Archive for the ‘Mining Hardware’ Category

zeusminer-lightning-x6

We have been playing with the new NX Gen X6 miners from ZeusMiner for a few days already and have already shared some interesting findings from our testing, but it is now time for detailed reviews focusing on each of the new lower power usage Scrypt ASIC miners. Beginning with the top model – ZeusMiner Lightning X6 Scrypt ASIC miner, rated officially as 40-42 MHS at 980W power consumption and being sold by ZeusMiner for $1399 USD, though you can already find it cheaper at their distributors. What Zeus did with their new X6 miners is to lower the voltage and operating frequency of their Scrypt ASIC chips and as a result we got the new lower power usage products. Now, if the reduction in power usage was not combined with new lower prices of the devices they might not have been so interesting with the continuing to increase Litecoin network hashrate. The lower power usage and the new lower price however do make the new X6 series of miner interesting option to consider if you are planning to invest in Scrypt ASICs. The ZeusMiner Lightning X6 might not be currently the most powerful in terms of hashrate Scrypt ASIC miner, however it is probably one of the best ones in terms of price per MHS from what is currently available on the market… especially if you manage to get it at sub $1000 USD price.

zeusminer-x6-chip-operating-voltage

Zeus has a strange policy of citing hashrate for their products, but not giving a recommended frequency for their products and the miner software does require the user to give an operating frequency for the chips. This is still valid for the new X6 miners, meaning that you need to find the best frequency yourself, though you should start at around 220-230 MHz for the Lightning X6 as per our tests results. With the smaller Thunder X6 and Lightning X6, and probably the Blizzard X6 as well a good start frequency seems to be 250 MHz and we have found out the reason for this frequency difference. The ZeusMiner Lightning X6 has the Scrypt ASIC chips inside operating at 1.05V and the other miners from the X6 series use voltage of 1.1V and as a result they are able to operate at higher frequency without problems. The lower voltage on the Lightning X6 was probably needed in order to bring the power usage level of the miner below 1000 Watts as otherwise the consumption could go as high as 1.1 KW or even higher and make it harder to find a suitable PSU for the device.

zeusminer-lightning-x6-power-usage

We have measured the power usage of the Lightning X6 operating at a frequency of 230 MHz and the result we’ve got was about 906-914 Watts as actual power usage and about 970W measured at the wall (taking into account the PSU efficiency). The actual power consumption measured at the wall could be higher depending on the PSU used, the 1200W Corsair 80 Plus Platinum power supply we have used does remain with high power efficiency at the measured load, so there is not much power wasted, but you can easily go above 1000W with other not so efficient one. So apparently Zeus did a good job in lowering the power usage almost two fold with just about 10-12 MHS lower power usage compared to the previous generation miners with 256 chips.

zeusminer-lightning-x6-hashrate-local

Our tests have shown that the optimal operating frequency of the ZeusMiner Lightning X6 Scrypt ASIC miner is around 220 MHz, though at that frequency the hashrate you can expect to get is slightly lower than what is being advertised – we are getting around 36-38 MHs with about 6.6% HW errors. If you increase the operating frequency of the chips to 230-235 MHz we can get about 39-40 MHS, but the level of the hardware errors does increase almost twice to about 12% which is even less acceptable. The higher the number of HW errors you get it means that you are pushing the hardware a bit too much, and even at 6.6% it is more that we would normally like to have. Zeus does advertise the Lightning X6 as a 40-42 MHS Scrypt ASIC, but we feel that a more adequate rating for the performance should be 38-40 MHS.

zeusminer-lightning-x6-hashrate-ltcrabbit

Here is the situation with the poolside reported hashrate at the LTC Rabbit Scrypt mining pool using the new fixed difficulty ports for Scrypt ASIC miners and the Lightning X6 running at 220 MHz operating frequency. We are getting an average hashrate of about 37.4 MHs with peaks to almost 38.5 MHS and drops to about 35.5 MHs. At 230 MHz the poolside result was slightly higher with an average of about 39.2 MHS and peaks going as high as 40.8 MHS and drops to about 37.5 MHs. Of course the poolside hashrate can vary a bit based on multiple factors, but the locally reported hashrate we are getting shows consistent results with what the pool is giving us out as performance data.

