AES: remove crypto_fl_tab and replace crypto_il_tab with isb_tab

Remove crypto_fl_tab from aes implementation. Because mix_col(1,n) = n, all information in cryto_fl_tab is in crypto_ft_tab too. crypto_il_tab is replaced by isb_tab, the byte shift is done during decryption.

These changes reduce the encryption cache footprint to 50% and decryption cache footprint to 53.1%. The code size is increased slightly. On my Intel CORE micro-architecture CPU, there is almost no performance penalty.

This patch has been built and tested against 2.6.25-rc3-mm1.

This patch is only for generic C implementation. If this kind of idea is desired, I will implement the ASM version for x86.

---
 crypto/aes_generic.c |   47 ++++++++++++++++-------------------------------
 include/crypto/aes.h |    2 --
 2 files changed, 16 insertions(+), 33 deletions(-)

@@ -63,18 +63,14 @@ static inline u8 byte(const u32 x, const
 static u8 pow_tab[256] __initdata;
 static u8 log_tab[256] __initdata;
 static u8 sbx_tab[256] __initdata;
-static u8 isb_tab[256] __initdata;
+static u8 isb_tab[256];
 static u32 rco_tab[10];
 
 u32 crypto_ft_tab[4][256];
-u32 crypto_fl_tab[4][256];
 u32 crypto_it_tab[4][256];
-u32 crypto_il_tab[4][256];
 
 EXPORT_SYMBOL_GPL(crypto_ft_tab);
-EXPORT_SYMBOL_GPL(crypto_fl_tab);
 EXPORT_SYMBOL_GPL(crypto_it_tab);
-EXPORT_SYMBOL_GPL(crypto_il_tab);
 
 static inline u8 __init f_mult(u8 a, u8 b)
 {

@@ -122,12 +118,6 @@ static void __init gen_tabs(void)
 	for (i = 0; i < 256; ++i) {
 		p = sbx_tab[i];
 
-		t = p;
-		crypto_fl_tab[0][i] = t;
-		crypto_fl_tab[1][i] = rol32(t, 8);
-		crypto_fl_tab[2][i] = rol32(t, 16);
-		crypto_fl_tab[3][i] = rol32(t, 24);
-
 		t = ((u32) ff_mult(2, p)) |
 		    ((u32) p << 8) |
 		    ((u32) p << 16) | ((u32) ff_mult(3, p) << 24);

@@ -139,12 +129,6 @@ static void __init gen_tabs(void)
 
 		p = isb_tab[i];
 
-		t = p;
-		crypto_il_tab[0][i] = t;
-		crypto_il_tab[1][i] = rol32(t, 8);
-		crypto_il_tab[2][i] = rol32(t, 16);
-		crypto_il_tab[3][i] = rol32(t, 24);
-
 		t = ((u32) ff_mult(14, p)) |
 		    ((u32) ff_mult(9, p) << 8) |
 		    ((u32) ff_mult(13, p) << 16) |

@@ -173,10 +157,10 @@ static void __init gen_tabs(void)
 } while (0)
 
 #define ls_box(x)		\
-	crypto_fl_tab[0][byte(x, 0)] ^	\
-	crypto_fl_tab[1][byte(x, 1)] ^	\
-	crypto_fl_tab[2][byte(x, 2)] ^	\
-	crypto_fl_tab[3][byte(x, 3)]
+	(crypto_ft_tab[2][byte(x, 0)] & 0x000000ff) ^	\
+	(crypto_ft_tab[3][byte(x, 1)] & 0x0000ff00) ^	\
+	(crypto_ft_tab[0][byte(x, 2)] & 0x00ff0000) ^	\
+	(crypto_ft_tab[1][byte(x, 3)] & 0xff000000)
 
 #define loop4(i)	do {		\
 	t = ror32(t, 8);		\

@@ -303,11 +287,12 @@ EXPORT_SYMBOL_GPL(crypto_aes_set_key);
 	k += 4;			\
 } while (0)
 
