Archive for the ‘Mining Hardware’ Category


The new AMD Radeon RX 480 is now officially out and we have some results from tests we have actually performed ourselves, results that pretty much confirm an earlier leak that was posted about the expected hashrate. As you will see the number of tested algorithms is only one – Ethereum’s Ethash as it seems that there are some problems at the moment running sgminer with most algorithms on RX 480… could be a driver issue, could be something else. Anyway, we are sure that most of you are interested in the Ethereum hashrate that the RX 480 is able to deliver and we have information about that. The hashrate for Ethereum mining that thr Radeon RX 480 delivers is good, though the power level is not that great, the cooling of the reference design RX 480 is also not that great for mining and the price is also not what some people in some regions expected and on top on shortages, though in many places you can buy the card at the price that AMD announced or close to it and there are enough cards.

Before we move to the tests we should note that there are two variants of the Radeon RX 480, one with 4GB video memory that apparently runs at 7 GHz and is with a lower price (the $199 USD price) and a more expensive 8GB model with memory running at 8 GHz. At launch it is hard to find 4GB versions as apparently they are much fewer as compared to the 8GB models, but if you are buying a card for mining the 8GB model is the one you would want. We have tested an ASUS reference design Radeon RX 480 card with 8GB and 8 GHz video memory, so the results below are for mining Ethereum with that card. The 4GB model with 7 GHz video memory should be slower by about 2-3 MHs compared to the 8GB 8 GHz version, so if you are buying these for mining Ethereum go for the 8 GB models.


We are using Claymore’s Dual Miner, starting with the default settings of 1266 MHz for the GPU and 8 GHz for the video memory withing the standard 150W TDP limit of the Radeon RX 480 GPU. The result we are getting is about 24.7-24.8 MHS at stock settings with GPU-Z reported power usage of about 108 W, but with actual power usage measured at the wall of 177 Watt. So taking into account the power conversion efficiency of an 80 Plus Gold PSU that was used the actual power usage of the card is apparently 150W at the default settings. This is further confirmed by the fact that while mining Ethereum with the default settings the GPU frequency hovers up to about 1240 MHz and does not go all the way up to 1266 MHz. Essentially hitting the power limit of 150W that the card has, increasing a bit the power limiter by a few percent over the default 100% allows the GPU to stay at the full 1266 MHz… not that it makes any difference for mining Ethereum, but it could be important for other not so video memory intensive crypto algorithms.

With dual mining mode enabled to mine both Ethereum and Decred at the same time the results show about 24.3 MHS for Ethereum and about 365 MHS for Decred (at default intensity of 30). Further increasing the Decred intensity to 35 decreased the Ethereum hashrate to about 24 MHS and increases the Decred mining hashrate to about 450 MHS. So essentially the RX 480 does pretty well in dual mining mode with not so much of a drop in the Ethereum hashrate while providing some good extra performance for mining Decred.


Moving to overclocking, the only thing we need to do is increase the video memory to the maximum and in this case this means 9 GHz (the card might be able to handle more, but the overclocking software or the driver is apparently capped at 9 GHz max). With the video memory at 9 GHz the RX 480 is capable of delivering about 28 MHS hashrate for mining Ethereum. It is highly possible that even higher hashrate could be achieved if you manage to get past the 9 GHz limit for the video memory, but for the moment AMD’s WattMan software or the Asus tool that we used do not currently allow for more. With the memory overclocked the power usage is increased with just about 7-8 Watt at the wall, so definitely worth overclocking it for mining Ethereum. We were a bit disappointed by the lack of overclockability headroom of the GPU that was able to get just about 1340 MHz with the stock voltage. Of course no point in overclocking it for Ethereum mining, in fact you might want to actually downclock it to get better power efficiency if mining only Ethereum.


What we are not too happy with is the stock cooling solution and how it handles the heat from the RX 480. AMD has optimized the default profile for the fan to keep the card silent and very hot, so obviously it is a no go for mining. What seems to work pretty well is setting up the fan to about 80% and getting decent temperatures with a moderately noisy card – good choice if noise from the mining rig can be an issue. With the cooling fan at 80% you can expect temperatures of about 72 degrees Celsius for the GPU, something that is a bit higher than you’d probably want for 24/7 mining. Going for 100% fan the noise increases significantly and the temperatures drops to about 68 degrees Celsius, so not by much.

