💀SON OF A TECH

⛏️ GPU Hashrates

Hashrate and measured power for 13 cards across 21 coins. Every number here came off our own bench, not an aggregator. Every row links to the video it came from, so you can go watch the run.

How to read these. Power is measured at the wall wherever possible, so it includes power supply losses and reads higher than the figure mining software reports for the card alone. That is deliberate. The wall number is what you actually pay for. A few rows are tagged card, meaning the figure came from the card's own reporting because no wall meter was available on that test. Those two are not directly comparable, so do not read a card-reported row as beating a wall-measured one. Efficiency is hashrate divided by power, and it only means something between rows using the same unit.
GPU Coin Hashrate Power Efficiency Miner Notes Tested
BC-250 (PS5 APU)AlephiumBlake3475 MH/s149 W3.19 MH/W—2024-01-14
BC-250 (PS5 APU)DynexDynexSolve2139 H/s150 W14.3 H/W—2024-01-14
BC-250 (PS5 APU)ErgoAutolykos v291.72 MH/s184 W0.498 MH/W—2024-01-14
BC-250 (PS5 APU)Ethereum ClassicEtchash47.51 MH/s197 W0.241 MH/WTeamRedMinerSame across every Ethash variant including ETC B3 for Hypra. Avoid GMiner and PhoenixMiner on these, both were well off the pace2024-01-14
BC-250 (PS5 APU)FluxZelHash20.5 Sol/s152 W0.135 Sol/W—Miner support on this board is patchy. Some miners run one algorithm and fail on the next depending on how they initialise OpenCL2024-01-14
BC-250 (PS5 APU)Iron FishBlake34.47 GH/s141 W0.0317 GH/WTeamRedMinerTeamRedMiner led lolMiner and BzMiner here2024-01-14
BC-250 (PS5 APU)KarlsenKarlsenHash136 MH/s142 W0.958 MH/W—2024-01-14
BC-250 (PS5 APU)NexaNexaPow2109 MH/s156 W13.5 MH/W—2024-01-14
BC-250 (PS5 APU)NovoSHA-256DT876.9 MH/s147 W5.97 MH/W—2024-01-14
BC-250 (PS5 APU)RavencoinProgPow / KawPow15.66 MH/s193 W0.0811 MH/W—2024-01-14
GTX 1660 SuperBeamBeamHashIII15.36 Sol/s91 W0.169 Sol/W—2021-08-23
GTX 1660 SuperConfluxOctopus28.75 MH/s112 W0.257 MH/W—Power could not be turned down far without losing hashrate2021-08-23
GTX 1660 SuperErgoAutolykos62.8 MH/s91 W0.69 MH/W—2021-08-23
GTX 1660 SuperEthereum retiredEthash32 MH/s91 W0.352 MH/W—Needs a NEGATIVE memory offset: -502 MHz in Afterburner, -1004 in HiveOS. Without it the card sits near 23 MH/s2021-08-23
GTX 1660 SuperRavencoinKawPow14 MH/s132 W0.106 MH/W—12 MH/s stock, 14 with overclocking. Most power hungry algorithm on this card2021-08-23
Intel Arc A380AlephiumBlake3432 MH/s119 W3.63 MH/WSRBMinerOverclocking is entirely greyed out on this card, so every algorithm drew the same 119W2023-03-24
Intel Arc A380DynexDynexSolve1 H/s119 W0.0084 H/WSRBMinerNot a typo. One hash per second, and adding an i9 11900K made it worse at 0.8. Effectively unusable on this card2023-03-24
Intel Arc A380Ethereum ClassicEtchash14.76 MH/s120 W0.123 MH/WSRBMiner2023-03-24
Intel Arc A380Iron FishBlake3936 MH/s120 W7.8 MH/WSRBMinerTested on testnet before launch, so no price existed to judge profitability against2023-03-24
Intel Arc A770 16GBAlephiumBlake31.71 GH/s210 W0.00814 GH/WSRBMinerCore locked to 2200 MHz, which is the ceiling before it starts throttling2023-04-29
Intel Arc A770 16GBDynexDynexSolve3.34 kH/s110 W0.0304 kH/WSRBMinerRequires running THREE instances of SRBMiner at once. One instance gets nowhere near this2023-04-29