-#define f_rl(bo, bi, n, k)	do {				\
-	bo[n] = crypto_fl_tab[0][byte(bi[n], 0)] ^			\
-		crypto_fl_tab[1][byte(bi[(n + 1) & 3], 1)] ^		\
-		crypto_fl_tab[2][byte(bi[(n + 2) & 3], 2)] ^		\
-		crypto_fl_tab[3][byte(bi[(n + 3) & 3], 3)] ^ *(k + n);	\
+#define f_rl(bo, bi, n, k)	do {					    \
+	bo[n] = (crypto_ft_tab[2][byte(bi[n], 0)] & 0x000000ff) ^ 	    \
+		(crypto_ft_tab[3][byte(bi[(n + 1) & 3], 1)] & 0x0000ff00) ^ \
+		(crypto_ft_tab[0][byte(bi[(n + 2) & 3], 2)] & 0x00ff0000) ^ \
+		(crypto_ft_tab[1][byte(bi[(n + 3) & 3], 3)] & 0xff000000) ^ \
+		*(k + n);						    \
 } while (0)
 
 #define f_lround(bo, bi, k)	do {\

@@ -375,11 +360,11 @@ static void aes_encrypt(struct crypto_tf
 	k += 4;			\
 } while (0)
 
-#define i_rl(bo, bi, n, k)	do {			\
-	bo[n] = crypto_il_tab[0][byte(bi[n], 0)] ^		\
-	crypto_il_tab[1][byte(bi[(n + 3) & 3], 1)] ^		\
-	crypto_il_tab[2][byte(bi[(n + 2) & 3], 2)] ^		\
-	crypto_il_tab[3][byte(bi[(n + 1) & 3], 3)] ^ *(k + n);	\
+#define i_rl(bo, bi, n, k)	do {					\
+	bo[n] = (u32)isb_tab[byte(bi[n], 0)] ^				\
+	((u32)isb_tab[byte(bi[(n + 3) & 3], 1)] << 8) ^			\
+	((u32)isb_tab[byte(bi[(n + 2) & 3], 2)] << 16) ^		\
+	((u32)isb_tab[byte(bi[(n + 1) & 3], 3)] << 24) ^ *(k + n);	\
 } while (0)
 
 #define i_lround(bo, bi, k)	do {\

@@ -22,9 +22,7 @@ struct crypto_aes_ctx {
 };
 
 extern u32 crypto_ft_tab[4][256];
-extern u32 crypto_fl_tab[4][256];
 extern u32 crypto_it_tab[4][256];
-extern u32 crypto_il_tab[4][256];
 
 int crypto_aes_set_key(struct crypto_tfm *tfm, const u8 *in_key,
 		unsigned int key_len);

Similar optimization is used by OpenSSL too. So I think this is reasonable.

Best Regards,
Huang Ying

nice.

Could you please run
| modprobe tcrypt mode=200
with and without the patch?

Sebastian

The files attached is the result.

Best Regards,
Huang Ying

testing speed of ecb(aes) encryption
test 0 (128 bit key, 16 byte blocks): 1 operation in 599 cycles (16 bytes) test 1 (128 bit key, 64 byte blocks): 1 operation in 1191 cycles (64 bytes) test 2 (128 bit key, 256 byte blocks): 1 operation in 3909 cycles (256 bytes) test 3 (128 bit key, 1024 byte blocks): 1 operation in 14758 cycles (1024 bytes) test 4 (128 bit key, 8192 byte blocks): 1 operation in 116484 cycles (8192 bytes) test 5 (192 bit key, 16 byte blocks): 1 operation in 546 cycles (16 bytes) test 6 (192 bit key, 64 byte blocks): 1 operation in 1349 cycles (64 bytes) test 7 (192 bit key, 256 byte blocks): 1 operation in 4550 cycles (256 bytes) test 8 (192 bit key, 1024 byte blocks): 1 operation in 17349 cycles (1024 bytes) test 9 (192 bit key, 8192 byte blocks): 1 operation in 138983 cycles (8192 bytes) test 10 (256 bit key, 16 byte blocks): 1 operation in 599 cycles (16 bytes) test 11 (256 bit key, 64 byte blocks): 1 operation in 1515 cycles (64 bytes) test 12 (256 bit key, 256 byte blocks): 1 operation in 5221 cycles (256 bytes) test 13 (256 bit key, 1024 byte blocks): 1 operation in 20040 cycles (1024 bytes) test 14 (256 bit key, 8192 byte blocks): 1 operation in 158428 cycles (8192 bytes)

