AMD Radeon R9 Fury graphics card - use of mining cryptocurrency in 2019

r9_fury_mining_2019Released in 2015, the High-End AMD R9 Fury X graphics card looks pretty good even by today's standards. With a clean performance of 8600Gflops and HBM memory with a bandwidth of 512Gb / s, Fury X surpasses even the recently released AMD RX 5700 graphics card in these parameters, although there is a  4-year difference between them. But what gives the age of R9 Fury is an outdated process in 28nm and correspondingly high power consumption compared to more recent cards at 14 nm and 7 nm technical processes, and in addition to this, this is the 3 generation GCN architecture. For comparison, the RX480 / RX580 uses GCN 4, the RX Vega 56/64 architecture is GCN 5, and the new RX5700XT and RX5700 have a new RDNA architecture. Knowing all these introductory data, it became interesting for us to learn how the Hight-End AMD video card of 2015 in the realities of 2019 behaves in cryptocurrency mining. Moreover, the Asus Strix Radeon R9 Fury 4Gb has been gathering dust on the shelf for a long time.

amd-radeon-r9-fury-mining-hashrate-2019

In more detail technical characteristics of AMD video cards of different generations can be compared in the table:

Specifications R9 Fury  R9 Fury X RX Vega 56 RX Vega 64 RX Vega 7 RX580 RX5700 RX5700XT
Release date  14.07.2015 24.06.2015  28.07.2017 28.07.2017 07.02.2019  18.04.2017  07.07.2019 07.07.2019
Frequency GPU (BOOST) MHz  1000 1050  1156(1471) 1247(1546) 1400(1700)  1257(1340)  1465(1725) 1605(1905)
ALU Cores  3584 4092 3584  4096 3840 2304  2304  2560
GFLOPS FP32  7168 8602 10544 12665 13440 6175  7950  9750
Memory  4Gb HBM 4Gb HBM 8 Gb HBM2  8 GB HBM2 16Gb HBM2 8 GB GDDR5  8GB GDDR6  8GB GDDR6
Memory bus width  4096bit 4096bit 2048bit  2048bit 4096bit 256bit 256bit  256bit
Total memory bandwidth  512Gbit/s 512Gbit/s 410Gbit/s  484Gbit/s 1024Gbit/s 256Gbit/s  448Gbit/s  448Gbit/s
Power usage  275W 275W 210W  295W 300W 185W  180W  225W
Announced price, USD  549USD 649USD 329USD  399 699USD 229USD  349USD 399USD

It can be noted that the R9 Fury X is not much inferior to the more recent model RX Vega 56 in terms of its technical characteristics, losing in power consumption, since uses 28nm process technology instead of 14nm. At the same time, the popular RX580 in the mining is inferior to the Fury X, the new 7nm mid-budget video card RX5700 only in 2019 reached the net performance of the flagships of 2015, but at the same time it has a noticeably smaller TDP.

Compared with the version of the R9 Fury X, the R9 Fury video card has a lower number of USM (Unified shader model) 3584 instead of 4092 for the Fury X, which reduces the overall GPU performance by 20% from 8600Gflops to 7168Gflops, the memory bandwidth of both models remains unchanged at 510GB/s.

Because It was the simplified version of R9 Fury that was tested in mining, then for owners of R9 Fury X, the obtained results should be increased by 20%, except for the Ethash (Ethereum) algorithm, since for him the memory subsystem is more important, not the power of the GPU.

Considering the fact that all modern miners rarely optimize for old video cards, or at all limited to supporting GCN version 4 and newer (RX480 / RX580), so this interest, how a video card released 4 years ago will behave in mining 2019, only becomes stronger.

Testing Asus Strix Radeon R9 Fury 4Gb graphics card in mining cryptocurrency 2019 year

 

Test stend:

MB: Gigabyte DA780T-D3L

CPU: Athlon2 X2 250

GPU: Asus Strix Radeon R9 Fury 4Gb (original BIOS) Power Limit -20%

Memory: 4Gb DDR3

OS:Windows 10 1803

Driver: Radeon Software 19.5.2 (Compute Mode)

 

