0.2.0 FastShiftInOut

2024-10-03 18:09:02 -04:00 · 2024-09-19 14:49:14 +02:00 · 2024-09-19 14:49:14 +02:00 · 188019a4fb
commit 188019a4fb
parent 4bb93dad77
8 changed files with 419 additions and 137 deletions
--- a/libraries/FastShiftInOut/CHANGELOG.md
+++ b/libraries/FastShiftInOut/CHANGELOG.md
@ -6,11 +6,18 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/)
 and this project adheres to [Semantic Versioning](http://semver.org/).


-## [0.1.3] - 2023-02-20
+## [0.2.0] - 2024-09-10
+- fix #7, loop unroll option, improving performance, kudos to nt314p
+- added flag to select LOOP UNROLL (is optional as it gives larger code size)
+- update readme.md
+- minor edits
+
+----
+
+## [0.1.4] - 2023-02-20
 - update readme.md
 - update keywords.txt

-
 ## [0.1.3] - 2023-02-20
 - optimized noInterrupts
 - add lastRead()
@ -20,7 +27,7 @@ and this project adheres to [Semantic Versioning](http://semver.org/).
 - minor edits

 ## [0.1.2] - 2022-11-06
- redo clock pulse to match fastShiftOut  
+- redo clock pulse to match fastShiftOut
  (after write and before read)

 ## [0.1.1] - 2022-11-05
--- a/libraries/FastShiftInOut/FastShiftInOut.cpp
+++ b/libraries/FastShiftInOut/FastShiftInOut.cpp
@ -1,7 +1,7 @@
 //
 //    FILE: FastShiftInOut.cpp
 //  AUTHOR: Rob Tillaart
-// VERSION: 0.1.4
+// VERSION: 0.2.0
 // PURPOSE: Arduino library for (AVR) optimized shiftInOut (simultaneously)
 //     URL: https://github.com/RobTillaart/FastShiftInOut

@ -40,6 +40,7 @@ FastShiftInOut::FastShiftInOut(uint8_t dataIn, uint8_t dataOut, uint8_t clockPin
  _clockPin   = clockPin;

 #endif
+
  _lastValue = 0;
  _lastRead  = 0;
 }
@ -55,111 +56,6 @@ uint8_t FastShiftInOut::write(uint8_t data)
 }