testing speed of ecb(aes) decryption
test 0 (128 bit key, 16 byte blocks): 1 operation in 627 cycles (16 bytes) test 1 (128 bit key, 64 byte blocks): 1 operation in 1175 cycles (64 bytes) test 2 (128 bit key, 256 byte blocks): 1 operation in 3854 cycles (256 bytes) test 3 (128 bit key, 1024 byte blocks): 1 operation in 14590 cycles (1024 bytes) test 4 (128 bit key, 8192 byte blocks): 1 operation in 115228 cycles (8192 bytes) test 5 (192 bit key, 16 byte blocks): 1 operation in 544 cycles (16 bytes) test 6 (192 bit key, 64 byte blocks): 1 operation in 1334 cycles (64 bytes) test 7 (192 bit key, 256 byte blocks): 1 operation in 4496 cycles (256 bytes) test 8 (192 bit key, 1024 byte blocks): 1 operation in 17144 cycles (1024 bytes) test 9 (192 bit key, 8192 byte blocks): 1 operation in 137100 cycles (8192 bytes) test 10 (256 bit key, 16 byte blocks): 1 operation in 589 cycles (16 bytes) test 11 (256 bit key, 64 byte blocks): 1 operation in 1496 cycles (64 bytes) test 12 (256 bit key, 256 byte blocks): 1 operation in 5144 cycles (256 bytes) test 13 (256 bit key, 1024 byte blocks): 1 operation in 19735 cycles (1024 bytes) test 14 (256 bit key, 8192 byte blocks): 1 operation in 156450 cycles (8192 bytes)

testing speed of cbc(aes) encryption
test 0 (128 bit key, 16 byte blocks): 1 operation in 654 cycles (16 bytes) test 1 (128 bit key, 64 byte blocks): 1 operation in 1381 cycles (64 bytes) test 2 (128 bit key, 256 byte blocks): 1 operation in 4309 cycles (256 bytes) test 3 (128 bit key, 1024 byte blocks): 1 operation in 16019 cycles (1024 bytes) test 4 (128 bit key, 8192 byte blocks): 1 operation in 126050 cycles (8192 bytes) test 5 (192 bit key, 16 byte blocks): 1 operation in 689 cycles (16 bytes) test 6 (192 bit key, 64 byte blocks): 1 operation in 1535 cycles (64 bytes) test 7 (192 bit key, 256 byte blocks): 1 operation in 4945 cycles (256 bytes) test 8 (192 bit key, 1024 byte blocks): 1 operation in 18602 cycles (1024 bytes) test 9 (192 bit key, 8192 byte blocks): 1 operation in 147854 cycles (8192 bytes) test 10 (256 bit key, 16 byte blocks): 1 operation in 724 cycles (16 bytes) test 11 (256 bit key, 64 byte blocks): 1 operation in 1696 cycles (64 bytes) test 12 (256 bit key, 256 byte blocks): 1 operation in 5604 cycles (256 bytes) test 13 (256 bit key, 1024 byte blocks): 1 operation in 21218 cycles (1024 bytes) test 14 (256 bit key, 8192 byte blocks): 1 operation in 167954 cycles (8192 bytes)

