diff --git a/.circleci/config.yml b/.circleci/config.yml
index 6c8193ae..de7d5dfa 100644
--- a/.circleci/config.yml
+++ b/.circleci/config.yml
@@ -373,12 +373,12 @@ commands:
             - /go/pkg/mod
 
 jobs:
-  test-llvm10-go115:
+  test-llvm11-go115:
     docker:
       - image: circleci/golang:1.15-buster
     steps:
       - test-linux:
-          llvm: "10"
+          llvm: "11"
   test-llvm11-go116:
     docker:
       - image: circleci/golang:1.16-buster
@@ -406,7 +406,7 @@ jobs:
 workflows:
   test-all:
     jobs:
-      - test-llvm10-go115
+      - test-llvm11-go115
       - test-llvm11-go116
       - build-linux
       - build-macos
diff --git a/Makefile b/Makefile
index 3af2c52f..6c569be8 100644
--- a/Makefile
+++ b/Makefile
@@ -10,9 +10,9 @@ LLD_SRC ?= $(LLVM_PROJECTDIR)/lld
 
 # Try to autodetect LLVM build tools.
 detect = $(shell command -v $(1) 2> /dev/null && echo $(1))
-CLANG ?= $(word 1,$(abspath $(call detect,llvm-build/bin/clang))$(call detect,clang-11)$(call detect,clang-10)$(call detect,clang))
-LLVM_AR ?= $(word 1,$(abspath $(call detect,llvm-build/bin/llvm-ar))$(call detect,llvm-ar-11)$(call detect,llvm-ar-10)$(call detect,llvm-ar))
-LLVM_NM ?= $(word 1,$(abspath $(call detect,llvm-build/bin/llvm-nm))$(call detect,llvm-nm-11)$(call detect,llvm-nm-10)$(call detect,llvm-nm))
+CLANG ?= $(word 1,$(abspath $(call detect,llvm-build/bin/clang))$(call detect,clang-11)$(call detect,clang))
+LLVM_AR ?= $(word 1,$(abspath $(call detect,llvm-build/bin/llvm-ar))$(call detect,llvm-ar-11)$(call detect,llvm-ar))
+LLVM_NM ?= $(word 1,$(abspath $(call detect,llvm-build/bin/llvm-nm))$(call detect,llvm-nm-11)$(call detect,llvm-nm))
 
 # Go binary and GOROOT to select
 GO ?= go
diff --git a/builder/sizes.go b/builder/sizes.go
index 750c4bd9..c88c9b7b 100644
--- a/builder/sizes.go
+++ b/builder/sizes.go
@@ -84,18 +84,7 @@ func loadProgramSize(path string) (*programSize, error) {
 		if section.Type != elf.SHT_PROGBITS && section.Type != elf.SHT_NOBITS {
 			continue
 		}
-		if section.Name == ".stack" {
-			// HACK: this works around a bug in ld.lld from LLVM 10. The linker
-			// marks sections with no input symbols (such as is the case for the
-			// .stack section) as SHT_PROGBITS instead of SHT_NOBITS. While it
-			// doesn't affect the generated binaries (.hex and .bin), it does
-			// affect the reported size.
-			// https://bugs.llvm.org/show_bug.cgi?id=45336
-			// https://reviews.llvm.org/D76981
-			// It has been merged in master, but it has not (yet) been
-			// backported to the LLVM 10 release branch.
-			sumBSS += section.Size
-		} else if section.Type == elf.SHT_NOBITS {
+		if section.Type == elf.SHT_NOBITS {
 			sumBSS += section.Size
 		} else if section.Flags&elf.SHF_EXECINSTR != 0 {
 			sumCode += section.Size
diff --git a/cgo/libclang.go b/cgo/libclang.go
index f27d7935..66f81c3d 100644
--- a/cgo/libclang.go
+++ b/cgo/libclang.go
@@ -16,7 +16,7 @@ import (
 )
 
 /*
-#include <clang-c/Index.h> // if this fails, install libclang-10-dev
+#include <clang-c/Index.h> // if this fails, install libclang-11-dev
 #include <stdlib.h>
 #include <stdint.h>
 
diff --git a/cgo/libclang_config.go b/cgo/libclang_config.go
index 78783db2..4b4ce2db 100644
--- a/cgo/libclang_config.go
+++ b/cgo/libclang_config.go
@@ -1,5 +1,4 @@
 // +build !byollvm
-// +build !llvm10
 
