diff options
Diffstat (limited to 'Documentation/mtd/nand_ecc.txt')
| -rw-r--r-- | Documentation/mtd/nand_ecc.txt | 12 |
1 files changed, 6 insertions, 6 deletions
diff --git a/Documentation/mtd/nand_ecc.txt b/Documentation/mtd/nand_ecc.txt index bdf93b7..274821b 100644 --- a/Documentation/mtd/nand_ecc.txt +++ b/Documentation/mtd/nand_ecc.txt @@ -50,7 +50,7 @@ byte 255: bit7 bit6 bit5 bit4 bit3 bit2 bit1 bit0 rp1 rp3 rp5 ... rp15 cp5 cp5 cp5 cp5 cp4 cp4 cp4 cp4 This figure represents a sector of 256 bytes. -cp is my abbreviaton for column parity, rp for row parity. +cp is my abbreviation for column parity, rp for row parity. Let's start to explain column parity. cp0 is the parity that belongs to all bit0, bit2, bit4, bit6. @@ -560,7 +560,7 @@ Measuring this code again showed big gain. When executing the original linux code 1 million times, this took about 1 second on my system. (using time to measure the performance). After this iteration I was back to 0.075 sec. Actually I had to decide to start measuring over 10 -million interations in order not to loose too much accuracy. This one +million iterations in order not to lose too much accuracy. This one definitely seemed to be the jackpot! There is a little bit more room for improvement though. There are three @@ -571,8 +571,8 @@ loop; This eliminates 3 statements per loop. Of course after the loop we need to correct by adding: rp4 ^= rp4_6; rp6 ^= rp4_6 -Furthermore there are 4 sequential assingments to rp8. This can be -encoded slightly more efficient by saving tmppar before those 4 lines +Furthermore there are 4 sequential assignments to rp8. This can be +encoded slightly more efficiently by saving tmppar before those 4 lines and later do rp8 = rp8 ^ tmppar ^ notrp8; (where notrp8 is the value of rp8 before those 4 lines). Again a use of the commutative property of xor. @@ -622,7 +622,7 @@ Not a big change, but every penny counts :-) Analysis 7 ========== -Acutally this made things worse. Not very much, but I don't want to move +Actually this made things worse. Not very much, but I don't want to move into the wrong direction. Maybe something to investigate later. Could have to do with caching again. @@ -642,7 +642,7 @@ Analysis 8 This makes things worse. Let's stick with attempt 6 and continue from there. Although it seems that the code within the loop cannot be optimised further there is still room to optimize the generation of the ecc codes. -We can simply calcualate the total parity. If this is 0 then rp4 = rp5 +We can simply calculate the total parity. If this is 0 then rp4 = rp5 etc. If the parity is 1, then rp4 = !rp5; But if rp4 = rp5 we do not need rp5 etc. We can just write the even bits in the result byte and then do something like |
