diff --git a/benchmarks/BM_cumsum.cpp b/benchmarks/BM_cumsum.cpp
index 08a49859..b91afe0e 100644
--- a/benchmarks/BM_cumsum.cpp
+++ b/benchmarks/BM_cumsum.cpp
@@ -29,12 +29,12 @@
 #include <iostream>
 #include "../sfizz/SIMDHelpers.h"
 
-class CumsumArray : public benchmark::Fixture {
+class CumArray : public benchmark::Fixture {
 public:
   void SetUp(const ::benchmark::State& state) {
     std::random_device rd { };
     std::mt19937 gen { rd() };
-    std::uniform_real_distribution<float> dist { 0, 1 };
+    std::uniform_real_distribution<float> dist { 0.1, 1 };
     input = std::vector<float>(state.range(0));
     output = std::vector<float>(state.range(0));
     std::generate(input.begin(), input.end(), [&]() { return dist(gen); });
@@ -49,36 +49,37 @@ public:
 };
 
 
-BENCHMARK_DEFINE_F(CumsumArray, Scalar)(benchmark::State& state) {
+BENCHMARK_DEFINE_F(CumArray, Sum_Scalar)(benchmark::State& state) {
     for (auto _ : state)
     {
         cumsum<float, false>(input, absl::MakeSpan(output));
     }
 }
 
-BENCHMARK_DEFINE_F(CumsumArray, SIMD)(benchmark::State& state) {
+BENCHMARK_DEFINE_F(CumArray, Sum_SIMD)(benchmark::State& state) {
     for (auto _ : state)
     {
         cumsum<float, true>(input, absl::MakeSpan(output));
     }
 }
 
-BENCHMARK_DEFINE_F(CumsumArray, Scalar_Unaligned)(benchmark::State& state) {
+BENCHMARK_DEFINE_F(CumArray, Sum_Scalar_Unaligned)(benchmark::State& state) {
     for (auto _ : state)
     {
         cumsum<float, false>(absl::MakeSpan(input).subspan(1), absl::MakeSpan(output).subspan(1));
     }
 }
 
-BENCHMARK_DEFINE_F(CumsumArray, SIMD_Unaligned)(benchmark::State& state) {
+BENCHMARK_DEFINE_F(CumArray, Sum_SIMD_Unaligned)(benchmark::State& state) {
     for (auto _ : state)
     {
         cumsum<float, true>(absl::MakeSpan(input).subspan(1), absl::MakeSpan(output).subspan(1));
     }
 }
 
-BENCHMARK_REGISTER_F(CumsumArray, Scalar)->RangeMultiplier(4)->Range(1 << 2, 1 << 12);
-BENCHMARK_REGISTER_F(CumsumArray, SIMD)->RangeMultiplier(4)->Range(1 << 2, 1 << 12);
-BENCHMARK_REGISTER_F(CumsumArray, Scalar_Unaligned)->RangeMultiplier(4)->Range(1 << 2, 1 << 12);
-BENCHMARK_REGISTER_F(CumsumArray, SIMD_Unaligned)->RangeMultiplier(4)->Range(1 << 2, 1 << 12);
+
+BENCHMARK_REGISTER_F(CumArray, Sum_Scalar)->RangeMultiplier(4)->Range(1 << 2, 1 << 12);
+BENCHMARK_REGISTER_F(CumArray, Sum_SIMD)->RangeMultiplier(4)->Range(1 << 2, 1 << 12);
+BENCHMARK_REGISTER_F(CumArray, Sum_Scalar_Unaligned)->RangeMultiplier(4)->Range(1 << 2, 1 << 12);
+BENCHMARK_REGISTER_F(CumArray, Sum_SIMD_Unaligned)->RangeMultiplier(4)->Range(1 << 2, 1 << 12);
 BENCHMARK_MAIN();
\ No newline at end of file
diff --git a/sfizz/SIMDDummy.cpp b/sfizz/SIMDDummy.cpp
index f91549de..cbb98be6 100644
--- a/sfizz/SIMDDummy.cpp
+++ b/sfizz/SIMDDummy.cpp
@@ -148,4 +148,4 @@ template <>
 void cumsum<float, true>(absl::Span<const float> input, absl::Span<float> output) noexcept
 {
     cumsum<float, false>(input, output);
-}
\ No newline at end of file
+}
diff --git a/sfizz/SIMDHelpers.h b/sfizz/SIMDHelpers.h
index c06ca20b..2647e709 100644
--- a/sfizz/SIMDHelpers.h
+++ b/sfizz/SIMDHelpers.h
@@ -474,4 +474,4 @@ void cumsum(absl::Span<const float> input, absl::Span<float> output) noexcept
 }
 
 template <>
-void cumsum<float, true>(absl::Span<const float> input, absl::Span<float> output) noexcept;
\ No newline at end of file
+void cumsum<float, true>(absl::Span<const float> input, absl::Span<float> output) noexcept;
diff --git a/sfizz/SIMDSSE.cpp b/sfizz/SIMDSSE.cpp
index f7900a81..8f702f3c 100644
--- a/sfizz/SIMDSSE.cpp
+++ b/sfizz/SIMDSSE.cpp
@@ -691,7 +691,7 @@ void cumsum<float, true>(absl::Span<const float> input, absl::Span<float> output
     while (in < lastAligned) {
         auto mmOffset = _mm_load_ps(in);
         mmOffset = _mm_add_ps(mmOffset, _mm_castsi128_ps(_mm_slli_si128(_mm_castps_si128(mmOffset), 4)));
-        mmOffset = _mm_add_ps(mmOffset, _mm_shuffle_ps(_mm_setzero_ps(), mmOffset, 0x40));
+        mmOffset = _mm_add_ps(mmOffset, _mm_shuffle_ps(_mm_setzero_ps(), mmOffset, _MM_SHUFFLE(1, 0, 0, 0)));
         mmOutput = _mm_add_ps(mmOutput, mmOffset);
         _mm_store_ps(out, mmOutput);
         mmOutput = _mm_shuffle_ps(mmOutput, mmOutput, _MM_SHUFFLE(3, 3, 3, 3));
@@ -701,4 +701,4 @@ void cumsum<float, true>(absl::Span<const float> input, absl::Span<float> output
 
     while (in < sentinel)
         snippetCumsum(in, out);
-}
\ No newline at end of file
+}