 package cgo
 
diff --git a/cgo/libclang_config_llvm10.go b/cgo/libclang_config_llvm10.go
deleted file mode 100644
index 9b993949..00000000
--- a/cgo/libclang_config_llvm10.go
+++ /dev/null
@@ -1,14 +0,0 @@
-// +build !byollvm
-// +build llvm10
-
-package cgo
-
-/*
-#cgo linux  CFLAGS: -I/usr/lib/llvm-10/include
-#cgo darwin CFLAGS: -I/usr/local/opt/llvm@10/include
-#cgo freebsd CFLAGS: -I/usr/local/llvm10/include
-#cgo linux  LDFLAGS: -L/usr/lib/llvm-10/lib -lclang
-#cgo darwin LDFLAGS: -L/usr/local/opt/llvm@10/lib -lclang -lffi
-#cgo freebsd LDFLAGS: -L/usr/local/llvm10/lib -lclang
-*/
-import "C"
diff --git a/cgo/libclang_stubs.c b/cgo/libclang_stubs.c
index 293ca602..ce9e164a 100644
--- a/cgo/libclang_stubs.c
+++ b/cgo/libclang_stubs.c
@@ -3,7 +3,7 @@
 // are slightly different from the ones defined in libclang.go, but they
 // should be ABI compatible.
 
-#include <clang-c/Index.h> // if this fails, install libclang-10-dev
+#include <clang-c/Index.h> // if this fails, install libclang-11-dev
 
 CXCursor tinygo_clang_getTranslationUnitCursor(CXTranslationUnit tu) {
 	return clang_getTranslationUnitCursor(tu);
diff --git a/interp/interp_test.go b/interp/interp_test.go
index 9702cb83..9080de89 100644
--- a/interp/interp_test.go
+++ b/interp/interp_test.go
@@ -3,7 +3,6 @@ package interp
 import (
 	"io/ioutil"
 	"os"
-	"regexp"
 	"strings"
 	"testing"
 
@@ -88,8 +87,6 @@ func runTest(t *testing.T, pathPrefix string) {
 	}
 }
 
-var alignRegexp = regexp.MustCompile(", align [0-9]+$")
-
 // fuzzyEqualIR returns true if the two LLVM IR strings passed in are roughly
 // equal. That means, only relevant lines are compared (excluding comments
 // etc.).
@@ -101,15 +98,6 @@ func fuzzyEqualIR(s1, s2 string) bool {
 	}
 	for i, line1 := range lines1 {
 		line2 := lines2[i]
-		match1 := alignRegexp.MatchString(line1)
-		match2 := alignRegexp.MatchString(line2)
-		if match1 != match2 {
-			// Only one of the lines has the align keyword. Remove it.
-			// This is a change to make the test work in both LLVM 10 and LLVM
-			// 11 (LLVM 11 appears to automatically add alignment everywhere).
-			line1 = alignRegexp.ReplaceAllString(line1, "")
-			line2 = alignRegexp.ReplaceAllString(line2, "")
-		}
 		if line1 != line2 {
 			return false
 		}
diff --git a/interp/testdata/basic.out.ll b/interp/testdata/basic.out.ll
index 51a3b3f4..174f073e 100644
--- a/interp/testdata/basic.out.ll
+++ b/interp/testdata/basic.out.ll
@@ -18,14 +18,14 @@ entry:
   call void @runtime.printint64(i64 5)
   call void @runtime.printnl()
   %value1 = call i64 @someValue()
-  store i64 %value1, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @main.nonConst1, i32 0, i32 0)
-  %value2 = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @main.nonConst1, i32 0, i32 0)
-  store i64 %value2, i64* @main.nonConst2
+  store i64 %value1, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @main.nonConst1, i32 0, i32 0), align 8
+  %value2 = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @main.nonConst1, i32 0, i32 0), align 8
+  store i64 %value2, i64* @main.nonConst2, align 8
   call void @modifyExternal(i32* getelementptr inbounds ([8 x { i16, i32 }], [8 x { i16, i32 }]* @main.someArray, i32 0, i32 3, i32 1))
   call void @modifyExternal(i32* bitcast ([1 x i16*]* @main.exportedValue to i32*))
-  store i16 5, i16* @main.exposedValue1
+  store i16 5, i16* @main.exposedValue1, align 2
   call void @modifyExternal(i32* bitcast (void ()* @willModifyGlobal to i32*))
-  store i16 7, i16* @main.exposedValue2
+  store i16 7, i16* @main.exposedValue2, align 2
   call void @modifyExternal(i32* bitcast (void ()* @hasInlineAsm to i32*))
   call void @runtime.printint64(i64 6)
   call void @runtime.printint64(i64 -1)
@@ -39,7 +39,7 @@ entry:
   %agg2.insertvalue2 = insertvalue { i64, i16 } %agg2.agg1, i64 5, 0
   %agg2.insertvalue1 = insertvalue { float, { i64, i16 } } %agg2.agg0, { i64, i16 } %agg2.insertvalue2, 1
   %agg2.insertvalue0 = insertvalue { i8, i32, { float, { i64, i16 } } } %agg, { float, { i64, i16 } } %agg2.insertvalue1, 2
-  store { i8, i32, { float, { i64, i16 } } } %agg2.insertvalue0, { i8, i32, { float, { i64, i16 } } }* @main.insertedValue
+  store { i8, i32, { float, { i64, i16 } } } %agg2.insertvalue0, { i8, i32, { float, { i64, i16 } } }* @main.insertedValue, align 8
   ret void
 }
 
