diff --git a/src/ftdiJtagMPSSE.cpp b/src/ftdiJtagMPSSE.cpp
index 9865045..c6c2fef 100644
--- a/src/ftdiJtagMPSSE.cpp
+++ b/src/ftdiJtagMPSSE.cpp
@@ -139,7 +139,8 @@ int FtdiJtagMPSSE::writeTMS(uint8_t *tms, int len, bool flush_buffer)
 		}
 		xfer -= bit_to_send;
 	}
-	mpsse_write();
+	if (flush_buffer)
+		mpsse_write();
 	if (_ch552WA) {
 		uint8_t c[len/8+1];
 		ftdi_read_data(_ftdi, c, len/8+1);
@@ -166,9 +167,6 @@ int FtdiJtagMPSSE::toggleClk(uint8_t tms, uint8_t tdi, uint32_t clk_len)
 			buf[1] = ((len / 8)     ) & 0xff;
 			buf[2] = ((len / 8) >> 8) & 0xff;
 			mpsse_store(buf, 3);
-			ret = mpsse_write();
-			if (ret < 0)
-				return ret;
 			len %= 8;
 		}
 
@@ -176,16 +174,13 @@ int FtdiJtagMPSSE::toggleClk(uint8_t tms, uint8_t tdi, uint32_t clk_len)
 			buf[0] = 0x8E;
 			buf[1] = len - 1;
 			mpsse_store(buf, 2);
-			ret = mpsse_write();
-			if (ret < 0)
-				return ret;
 		}
 		ret = clk_len;
 	} else {
 		int byteLen = (len+7)/8;
 		uint8_t buf_tms[byteLen];
 		memset(buf_tms, (tms) ? 0xff : 0x00, byteLen);
-		ret = writeTMS(buf_tms, len, true);
+		ret = writeTMS(buf_tms, len, false);
 	}
 
 	return ret;
@@ -221,6 +216,9 @@ int FtdiJtagMPSSE::writeTDI(uint8_t *tdi, uint8_t *tdo, uint32_t len, bool last)
 	display("%s len : %d %d %d %d\n", __func__, len, real_len, nb_byte,
 		nb_bit);
 
+	if ((nb_byte + _num + 3) > _buffer_size)
+		mpsse_write();
+
 	if ((nb_byte * 8) + nb_bit != real_len) {
 		printf("pas cool\n");
 		throw std::exception();
diff --git a/src/ftdipp_mpsse.hpp b/src/ftdipp_mpsse.hpp
index b4a5d5d..d2bc5f6 100644
--- a/src/ftdipp_mpsse.hpp
+++ b/src/ftdipp_mpsse.hpp
@@ -50,8 +50,8 @@ class FTDIpp_MPSSE {
 		int _clkHZ;
 		struct ftdi_context *_ftdi;
 		int _buffer_size;
-	private:
 		int _num;
+	private:
 		unsigned char *_buffer;
 };
 
diff --git a/src/gowin.cpp b/src/gowin.cpp
index 70db72e..0cf6ba6 100644
--- a/src/gowin.cpp
+++ b/src/gowin.cpp
@@ -139,15 +139,15 @@ void Gowin::programFlash()
 	wr_rd(NOOP, NULL, 0, NULL, 0);
 
 	/* wait for reload */
-	usleep(150*1000);
+	usleep(2*150*1000);
 
 	/* check if file checksum == checksum in FPGA */
 	status = readUserCode();
 	if (_fs->checksum() != status) {
-		printError("SRAM Flash: FAIL");
+		printError("CRC check : FAIL");
 		printf("%04x %04x\n", _fs->checksum(), status);
 	} else
-		printSuccess("SRAM Flash: Success");
+		printSuccess("CRC check: Success");
 
 	if (_verbose)
 		displayReadReg(readStatusReg());
@@ -265,6 +265,7 @@ bool Gowin::wr_rd(uint8_t cmd,
 	if (rx || tx) {
 		_jtag->shiftDR(xfer_tx, (rx) ? xfer_rx : NULL, 8 * xfer_len);
 		_jtag->toggleClk(6);
+		_jtag->flush();
 	}
 	if (rx) {
 		if (verbose) {
@@ -375,7 +376,7 @@ bool Gowin::flashFLASH(uint8_t *data, int length)
 	memcpy(buffer+6*4, data, byte_length);
 