Intel Arc A770 16GBErgoAutolykos v2118 MH/s80 W1.48 MH/WSRBMinerCore locked to 1400 MHz. Unlocking it reaches 128 MH/s but the extra power is not worth it. An A750 hits the same rate for up to 10W less2023-04-29
Intel Arc A770 16GBEthereum retiredEthash48.76 MH/s170 W0.287 MH/WlolMinerCore at 2100 MHz. Lightly tuned only2023-04-29
Intel Arc A770 16GBKaspakHeavyHash410 MH/s210 W1.95 MH/WSRBMinerCore must be locked to 2200 MHz or it errors trying to push past 24002023-04-29
Intel Arc A770 16GBRadiantSHA-512/256d608 MH/s210 W2.9 MH/WSRBMiner2023-04-29
Intel Arc B570Ethereum ClassicEtchash9 MH/s136 W0.0662 MH/WSRBMinerPoor for a memory intensive algorithm, and far below the ETC B3 result on the same card2025-01-29
Intel Arc B570Ethereum ClassicETC B318 MH/s125 W0.144 MH/WSRBMinerTwice the Etchash result on identical hardware. Both are Ethash family, so the older Ethash support on Intel looks like the problem rather than the card2025-01-29
Intel Arc B580DynexDynexSolve4470 H/s117 W38.2 H/WBzMinerProof of work only mode2025-01-14
Intel Arc B580Ethereum ClassicETC B321.5 MH/s150 W0.143 MH/WSRBMiner2025-01-14
Intel Arc B580Iron FishFishHash17.5 MH/s157 W0.111 MH/WBzMiner2025-01-14
Intel Arc B580KarlsenKarlsenHash18.2 MH/s167 W0.109 MH/WBzMiner2025-01-14
Intel Arc B580RavencoinKawPow21 MH/s215 W0.0977 MH/WWildRig2025-01-14
RTX 3050ConfluxOctopus26 MH/s110 W0.236 MH/W—+1500 MHz memory offset2022-01-28
RTX 3050ErgoAutolykos40 MH/s90 W0.444 MH/W—2022-01-28
RTX 3050Ethereum retiredEthash13.69 MH/s90 W0.152 MH/WlolMiner2022-01-28
RTX 3050FiroFiroPoW12.43 MH/s140 W0.0888 MH/W—Matched Ravencoin almost exactly2022-01-28
RTX 3050FluxZelHash22 Sol/s140 W0.157 Sol/W—2022-01-28
RTX 3050MoneroCryptoNight1550 H/s90 W17.2 H/WXMR-StakMiner last updated 2019 at time of test2022-01-28
RTX 3050RavencoinKawPow12.43 MH/s140 W0.0888 MH/W—2022-01-28
RTX 3050VertcoinVerthash264 kH/s80 W3.3 kH/WOne Click MinerMost power efficient algorithm on this card. ccminer would not reach full rate2022-01-28
RTX 3070 TiConfluxOctopus78 MH/s350 W0.223 MH/W—77 to 78 MH/s2021-12-28
RTX 3070 TiErgoAutolykos181 MH/s214 W0.846 MH/W—+1800 memory, -200 core, 80% power. The video's own recap quotes 186 MH/s at 300W instead2021-12-28
RTX 3070 TiEthereum retiredEthash53 MH/s200 W0.265 MH/WlolMiner 1.3848 MH/s at 350W untuned. Tuning the power limit got the same rate for 150W less2021-12-28
RTX 3070 TiFiroFiroPoW41 MH/s350 W0.117 MH/W—2021-12-28
RTX 3070 TiFluxZelHash35 Sol/s180 W0.194 Sol/WlolMiner 1.38Aggregator sites listed 61 for this card. It never went above 35 in testing2021-12-28
RTX 3070 TiRavencoinKawPow43 MH/s353 W0.122 MH/W—42 to 43 MH/s. The video's own recap quotes 63 MH/s, which does not match the detailed testing2021-12-28
RTX 3080 TiBeamBeamHashIII35.3 Sol/s318 W0.111 Sol/W—Lowest power draw of the algorithms tested2021-08-20
RTX 3080 TiConfluxOctopus99.4 MH/s307 W0.324 MH/W—2021-08-20
RTX 3080 TiErgoAutolykos262 MH/s371 W0.706 MH/W—2021-08-20
RTX 3080 TiEthereum retiredEthash81 MH/s320 W0.253 MH/WNBMinerWith the NBMiner unlock. Close to a fully overclocked 30802021-08-20
RTX 3080 TiRavencoinKawPow53 MH/s371 W0.143 MH/W—2021-08-20
RTX 5070 TiErgoAutolykos v2311 MH/s131 W card2.37 MH/WRigel+3000 memory, core tuned down to 1700 MHz. Took the efficiency crown from the CMP 170HX2025-02-22
RTX 5070 TiEthereum ClassicEtchash88 MH/s141 W card0.624 MH/WRigelCore bottlenecked for peak hashrate but still very efficient. +3000 memory, 2100 MHz core lock2025-02-22
