AMD Reportedly Working for Navi 10/12 Based Mining Cards: Up to 55 MH/s Hash Rate
According to a few recent patches to the AMD Linux driver, Team Red might be working on a couple of miner-centric graphics cards based on the older RDNA 1 architecture. The details indicate that there will be two models, one based on the Navi 10 die and the other on Navi 12. While the former would make for a capable Ether mining GPU with a hash rate between 54-56 MH/s, roughly the same as the RTX 3070 and the RX 6800 XT, the Navi 12 die is limited to a 128-bit bus and a bandwidth of just 224GB/s.
Furthermore, the 4GB variant will not worth be able to miner Ether in most cases due to the algorithm DAG size being 4.117GB. Even if it does with a multi-algorithm protocol, the returns will be pretty low. Other currencies like KAWPAW should be compatible but that will net you just a dollar worth of profit for an entire day of mining.
The Navi 10 is another story altogether. With a bandwidth same as that of the RTX 3070 and the RX 6800 XT, it can manage a respectable 54-55 MH/s which can amount to profits of nearly $5/day per GPU. With a memory overclock, you can extract up to $6-6.5 from the card with little to no increase in power.
| 30X | 40X | 50X | 90X | |
|---|---|---|---|---|
| Architecture | Turing (TU116) | Turing (TU106) | Turing (TU102) | Ampere (GA102) |
| Ethereum Hash Rate(1) | 26 MH/s | 36 MH/s | 45 MH/s | 86 MH/s |
| Rated Power(2) | 125 W | 185 W | 250 W | 320 W |
| Power Connectors(2) | 1x 8-pin | 1x 8-pin | 2x 8-pin | 2x 8-pin |
| Memory Size | 6GB | 8GB | 10GB | 10GB |
| Starting Availability | Q1 | Q1 | Q2 | Q2 |
NVIDIA’s CMP 50X miner has an Ether hash rate of 45 MH/s, around 15-20% lower than the RX 5700 XT (full-fledged Navi 10). Only the 90X which is based on the newer GA102 (Ampere) die offers a higher hash rate of 86 MH/s.
Considering that AMD can barely supply a dozen graphics cards a month to a large state like the US, it’ll be interesting to see their reasoning on starting a line of miner-center graphics cards based on the same process node.
Via: Phoronix