diff --git a/src/Cafe/HW/Latte/Renderer/Vulkan/TextureReadbackVk.cpp b/src/Cafe/HW/Latte/Renderer/Vulkan/TextureReadbackVk.cpp
index bce23b59..1fe90abe 100644
--- a/src/Cafe/HW/Latte/Renderer/Vulkan/TextureReadbackVk.cpp
+++ b/src/Cafe/HW/Latte/Renderer/Vulkan/TextureReadbackVk.cpp
@@ -124,12 +124,8 @@ void LatteTextureReadbackInfoVk::StartTransfer()
 
 	renderer->barrier_image<VulkanRenderer::ANY_TRANSFER | VulkanRenderer::IMAGE_WRITE, VulkanRenderer::TRANSFER_READ>(baseTexture, region.imageSubresource, VK_IMAGE_LAYOUT_GENERAL);
 
-	renderer->barrier_sequentializeTransfer();
-
 	vkCmdCopyImageToBuffer(renderer->getCurrentCommandBuffer(), baseTexture->GetImageObj()->m_image, VK_IMAGE_LAYOUT_GENERAL, m_buffer, 1, &region);
 
-	renderer->barrier_sequentializeTransfer();
-
 	renderer->barrier_image<VulkanRenderer::TRANSFER_READ, VulkanRenderer::ANY_TRANSFER | VulkanRenderer::IMAGE_WRITE>(baseTexture, region.imageSubresource, VK_IMAGE_LAYOUT_GENERAL); // make sure transfer is finished before image is modified
 	renderer->barrier_bufferRange<VulkanRenderer::TRANSFER_WRITE, VulkanRenderer::HOST_READ>(m_buffer, m_buffer_offset, m_image_size); // make sure transfer is finished before result is read
 
diff --git a/src/Cafe/HW/Latte/Renderer/Vulkan/VulkanRenderer.cpp b/src/Cafe/HW/Latte/Renderer/Vulkan/VulkanRenderer.cpp
index 4dfc763f..42571aa5 100644
--- a/src/Cafe/HW/Latte/Renderer/Vulkan/VulkanRenderer.cpp
+++ b/src/Cafe/HW/Latte/Renderer/Vulkan/VulkanRenderer.cpp
@@ -3709,9 +3709,9 @@ void VulkanRenderer::bufferCache_upload(uint8* buffer, sint32 size, uint32 buffe
 
 	vkMemAllocator.FlushReservation(uploadResv);
 
-	barrier_bufferRange<ANY_TRANSFER | HOST_WRITE, ANY_TRANSFER,
-		BUFFER_SHADER_READ, TRANSFER_WRITE>(
-			uploadResv.vkBuffer, uploadResv.bufferOffset, uploadResv.size, // make sure any in-flight transfers are completed
+	barrier_bufferRange<TRANSFER_WRITE | HOST_WRITE, TRANSFER_READ,
+						TRANSFER_READ | BUFFER_SHADER_READ, TRANSFER_WRITE>(
+			uploadResv.vkBuffer, uploadResv.bufferOffset, uploadResv.size, // make sure source data is visible
 			m_bufferCache, bufferOffset, size); // make sure all reads are completed before we overwrite the data
 
 	VkBufferCopy region;
@@ -3720,7 +3720,7 @@ void VulkanRenderer::bufferCache_upload(uint8* buffer, sint32 size, uint32 buffe
 	region.size = size;
 	vkCmdCopyBuffer(m_state.currentCommandBuffer, uploadResv.vkBuffer, m_bufferCache, 1, &region);
 
-	barrier_sequentializeTransfer();
+	barrier_bufferRange<TRANSFER_WRITE, BUFFER_SHADER_READ>(m_bufferCache, bufferOffset, size);
 }
 
 void VulkanRenderer::bufferCache_copy(uint32 srcOffset, uint32 dstOffset, uint32 size)
@@ -3728,7 +3728,10 @@ void VulkanRenderer::bufferCache_copy(uint32 srcOffset, uint32 dstOffset, uint32
 	cemu_assert_debug(!m_useHostMemoryForCache);
 	draw_endRenderPass();
 
-	barrier_sequentializeTransfer();
+	barrier_bufferRange<BUFFER_SHADER_WRITE | TRANSFER_WRITE, TRANSFER_READ,
+		TRANSFER_READ | BUFFER_SHADER_READ, TRANSFER_WRITE>(
+			m_bufferCache, srcOffset, size, // make sure source data is visible
+			m_bufferCache, dstOffset, size); // make sure all reads are completed before we overwrite the data
 
 	bool isOverlapping = (srcOffset + size) > dstOffset && (srcOffset) < (dstOffset + size);
 	cemu_assert_debug(!isOverlapping);
@@ -3739,7 +3742,7 @@ void VulkanRenderer::bufferCache_copy(uint32 srcOffset, uint32 dstOffset, uint32
 	bufferCopy.size = size;
 	vkCmdCopyBuffer(m_state.currentCommandBuffer, m_bufferCache, m_bufferCache, 1, &bufferCopy);
 