testing speed of cbc(aes) decryption
test 0 (128 bit key, 16 byte blocks): 1 operation in 814 cycles (16 bytes) test 1 (128 bit key, 64 byte blocks): 1 operation in 1466 cycles (64 bytes) test 2 (128 bit key, 256 byte blocks): 1 operation in 4378 cycles (256 bytes) test 3 (128 bit key, 1024 byte blocks): 1 operation in 16025 cycles (1024 bytes) test 4 (128 bit key, 8192 byte blocks): 1 operation in 126127 cycles (8192 bytes) test 5 (192 bit key, 16 byte blocks): 1 operation in 769 cycles (16 bytes) test 6 (192 bit key, 64 byte blocks): 1 operation in 1616 cycles (64 bytes) test 7 (192 bit key, 256 byte blocks): 1 operation in 5012 cycles (256 bytes) test 8 (192 bit key, 1024 byte blocks): 1 operation in 18579 cycles (1024 bytes) test 9 (192 bit key, 8192 byte blocks): 1 operation in 147793 cycles (8192 bytes) test 10 (256 bit key, 16 byte blocks): 1 operation in 812 cycles (16 bytes) test 11 (256 bit key, 64 byte blocks): 1 operation in 1783 cycles (64 bytes) test 12 (256 bit key, 256 byte blocks): 1 operation in 5654 cycles (256 bytes) test 13 (256 bit key, 1024 byte blocks): 1 operation in 21182 cycles (1024 bytes) test 14 (256 bit key, 8192 byte blocks): 1 operation in 167341 cycles (8192 bytes)

testing speed of lrw(aes) encryption
test 0 (256 bit key, 16 byte blocks): 1 operation in 802 cycles (16 bytes) test 1 (256 bit key, 64 byte blocks): 1 operation in 1515 cycles (64 bytes) test 2 (256 bit key, 256 byte blocks): 1 operation in 4550 cycles (256 bytes) test 3 (256 bit key, 1024 byte blocks): 1 operation in 16583 cycles (1024 bytes) test 4 (256 bit key, 8192 byte blocks): 1 operation in 130258 cycles (8192 bytes) test 5 (320 bit key, 16 byte blocks): 1 operation in 780 cycles (16 bytes) test 6 (320 bit key, 64 byte blocks): 1 operation in 1678 cycles (64 bytes) test 7 (320 bit key, 256 byte blocks): 1 operation in 5194 cycles (256 bytes) test 8 (320 bit key, 1024 byte blocks): 1 operation in 19213 cycles (1024 bytes) test 9 (320 bit key, 8192 byte blocks): 1 operation in 152362 cycles (8192 bytes) test 10 (384 bit key, 16 byte blocks): 1 operation in 812 cycles (16 bytes) test 11 (384 bit key, 64 byte blocks): 1 operation in 1843 cycles (64 bytes) test 12 (384 bit key, 256 byte blocks): 1 operation in 5860 cycles (256 bytes) test 13 (384 bit key, 1024 byte blocks): 1 operation in 21879 cycles (1024 bytes) test 14 (384 bit key, 8192 byte blocks): 1 operation in 172419 cycles (8192 bytes)

testing speed of lrw(aes) decryption
test 0 (256 bit key, 16 byte blocks): 1 operation in 799 cycles (16 bytes) test 1 (256 bit key, 64 byte blocks): 1 operation in 1495 cycles (64 bytes) test 2 (256 bit key, 256 byte blocks): 1 operation in 4482 cycles (256 bytes) test 3 (256 bit key, 1024 byte blocks): 1 operation in 16413 cycles (1024 bytes) test 4 (256 bit key, 8192 byte blocks): 1 operation in 128887 cycles (8192 bytes) test 5 (320 bit key, 16 byte blocks): 1 operation in 771 cycles (16 bytes) test 6 (320 bit key, 64 byte blocks): 1 operation in 1658 cycles (64 bytes) test 7 (320 bit key, 256 byte blocks): 1 operation in 5136 cycles (256 bytes) test 8 (320 bit key, 1024 byte blocks): 1 operation in 19011 cycles (1024 bytes) test 9 (320 bit key, 8192 byte blocks): 1 operation in 150736 cycles (8192 bytes) test 10 (384 bit key, 16 byte blocks): 1 operation in 814 cycles (16 bytes) test 11 (384 bit key, 64 byte blocks): 1 operation in 1821 cycles (64 bytes) test 12 (384 bit key, 256 byte blocks): 1 operation in 5784 cycles (256 bytes) test 13 (384 bit key, 1024 byte blocks): 1 operation in 21629 cycles (1024 bytes) test 14 (384 bit key, 8192 byte blocks): 1 operation in 170319 cycles (8192 bytes)