@@ -56,7 +56,7 @@ declare void @modifyExternal(i32*) local_unnamed_addr
 
 define void @willModifyGlobal() {
 entry:
-  store i16 8, i16* @main.exposedValue2
+  store i16 8, i16* @main.exposedValue2, align 2
   ret void
 }
 
diff --git a/interp/testdata/slice-copy.out.ll b/interp/testdata/slice-copy.out.ll
index f3b15950..9f88df09 100644
--- a/interp/testdata/slice-copy.out.ll
+++ b/interp/testdata/slice-copy.out.ll
@@ -1,7 +1,7 @@
 target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
 target triple = "x86_64--linux"
 
-@"main$alloc.1" = internal unnamed_addr constant [6 x i8] c"\05\00{\00\00\04"
+@"main$alloc.1" = internal unnamed_addr constant [6 x i8] c"\05\00{\00\00\04", align 8
 
 declare void @runtime.printuint8(i8) local_unnamed_addr
 
@@ -17,7 +17,7 @@ entry:
   call void @runtime.printuint8(i8 3)
   call void @runtime.printuint8(i8 3)
   call void @runtime.printint16(i16 5)
-  %int16SliceDst.val = load i16, i16* bitcast ([6 x i8]* @"main$alloc.1" to i16*)
+  %int16SliceDst.val = load i16, i16* bitcast ([6 x i8]* @"main$alloc.1" to i16*), align 2
   call void @runtime.printint16(i16 %int16SliceDst.val)
   ret void
 }
diff --git a/src/runtime/print.go b/src/runtime/print.go
index f7d3a285..04336cce 100644
--- a/src/runtime/print.go
+++ b/src/runtime/print.go
@@ -48,16 +48,6 @@ func printint16(n int16) {
 }
 
 func printuint32(n uint32) {
-	if TargetBits == 8 {
-		// AVR-specific workaround on LLVM 10. Should be removed when we switch
-		// to LLVM 11.
-		prevdigits := n / 10
-		if prevdigits != 0 {
-			printuint32(prevdigits)
-		}
-		putchar(byte((n % 10) + '0'))
-		return
-	}
 	printuint64(uint64(n))
 }
 
diff --git a/transform/testdata/allocs.out.ll b/transform/testdata/allocs.out.ll
index eaddd464..4543ac17 100644
--- a/transform/testdata/allocs.out.ll
+++ b/transform/testdata/allocs.out.ll
@@ -6,21 +6,21 @@ target triple = "armv7m-none-eabi"
 declare nonnull i8* @runtime.alloc(i32)
 
 define void @testInt() {
-  %stackalloc.alloca = alloca [1 x i32]
-  store [1 x i32] zeroinitializer, [1 x i32]* %stackalloc.alloca
+  %stackalloc.alloca = alloca [1 x i32], align 4
+  store [1 x i32] zeroinitializer, [1 x i32]* %stackalloc.alloca, align 4
   %stackalloc = bitcast [1 x i32]* %stackalloc.alloca to i32*
-  store i32 5, i32* %stackalloc
+  store i32 5, i32* %stackalloc, align 4
   ret void
 }
 
 define i16 @testArray() {
-  %stackalloc.alloca = alloca [2 x i32]
-  store [2 x i32] zeroinitializer, [2 x i32]* %stackalloc.alloca
+  %stackalloc.alloca = alloca [2 x i32], align 4
+  store [2 x i32] zeroinitializer, [2 x i32]* %stackalloc.alloca, align 4
   %stackalloc = bitcast [2 x i32]* %stackalloc.alloca to i16*
   %1 = getelementptr i16, i16* %stackalloc, i32 1
-  store i16 5, i16* %1
+  store i16 5, i16* %1, align 2
   %2 = getelementptr i16, i16* %stackalloc, i32 2
-  %3 = load i16, i16* %2
+  %3 = load i16, i16* %2, align 2
   ret i16 %3
 }
 
@@ -39,8 +39,8 @@ define void @testEscapingCall2() {
 }
 
 define void @testNonEscapingCall() {
-  %stackalloc.alloca = alloca [1 x i32]
-  store [1 x i32] zeroinitializer, [1 x i32]* %stackalloc.alloca
+  %stackalloc.alloca = alloca [1 x i32], align 4
+  store [1 x i32] zeroinitializer, [1 x i32]* %stackalloc.alloca, align 4
   %stackalloc = bitcast [1 x i32]* %stackalloc.alloca to i32*
   %1 = call i32* @noescapeIntPtr(i32* %stackalloc)
   ret void
@@ -54,11 +54,11 @@ define i32* @testEscapingReturn() {
 
 define void @testNonEscapingLoop() {
 entry:
-  %stackalloc.alloca = alloca [1 x i32]
+  %stackalloc.alloca = alloca [1 x i32], align 4
   br label %loop
 
