setup, NULL, &data, 10, 200);
+    if (have_flag(argc, argv, "context") || have_flag(argc, argv, "verify")) run_benchmark("context_verify", bench_context_verify, bench_setup, NULL, &data, 10, 1 + iters/1000);
+    if (have_flag(argc, argv, "context") || have_flag(argc, argv, "sign")) run_benchmark("context_sign", bench_context_sign, bench_setup, NULL, &data, 10, 1 + iters/100);
 
 #ifndef USE_NUM_NONE
-    if (have_flag(argc, argv, "num") || have_flag(argc, argv, "jacobi")) run_benchmark("num_jacobi", bench_num_jacobi, bench_setup, NULL, &data, 10, 200000);
+    if (have_flag(argc, argv, "num") || have_flag(argc, argv, "jacobi")) run_benchmark("num_jacobi", bench_num_jacobi, bench_setup, NULL, &data, 10, iters*10);
 #endif
     return 0;
 }
diff --git a/src/secp256k1/src/bench_recover.c b/src/secp256k1/src/bench_recover.c
--- a/src/secp256k1/src/bench_recover.c
+++ b/src/secp256k1/src/bench_recover.c
@@ -15,13 +15,13 @@
     unsigned char sig[64];
 } bench_recover_data;
 
-void bench_recover(void* arg) {
+void bench_recover(void* arg, int iters) {
     int i;
     bench_recover_data *data = (bench_recover_data*)arg;
     secp256k1_pubkey pubkey;
     unsigned char pubkeyc[33];
 
-    for (i = 0; i < 20000; i++) {
+    for (i = 0; i < iters; i++) {
         int j;
         size_t pubkeylen = 33;
         secp256k1_ecdsa_recoverable_signature sig;
@@ -51,9 +51,11 @@
 int main(void) {
     bench_recover_data data;
 
+    int iters = get_iters(20000);
+
     data.ctx = secp256k1_context_create(SECP256K1_CONTEXT_VERIFY);
 
-    run_benchmark("ecdsa_recover", bench_recover, bench_recover_setup, NULL, &data, 10, 20000);
+    run_benchmark("ecdsa_recover", bench_recover, bench_recover_setup, NULL, &data, 10, iters);
 
     secp256k1_context_destroy(data.ctx);
     return 0;
diff --git a/src/secp256k1/src/bench_sign.c b/src/secp256k1/src/bench_sign.c
--- a/src/secp256k1/src/bench_sign.c
+++ b/src/secp256k1/src/bench_sign.c
@@ -31,12 +31,12 @@
     }
 }
 
-static void bench_sign_run(void* arg) {
+static void bench_sign_run(void* arg, int iters) {
     int i;
     bench_sign *data = (bench_sign*)arg;
 
     unsigned char sig[74];
-    for (i = 0; i < 20000; i++) {
+    for (i = 0; i < iters; i++) {
         size_t siglen = 74;
         int j;
         secp256k1_ecdsa_signature signature;
@@ -50,12 +50,12 @@
 }
 
 #ifdef ENABLE_MODULE_SCHNORR
-static void bench_schnorr_sign_run(void* arg) {
+static void bench_schnorr_sign_run(void* arg, int iters) {
     int i,j;
     bench_sign *data = (bench_sign*)arg;
 
     unsigned char sig[64];
-    for (i = 0; i < 20000; i++) {
+    for (i = 0; i < iters; i++) {
         CHECK(secp256k1_schnorr_sign(data->ctx, sig, data->msg, data->key, NULL, NULL));
         for (j = 0; j < 32; j++) {
             data->msg[j] = sig[j];
@@ -68,11 +68,13 @@
 int main(void) {
     bench_sign data;
 
+    int iters = get_iters(20000);
+
     data.ctx = secp256k1_context_create(SECP256K1_CONTEXT_SIGN);
 
-    run_benchmark("ecdsa_sign", bench_sign_run, bench_sign_setup, NULL, &data, 10, 20000);
+    run_benchmark("ecdsa_sign", bench_sign_run, bench_sign_setup, NULL, &data, 10, iters);
 #ifdef ENABLE_MODULE_SCHNORR
-    run_benchmark("schnorr_sign", bench_schnorr_sign_run, bench_sign_setup, NULL, &data, 10, 20000);
+    run_benchmark("schnorr_sign", bench_schnorr_sign_run, bench_sign_setup, NULL, &data, 10, iters);
 #endif
 
     secp256k1_context_destroy(data.ctx);
diff --git a/src/secp256k1/src/bench_verify.c b/src/secp256k1/src/bench_verify.c
--- a/src/secp256k1/src/bench_verify.c
+++ b/src/secp256k1/src/bench_verify.c
@@ -21,6 +21,7 @@
 #include <openssl/obj_mac.h>
 #endif
 
+
 typedef struct {
     secp256k1_context *ctx;
     unsigned char msg[32];
@@ -34,11 +35,11 @@
 #endif
 } benchmark_verify_t;
 