testing speed of xts(aes) encryption
test 0 (256 bit key, 16 byte blocks): 1 operation in 833 cycles (16 bytes) test 1 (256 bit key, 64 byte blocks): 1 operation in 1509 cycles (64 bytes) test 2 (256 bit key, 256 byte blocks): 1 operation in 4458 cycles (256 bytes) test 3 (256 bit key, 1024 byte blocks): 1 operation in 16232 cycles (1024 bytes) test 4 (256 bit key, 8192 byte blocks): 1 operation in 126195 cycles (8192 bytes) test 5 (384 bit key, 16 byte blocks): 1 operation in 852 cycles (16 bytes) test 6 (384 bit key, 64 byte blocks): 1 operation in 1708 cycles (64 bytes) test 7 (384 bit key, 256 byte blocks): 1 operation in 5135 cycles (256 bytes) test 8 (384 bit key, 1024 byte blocks): 1 operation in 18802 cycles (1024 bytes) test 9 (384 bit key, 8192 byte blocks): 1 operation in 147053 cycles (8192 bytes) test 10 (512 bit key, 16 byte blocks): 1 operation in 938 cycles (16 bytes) test 11 (512 bit key, 64 byte blocks): 1 operation in 1924 cycles (64 bytes) test 12 (512 bit key, 256 byte blocks): 1 operation in 5850 cycles (256 bytes) test 13 (512 bit key, 1024 byte blocks): 1 operation in 21524 cycles (1024 bytes) test 14 (512 bit key, 8192 byte blocks): 1 operation in 168333 cycles (8192 bytes)

testing speed of xts(aes) decryption
test 0 (256 bit key, 16 byte blocks): 1 operation in 855 cycles (16 bytes) test 1 (256 bit key, 64 byte blocks): 1 operation in 1490 cycles (64 bytes) test 2 (256 bit key, 256 byte blocks): 1 operation in 4379 cycles (256 bytes) test 3 (256 bit key, 1024 byte blocks): 1 operation in 15931 cycles (1024 bytes) test 4 (256 bit key, 8192 byte blocks): 1 operation in 124614 cycles (8192 bytes) test 5 (384 bit key, 16 byte blocks): 1 operation in 847 cycles (16 bytes) test 6 (384 bit key, 64 byte blocks): 1 operation in 1684 cycles (64 bytes) test 7 (384 bit key, 256 byte blocks): 1 operation in 5062 cycles (256 bytes) test 8 (384 bit key, 1024 byte blocks): 1 operation in 18574 cycles (1024 bytes) test 9 (384 bit key, 8192 byte blocks): 1 operation in 145296 cycles (8192 bytes) test 10 (512 bit key, 16 byte blocks): 1 operation in 932 cycles (16 bytes) test 11 (512 bit key, 64 byte blocks): 1 operation in 1892 cycles (64 bytes) test 12 (512 bit key, 256 byte blocks): 1 operation in 5756 cycles (256 bytes) test 13 (512 bit key, 1024 byte blocks): 1 operation in 21227 cycles (1024 bytes) test 14 (512 bit key, 8192 byte blocks): 1 operation in 166316 cycles (8192 bytes)