-uint8_t FastShiftInOut::writeLSBFIRST(uint8_t data)
-{
-  uint8_t rv    = 0;
-  uint8_t value = data;
-  _lastValue = value;
-
-#if defined(ARDUINO_ARCH_AVR) || defined(ARDUINO_ARCH_MEGAAVR)
-
-  uint8_t cbmask1  = _clockBit;
-  uint8_t cbmask2  = ~_clockBit;
-  uint8_t inmask1  = _dataInBit;
-  uint8_t outmask1 = _dataOutBit;
-  uint8_t outmask2 = ~_dataOutBit;
-
-  uint8_t oldSREG = SREG;
-  noInterrupts();
-  for (uint8_t m = 1; m > 0; m <<= 1)
-  {
-    //  write one bit
-    if ((value & m) == 0) *_dataOutRegister &= outmask2;
-    else                  *_dataOutRegister |= outmask1;
-    //  clock pulse HIGH
-    *_clockRegister |= cbmask1;
-    //  read one bit
-    if ((*_dataInRegister & inmask1) > 0) rv |= m;
-    //  clock pulse LOW
-    *_clockRegister &= cbmask2;
-  }
-  SREG = oldSREG;
-
-#else
-
-  for (uint8_t i = 0; i < 8; i++)
-  {
-    //  write one bit
-    digitalWrite(_dataPinOut, value & 0x01);
-    value >>= 1;
-    //  clock pulse
-    digitalWrite(_clockPin, HIGH);
-    //  read one bit
-    rv >>= 1;
-    if (digitalRead(_dataPinIn) == HIGH)  rv |= 0x80;
-    //  clock pulse
-    digitalWrite(_clockPin, LOW);
-  }
-
-#endif
-
-  _lastRead = rv;
-  return rv;
-}
-
-
-uint8_t FastShiftInOut::writeMSBFIRST(uint8_t data)
-{
-  uint8_t rv    = 0;
-  uint8_t value = data;
-  _lastValue = value;
-
-#if defined(ARDUINO_ARCH_AVR) || defined(ARDUINO_ARCH_MEGAAVR)
-
-  uint8_t cbmask1  = _clockBit;
-  uint8_t cbmask2  = ~_clockBit;
-  uint8_t inmask1  = _dataInBit;
-  uint8_t outmask1 = _dataOutBit;
-  uint8_t outmask2 = ~_dataOutBit;
-
-  uint8_t oldSREG = SREG;
-  noInterrupts();
-  for (uint8_t m = 0x80; m > 0; m >>= 1)
-  {
-    //  write one bit
-    if ((value & m) == 0) *_dataOutRegister &= outmask2;
-    else                  *_dataOutRegister |= outmask1;
-    //  clock pulse HIGH
-    *_clockRegister |= cbmask1;
-    //  read one bit
-    if ((*_dataInRegister & inmask1) > 0) rv |= m;
-    //  clock pulse LOW
-    *_clockRegister &= cbmask2;
-  }
-  SREG = oldSREG;
-
-#else
-
-  for (uint8_t i = 0; i < 8; i++)
-  {
-    // write one bit
-    digitalWrite(_dataPinOut, value & 0x80);
-    value <<= 1;
-    //  clock pulse
-    digitalWrite(_clockPin, HIGH);
-    //  read one bit
-    rv <<= 1;
-    if (digitalRead(_dataPinIn) == HIGH) rv |= 1;
-    //  clock pulse
-    digitalWrite(_clockPin, LOW);
-  }
-
-#endif
-
-  _lastRead = rv;
-  return rv;
-}
-

 uint8_t FastShiftInOut::lastWritten(void)
 {
@ -190,5 +86,287 @@ uint8_t FastShiftInOut::getBitOrder(void)
 };