-static void benchmark_verify(void* arg) {
+static void benchmark_verify(void* arg, int iters) {
     int i;
     benchmark_verify_t* data = (benchmark_verify_t*)arg;
 
-    for (i = 0; i < 20000; i++) {
+    for (i = 0; i < iters; i++) {
         secp256k1_pubkey pubkey;
         secp256k1_ecdsa_signature sig;
         data->sig[data->siglen - 1] ^= (i & 0xFF);
@@ -54,11 +55,11 @@
 }
 
 #ifdef ENABLE_OPENSSL_TESTS
-static void benchmark_verify_openssl(void* arg) {
+static void benchmark_verify_openssl(void* arg, int iters) {
     int i;
     benchmark_verify_t* data = (benchmark_verify_t*)arg;
 
-    for (i = 0; i < 20000; i++) {
+    for (i = 0; i < iters; i++) {
         data->sig[data->siglen - 1] ^= (i & 0xFF);
         data->sig[data->siglen - 2] ^= ((i >> 8) & 0xFF);
         data->sig[data->siglen - 3] ^= ((i >> 16) & 0xFF);
@@ -84,11 +85,11 @@
 #endif
 
 #ifdef ENABLE_MODULE_SCHNORR
-static void benchmark_schnorr_verify(void* arg) {
+static void benchmark_schnorr_verify(void* arg, int iters) {
     int i;
     benchmark_verify_t* data = (benchmark_verify_t*)arg;
 
-    for (i = 0; i < 20000; i++) {
+    for (i = 0; i < iters; i++) {
         secp256k1_pubkey pubkey;
         data->sig[data->siglen - 1] ^= (i & 0xFF);
         data->sig[data->siglen - 2] ^= ((i >> 8) & 0xFF);
@@ -108,6 +109,8 @@
     secp256k1_ecdsa_signature sig;
     benchmark_verify_t data;
 
+    int iters = get_iters(20000);
+
     data.ctx = secp256k1_context_create(SECP256K1_CONTEXT_SIGN | SECP256K1_CONTEXT_VERIFY);
 
     for (i = 0; i < 32; i++) {
@@ -123,16 +126,16 @@
     data.pubkeylen = 33;
     CHECK(secp256k1_ec_pubkey_serialize(data.ctx, data.pubkey, &data.pubkeylen, &pubkey, SECP256K1_EC_COMPRESSED) == 1);
 
-    run_benchmark("ecdsa_verify", benchmark_verify, NULL, NULL, &data, 10, 20000);
+    run_benchmark("ecdsa_verify", benchmark_verify, NULL, NULL, &data, 10, iters);
 #ifdef ENABLE_OPENSSL_TESTS
     data.ec_group = EC_GROUP_new_by_curve_name(NID_secp256k1);
-    run_benchmark("ecdsa_verify_openssl", benchmark_verify_openssl, NULL, NULL, &data, 10, 20000);
+    run_benchmark("ecdsa_verify_openssl", benchmark_verify_openssl, NULL, NULL, &data, 10, iters);
     EC_GROUP_free(data.ec_group);
 #endif
 #ifdef ENABLE_MODULE_SCHNORR
     CHECK(secp256k1_schnorr_sign(data.ctx, data.sig, data.msg, data.key, NULL, NULL));
     data.siglen = 64;
-    run_benchmark("schnorr_verify", benchmark_schnorr_verify, NULL, NULL, &data, 10, 20000);
+    run_benchmark("schnorr_verify", benchmark_schnorr_verify, NULL, NULL, &data, 10, iters);
 #endif
 
     secp256k1_context_destroy(data.ctx);
diff --git a/src/secp256k1/travis/build_autotools.sh b/src/secp256k1/travis/build_autotools.sh
--- a/src/secp256k1/travis/build_autotools.sh
+++ b/src/secp256k1/travis/build_autotools.sh
@@ -39,11 +39,28 @@
   cat tests.log || :
   cat exhaustive_tests.log || :
   cat valgrind_ctime_test.log || :
+  cat bench.log || :
 }
 trap 'print_logs' ERR
 
 make -j2 $AUTOTOOLS_TARGET
 
+if [ -n "$BENCH" ]; then
+  if [ -n "$VALGRIND" ]; then
+    EXEC='libtool --mode=execute valgrind --error-exitcode=42';
+  else
+    EXEC= ;
+  fi
+  $EXEC ./bench_ecmult &>> bench.log
+  $EXEC ./bench_internal &>> bench.log
+  $EXEC ./bench_sign &>> bench.log
+  $EXEC ./bench_verify &>> bench.log
+  if [ "$RECOVERY" == "yes" ]; then
+    $EXEC ./bench_recover &>> bench.log
+  fi
+  if [ "$ECDH" == "yes" ]; then
+    $EXEC ./bench_ecdh &>> bench.log
+  fi
 if [ -n "$CTIMETEST" ]; then
   libtool --mode=execute valgrind  ./valgrind_ctime_test &> valgrind_ctime_test.log
 fi