Verified ×746@peakminerPearlHash · Pearl (PRL)2026-10-16
Measured by other miners
205.2TH/s
Author reported
205.6-0.4
Efficiency
774.3GH/W
783.2 GH/W at the 262 W testers measured at the card
Trust
100
165.3–168.4 TH/s2 miners of 746 reproduced this@SandraA52975 167.9@duongtam2787 165.7171.0 TH/s1 miner of 746 reproduced this@EtiniPaul715 171.0171.7 TH/s1 miner of 746 reproduced this@Speaking77 171.7174.6–177.7 TH/s2 miners of 746 reproduced this@Simoncc65919245 175.9@samuelmatty745 175.6177.7–180.7 TH/s2 miners of 746 reproduced this@quinn90958 179.1@ChukwuMatt19378 179.2180.7–183.8 TH/s6 miners of 746 reproduced this183.8–186.9 TH/s7 miners of 746 reproduced this186.9–190.0 TH/s7 miners of 746 reproduced this190.0–193.0 TH/s9 miners of 746 reproduced this193.0–196.1 TH/s15 miners of 746 reproduced this196.1–199.2 TH/s21 miners of 746 reproduced this199.2–202.3 TH/s65 miners of 746 reproduced this202.3–205.4 TH/s252 miners of 746 reproduced this205.4–208.4 TH/s284 miners of 746 reproduced this208.4–211.5 TH/s58 miners of 746 reproduced this211.5–214.6 TH/s7 miners of 746 reproduced this214.6–217.7 TH/s5 miners of 746 reproduced this217.7–220.7 TH/s2 miners of 746 reproduced this@anonymo83861318 220.4@atoh_web3_life 218.5
143.8242.3
Claimed205.6FewerMore minersMedian205.2
746 independent readings. Spread of 54.7 TH/s reflects normal card-to-card variation.
How to read the scale+−
One block of the rangeThe range of hashrates miners measured. The darker a block, the more miners got that figure.
ClaimedThe hashrate the author reported.
MedianThe middle result from all testers. Profiles are ranked on this.
One dark block means most miners got the same hashrate, so you can expect it too. Dark blocks well to the left of the dashed tick mean miners got less than the author reported.
Settings
Core offset
n/a
Core lock
1110 MHz
Memory offset
n/a
Memory lock
1500 MHz
Power limit
265 W
Power used
256 W
Temperature
75 °C
Applied with
PeakMiner 2.16.2
Before you apply thisTwo cards of the same model do not overclock the same. Raise the memory offset in steps and watch for invalid shares before going to the full figure.