 loop:                                             ; preds = %loop, %entry
-  store [1 x i32] zeroinitializer, [1 x i32]* %stackalloc.alloca
+  store [1 x i32] zeroinitializer, [1 x i32]* %stackalloc.alloca, align 4
   %stackalloc = bitcast [1 x i32]* %stackalloc.alloca to i32*
   %0 = call i32* @noescapeIntPtr(i32* %stackalloc)
   %1 = icmp eq i32* null, %0
diff --git a/transform/testdata/gc-globals.out.ll b/transform/testdata/gc-globals.out.ll
index 905e8eaf..78b182fb 100644
--- a/transform/testdata/gc-globals.out.ll
+++ b/transform/testdata/gc-globals.out.ll
@@ -14,18 +14,18 @@ target triple = "wasm32-unknown-unknown-wasm"
 @runtime.trackedGlobalsBitmap.1 = internal global [1 x i8] c"\09"
 
 define void @main() {
-  %1 = load i32, i32* @globalInt
-  %2 = load %runtime._string, %runtime._string* getelementptr inbounds ({ %runtime._string, %runtime._interface }, { %runtime._string, %runtime._interface }* @tinygo.trackedGlobals, i32 0, i32 0)
-  %3 = load %runtime._interface, %runtime._interface* getelementptr inbounds ({ %runtime._string, %runtime._interface }, { %runtime._string, %runtime._interface }* @tinygo.trackedGlobals, i32 0, i32 1)
-  %4 = load %runtime._string, %runtime._string* @constString
-  %5 = load %runtime._interface, %runtime._interface* @constInterface
+  %1 = load i32, i32* @globalInt, align 4
+  %2 = load %runtime._string, %runtime._string* getelementptr inbounds ({ %runtime._string, %runtime._interface }, { %runtime._string, %runtime._interface }* @tinygo.trackedGlobals, i32 0, i32 0), align 4
+  %3 = load %runtime._interface, %runtime._interface* getelementptr inbounds ({ %runtime._string, %runtime._interface }, { %runtime._string, %runtime._interface }* @tinygo.trackedGlobals, i32 0, i32 1), align 4
+  %4 = load %runtime._string, %runtime._string* @constString, align 4
+  %5 = load %runtime._interface, %runtime._interface* @constInterface, align 4
   ret void
 }
 
 define void @runtime.markGlobals() {
-  %1 = load i32, i32* @runtime.trackedGlobalsStart
-  %2 = load i32, i32* @runtime.trackedGlobalsLength
+  %1 = load i32, i32* @runtime.trackedGlobalsStart, align 4
+  %2 = load i32, i32* @runtime.trackedGlobalsLength, align 4
   %3 = getelementptr inbounds [0 x i8], [0 x i8]* bitcast ([1 x i8]* @runtime.trackedGlobalsBitmap.1 to [0 x i8]*), i32 0, i32 0
-  %4 = load i8, i8* %3
+  %4 = load i8, i8* %3, align 1
   ret void
 }
diff --git a/transform/testdata/gc-stackslots.out.ll b/transform/testdata/gc-stackslots.out.ll
index 0edb9168..af474534 100644
--- a/transform/testdata/gc-stackslots.out.ll
+++ b/transform/testdata/gc-stackslots.out.ll
@@ -15,44 +15,44 @@ define i8* @getPointer() {
 }
 
 define i8* @needsStackSlots() {
-  %gc.stackobject = alloca { %runtime.stackChainObject*, i32, i8* }
-  store { %runtime.stackChainObject*, i32, i8* } { %runtime.stackChainObject* null, i32 1, i8* null }, { %runtime.stackChainObject*, i32, i8* }* %gc.stackobject
-  %1 = load %runtime.stackChainObject*, %runtime.stackChainObject** @runtime.stackChainStart
+  %gc.stackobject = alloca { %runtime.stackChainObject*, i32, i8* }, align 8
+  store { %runtime.stackChainObject*, i32, i8* } { %runtime.stackChainObject* null, i32 1, i8* null }, { %runtime.stackChainObject*, i32, i8* }* %gc.stackobject, align 4
+  %1 = load %runtime.stackChainObject*, %runtime.stackChainObject** @runtime.stackChainStart, align 4
   %2 = getelementptr { %runtime.stackChainObject*, i32, i8* }, { %runtime.stackChainObject*, i32, i8* }* %gc.stackobject, i32 0, i32 0
-  store %runtime.stackChainObject* %1, %runtime.stackChainObject** %2
+  store %runtime.stackChainObject* %1, %runtime.stackChainObject** %2, align 4
   %3 = bitcast { %runtime.stackChainObject*, i32, i8* }* %gc.stackobject to %runtime.stackChainObject*
-  store %runtime.stackChainObject* %3, %runtime.stackChainObject** @runtime.stackChainStart
+  store %runtime.stackChainObject* %3, %runtime.stackChainObject** @runtime.stackChainStart, align 4
   %ptr = call i8* @runtime.alloc(i32 4)
   %4 = getelementptr { %runtime.stackChainObject*, i32, i8* }, { %runtime.stackChainObject*, i32, i8* }* %gc.stackobject, i32 0, i32 2
-  store i8* %ptr, i8** %4
-  store %runtime.stackChainObject* %1, %runtime.stackChainObject** @runtime.stackChainStart
+  store i8* %ptr, i8** %4, align 4
+  store %runtime.stackChainObject* %1, %runtime.stackChainObject** @runtime.stackChainStart, align 4
   call void @someArbitraryFunction()
-  %val = load i8, i8* @someGlobal
+  %val = load i8, i8* @someGlobal, align 1
   ret i8* %ptr
 }
 