testing speed of ecb(aes) encryption
test 0 (128 bit key, 16 byte blocks): 1 operation in 528 cycles (16 bytes) test 1 (128 bit key, 64 byte blocks): 1 operation in 1213 cycles (64 bytes) test 2 (128 bit key, 256 byte blocks): 1 operation in 3984 cycles (256 bytes) test 3 (128 bit key, 1024 byte blocks): 1 operation in 15060 cycles (1024 bytes) test 4 (128 bit key, 8192 byte blocks): 1 operation in 118884 cycles (8192 bytes) test 5 (192 bit key, 16 byte blocks): 1 operation in 558 cycles (16 bytes) test 6 (192 bit key, 64 byte blocks): 1 operation in 1362 cycles (64 bytes) test 7 (192 bit key, 256 byte blocks): 1 operation in 4618 cycles (256 bytes) test 8 (192 bit key, 1024 byte blocks): 1 operation in 17634 cycles (1024 bytes) test 9 (192 bit key, 8192 byte blocks): 1 operation in 141172 cycles (8192 bytes) test 10 (256 bit key, 16 byte blocks): 1 operation in 601 cycles (16 bytes) test 11 (256 bit key, 64 byte blocks): 1 operation in 1535 cycles (64 bytes) test 12 (256 bit key, 256 byte blocks): 1 operation in 5263 cycles (256 bytes) test 13 (256 bit key, 1024 byte blocks): 1 operation in 20236 cycles (1024 bytes) test 14 (256 bit key, 8192 byte blocks): 1 operation in 160585 cycles (8192 bytes)

testing speed of ecb(aes) decryption
test 0 (128 bit key, 16 byte blocks): 1 operation in 521 cycles (16 bytes) test 1 (128 bit key, 64 byte blocks): 1 operation in 1191 cycles (64 bytes) test 2 (128 bit key, 256 byte blocks): 1 operation in 3895 cycles (256 bytes) test 3 (128 bit key, 1024 byte blocks): 1 operation in 14735 cycles (1024 bytes) test 4 (128 bit key, 8192 byte blocks): 1 operation in 116355 cycles (8192 bytes) test 5 (192 bit key, 16 byte blocks): 1 operation in 554 cycles (16 bytes) test 6 (192 bit key, 64 byte blocks): 1 operation in 1352 cycles (64 bytes) test 7 (192 bit key, 256 byte blocks): 1 operation in 4542 cycles (256 bytes) test 8 (192 bit key, 1024 byte blocks): 1 operation in 17295 cycles (1024 bytes) test 9 (192 bit key, 8192 byte blocks): 1 operation in 138497 cycles (8192 bytes) test 10 (256 bit key, 16 byte blocks): 1 operation in 594 cycles (16 bytes) test 11 (256 bit key, 64 byte blocks): 1 operation in 1507 cycles (64 bytes) test 12 (256 bit key, 256 byte blocks): 1 operation in 5178 cycles (256 bytes) test 13 (256 bit key, 1024 byte blocks): 1 operation in 19855 cycles (1024 bytes) test 14 (256 bit key, 8192 byte blocks): 1 operation in 157464 cycles (8192 bytes)

testing speed of cbc(aes) encryption
test 0 (128 bit key, 16 byte blocks): 1 operation in 657 cycles (16 bytes) test 1 (128 bit key, 64 byte blocks): 1 operation in 1396 cycles (64 bytes) test 2 (128 bit key, 256 byte blocks): 1 operation in 4406 cycles (256 bytes) test 3 (128 bit key, 1024 byte blocks): 1 operation in 16393 cycles (1024 bytes) test 4 (128 bit key, 8192 byte blocks): 1 operation in 129278 cycles (8192 bytes) test 5 (192 bit key, 16 byte blocks): 1 operation in 691 cycles (16 bytes) test 6 (192 bit key, 64 byte blocks): 1 operation in 1641 cycles (64 bytes) test 7 (192 bit key, 256 byte blocks): 1 operation in 5187 cycles (256 bytes) test 8 (192 bit key, 1024 byte blocks): 1 operation in 19155 cycles (1024 bytes) test 9 (192 bit key, 8192 byte blocks): 1 operation in 151310 cycles (8192 bytes) test 10 (256 bit key, 16 byte blocks): 1 operation in 736 cycles (16 bytes) test 11 (256 bit key, 64 byte blocks): 1 operation in 1794 cycles (64 bytes) test 12 (256 bit key, 256 byte blocks): 1 operation in 5819 cycles (256 bytes) test 13 (256 bit key, 1024 byte blocks): 1 operation in 21769 cycles (1024 bytes) test 14 (256 bit key, 8192 byte blocks): 1 operation in 170831 cycles (8192 bytes)