With the ZeusMiner Lightning X6 we see a good improvement in terms of lowering the power consumption for a 256-chip miner without too much of performance loss. The price of the miner is also very good considering that you can already find the Scrypt ASIC miner with a price per MHS of about $25 USD. We did not like very much the high level of HW errors that we are getting in order to be able to push a performance close to the one officially advertised hashrate, but the difference isn’t that huge. And only if the miner could not be so noisy, but at that power usage you can expect to have a lot of noise coming from the cooling fans used to keep the device cool.

zeusminer-x6-scrypt-asic-miners-1

We have been testing the new Next Generation of ZeusMiner Scrypt ASICs (Zeus Lightning X6, Zeus Thunder X6 and the Zeus Hurricane X6) for a few days already and have already reported our very first impressions. Now we have some more interesting findings and information available to share with you based on our continuing tests, this time we are going to focus on the power usage of the new X6 series of miners that are supposed to be more power efficient – their key feature. What Zeus has done with their ZeusMiner X6 Scrypt ASICs is essentially lowering the operating voltage and power and thus achieving lower power consumption. If this was not also combined with new lower prices as well the new X6 miners would not be so interesting, but their new lower prices (even though you cannot yet take advantage from them directly buying from Zeus due to their pre-order 1+1 bonus promotion) at their distributors are actually quite attractive combined with the lower power consumption.

zeusminer-x6-chip-operating-voltage

We have already discovered that the best operating frequencies of the new miners (there is no recommended frequency published on the official website for the advertised hashrates) seems to be at about 245-250 MHz for the ZeusMiner Thunder X6 and the ZeusMiner Hurricane X6, however the ZeusMiner Lightning X6 does not appear to be handling so well at these operating frequencies. In fact even at 230-235 MHz we are still seeing HW error rate of about 10% or more which in our opinion is a bit too high, but going lower in terms of operating frequency does lead to decrease in the hasrate of about 40-42 MHS. So what is the reason for this difference? We have already suspected that the operating voltage of the Lightning X6 is even lower than what it is on the smaller Thunder X6 and Hurricane X6, but checking out the actual power consumption of the miners and opening them up to inspect them has confirmed our suspicion. Zeus has even made it very easy to spot the difference by putting a label of the actual voltage used by each board and as you can see on the photos, the Thunder and Hurricane X6 miners have 1.1V used and the Lightning X6 has the chip voltage lowered to 1.05V. This is probably done in order to get the total power usage of the Scrypt ASIC miner below 1 KW, but as a result you will need to use an operating frequency of about 220 MHz for low HW error rate and the hashrate you can expect is more like 36-38 MHS and not 40-42 MHs. But this is not the only concern we have with the new miners as Zeus has done some more optimizations, probably with the idea to cut costs further, that we do not like very much.

zeusminer-hurricane-x6-power-usage

We are stating with the power usage of the 9-10 MHS ZeusMiner Hurricane X6, according to the official specifications is should use about 230W of power and a good 250W power supply is recommended, though you probably will not be able to find a 250-300W 80 Plus Gold or Platinum PSU, you can use a more powerful one. Running the Hurricane X6 miner at 250 MHz operating frequency we have measured an actual power consumption of about 276-278 Watts. This is a bit higher than the claimed power usage, and though it may not be such a big of a deal, when you consider that the miner is powered via a single 6-pin PCI-E power connector this raises some concerns.

zeusminer-thunder-x6-power-usage

Moving to the power usage of the 18-20 MHS ZeusMiner Thunder X6 we see on the official website a power consumption of 450W claimed with a good quality 500W power supply recommended. Our tests have shown a different number running the miner at 250 MHz, the number we got is more like 542-546 Watts and that is like 100W more. This is essentially double the power usage of the Hurricane X6 as the Thunder X6 does have twice the number of chips inside, so it is to be expected if both miner run at the same frequency and use the same voltage for the ASIC chips.

