Wow, actually this looks like a perfect temperature handling for me.
At the beginning for both safety reasons and maybe because your card was already a little bit warm/hot, the fan speed was set to a reasonable high value. After that it recognizes that the temperature is still a little bit below the --gpu-temp-retain value and acts accordingly.
Since the temperature raised again (now because of the cracking), fan speed was increases from 15 to 30, than 45, 51, 88 and finally 100% because the retain value was reached...
At some point in time the temperature will almost exactly hit the --gpu-temp-retain value and keep that value, while the fan speed might lower a little bit again (but in some cases - depending on GPU, cooling setup etc - it also could need to keep the fan at 100%.... in most cases, in my experience, it won't keep a constant 100% fan speed, at least for NVidia GPUs)
At the beginning for both safety reasons and maybe because your card was already a little bit warm/hot, the fan speed was set to a reasonable high value. After that it recognizes that the temperature is still a little bit below the --gpu-temp-retain value and acts accordingly.
Since the temperature raised again (now because of the cracking), fan speed was increases from 15 to 30, than 45, 51, 88 and finally 100% because the retain value was reached...
At some point in time the temperature will almost exactly hit the --gpu-temp-retain value and keep that value, while the fan speed might lower a little bit again (but in some cases - depending on GPU, cooling setup etc - it also could need to keep the fan at 100%.... in most cases, in my experience, it won't keep a constant 100% fan speed, at least for NVidia GPUs)