+uint8_t FastShiftInOut::writeLSBFIRST(uint8_t data)
+{
+  uint8_t rv    = 0;
+  uint8_t value = data;
+  _lastValue = value;
+
+#if defined(ARDUINO_ARCH_AVR) || defined(ARDUINO_ARCH_MEGAAVR)
+
+#if defined(FASTSHIFTINOUT_AVR_LOOP_UNROLLED)  //  AVR SPEED OPTIMIZED
+
+  uint8_t cbmask1  = _clockBit;
+  uint8_t inmask1  = _dataInBit;
+  uint8_t outmask1 = _dataOutBit;
+  uint8_t outmask2 = ~_dataOutBit;
+
+  volatile uint8_t* localDataInRegister  = _dataInRegister;
+  volatile uint8_t* localDataOutRegister = _dataOutRegister;
+  volatile uint8_t* localClockRegister   = _clockRegister;
+
+  //  disable interrupts (for all bits)
+  uint8_t oldSREG = SREG;
+  noInterrupts();
+
+  if ((value & 0x01) == 0) *localDataOutRegister &= outmask2;
+  else                     *localDataOutRegister |= outmask1;
+  // *localClockRegister |= cbmask1;
+  //  if ((*localDataInRegister & inmask1) > 0) rv |= 0x01;
+  // *localClockRegister &= cbmask2;  // ~_clockBit;
+  //  following code is allowed as interrupts are disabled.
+  //  so register can not change
+  uint8_t r = *localClockRegister;
+  *localClockRegister = r | cbmask1;  //  set one bit
+  if ((*localDataInRegister & inmask1) > 0) rv |= 0x01;
+  *localClockRegister = r;            //  reset it
+
+  if ((value & 0x02) == 0) *localDataOutRegister &= outmask2;
+  else                     *localDataOutRegister |= outmask1;
+  r = *localClockRegister;
+  *localClockRegister = r | cbmask1;  //  set one bit
+  if ((*localDataInRegister & inmask1) > 0) rv |= 0x02;
+  *localClockRegister = r;            //  reset it
+
+  if ((value & 0x04) == 0) *localDataOutRegister &= outmask2;
+  else                     *localDataOutRegister |= outmask1;
+  r = *localClockRegister;
+  *localClockRegister = r | cbmask1;  //  set one bit
+  if ((*localDataInRegister & inmask1) > 0) rv |= 0x04;
+  *localClockRegister = r;            //  reset it
+
+  if ((value & 0x08) == 0) *localDataOutRegister &= outmask2;
+  else                     *localDataOutRegister |= outmask1;
+  r = *localClockRegister;
+  *localClockRegister = r | cbmask1;  //  set one bit
+  if ((*localDataInRegister & inmask1) > 0) rv |= 0x08;
+  *localClockRegister = r;            //  reset it
+
+  if ((value & 0x10) == 0) *localDataOutRegister &= outmask2;
+  else                     *localDataOutRegister |= outmask1;
+  r = *localClockRegister;
+  *localClockRegister = r | cbmask1;  //  set one bit
+  if ((*localDataInRegister & inmask1) > 0) rv |= 0x10;
+  *localClockRegister = r;            //  reset it
+
+  if ((value & 0x20) == 0) *localDataOutRegister &= outmask2;
+  else                     *localDataOutRegister |= outmask1;
+  r = *localClockRegister;
+  *localClockRegister = r | cbmask1;  //  set one bit
+  if ((*localDataInRegister & inmask1) > 0) rv |= 0x20;
+  *localClockRegister = r;            //  reset it
+
+  if ((value & 0x40) == 0) *localDataOutRegister &= outmask2;
+  else                     *localDataOutRegister |= outmask1;
+  r = *localClockRegister;
+  *localClockRegister = r | cbmask1;  //  set one bit
+  if ((*localDataInRegister & inmask1) > 0) rv |= 0x40;
+  *localClockRegister = r;            //  reset it
+
+  if ((value & 0x80) == 0) *localDataOutRegister &= outmask2;
+  else                     *localDataOutRegister |= outmask1;
+  r = *localClockRegister;
+  *localClockRegister = r | cbmask1;  //  set one bit
+  if ((*localDataInRegister & inmask1) > 0) rv |= 0x80;
+  *localClockRegister = r;            //  reset it
+
+  SREG = oldSREG;
+
+#else  //  AVR SIZE OPTIMIZED
+
+  uint8_t cbmask1  = _clockBit;
+  uint8_t inmask1  = _dataInBit;
+  uint8_t outmask1 = _dataOutBit;
+  uint8_t outmask2 = ~_dataOutBit;
+
+  volatile uint8_t* localDataInRegister  = _dataInRegister;
+  volatile uint8_t* localDataOutRegister = _dataOutRegister;
+  volatile uint8_t* localClockRegister   = _clockRegister;
+ 
+  uint8_t oldSREG = SREG;
+  noInterrupts();
+  
+  uint8_t r = *localClockRegister;
+  
+  for (uint8_t m = 1; m > 0; m <<= 1)
+  {
+    //  write one bit
+    if ((value & m) == 0) *localDataOutRegister &= outmask2;
+    else                  *localDataOutRegister |= outmask1;
+    //  clock pulse HIGH
+    *localClockRegister |= cbmask1;
+    //  read one bit
+    if ((*localDataInRegister & inmask1) > 0) rv |= m;
+    //  clock pulse LOW
+    *localClockRegister = r;
+  }
+  SREG = oldSREG;
+
+#endif  //  if (AVR)
+
+#else   //  other platforms reference implementation
+
+  for (uint8_t i = 0; i < 8; i++)
+  {
+    //  write one bit
+    digitalWrite(_dataPinOut, value & 0x01);
+    value >>= 1;
+    //  clock pulse
+    digitalWrite(_clockPin, HIGH);
+    //  read one bit
+    rv >>= 1;
+    if (digitalRead(_dataPinIn) == HIGH)  rv |= 0x80;
+    //  clock pulse
+    digitalWrite(_clockPin, LOW);
+  }
+
+#endif
+
+  _lastRead = rv;
+  return rv;
+}
+
+