zeusminer-lightning-x6-power-usage

Next up is the power usage we have measured for the 40-42 MHS ZeusMiner Lightning X6 Scrypt ASIC. The officially cited power consumption is 980W with a good 1200W PSU as a recommendation. Our tests have shown actual power consumption for the miner at just about 906-914W at an operating frequency of 230 MHz and about 970W measured at the wall (taking into account the PSU efficiency). Here the power consumption is not double the one from the Thunder X6 even though the number of chips inside is doubled and the reason for that is the lower voltage of the chips that is 1.05V instead of 1.1V. It has been lowered as otherwise the actual power consumption should’ve been something like 1100 Watts at 1.1V.

We’ve confirmed that the power consumption numbers on the ZeusMiner website are not entirely correct, they are lower on the smaller miners and higher ton the top miner compared to what we have measured. That however is not the actual problem here, the problem and the thing we did not like much is the fact that Zeus has reduced in half the number of PCI-E power connectors available to power the new miners. As you should know by specifications power supplies are designed to supply 75W over the 6-pin PCI-E power connector and 150W over the 8-pin PCI-E power, even though the actual hardware is usually capable of supporting higher power as these were originally designed to power video cards. The 18 AWG wires used normally in most power supplies (some use 16 AWG) and the connectors should be capable of 8-10A over a single 12V wire and with 3x 12V lines on both 6-pin and 8-pin PCI-E power connectors this should in theory easily cover up to 300W. The problem however comes from the fact that PSU manufacturers are not required to design their products to handle so much power over a single rail for PCI-E power connector and most don’t do it, even if they do supply some extra over the 75W/150W requirements they have. So 300W per single 6-pin PCI-E power line can be a bit overkill for an ASIC miner running 24/7 and the easiest way to confirm that without special equipment is to check the wire by touching it after the miner is started. if it quickly gets hot, then things are definitely not Ok, though it is just a bit warmer to the touch it might be fine.

Having 278 Watts of power provided to the ZeusMiner Hurricane X6 over a single 6-pin PCI-E power connector is something that raises some concerns, so you should be extra careful what power supply you are using. With the ZeusMiner Thunder X6 the situation is very similar, though here you have two 6-pin PCI-E power connectors with each transferring about 273 Watts of power. Things are a bit better with the ZeusMiner Lightning X6 that has 4x 6-pin PCI-E power connectors with each one getting just about 228.5 Watts. So what ZeusMiner should have done is actually double the number of PCI-E power connectors on their miner to distribute power more evenly to each board and to lower the load on each of the PCI-E power lines. We do recommend to be extra cautions when you are setting up the new miners and check the cables to make sure that they are not getting very host quickly after you turn on the miner as this may potentially lead to damage of the hardware – both the miner and the power supply and this is something that you will want to avoid.

silverfish-blade-1

Silver Fish is a Chinese company making Scrypt ASIC miners based on 55nm Scrypt ASIC chips that they have apparently developed in-house called SF300K01 – 55nm chips with lower power consumption and rated at 625 KHS at 7W per chip. The main target for their Srypt ASIC miners available in the smaller Blade version we tested and the bigger 70 MHS models that Silver Fish makes is the Chinese market as they do not even have an English version of their official website. Their Scrypt ASIC miners, especially the smaller 25-28 MHS Silver Fish Blade miner, are already finding their way in other parts of the world and that is precisely why we have decided to get a unit and review it. It is certainly an interesting product with some specifics that you can also consider as disadvantages if you compare to what we have used with traditional ASIC devices, but they are not necessarily such. In fact the approach that Silver Fish has taken into their products is quite interesting even though it might be considered somewhat limiting the user.