 
-	ProgressBar progress("Flash SRAM", buffer_length, 50);
+	ProgressBar progress("write Flash", buffer_length, 50);
 
 	for (int i=0, xpage=0; xpage < nb_xpage; i+=(nb_iter*4), xpage++) {
 		wr_rd(CONFIG_ENABLE, NULL, 0, NULL, 0);
diff --git a/src/jtag.cpp b/src/jtag.cpp
index f12132e..2ba16c2 100644
--- a/src/jtag.cpp
+++ b/src/jtag.cpp
@@ -132,6 +132,7 @@ int Jtag::detectChain(vector<int> &devices, int max_dev)
 			devices.push_back(tmp);
 	}
 	go_test_logic_reset();
+	flushTMS(true);
 	return devices.size();
 }
 
@@ -139,7 +140,7 @@ void Jtag::setTMS(unsigned char tms)
 {
 	display("%s %x %d %d\n", __func__, tms, _num_tms, (_num_tms >> 3));
 	if (_num_tms+1 == _tms_buffer_size * 8)
-		flushTMS();
+		flushTMS(false);
 	if (tms != 0)
 		_tms_buffer[_num_tms>>3] |= (0x1) << (_num_tms & 0x7);
 	_num_tms++;
@@ -175,13 +176,13 @@ void Jtag::go_test_logic_reset()
 	/* idenpendly to current state 5 clk with TMS high is enough */
 	for (int i = 0; i < 6; i++)
 		setTMS(0x01);
-	flushTMS(true);
+	flushTMS(false);
 	_state = TEST_LOGIC_RESET;
 }
 
 int Jtag::read_write(unsigned char *tdi, unsigned char *tdo, int len, char last)
 {
-	flushTMS(true);
+	flushTMS(false);
 	_jtag->writeTDI(tdi, tdo, len, last);
 	if (last == 1)
 		_state = (_state == SHIFT_DR) ? EXIT1_DR : EXIT1_IR;
@@ -191,7 +192,7 @@ int Jtag::read_write(unsigned char *tdi, unsigned char *tdo, int len, char last)
 void Jtag::toggleClk(int nb)
 {
 	unsigned char c = (TEST_LOGIC_RESET == _state) ? 1 : 0;
-	flushTMS(true);
+	flushTMS(false);
 	if (_jtag->toggleClk(c, 0, nb) >= 0)
 		return;
 	throw std::exception();
@@ -202,7 +203,7 @@ int Jtag::shiftDR(unsigned char *tdi, unsigned char *tdo, int drlen, int end_sta
 {
 	set_state(SHIFT_DR);
 	// force transmit tms state
-	flushTMS(true);
+	flushTMS(false);
 	// currently don't care about multiple device in the chain
 	read_write(tdi, tdo, drlen, 1);// 1 since only one device
 
@@ -223,7 +224,7 @@ int Jtag::shiftIR(unsigned char *tdi, unsigned char *tdo, int irlen, int end_sta
 {
 	display("%s: avant shiftIR\n", __func__);
 	set_state(SHIFT_IR);
-	flushTMS(true);
+	flushTMS(false);
 	// currently don't care about multiple device in the chain
 
 	display("%s: envoi ircode\n", __func__);
@@ -421,7 +422,7 @@ void Jtag::set_state(int newState)
 			_tms_buffer[(_num_tms-1) / 8]);
 	}
 	/* force write buffer */
-	flushTMS();
+	flushTMS(false);
 }
 
 const char *Jtag::getStateName(tapState_t s)