+uint8_t FastShiftInOut::writeMSBFIRST(uint8_t data)
+{
+  uint8_t rv    = 0;
+  uint8_t value = data;
+  _lastValue = value;
+
+#if defined(ARDUINO_ARCH_AVR) || defined(ARDUINO_ARCH_MEGAAVR)
+
+#if defined(FASTSHIFTINOUT_AVR_LOOP_UNROLLED)  //  AVR SPEED OPTIMIZED
+
+  uint8_t cbmask1  = _clockBit;
+  uint8_t inmask1  = _dataInBit;
+  uint8_t outmask1 = _dataOutBit;
+  uint8_t outmask2 = ~_dataOutBit;
+
+  volatile uint8_t* localDataInRegister  = _dataInRegister;
+  volatile uint8_t* localDataOutRegister = _dataOutRegister;
+  volatile uint8_t* localClockRegister   = _clockRegister;
+ 
+  uint8_t oldSREG = SREG;
+  noInterrupts();
+
+  if ((value & 0x80) == 0) *localDataOutRegister &= outmask2;
+  else                     *localDataOutRegister |= outmask1;
+  // *localClockRegister |= cbmask1;
+  //  if ((*localDataInRegister & inmask1) > 0) rv |= 0x80;
+  // *localClockRegister &= cbmask2;  // ~_clockBit;
+  //  following code is allowed as interrupts are disabled.
+  //  so register can not change
+  uint8_t r = *localClockRegister;
+  *localClockRegister = r | cbmask1;  //  set one bit
+  if ((*localDataInRegister & inmask1) > 0) rv |= 0x80;
+  *localClockRegister = r;            //  reset it
+
+  if ((value & 0x40) == 0) *localDataOutRegister &= outmask2;
+  else                     *localDataOutRegister |= outmask1;
+  r = *localClockRegister;
+  *localClockRegister = r | cbmask1;  //  set one bit
+  if ((*localDataInRegister & inmask1) > 0) rv |= 0x40;
+  *localClockRegister = r;            //  reset it
+
+  if ((value & 0x20) == 0) *localDataOutRegister &= outmask2;
+  else                     *localDataOutRegister |= outmask1;
+  r = *localClockRegister;
+  *localClockRegister = r | cbmask1;  //  set one bit
+  if ((*localDataInRegister & inmask1) > 0) rv |= 0x20;
+  *localClockRegister = r;            //  reset it
+
+  if ((value & 0x10) == 0) *localDataOutRegister &= outmask2;
+  else                     *localDataOutRegister |= outmask1;
+  r = *localClockRegister;
+  *localClockRegister = r | cbmask1;  //  set one bit
+  if ((*localDataInRegister & inmask1) > 0) rv |= 0x10;
+  *localClockRegister = r;            //  reset it
+
+  if ((value & 0x08) == 0) *localDataOutRegister &= outmask2;
+  else                     *localDataOutRegister |= outmask1;
+  r = *localClockRegister;
+  *localClockRegister = r | cbmask1;  //  set one bit
+  if ((*localDataInRegister & inmask1) > 0) rv |= 0x08;
+  *localClockRegister = r;            //  reset it
+
+  if ((value & 0x04) == 0) *localDataOutRegister &= outmask2;
+  else                     *localDataOutRegister |= outmask1;
+  r = *localClockRegister;
+  *localClockRegister = r | cbmask1;  //  set one bit
+  if ((*localDataInRegister & inmask1) > 0) rv |= 0x04;
+  *localClockRegister = r;            //  reset it
+
+  if ((value & 0x02) == 0) *localDataOutRegister &= outmask2;
+  else                     *localDataOutRegister |= outmask1;
+  r = *localClockRegister;
+  *localClockRegister = r | cbmask1;  //  set one bit
+  if ((*localDataInRegister & inmask1) > 0) rv |= 0x02;
+  *localClockRegister = r;            //  reset it
+
+  if ((value & 0x01) == 0) *localDataOutRegister &= outmask2;
+  else                     *localDataOutRegister |= outmask1;
+  r = *localClockRegister;
+  *localClockRegister = r | cbmask1;  //  set one bit
+  if ((*localDataInRegister & inmask1) > 0) rv |= 0x01;
+  *localClockRegister = r;            //  reset it
+
+  SREG = oldSREG;
+
+#else  //  AVR SIZE OPTIMIZED
+
+  uint8_t cbmask1  = _clockBit;
+  uint8_t inmask1  = _dataInBit;
+  uint8_t outmask1 = _dataOutBit;
+  uint8_t outmask2 = ~_dataOutBit;
+
+  volatile uint8_t* localDataInRegister  = _dataInRegister;
+  volatile uint8_t* localDataOutRegister = _dataOutRegister;
+  volatile uint8_t* localClockRegister   = _clockRegister;
+
+  uint8_t oldSREG = SREG;
+  noInterrupts();
+
+  uint8_t r = *localClockRegister;
+  for (uint8_t m = 0x80; m > 0; m >>= 1)
+  {
+    //  write one bit
+    if ((value & m) == 0) *localDataOutRegister &= outmask2;
+    else                  *localDataOutRegister |= outmask1;
+    //  clock pulse HIGH
+    *localClockRegister |= cbmask1;
+    //  read one bit
+    if ((*localDataInRegister & inmask1) > 0) rv |= m;
+    //  clock pulse LOW
+    *localClockRegister = r;
+  }
+  
+  //  reset interrupts flag to previous state
+  SREG = oldSREG;
+
+#endif  //  if (AVR)
+
+#else   //  other platforms reference implementation
+
+  for (uint8_t i = 0; i < 8; i++)
+  {
+    // write one bit
+    digitalWrite(_dataPinOut, value & 0x80);
+    value <<= 1;
+    //  clock pulse
+    digitalWrite(_clockPin, HIGH);
+    //  read one bit
+    rv <<= 1;
+    if (digitalRead(_dataPinIn) == HIGH) rv |= 1;
+    //  clock pulse
+    digitalWrite(_clockPin, LOW);
+  }
+
+#endif
+
+  _lastRead = rv;
+  return rv;
+}
+
+
 //  -- END OF FILE --