silverfish-blade-2
silverfish-blade-3

The Silver Fish Blade consists of two independent PCBs, each with 20 SF300K01 Scyrpt ASIC chips grouped in blocks of 4 with independent power blocks. So the whole Blade miner has 40 chips for a total hashrate of about 25-28 MHS for mining Scrypt crypto currencies such as Litecoin. The device is cooled by a single large 120mm Delta fan that is quite noisy, even though it apparently should be controlled through a PWM from the device based on the reported temperature from the available onboard sensors. The reason for the high noise level is probably a direct result from the open design of the device that certainly is not optimal for proper cooling with lower RPM of the fan, though apparently the device does manage to keep things in normal operating temperatures even this way.

silverfish-blade-4

Notice that each of the PCBs is completely independent and has a single 6-pin PCI-E power connector as well as two LAN ports. The advertised power usage at wall for about 28 MHS performance is 470 Watts, meaning that each PCB needs about 235W, but we’ll get to the actual power usage in a bit. The LAN connectors on the back are used to connect the two PCBs together and you only need one cable to go to a computer or a network appliance in order for your device to be accessible from the mining software. According to Silver Fish this approach for connection allows for up to 100 PCBs to be connected together with a single connection to your network, so apparently these miners have been designed to be used and deployed in larger mining farms and nor for small home-based miners.

silverfish-software

The software that is used to control and mine Scrypt with the Silver Fish Blade miners is only Windows-based and needs to be run on a PC that is connected on the same network that the Scrypt ASIC miner is connected in. The software should be able to automatically find the miners in the network and start mining with the pools you have set to be used, so it is pretty straight forward thing to do. Unfortunately the latest version of the software R3.04 is still not available in English, though you should be able to use the previous SFMiner 2.1 English just fine, even though the configuration utility inside it is not translated in English.

silverfish-power-usage

A quick look on the actual power usage of the Silver Fish Blade miner supposedly running at 28 MHS with a 1200W power supply attached to the miner. Apparently the actual power consumption is is around 430W and thanks to the fact that we are using a good power supply at its sweet spot in terms of efficiency (about 99%) the power usage at the wall is just slightly higher. With a not so good power supply you can probably expect to get close to the 470W advertised as actual power usage at the wall. With this power usage the Silver Fish Blade is probably the most power efficient 55nm-based Scrypt ASIC miner currently available on the market, beating even the new lower power usage ZeusMiner models.

silverfish-thermal-images

We’ve already mentioned that we do not like that much the open design of the Silver Fish Blade miner, even though apparently it has no problems with keeping the things in acceptable temperature levels. As you can see from the thermal images we’ve captured of the miner in action the temperature level increases as we get further away from the cooling fan and the temperature difference can be higher than about 10 degrees between the two ends of the PCBs. If you want to improve things with the cooling you can easily create a case for the miner with handy material in order to improve the airflow and thus the cooling efficiency of the device. What we could not find a way to do is to overclock the miner to see if we can squeeze some more performance without significant increase in power usage.

silverfish-hashrate-ltcrabbit

Time to see what is the performance we are getting with the Silver Fish Blade Scrypt ASIC miner, officially the manufacturer rates these at 25 MHS (40 x 625 KHS), but the unit we got was actually rated at 28 MHS, so apparently slightly overclocked. We have tested the poolside performance at the LTC Rabbit Scrypt mining pool as it has recently opened new fixed difficulty ports for Scrypt ASIC miners. The results we’ve got have shown hashrate peaking at a bit over 28 MHS and with an average of about 27.4 MHS, pretty similar to what the software used for control and mining also reports in terms of performance. Trying out with other Scrypt mining pools such as CoinKing for example has shown us pretty similar results, or you can expect an average hashrate of a bit less than 28 MHs.

So what is our conclusion about the Silver Fish Blade Scrypt ASIC miner? There are some things that we did not like such as the open case design and not so efficient and nosier cooling or the availability only of Windows-based software that is not completely translated in English. As well as some things that we liked, such as the compact size, the lower power usage and the network connectivity that allows for easy scalability. Good performance and stable operation, combined with lower power usage than competing 55nm-based alternatives makes this an interesting option if you manage to get a good deal in terms of price. If you manage to find these at sub $900-1000 USD price they may still be a good investment, even though with the current network difficulty and price of LTC you should still consider carefully.


top