 define i8* @needsStackSlots2() {
-  %gc.stackobject = alloca { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }
-  store { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* } { %runtime.stackChainObject* null, i32 5, i8* null, i8* null, i8* null, i8* null, i8* null }, { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }* %gc.stackobject
-  %1 = load %runtime.stackChainObject*, %runtime.stackChainObject** @runtime.stackChainStart
+  %gc.stackobject = alloca { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }, align 8
+  store { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* } { %runtime.stackChainObject* null, i32 5, i8* null, i8* null, i8* null, i8* null, i8* null }, { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }* %gc.stackobject, align 4
+  %1 = load %runtime.stackChainObject*, %runtime.stackChainObject** @runtime.stackChainStart, align 4
   %2 = getelementptr { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }, { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }* %gc.stackobject, i32 0, i32 0
-  store %runtime.stackChainObject* %1, %runtime.stackChainObject** %2
+  store %runtime.stackChainObject* %1, %runtime.stackChainObject** %2, align 4
   %3 = bitcast { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }* %gc.stackobject to %runtime.stackChainObject*
-  store %runtime.stackChainObject* %3, %runtime.stackChainObject** @runtime.stackChainStart
+  store %runtime.stackChainObject* %3, %runtime.stackChainObject** @runtime.stackChainStart, align 4
   %ptr1 = call i8* @getPointer()
   %4 = getelementptr { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }, { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }* %gc.stackobject, i32 0, i32 4
-  store i8* %ptr1, i8** %4
+  store i8* %ptr1, i8** %4, align 4
   %5 = getelementptr { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }, { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }* %gc.stackobject, i32 0, i32 3
-  store i8* %ptr1, i8** %5
+  store i8* %ptr1, i8** %5, align 4
   %6 = getelementptr { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }, { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }* %gc.stackobject, i32 0, i32 2
-  store i8* %ptr1, i8** %6
+  store i8* %ptr1, i8** %6, align 4
   %ptr2 = getelementptr i8, i8* @someGlobal, i32 0
   %7 = getelementptr { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }, { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }* %gc.stackobject, i32 0, i32 5
-  store i8* %ptr2, i8** %7
+  store i8* %ptr2, i8** %7, align 4
   %unused = call i8* @runtime.alloc(i32 4)
   %8 = getelementptr { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }, { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }* %gc.stackobject, i32 0, i32 6
-  store i8* %unused, i8** %8
-  store %runtime.stackChainObject* %1, %runtime.stackChainObject** @runtime.stackChainStart
+  store i8* %unused, i8** %8, align 4
+  store %runtime.stackChainObject* %1, %runtime.stackChainObject** @runtime.stackChainStart, align 4
   ret i8* %ptr1
 }
 
@@ -62,79 +62,79 @@ define i8* @noAllocatingFunction() {
 }
 
 define i8* @fibNext(i8* %x, i8* %y) {
-  %gc.stackobject = alloca { %runtime.stackChainObject*, i32, i8* }
-  store { %runtime.stackChainObject*, i32, i8* } { %runtime.stackChainObject* null, i32 1, i8* null }, { %runtime.stackChainObject*, i32, i8* }* %gc.stackobject
-  %1 = load %runtime.stackChainObject*, %runtime.stackChainObject** @runtime.stackChainStart
+  %gc.stackobject = alloca { %runtime.stackChainObject*, i32, i8* }, align 8
+  store { %runtime.stackChainObject*, i32, i8* } { %runtime.stackChainObject* null, i32 1, i8* null }, { %runtime.stackChainObject*, i32, i8* }* %gc.stackobject, align 4
+  %1 = load %runtime.stackChainObject*, %runtime.stackChainObject** @runtime.stackChainStart, align 4
   %2 = getelementptr { %runtime.stackChainObject*, i32, i8* }, { %runtime.stackChainObject*, i32, i8* }* %gc.stackobject, i32 0, i32 0
-  store %runtime.stackChainObject* %1, %runtime.stackChainObject** %2
+  store %runtime.stackChainObject* %1, %runtime.stackChainObject** %2, align 4
   %3 = bitcast { %runtime.stackChainObject*, i32, i8* }* %gc.stackobject to %runtime.stackChainObject*
-  store %runtime.stackChainObject* %3, %runtime.stackChainObject** @runtime.stackChainStart
-  %x.val = load i8, i8* %x
-  %y.val = load i8, i8* %y
+  store %runtime.stackChainObject* %3, %runtime.stackChainObject** @runtime.stackChainStart, align 4
+  %x.val = load i8, i8* %x, align 1
+  %y.val = load i8, i8* %y, align 1
   %out.val = add i8 %x.val, %y.val
   %out.alloc = call i8* @runtime.alloc(i32 1)
   %4 = getelementptr { %runtime.stackChainObject*, i32, i8* }, { %runtime.stackChainObject*, i32, i8* }* %gc.stackobject, i32 0, i32 2
-  store i8* %out.alloc, i8** %4
-  store %runtime.stackChainObject* %1, %runtime.stackChainObject** @runtime.stackChainStart
-  store i8 %out.val, i8* %out.alloc
+  store i8* %out.alloc, i8** %4, align 4
+  store %runtime.stackChainObject* %1, %runtime.stackChainObject** @runtime.stackChainStart, align 4
+  store i8 %out.val, i8* %out.alloc, align 1
   ret i8* %out.alloc
 }
 