-	barrier_sequentializeTransfer();
+	barrier_bufferRange<TRANSFER_WRITE, BUFFER_SHADER_READ>(m_bufferCache, dstOffset, size);
 }
 
 void VulkanRenderer::bufferCache_copyStreamoutToMainBuffer(uint32 srcOffset, uint32 dstOffset, uint32 size)
@@ -3756,12 +3759,10 @@ void VulkanRenderer::bufferCache_copyStreamoutToMainBuffer(uint32 srcOffset, uin
 	else
 		dstBuffer = m_bufferCache;
 
-	barrier_bufferRange<BUFFER_SHADER_WRITE, TRANSFER_READ,
-		ANY_TRANSFER | BUFFER_SHADER_READ, TRANSFER_WRITE>(
-			m_xfbRingBuffer, srcOffset, size, // wait for all writes to finish
-			dstBuffer, dstOffset, size); // wait for all reads to finish
-
-	barrier_sequentializeTransfer();
+	barrier_bufferRange<BUFFER_SHADER_WRITE | TRANSFER_WRITE, TRANSFER_READ,
+							TRANSFER_READ | BUFFER_SHADER_READ, TRANSFER_WRITE>(
+			m_xfbRingBuffer, srcOffset, size, // make sure source data is visible
+			dstBuffer, dstOffset, size); // make sure all reads are completed before we overwrite the data
 
 	VkBufferCopy bufferCopy{};
 	bufferCopy.srcOffset = srcOffset;
@@ -3769,7 +3770,7 @@ void VulkanRenderer::bufferCache_copyStreamoutToMainBuffer(uint32 srcOffset, uin
 	bufferCopy.size = size;
 	vkCmdCopyBuffer(m_state.currentCommandBuffer, m_xfbRingBuffer, dstBuffer, 1, &bufferCopy);
 
-	barrier_sequentializeTransfer();
+	barrier_bufferRange<TRANSFER_WRITE, BUFFER_SHADER_READ>(dstBuffer, dstOffset, size); // make sure writes are visible to host
 }
 
 void VulkanRenderer::AppendOverlayDebugInfo()
diff --git a/src/Cafe/HW/Latte/Renderer/Vulkan/VulkanRenderer.h b/src/Cafe/HW/Latte/Renderer/Vulkan/VulkanRenderer.h
index 0a329c28..c819134f 100644
--- a/src/Cafe/HW/Latte/Renderer/Vulkan/VulkanRenderer.h
+++ b/src/Cafe/HW/Latte/Renderer/Vulkan/VulkanRenderer.h
@@ -871,34 +871,6 @@ private:
 		performanceMonitor.vk.numDrawBarriersPerFrame.increment();
 	}
 
-	void barrier_sequentializeTransfer()
-	{
-		VkMemoryBarrier memBarrier{};
-		memBarrier.sType = VK_STRUCTURE_TYPE_MEMORY_BARRIER;
-		memBarrier.pNext = nullptr;
-
-		VkPipelineStageFlags srcStages = VK_PIPELINE_STAGE_TRANSFER_BIT;
-		VkPipelineStageFlags dstStages = VK_PIPELINE_STAGE_ALL_COMMANDS_BIT;
-
-		memBarrier.srcAccessMask = VK_ACCESS_TRANSFER_READ_BIT | VK_ACCESS_TRANSFER_WRITE_BIT;
-		memBarrier.dstAccessMask = 0;
-
-		memBarrier.srcAccessMask |= (VK_ACCESS_MEMORY_READ_BIT | VK_ACCESS_MEMORY_WRITE_BIT);
-		memBarrier.dstAccessMask |= (VK_ACCESS_MEMORY_READ_BIT | VK_ACCESS_MEMORY_WRITE_BIT);
-
-		vkCmdPipelineBarrier(m_state.currentCommandBuffer, srcStages, dstStages, 0, 1, &memBarrier, 0, nullptr, 0, nullptr);
-		performanceMonitor.vk.numDrawBarriersPerFrame.increment();
-	}
-
-	void barrier_sequentializeCommand()
-	{
-		VkPipelineStageFlags srcStages = VK_PIPELINE_STAGE_BOTTOM_OF_PIPE_BIT;
-		VkPipelineStageFlags dstStages = VK_PIPELINE_STAGE_TOP_OF_PIPE_BIT;
-
-		vkCmdPipelineBarrier(m_state.currentCommandBuffer, srcStages, dstStages, 0, 0, nullptr, 0, nullptr, 0, nullptr);
-		performanceMonitor.vk.numDrawBarriersPerFrame.increment();
-	}
-
 	template<uint32 TSrcSyncOp, uint32 TDstSyncOp>
 	void barrier_image(VkImage imageVk, VkImageSubresourceRange& subresourceRange, VkImageLayout oldLayout, VkImageLayout newLayout)
 	{