--- a/libraries/FastShiftInOut/FastShiftInOut.h
+++ b/libraries/FastShiftInOut/FastShiftInOut.h
@ -2,7 +2,7 @@
 //
 //    FILE: FastShiftInOut.cpp
 //  AUTHOR: Rob Tillaart
-// VERSION: 0.1.4
+// VERSION: 0.2.0
 // PURPOSE: Arduino library for (AVR) optimized shiftInOut (simultaneously)
 //     URL: https://github.com/RobTillaart/FastShiftInOut

@ -10,7 +10,10 @@
 #include "Arduino.h"


-#define FASTSHIFTINOUT_LIB_VERSION        (F("0.1.4"))
+#define FASTSHIFTINOUT_LIB_VERSION        (F("0.2.0"))
+
+//  uncomment next line to get SPEED OPTIMIZED CODE
+#define FASTSHIFTINOUT_AVR_LOOP_UNROLLED    1


 class FastShiftInOut
@ -28,7 +31,7 @@ public:
  bool    setBitOrder(uint8_t bitOrder);
  uint8_t getBitOrder(void);

-  // overrule bitOrder (most optimized).
+  //  overrule bitOrder (most optimized).
  uint8_t writeLSBFIRST(uint8_t data);
  uint8_t writeMSBFIRST(uint8_t data);

--- a/libraries/FastShiftInOut/README.md
+++ b/libraries/FastShiftInOut/README.md
@ -13,33 +13,78 @@

 Arduino library for **AVR** optimized shiftInOut (simultaneously).