 define i8* @allocLoop() {
 entry:
-  %gc.stackobject = alloca { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }
-  store { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* } { %runtime.stackChainObject* null, i32 5, i8* null, i8* null, i8* null, i8* null, i8* null }, { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }* %gc.stackobject
-  %0 = load %runtime.stackChainObject*, %runtime.stackChainObject** @runtime.stackChainStart
+  %gc.stackobject = alloca { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }, align 8
+  store { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* } { %runtime.stackChainObject* null, i32 5, i8* null, i8* null, i8* null, i8* null, i8* null }, { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }* %gc.stackobject, align 4
+  %0 = load %runtime.stackChainObject*, %runtime.stackChainObject** @runtime.stackChainStart, align 4
   %1 = getelementptr { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }, { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }* %gc.stackobject, i32 0, i32 0
-  store %runtime.stackChainObject* %0, %runtime.stackChainObject** %1
+  store %runtime.stackChainObject* %0, %runtime.stackChainObject** %1, align 4
   %2 = bitcast { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }* %gc.stackobject to %runtime.stackChainObject*
-  store %runtime.stackChainObject* %2, %runtime.stackChainObject** @runtime.stackChainStart
+  store %runtime.stackChainObject* %2, %runtime.stackChainObject** @runtime.stackChainStart, align 4
   %entry.x = call i8* @runtime.alloc(i32 1)
   %3 = getelementptr { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }, { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }* %gc.stackobject, i32 0, i32 2
-  store i8* %entry.x, i8** %3
+  store i8* %entry.x, i8** %3, align 4
   %entry.y = call i8* @runtime.alloc(i32 1)
   %4 = getelementptr { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }, { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }* %gc.stackobject, i32 0, i32 3
-  store i8* %entry.y, i8** %4
-  store i8 1, i8* %entry.y
+  store i8* %entry.y, i8** %4, align 4
+  store i8 1, i8* %entry.y, align 1
   br label %loop
 
 loop:                                             ; preds = %loop, %entry
   %prev.y = phi i8* [ %entry.y, %entry ], [ %prev.x, %loop ]
   %prev.x = phi i8* [ %entry.x, %entry ], [ %next.x, %loop ]
   %5 = getelementptr { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }, { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }* %gc.stackobject, i32 0, i32 5
-  store i8* %prev.y, i8** %5
+  store i8* %prev.y, i8** %5, align 4
   %6 = getelementptr { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }, { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }* %gc.stackobject, i32 0, i32 4
-  store i8* %prev.x, i8** %6
+  store i8* %prev.x, i8** %6, align 4
   %next.x = call i8* @fibNext(i8* %prev.x, i8* %prev.y)
   %7 = getelementptr { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }, { %runtime.stackChainObject*, i32, i8*, i8*, i8*, i8*, i8* }* %gc.stackobject, i32 0, i32 6
-  store i8* %next.x, i8** %7
-  %next.x.val = load i8, i8* %next.x
+  store i8* %next.x, i8** %7, align 4
+  %next.x.val = load i8, i8* %next.x, align 1
   %loop.done = icmp ult i8 40, %next.x.val
   br i1 %loop.done, label %end, label %loop
 
 end:                                              ; preds = %loop
-  store %runtime.stackChainObject* %0, %runtime.stackChainObject** @runtime.stackChainStart
+  store %runtime.stackChainObject* %0, %runtime.stackChainObject** @runtime.stackChainStart, align 4
   ret i8* %next.x
 }
 
 declare [32 x i8]* @arrayAlloc()
 
