You really should google the Maximum Operating Temperature of the 5000 and newer GPU cards.
It is, in fact, 110c and not 99c. Just because you received errors at 99c in your specific PC doesn't verify that 99c is the Maximum Operating Temperature of the GPU card. Plus you are overclocking the GPU card which could be a reason for your errors at 99c.
Was finally able to find a Tom's Hardware article about the 5700XT hotspot being 110c: https://www.tomshardware.com/news/amd-rx-5700-graphics-card-thermal-management,40144.html
From the above linked article from Tom's Hardware:
We noted in our review of the AMD Radeon RX 5700 and Radeon RX 5700 XT that junction temperatures peaked above 100 degrees Celsius under load. (That's 230 degrees Fahrenheit for those of us forced to live with the clearly inferior temperature scale.) AMD published a blog post yesterday to explain why those high temperatures are "in spec" and how the RX 5700's approach to thermal management is supposed to help enthusiasts get the best performance from their cards.
AMD said that it previously relied on "a single sensor that was placed in the vicinity of the legacy thermal diode" to measure a GPU's core temperature. That changed with the release of the AMD Radeon VII in February. The graphics card has "enhanced thermal monitoring," according to AMD, and laid the foundation for the RX 5700's "extensive network of thermal sensors distributed across the entire GPU die." Where once there was one sensor, now there are many.
The company said its previous GPUs were "often leaving significant thermal headroom – and resulting performance – on the table" by adjusting their operation based on a single measurement. These additional sensors are supposed to help RX 5700 graphics cards balance operating temperatures and performance. (Kind of like checking a hunk of meat's temperature in multiple spots to allow it to be cooked as fast as possible without any burning.)
Relying on numerous thermal sensors will become more important as companies rely on ever-smaller manufacturing processes. Those smaller nodes lead to greater thermal density, which in turn makes hotspots more common. Basing performance on the average temperature of the die, like AMD did before it introduced the Radeon VII and RX 5700, would force the GPUs to squander their potential. There's where more precise measurements come in.
AMD explained:
"Instead of setting a conservative, ‘worst case’ throttling temperature for the entire die, the Radeon RX 5700 series GPUs will continue to opportunistically and aggressively ramp clocks until any one of the many available sensors hits the ‘hotspot’ or ‘Junction’ temperature of 110 degrees Celsius. Operating at up to 110C Junction Temperature during typical gaming usage is expected and within spec. This enables the Radeon RX 5700 series GPUs to offer much higher performance and clocks out of the box, while maintaining acoustic and reliability targets."
That won't stop the company's add-in board partners from introducing cooling systems powerful enough to keep junction temperatures below triple digits. Sapphire did just that with the Pulse Radeon RX 5700 XT we reviewed yesterday, actually, and we sincerely doubt it will be the only one. But at least the company made an effort to explain why temperatures high enough to boil water are considered "in spec" for its latest GPUs.
By the way, the Maximum Operating Temperature of 110c is the same for the newer AMD GPU Cards.