RTX 5070 TiRavencoinKawPow46 MH/s233 W card0.197 MH/WRigelDoes not beat AMD here. +3000 memory, 2100 MHz core lock2025-02-22
RTX 5080CryptoNight GPUCryptoNight GPU8677 H/s385 W22.5 H/WSRBMinerAn RTX 3090 Ti sits just under 5,500 H/s. A second tuned run in the same video recorded 6,778 H/s at 266W2025-02-01
RTX 5080DecredBlake323 GH/s272 W0.0846 GH/WSRBMinerTested for completeness. Centralised mining makes it impractical2025-02-01
RTX 5080ErgoAutolykos v2271 MH/s210 W1.29 MH/WSRBMinerOutperformed every previous generation non-HBM card tested2025-02-01
RTX 5080Ethereum ClassicEtchash113 MH/s272 W0.415 MH/WSRBMinerETC and ETC B3 gave the same result2025-02-01
RTX 5080Iron FishFishHash74 MH/s248 W0.298 MH/WSRBMinerOthers report 90 MH/s. Could not be reproduced here2025-02-01
RTX 5080KarlsenKarlsenHash74 MH/s250 W0.296 MH/WSRBMinerBehaves much like FishHash on this card2025-02-01
RTX 5080KaspakHeavyHash1545 MH/s305 W5.07 MH/WSRBMiner2025-02-01
RTX 5080NexaNexaPow1517 MH/s324 W4.68 MH/WSRBMiner2025-02-01
RTX 5080NovoSHA-256DT6051 MH/s279 W21.7 MH/WSRBMiner2025-02-01
RTX 5080RavencoinKawPow56 MH/s373 W0.15 MH/WSRBMiner2025-02-01
RTX 5080VertcoinVerthash292 kH/s205 W1.42 kH/WSRBMinerRoughly RTX 4090 performance with no overclock applied2025-02-01
RX 6900 XTErgoAutolykos120 MH/s227 W0.529 MH/W—Memory 2138 MHz, core to 2521 MHz. Drops to ~113 MH/s for noticeably less power2022-01-12
RX 6900 XTFluxZelHash67 Sol/s263 W0.255 Sol/W—Weak for this card. A 2080 Ti reaches about 95 Sol/s2022-01-12
RX 6900 XTTONTON4.7 GH/s263 W0.0179 GH/W—Core to 2196 MHz. An RTX 3090 lands in the same 4.7 to 5 GH/s range2022-01-12
RX 7900 XTXAlephiumBlake33402 MH/s419 W8.12 MH/WSRBMiner2024-01-28
RX 7900 XTXBeamBeamHash47.3 Sol/s410 W0.115 Sol/WlolMinerOnly lolMiner worked2024-01-28
RX 7900 XTXDynexDynexSolve7660 H/s395 W19.4 H/WBzMinerOnly BzMiner worked2024-01-28
RX 7900 XTXErgoAutolykos v2193 MH/s416 W0.464 MH/WTeamRedMinerBeat BzMiner, NanoMiner and SRBMiner on this algorithm2024-01-28
RX 7900 XTXEthereum retiredEthash110 MH/s416 W0.264 MH/WNanoMiner2024-01-28
RX 7900 XTXEthereum ClassicEtchash110 MH/s414 W0.266 MH/WNanoMinerSeven miners worked. NanoMiner led by a wide margin but fluctuates between 105 and 115, so this is the average. Verified against the pool2024-01-28
RX 7900 XTXEthereum ClassicETC B3102.9 MH/s414 W0.249 MH/WSRBMiner2024-01-28
RX 7900 XTXIron FishBlake33.42 GH/s420 W0.00814 GH/WTeamRedMinerFour miners supported it. TeamRedMiner led2024-01-28
RX 7900 XTXKarlsenKarlsenHash2319 MH/s419 W5.53 MH/WlolMiner2024-01-28
RX 7900 XTXNexaNexaPow118.13 MH/s418 W0.283 MH/WlolMinerOnly lolMiner was supported2024-01-28
RX 7900 XTXRavencoinKawPow57.12 MH/s411 W0.139 MH/WSRBMiner2024-01-28
These are dated measurements, not live figures. A hashrate is only true for the driver, the miner version and the overclock it was taken with, on that day. Several of these chains have moved on since. Ethereum is marked retired because it stopped using proof of work entirely, so those rows are history rather than something you can go and do. Treat every number as a starting point you verify on your own card, not a promise.
Where the source contradicts itself, the note says so. A couple of these videos quote one figure during the detailed testing and a different one in the closing summary. Rather than quietly picking the nicer number, both are recorded in the note. The detailed testing figure is the one in the table.