-Related libraries
- https://github.com/RobTillaart/FastShiftIn
- https://github.com/RobTillaart/FastShiftOut
- https://github.com/RobTillaart/ShiftInSlow
- https://github.com/RobTillaart/ShiftOutSlow
-
-
 ## Description

-FastShiftInOut is a class that can send and receive bytes simultaneously.
-In that sense it mimics a SPI bus.
+**Experimental**

-Experimental.
+FastShiftInOut is a class that has optimized code (**AVR** only) to send and receive 
+bytes simultaneously. In that sense it mimics a SPI bus.
+It speeds up the shift using low level ports and masks. These are predetermined
+in the constructor of the FastShiftOut object.
+
+If not an **ARDUINO_ARCH_AVR** or **ARDUINO_ARCH_MEGAAVR** the class falls back 
+to a default non optimized implementation. 
+
+The library allows to set (and get) the bitOrder and apply this to multiple write()
+calls. It also provide access to **writeLSBFIRST()** and **writeMSBFIRST()** which 
+are the low level workers and most optimized code (so far).
+
+Note: the bitOrder of the byte read and the byte written are the same.


-## Performance
+### 0.2.0 breaking changes

-performance of **write()**
+The 0.2.0 version has a flag to unroll the inner loop in **writeLSBFIRST()**
+and **writeMSBFIRST()**. The unrolled loop blocks the interrupts per byte.
+
+Note: this optimization is new and thus experimental.
+Feedback, including improvements, is welcome.
+
+
+### Performance
+
+
+#### Measurements (pre 0.2.0)
+
+Performance of **write()**

 |  version  |  UNO (us)  |  ESP32 (us)  |
 |:---------:|-----------:|-------------:|
 |   0.1.0   |   181.08   |     4.32     |
 |   0.1.1   |    26.84   |     4.32     |
-|   0.1.2   |    26.84   |   no data    |
+|   0.1.2   |    26.84   |    no data   |
 |   0.1.3   |    25.52   |     4.32     |