testing speed of cbc(aes) decryption
test 0 (128 bit key, 16 byte blocks): 1 operation in 738 cycles (16 bytes) test 1 (128 bit key, 64 byte blocks): 1 operation in 1475 cycles (64 bytes) test 2 (128 bit key, 256 byte blocks): 1 operation in 4421 cycles (256 bytes) test 3 (128 bit key, 1024 byte blocks): 1 operation in 16220 cycles (1024 bytes) test 4 (128 bit key, 8192 byte blocks): 1 operation in 127525 cycles (8192 bytes) test 5 (192 bit key, 16 byte blocks): 1 operation in 771 cycles (16 bytes) test 6 (192 bit key, 64 byte blocks): 1 operation in 1626 cycles (64 bytes) test 7 (192 bit key, 256 byte blocks): 1 operation in 5045 cycles (256 bytes) test 8 (192 bit key, 1024 byte blocks): 1 operation in 18753 cycles (1024 bytes) test 9 (192 bit key, 8192 byte blocks): 1 operation in 149204 cycles (8192 bytes) test 10 (256 bit key, 16 byte blocks): 1 operation in 809 cycles (16 bytes) test 11 (256 bit key, 64 byte blocks): 1 operation in 1785 cycles (64 bytes) test 12 (256 bit key, 256 byte blocks): 1 operation in 5691 cycles (256 bytes) test 13 (256 bit key, 1024 byte blocks): 1 operation in 21275 cycles (1024 bytes) test 14 (256 bit key, 8192 byte blocks): 1 operation in 168499 cycles (8192 bytes)

testing speed of lrw(aes) encryption
test 0 (256 bit key, 16 byte blocks): 1 operation in 745 cycles (16 bytes) test 1 (256 bit key, 64 byte blocks): 1 operation in 1547 cycles (64 bytes) test 2 (256 bit key, 256 byte blocks): 1 operation in 4614 cycles (256 bytes) test 3 (256 bit key, 1024 byte blocks): 1 operation in 16885 cycles (1024 bytes) test 4 (256 bit key, 8192 byte blocks): 1 operation in 132411 cycles (8192 bytes) test 5 (320 bit key, 16 byte blocks): 1 operation in 778 cycles (16 bytes) test 6 (320 bit key, 64 byte blocks): 1 operation in 1689 cycles (64 bytes) test 7 (320 bit key, 256 byte blocks): 1 operation in 5260 cycles (256 bytes) test 8 (320 bit key, 1024 byte blocks): 1 operation in 19462 cycles (1024 bytes) test 9 (320 bit key, 8192 byte blocks): 1 operation in 154419 cycles (8192 bytes) test 10 (384 bit key, 16 byte blocks): 1 operation in 816 cycles (16 bytes) test 11 (384 bit key, 64 byte blocks): 1 operation in 1856 cycles (64 bytes) test 12 (384 bit key, 256 byte blocks): 1 operation in 5916 cycles (256 bytes) test 13 (384 bit key, 1024 byte blocks): 1 operation in 22159 cycles (1024 bytes) test 14 (384 bit key, 8192 byte blocks): 1 operation in 174627 cycles (8192 bytes)

testing speed of lrw(aes) decryption
test 0 (256 bit key, 16 byte blocks): 1 operation in 740 cycles (16 bytes) test 1 (256 bit key, 64 byte blocks): 1 operation in 1502 cycles (64 bytes) test 2 (256 bit key, 256 byte blocks): 1 operation in 4516 cycles (256 bytes) test 3 (256 bit key, 1024 byte blocks): 1 operation in 16549 cycles (1024 bytes) test 4 (256 bit key, 8192 byte blocks): 1 operation in 129912 cycles (8192 bytes) test 5 (320 bit key, 16 byte blocks): 1 operation in 774 cycles (16 bytes) test 6 (320 bit key, 64 byte blocks): 1 operation in 1659 cycles (64 bytes) test 7 (320 bit key, 256 byte blocks): 1 operation in 5175 cycles (256 bytes) test 8 (320 bit key, 1024 byte blocks): 1 operation in 19139 cycles (1024 bytes) test 9 (320 bit key, 8192 byte blocks): 1 operation in 151969 cycles (8192 bytes) test 10 (384 bit key, 16 byte blocks): 1 operation in 809 cycles (16 bytes) test 11 (384 bit key, 64 byte blocks): 1 operation in 1821 cycles (64 bytes) test 12 (384 bit key, 256 byte blocks): 1 operation in 5817 cycles (256 bytes) test 13 (384 bit key, 1024 byte blocks): 1 operation in 21775 cycles (1024 bytes) test 14 (384 bit key, 8192 byte blocks): 1 operation in 171423 cycles (8192 bytes)