 define void @testGEPBitcast() {
-  %gc.stackobject = alloca { %runtime.stackChainObject*, i32, i8*, i8* }
-  store { %runtime.stackChainObject*, i32, i8*, i8* } { %runtime.stackChainObject* null, i32 2, i8* null, i8* null }, { %runtime.stackChainObject*, i32, i8*, i8* }* %gc.stackobject
-  %1 = load %runtime.stackChainObject*, %runtime.stackChainObject** @runtime.stackChainStart
+  %gc.stackobject = alloca { %runtime.stackChainObject*, i32, i8*, i8* }, align 8
+  store { %runtime.stackChainObject*, i32, i8*, i8* } { %runtime.stackChainObject* null, i32 2, i8* null, i8* null }, { %runtime.stackChainObject*, i32, i8*, i8* }* %gc.stackobject, align 4
+  %1 = load %runtime.stackChainObject*, %runtime.stackChainObject** @runtime.stackChainStart, align 4
   %2 = getelementptr { %runtime.stackChainObject*, i32, i8*, i8* }, { %runtime.stackChainObject*, i32, i8*, i8* }* %gc.stackobject, i32 0, i32 0
-  store %runtime.stackChainObject* %1, %runtime.stackChainObject** %2
+  store %runtime.stackChainObject* %1, %runtime.stackChainObject** %2, align 4
   %3 = bitcast { %runtime.stackChainObject*, i32, i8*, i8* }* %gc.stackobject to %runtime.stackChainObject*
-  store %runtime.stackChainObject* %3, %runtime.stackChainObject** @runtime.stackChainStart
+  store %runtime.stackChainObject* %3, %runtime.stackChainObject** @runtime.stackChainStart, align 4
   %arr = call [32 x i8]* @arrayAlloc()
   %arr.bitcast = getelementptr [32 x i8], [32 x i8]* %arr, i32 0, i32 0
   %4 = getelementptr { %runtime.stackChainObject*, i32, i8*, i8* }, { %runtime.stackChainObject*, i32, i8*, i8* }* %gc.stackobject, i32 0, i32 2
-  store i8* %arr.bitcast, i8** %4
+  store i8* %arr.bitcast, i8** %4, align 4
   %other = call i8* @runtime.alloc(i32 1)
   %5 = getelementptr { %runtime.stackChainObject*, i32, i8*, i8* }, { %runtime.stackChainObject*, i32, i8*, i8* }* %gc.stackobject, i32 0, i32 3
-  store i8* %other, i8** %5
-  store %runtime.stackChainObject* %1, %runtime.stackChainObject** @runtime.stackChainStart
+  store i8* %other, i8** %5, align 4
+  store %runtime.stackChainObject* %1, %runtime.stackChainObject** @runtime.stackChainStart, align 4
   ret void
 }
 
diff --git a/transform/testdata/maps.out.ll b/transform/testdata/maps.out.ll
index 925a61b0..81f9bada 100644
--- a/transform/testdata/maps.out.ll
+++ b/transform/testdata/maps.out.ll
@@ -17,14 +17,14 @@ define void @testUnused() {
 
 define i32 @testReadonly() {
   %map = call %runtime.hashmap* @runtime.hashmapMake(i8 4, i8 4, i32 0)
-  %hashmap.value = alloca i32
-  store i32 42, i32* %hashmap.value
+  %hashmap.value = alloca i32, align 4
+  store i32 42, i32* %hashmap.value, align 4
   %hashmap.value.bitcast = bitcast i32* %hashmap.value to i8*
   call void @runtime.hashmapStringSet(%runtime.hashmap* %map, i8* getelementptr inbounds ([6 x i8], [6 x i8]* @answer, i32 0, i32 0), i32 6, i8* %hashmap.value.bitcast)
-  %hashmap.value2 = alloca i32
+  %hashmap.value2 = alloca i32, align 4
   %hashmap.value2.bitcast = bitcast i32* %hashmap.value2 to i8*
   %commaOk = call i1 @runtime.hashmapStringGet(%runtime.hashmap* %map, i8* getelementptr inbounds ([6 x i8], [6 x i8]* @answer, i32 0, i32 0), i32 6, i8* %hashmap.value2.bitcast)
-  %loadedValue = load i32, i32* %hashmap.value2
+  %loadedValue = load i32, i32* %hashmap.value2, align 4
   ret i32 %loadedValue
 }
 
diff --git a/transform/testdata/stacksize.out.ll b/transform/testdata/stacksize.out.ll
index 08ad3022..29f51e34 100644
--- a/transform/testdata/stacksize.out.ll
+++ b/transform/testdata/stacksize.out.ll
@@ -11,7 +11,7 @@ declare void @"internal/task.start"(i32, i8*, i32)
 
