This is a pretty old thread now, but having run into the same problem myself trying to recover an MS-7621 motherboard (a custom board Microstar made for Medion's all-in-one PC range, specifically the Akoya P4020D), I thought my findings might be of help to other lost souls. This motherboard uses the Winbond W25Q80 flash chip (1 MB), and I was programming it using MSI's JSPI1 connector, pinouts verified as described above.
I covered most of the ground discussed above earlier, arriving at the same point where everything _should_ work but nothing actually did.
The mini breakthrough came when I measured the actual voltage on the Winbond between VCC and Gnd - using power from the Blackcat, it was only measuring about 1.7V. The Winbond is quite flexible, operating on anything from 2.5V to 3.7V, but not quite that low. I expect all the other devices on the board, also linked to the same VCC feed, were pulling enough current to overload the poor Blackcat board.
The fix was easy: I just powered on the PC (I figured nothing to lose). I also removed VCC from the link cable, though left everything else including GND connected. Once done, the Blackcat software identified the Winbond immediately and I was able to reliably read and write the flash contents with no further tweaking needed.
(Worth noting: I didn't need to do anything fancy with HOLD - it was already pulled high, either by the mainboard or the chip itself.)