testing speed of xts(aes) encryption
test 0 (256 bit key, 16 byte blocks): 1 operation in 788 cycles (16 bytes) test 1 (256 bit key, 64 byte blocks): 1 operation in 1537 cycles (64 bytes) test 2 (256 bit key, 256 byte blocks): 1 operation in 4526 cycles (256 bytes) test 3 (256 bit key, 1024 byte blocks): 1 operation in 16521 cycles (1024 bytes) test 4 (256 bit key, 8192 byte blocks): 1 operation in 128545 cycles (8192 bytes) test 5 (384 bit key, 16 byte blocks): 1 operation in 866 cycles (16 bytes) test 6 (384 bit key, 64 byte blocks): 1 operation in 1740 cycles (64 bytes) test 7 (384 bit key, 256 byte blocks): 1 operation in 5183 cycles (256 bytes) test 8 (384 bit key, 1024 byte blocks): 1 operation in 19048 cycles (1024 bytes) test 9 (384 bit key, 8192 byte blocks): 1 operation in 149163 cycles (8192 bytes) test 10 (512 bit key, 16 byte blocks): 1 operation in 945 cycles (16 bytes) test 11 (512 bit key, 64 byte blocks): 1 operation in 1931 cycles (64 bytes) test 12 (512 bit key, 256 byte blocks): 1 operation in 5903 cycles (256 bytes) test 13 (512 bit key, 1024 byte blocks): 1 operation in 21744 cycles (1024 bytes) test 14 (512 bit key, 8192 byte blocks): 1 operation in 170528 cycles (8192 bytes)

testing speed of xts(aes) decryption
test 0 (256 bit key, 16 byte blocks): 1 operation in 783 cycles (16 bytes) test 1 (256 bit key, 64 byte blocks): 1 operation in 1507 cycles (64 bytes) test 2 (256 bit key, 256 byte blocks): 1 operation in 4436 cycles (256 bytes) test 3 (256 bit key, 1024 byte blocks): 1 operation in 16192 cycles (1024 bytes) test 4 (256 bit key, 8192 byte blocks): 1 operation in 125962 cycles (8192 bytes) test 5 (384 bit key, 16 byte blocks): 1 operation in 853 cycles (16 bytes) test 6 (384 bit key, 64 byte blocks): 1 operation in 1701 cycles (64 bytes) test 7 (384 bit key, 256 byte blocks): 1 operation in 5120 cycles (256 bytes) test 8 (384 bit key, 1024 byte blocks): 1 operation in 18779 cycles (1024 bytes) test 9 (384 bit key, 8192 byte blocks): 1 operation in 146792 cycles (8192 bytes) test 10 (512 bit key, 16 byte blocks): 1 operation in 937 cycles (16 bytes) test 11 (512 bit key, 64 byte blocks): 1 operation in 1906 cycles (64 bytes) test 12 (512 bit key, 256 byte blocks): 1 operation in 5804 cycles (256 bytes) test 13 (512 bit key, 1024 byte blocks): 1 operation in 21413 cycles (1024 bytes) test 14 (512 bit key, 8192 byte blocks): 1 operation in 167733 cycles (8192 bytes)

According to the tcrypt numbers you've posted it is getting "slightly" slower on encryption/decryption of 64+ bytes. How did you measure your "almost no performance penalty"?

Sebastian