 define void @Reset_Handler() {
 entry:
-  %stacksize1 = load i32, i32* getelementptr inbounds ([1 x i32], [1 x i32]* @"internal/task.stackSizes", i32 0, i32 0)
+  %stacksize1 = load i32, i32* getelementptr inbounds ([1 x i32], [1 x i32]* @"internal/task.stackSizes", i32 0, i32 0), align 4
   call void @"internal/task.start"(i32 ptrtoint (void (i8*)* @"runtime.run$1$gowrapper" to i32), i8* undef, i32 %stacksize1)
   ret void
 }
diff --git a/transform/testdata/wasm-abi.out.ll b/transform/testdata/wasm-abi.out.ll
index 68bed05e..cf63c3d7 100644
--- a/transform/testdata/wasm-abi.out.ll
+++ b/transform/testdata/wasm-abi.out.ll
@@ -4,24 +4,24 @@ target triple = "wasm32-unknown-unknown-wasm"
 declare i64 @"externalCall$i64wrap"(i8*, i32, i64)
 
 define internal i64 @testCall(i8* %ptr, i32 %len, i64 %foo) {
-  %i64asptr = alloca i64
-  %i64asptr1 = alloca i64
-  store i64 %foo, i64* %i64asptr1
+  %i64asptr = alloca i64, align 8
+  %i64asptr1 = alloca i64, align 8
+  store i64 %foo, i64* %i64asptr1, align 8
   call void @externalCall(i64* %i64asptr, i8* %ptr, i32 %len, i64* %i64asptr1)
-  %retval = load i64, i64* %i64asptr
+  %retval = load i64, i64* %i64asptr, align 8
   ret i64 %retval
 }
 
 define internal i64 @testCallNonEntry(i8* %ptr, i32 %len) {
 entry:
-  %i64asptr = alloca i64
-  %i64asptr1 = alloca i64
+  %i64asptr = alloca i64, align 8
+  %i64asptr1 = alloca i64, align 8
   br label %bb1
 
 bb1:                                              ; preds = %entry
-  store i64 3, i64* %i64asptr1
+  store i64 3, i64* %i64asptr1, align 8
   call void @externalCall(i64* %i64asptr, i8* %ptr, i32 %len, i64* %i64asptr1)
-  %retval = load i64, i64* %i64asptr
+  %retval = load i64, i64* %i64asptr, align 8
   ret i64 %retval
 }
 
@@ -39,7 +39,7 @@ declare void @externalCall(i64*, i8*, i32, i64*)
 
 define void @exportedFunction(i64* %0) {
 entry:
-  %i64 = load i64, i64* %0
+  %i64 = load i64, i64* %0, align 8
   call void @"exportedFunction$i64wrap"(i64 %i64)
   ret void
 }
diff --git a/transform/transform_test.go b/transform/transform_test.go
index f16baabd..9689a063 100644
--- a/transform/transform_test.go
+++ b/transform/transform_test.go
@@ -9,7 +9,6 @@ import (
 	"io/ioutil"
 	"os"
 	"path/filepath"
-	"regexp"
 	"strconv"
 	"strings"
 	"testing"
@@ -67,8 +66,6 @@ func testTransform(t *testing.T, pathPrefix string, transform func(mod llvm.Modu
 	}
 }
 
-var alignRegexp = regexp.MustCompile(", align [0-9]+$")
-
 // fuzzyEqualIR returns true if the two LLVM IR strings passed in are roughly
 // equal. That means, only relevant lines are compared (excluding comments
 // etc.).
@@ -80,15 +77,6 @@ func fuzzyEqualIR(s1, s2 string) bool {
 	}
 	for i, line1 := range lines1 {
 		line2 := lines2[i]
-		match1 := alignRegexp.MatchString(line1)
-		match2 := alignRegexp.MatchString(line2)
-		if match1 != match2 {
-			// Only one of the lines has the align keyword. Remove it.
-			// This is a change to make the test work in both LLVM 10 and LLVM
-			// 11 (LLVM 11 appears to automatically add alignment everywhere).
-			line1 = alignRegexp.ReplaceAllString(line1, "")
-			line2 = alignRegexp.ReplaceAllString(line2, "")
-		}
 		if line1 != line2 {
 			return false
 		}
@@ -117,19 +105,10 @@ func filterIrrelevantIRLines(lines []string) []string {
 		if strings.HasPrefix(line, "source_filename = ") {
 			continue
 		}
-		if llvmVersion < 10 && strings.HasPrefix(line, "attributes ") {
+		if llvmVersion < 11 && strings.HasPrefix(line, "attributes ") {
 			// Ignore attribute groups. These may change between LLVM versions.
-			// Right now test outputs are for LLVM 10.
 			continue
 		}
-		if llvmVersion < 10 && strings.HasPrefix(line, "define ") {
-			// Remove parameter values such as %0 in function definitions. These
-			// were added in LLVM 10 so to get the tests to pass on older
-			// versions, ignore them there (there are other tests that verify
-			// correct behavior).
-			re := regexp.MustCompile(` %[0-9]+(\)|,)`)
-			line = re.ReplaceAllString(line, "$1")
-		}
 		out = append(out, line)
 	}
 	return out