Algorithm Miner Hashrate Power
Ethash Claymore 14.7 29,8Mh/s 170W
Equihash 150/5 Lolminer 0.8.3 16,5Mh/s 200W
Equihash 210/9  Lolminer 0.8.3 110H/s 170W
Equihash 144/5 Lolminer 0.8.3 36,5Mh/s 190W
Equihash 96/5 Lolminer 0.8.3 15,5MH/s 170W
Equihash 192/7 Lolminer 0.8.3 24H/s 180W
Equihash 125/4 Lolminer 0.8.3 22H/s 150W
GRIN29* Lolminer 0.8.3 - -
GRIN31 Lolminer 0.8.3 0,33 150W
CryptonightR TeamredMiner 0.5.5 890H/s 200W
PHI2 TeamRedMiner 0.5.5 5MH/s 240W
aergo WildRigMulti 0.17.9 3,25Mh/s 125W
anime WildRigMulti 0.17.9 19Mh/s 200W
BCD WildRigMulti 0.17.9 11MH/s 220W
bitcore WildRigMulti 0.17.9 14,7MH/s 200W
blake2b-btcc WildRigMulti 0.17.9 1550Mh/s 230W
blake2b-glt WildRigMulti 0.17.9 1550Mh/s 210W
blake2s WildRigMulti 0.17.9 3100Mh/s 250W
BMW512 WildRigMulti 0.17.9 505MH/s 180W
C11 WildRigMulti 0.17.9 12,8MH/s 210W
dedal WildRigMulti 0.17.9 10,3MH/s 180W
exosis WildRigMulti 0.17.9 24,5MH/s 200W
geek WildRigMulti 0.17.9 15MH/s 160W
glt-astsralhash WildRigMulti 0.17.9 14,7MH/s 200W
glt-jeonghash WildRigMulti 0.17.9 5,7MH/s 100W
glt-pawelhash WildRigMulti 0.17.9 6,7Mh/s 220W
HEX WildRigMulti 0.17.9 6,1MH/s 130W
hmq1725 WildRigMulti 0.17.9 6Mh/s 170W
honeycomb WildRigMulti 0.17.9 17,7MH/s 160W
lyra2v3 WildRigMulti 0.17.9 51,2MH/s 220W
lyra2vc0ban WildRigMulti 0.17.9 44,4MH/s 150W
PHI WildRigMulti 0.17.9 14,1MH/s 180W
polytimos WildRigMulti 0.17.9 19,1MH/s 210W
renesis WildRigMulti 0.17.9 9,8Mh/s 140W
sha256q WildRigMulti 0.17.9 470Mh/s 240W
sha256t WildRigMulti 0.17.9 630Mh/s 240W
skein2 WildRigMulti 0.17.9 390MH/s 220W
timetravel WildRigMulti 0.17.9 24,4Mh/s 230W
timetravel10 WildRigMulti 0.17.9 14,7MH/s 160W
tribus WildRigMulti 0.17.9 47Mh/s 240W
wildkeccak* WildRigMulti 0.17.9 - -
X16r WildRigMulti 0.17.9 10Mh/s 150W
X16rt WildRigMulti 0.17.9 10MH/s 130W
X16s WildRigMulti 0.17.9 10MH/s 160W
X17 WildRigMulti 0.17.9 9,7Mh/s 210W
X18 WildRigMulti 0.17.9 7,4MH/s 250W
X20r WildRigMulti 0.17.9 9,5MH/s 200W
X22i WildRigMulti 0.17.9 4,5MH/s 160W
X25x WildRigMulti 0.17.9 1,1MH/s 100W
xevan WildRigMulti 0.17.9 3MH/s 180W

 *Miner error

In testing, we didn’t use the AMDMemoryTweak utility due to the fact that we couldn’t pick up the HBM 1 generation memory timings without loss of stability of the video card, and in the Claymores miner version 14.7 this video card is not supported to change the memory timings.

It should be noted that since 2016, the hash rate on the Ethash algorithm has not decreased and it is still equal to 29-30MH/s in stock settings, which cannot be said of other video card models of the same generation, which have a noticeable gradual decrease in hashes. If you look again at the table with the characteristics of video cards, then we can assume that the selection of optimal memory timings for the R9 Fury video card can significantly increase its performance on the Ethash algorithm, at least by 20-30%. But this already requires separate testing and verification.

Now let's turn to the calculation of the profitability of mining on AMD R9 Fury graphics cards and here it will not be a secret to anyone that the cost of electricity will be in this case a decisive value.

The calculation was made at a cost of 0.05 USD per kilowatt; you can calculate your own electricity price on the Profit-mine website, where the results of this testing are already reflected.

18.07.2019 Bitcoin 9800USD, Ethereum 217USD, Monero 77USD

fury-mining-ROI

Conclusion: The AMD R9 Fury 2015 video card still shows itself well in mining on the Ethash algorithm yielding 30MH/s, but with a power consumption of 170W, which is of course much more than modern graphics cards, but even with such power consumption, R9 Fury can generate a profit in mining When choosing such a video card for mining, first of all you need to pay attention to the cost of your electricity, because This will be the main item of your expenses due to the low energy efficiency of the 28nm process used in manufacturing AMD Radeon R9 Fury, Fury X and Nano video cards.