Overclocking RTX 2060 memory for mining purposes
Overclocking RTX 2060 memory for mining purposes
Gpu - rtx2060. With default settings when I run miner I get around 26.5mh/s and the GPU temperature reaches 82°C. After undervolting the GPU from 1.1V to about 850mV, lowering the core clock by 500mHz and boosting GPU memory speed by 800mHz (it crashes when trying to increase further), I achieved roughly 30.5Mh/s with a maximum temperature of 71°C. Have I made a mistake so far?
As I understand, there’s no temperature sensor on the GPU memory, but can I still cause overheating? I have poor airflow in my case, but I managed to bring the GPU down to around 70°C.
The memory is unlikely to trigger a complete system failure; instead, it may generate errors and lead to invalid shares as stability decreases. The problem seems tied to your central clock and power supply, which might not be consistent at that voltage level. Gradually increase the voltage until it reaches a stable core clock. This challenge stems from inherent limitations in the silicon material—known as the Silicon lottery.
I had already adjusted the afterburner settings earlier, only slightly increasing the voltage and setting up two profiles.
I’ve tried these configurations (both with the GPU at 700mV and it still performs similarly to 850mV).
1. Core clock (-500) Memory clock (+800) Power limit (73%) Fans: auto
Average GPU temperature - 64.8°C, average hotspot temp - 81.7°C
Average hashrate --- 30.40 mh/s
2. Core clock (-500) Memory clock (-500) Power limit (73%) Fans: auto
Average GPU temperature - 62.3°C, average hotspot temp - 78.8°C
Average hashrate --- 27.4 mh/s
These readings from hwinfo now show results when using the second profile. After adjusting the fan speed to 85%, temperatures dropped to around 56°C, which seems safe.
My question is: should I increase memory clock by +800mHz to potentially boost speed to about 3mh/s (with fans at max or 65°C with auto), or should I stick with the lower settings because I can’t monitor VRAM temps closely enough for safety?
Thanks in advance.
P.S. It’s conceivable to reach over 32mh/s, but I currently lack sufficient airflow. If I revert voltage and core clock back, temperatures will likely rise significantly.
Keep pushing the VRAM beyond its limits until you notice problems or irregularities in performance or valid data. If the temperature rises significantly, you'll likely experience a decline in hashrate or an increase in invalid shares. The VRAM tends to handle minor overclocks well—it usually just becomes more sensitive. However, the GPU itself is much stricter; excessive overclocking can cause system instability and crashes. I increased my 2080Ti's VRAM to 1000+MHz, achieving 65MH/s, though it introduced a few extra errors per hour. Further boosting it to 1100 resulted in erratic performance ranging from 57MH/s to 63MH/s. With GDDR6 on RTX, memory correction is built in, so the more unstable the overclock, the greater the effort the memory must perform to fix errors. Every card is unique—what works for one might not work for another. Keep experimenting until you discover the optimal balance for your hardware.
it is becoming obvious now that after six hours of mining I achieved a stable speed of 30.4mh/s with an 800 memory overclock, indicating it’s safe to operate normally without risking the GPU temperature to between 58-66°C.
Yes it's fine. That temp is mostly for the GPU die itself, not the VRAM. If you can't see the VRAM's temp then that means there's no sensor for it, but normally MAX temp for GDDR6 VRAM is 95C, anything more will cause damage, but these days manufacturers program these devices so if they do reach those temps the system automatically downclocks itself (thermal throttling) to prevent further damage. We live in a convenient era. Sort-of....
haha, i hope so)) Thanks for your answers, the last question for you, should i run it with fans set to auto (66% speed) or set custom speed to 85% There is 5°C difference(59C and 64C) I mean this is not such a big temp difference but i can decrease noise.
As long as the GPU stays under 70°C, pick any fan profile that suits your needs. Usually fans can last many years even when running at full capacity continuously, so pushing them to 100% all the time isn't necessary. It's better to opt for a fan replacement than to replace the entire GPU. However, keeping it below 70°C is ideal.
I just found out that running the miner causes my memory clock to drop by 200 instantly. Whether it's slowed down or sped up, it affects me. If I keep the default at 7000 and lower it by 500 MHz, it becomes 7000 - 500 - 200, and if I raise it by 800 MHz, it becomes 7000 + 800 - 200. The hashrate stays consistent at both 6300 MHz and 7600 MHz. What might be the problem?