We are not too happy with the power usage and the hashrate you get from the RX 480 for mining Ethereum, we kind of expected better results in that area. But then again by lowering the GPU operating voltage and frequency you might be able to get better power efficiency without actually loosing any performance and this is a must if you are mining Ethereum… that is if you are actually able to downclock the GPU, a modified video BIOS with the right settings might be a good solution to that problem, but that would also need some time.

We are going to be doing some more tests to see if things with the sgminer crashing on most algorithms and looking for other options to improve the power usage when mining Ethereum etc., so stay tuned for more info about the RX 480. For the moment however we are not very impressed by the new AMD Radeon RX 480, though a non-reference design with some extra tweaking might be able to change our opinion.


We have been keeping track on the development of the new Innosilicon A4 Dominator Scrypt ASIC miners since the initial announcement back in November last year and it is time for another update. It seems that Innosilicon is starting to take pre-orders for the final ASIC miners with shipping expected to start in September and there is already a thread on Bitcointalk about a group pre-order that you might want to check out if interested. In the past we have tested the previous generation of Innosilicon A2 Scrypt ASIC miners in the form of the 86 MHS A2BOX Innosilicon A2 Scrypt ASIC Miner that was consuming about 1000 Watts of power, so we can see what the development is and how good actually the new generation is going to be. A solid first generation Scrypt ASIC miner that was built well and performed really good when it was released with the only drawback being the very high initial price of the device, so we were anticipating another good solid product with much better specs when the first A4 details surfaced.

The initial announcement from last year claimed that the A4 Dominator is so efficient that it should be able to deliver about 850 MHS for Scrypt mining with 1020W of power usage. This was really impressive number and really caught our attention, especially considering that the first generation A2 Terminator miners were able to achieve just about 1/10 of that hashrate with the same power usage. A little bit later, in March this year, there was an update on the power usage and efficiency of the new A4 ASIC chips that has significantly lowered the initial numbers – down to 400 MHS from 850 MHS per kilowatt of power. Although the expected performance was lowered more than half it was still a pretty impressive number, but unfortunately this was not the final change and now we have a new number available. Apparently the final product will only be capable of about 260 MHs per 1 KW of power and be available for about $1800 USD without shipping and with power supply not included in the price. While the final performance per KW of power usage is not that bad either, we are not as excited as we were with the initial announcement or the March update. At 260 MHS per 1000 Watts things aren’t that bad, but with these specs we are not going to be pre-ordering one of these miners for sure. Maybe we are going to try to get one unit to test when they come out if it is not a pain in the ass to secure a single unit order.

If you are interested in the A4 Dominator ASICs group pre-order thread on Bitcointalk…


One more announcement fro Bitmain, the AntMiner S9 ASIC Bitcoin miner is no official and will start shipping on June 12th for $2100 USD without the shipping cost (power supply is not included). The device uses the company’s new 16nm ASIC chips BM1387 and essentially triples the hashrate of the previous generation S7 miners while offering the same level of power usage. The Bitmain AntMiner S9 should be capable of 14 THS SHA-256 mining hashrate with a power usage of 1375 Watts at the wall, so definitely a good improvement over the S7.

Bitmain AntMiner S9 Specifications:
– Hash Rate: 14.0 THS ±5%
– Power Consumption: 1375W + 7% (at the wall, with APW3 ,93% efficiency, 25C ambient temp)
– Power Efficiency: 0.098 J/GH + 7% (at the wall, with APW3 93% efficiency, 25°C ambient temp)
– Rated Voltage: 11.60~13.00V
– Chip quantity per unit: 189x BM1387
– Dimensions: 350mm(L)*135mm(W)*158mm(H)
– Cooling: 2x 12038 fan
– Operating Temperature: 0 °C to 40 °C
– Network Connection: Ethernet
– Default Frequency: 650 MHz

The big question however is if now was the right time to announce the new mining hardware, as of you order now from the first batch you might be getting your units delivered right before the Bitcoin block reward halving. So it might be wise to wait for a later batch of S9 miners after the halving to see how the situation will change and if you might get a better deal in terms of price for the hardware. Then again if you are using S7 miners you might already get to the planning for switching to S9 now that the previous generation S5 is no longer profitable barely covering the electricity costs.