+#### Measurements
+
+(0.2.0)  
+Indicative time in microseconds, Arduino UNO, IDE 1.8.19, measured over 1000 calls.  
+(delta between 2 calls and 1 call to eliminate overhead)
+
+|  function                |   0.1.3  |   0.2.0  |   0.2.0L  |
+|:-------------------------|---------:|---------:|----------:|
+|  write() (reference)     | no data  |  158.24  |  no data  |
+|  write()                 |   25.52  |   17.61  |    12.26  |
+|  writeLSBFIRST()         |   25.52  |   17.61  |    12.26  |
+|  writeMSBFIRST()         |   25.52  |   17.60  |    12.20  |
+
+
+- Note: 0.1.3 added from old table.
+- Note: reference run on AVR by commenting all optimizations.
+- Note: 0.2.0 measured with loop unroll flag disabled.
+- Note: 0.2.0L measured with loop unrolled flag enabled.
+
+
+### Related
+
+- https://github.com/RobTillaart/FastShiftIn
+- https://github.com/RobTillaart/FastShiftInOut
+- https://github.com/RobTillaart/FastShiftOut
+- https://github.com/RobTillaart/ShiftInSlow
+- https://github.com/RobTillaart/ShiftOutSlow
+
+
 ## Interface

 ```cpp
@ -54,29 +99,34 @@ bitOrder = { LSBFIRST, MSBFIRST };
 - **uint8_t write(uint8_t data)** reads and writes simultaneously.
 - **uint8_t lastWritten(void)** returns last byte written.
 - **uint8_t lastRead(void)** returns last byte read.
+- **uint8_t writeLSBFIRST(uint8_t data)** lowest level function, optimized for LSB.
+- **uint8_t writeMSBFIRST(uint8_t data)** lowest level function, optimized for MSB.
+
+
+### BitOrder
+
 - **bool setBitOrder(uint8_t bitOrder)** bitOrder must be LSBFIRST or MSBFIRST.
 - **uint8_t getBitOrder(void)** idem.
- **uint8_t writeLSBFIRST(uint8_t data)** optimized version, in practice almost no difference.
- **uint8_t writeMSBFIRST(uint8_t data)** optimized version, in practice almost no difference.


 ## Future

 #### Must

- documentation
- follow FastShiftIn and FastShiftOut
+- Update documentation
+- Follow FastShiftIn and FastShiftOut

 #### Should


 #### Could

- **void ignoreRead()**
- add Print interface?

 #### Wont

+- **void ignoreRead()** => would in effect be FastShiftIn()
+- add Print interface?
+  - meaning of the return value is not defined.

 ## Support

--- a/libraries/FastShiftInOut/examples/FastShiftInOut_performance/FastShiftInOut_performance.ino
+++ b/libraries/FastShiftInOut/examples/FastShiftInOut_performance/FastShiftInOut_performance.ino
@ -37,7 +37,8 @@ void test1()
  duration1 = micros() - start;
  Serial.print("        write: ");
  Serial.println(duration1 * 0.001);
-
+  delay(100);
+  
  start = micros();
  for (int i = 0; i < 1000; i++)
  {
@ -64,7 +65,8 @@ void test2()
  duration1 = micros() - start;
  Serial.print("writeLSBFIRST: ");
  Serial.println(duration1 * 0.001);
-
+  delay(100);
+  
  start = micros();
  for (int i = 0; i < 1000; i++)
  {
@ -91,7 +93,8 @@ void test3()
  duration1 = micros() - start;
  Serial.print("writeMSBFIRST: ");
  Serial.println(duration1 * 0.001);
-
+  delay(100);
+  
  start = micros();
  for (int i = 0; i < 1000; i++)
  {
@ -113,4 +116,4 @@ void loop()
 }


-// -- END OF FILE --
+// -- END OF FILE --
--- a/libraries/FastShiftInOut/examples/FastShiftInOut_performance/performance_0.2.0.txt
+++ b/libraries/FastShiftInOut/examples/FastShiftInOut_performance/performance_0.2.0.txt
@ -0,0 +1,41 @@
+IDE:      1.8.19
+Board:    UNO
+FASTSHIFTINOUT_LIB_VERSION: 0.2.0
+
+Performance - time in us
+        write: 18.74
+        write: 36.35
+        Delta: 17.61
+
+writeLSBFIRST: 17.86
+writeLSBFIRST: 35.46
+        Delta: 17.61
+
+writeMSBFIRST: 17.86
+writeMSBFIRST: 35.46
+        Delta: 17.60
+
+
+done ...
+
+
+# loop unrolled.
+
+FASTSHIFTINOUT_LIB_VERSION: 0.2.0
+
+Performance - time in us
+        write: 13.40
+        write: 25.65
+        Delta: 12.26
+
+writeLSBFIRST: 12.52
+writeLSBFIRST: 24.77
+        Delta: 12.26
+
+writeMSBFIRST: 12.45
+writeMSBFIRST: 24.65
+        Delta: 12.20
+
+
+done ...
+
--- a/libraries/FastShiftInOut/library.json
+++ b/libraries/FastShiftInOut/library.json
@ -15,7 +15,7 @@
    "type": "git",
    "url": "https://github.com/RobTillaart/FastShiftInOut.git"
  },
-  "version": "0.1.4",
+  "version": "0.2.0",
  "license": "MIT",
  "frameworks": "*",
  "platforms": "*",
--- a/libraries/FastShiftInOut/library.properties
+++ b/libraries/FastShiftInOut/library.properties
@ -1,5 +1,5 @@
 name=FastShiftInOut
-version=0.1.4
+version=0.2.0
 author=Rob Tillaart <rob.tillaart@gmail.com>
 maintainer=Rob Tillaart <rob.tillaart@gmail.com>
 sentence=Arduino library for (AVR) optimized shiftInOut (simultaneously)