Archive for the ‘Mining Hardware’ Category


Today we did took thermal images in order to be able to see how well does the GAWMiners Falcon Scrypt ASIC miner deal with the heat, since we have an unbranded Falcon unit it is essentially the same as the standard ZeusMiner THUNDER X3 inside. With a power consumption of over 900W you might worry a bit about the thermal performance of the device, or at least be a bit curious how good is the cooling and if you are able to possibly improve it would that bring the level of HW errors down a bit.


As you can see from the thermal images on the outside the Falcon is pretty cool with the hottest areas barely getting up to about 42 degrees Celsius. The device uses two large 120mm fans that pull the hot air through the case of the miner over the large aluminum coolers inside the case, these fans are very powerful and a bit noisy. In fact they are so powerful that they cause some vibrations of the whole case of the miner and since there are no rubber feet at the bottom of the case the miner might sound noisier if the surface it is placed on picks up the vibrations and amplifies them. Not that the fans are that cool, they are a bit noisy, butt they do their job very well in keeping the device cooler and operating without heat problems of any kind. Actually what you should be more careful with is the cables from the power supply that go to the miner as they may get hot due to the significant power being transferred through them. If they are cool or just slightly hot to the touch they are most likely fine, but if they start to get hotter after you turn on the miner, then you might need to think about getting a better power supply.


Opening the case of the Falcon Scrypt ASIC while it is operating may not be the wisest thing to do, especially for longer periods of time as it may permanently damage the device due to overheating. The fact that you have fans pulling air from inside the case means that when you open the top the efficiency of the cooling of the device drops a lot and everything starts to get hotter and hotter very quickly. We did it remove it very quickly just to take the thermal photos and put back together the case in order to measure the temperatures as close as possible to the actual ones while the miner is working normally. As you can see on the left image the four big aluminum heatsinks remain very cool at about 30-35 degrees Celsius while the fans are pulling a lot of air though them. Looking at the boards with the ASIC chips on them things do seem a bit hotter as to be expected, the chips are pretty hot as the design of the miner uses the back of the PCB to transfer the heat to the heatsinks and the chips do not have coolers placed on top of them. Temperatures of about 60-70 degrees Celsius are apparently something that is not a problem for them, though you should be careful not to get the temperatures much higher like for example leaving the miner to work open for some time. The power modules of the boards do seem cooler than the ASIC chips, though they do get a bit hot as well. So there is the possibility for some improvements to the cooling of the device to be improved further and now that we have some thermal images we know where we should focus on, you are also welcome to use these images to work on improving the cooling and possibly the performance of the device.


We’ve read some comments about people complaining from the stock ZeusMiner Blizzard Scrypt ASICs’ like them being crappy and getting too hot. From our personal experience we can also confirm that the 60W power adapters provided with the miners do get hot, but this is to be expected with a power draw of the device of about 44W at 300 MHz and 48W at 328 MHz. Our unit that we are currently testing is hot to the touch, but only using that to judge can be misleading as any temperature higher the one of our body is perceived as hot. Anyway, we did decide to try using the Blizzard miner with a high-quality ATX power supply and measure the actual power draw we are getting…

We have attached the ZeusMiner Blizzard to an 80Plus Platinum power supply and the result was a bit surprising – very low efficiency of the power supply due to the low load. Clearly the 1200W Corsair power supply is not designed to be very efficient with a load of just about 50W, actually 48-49W measured as used by the miner, so not much different than what we got from the standard PSU. The difference here however is that due to the low efficiency that the PSU is running at the actual power consumption of the miner off the power socket is about 64W. We did measure 48W of power usage with the stock power adapter supplied with the miner at the power socket, so it seems that these 60W power adapters are quite efficient in converting the 220V power to 12V. Of course by increasing the number of miners connected to the power supply and raising the load to at least 10% or more the efficiency should increase and make things right, though no point in running just a single Blizzard off an ATX computer power supply – better stick to the power adapter supplied with the miner.


Our testing of the 27 MHS GAWMiners Falcon Scrypt ASIC continues and in the last few days we’ve been moving the ASIC miner through different pools in order to check what is the actual performance that you can expect from it. If judging the performance based on what cgminer reports you should be looking at the WU reported speed as the actual one and not the one for the last 5 seconds or the average reported by the software. Though do note that the WU reported speed is for the last minute based on the shares submitted. To get a more accurate result we need an average that covers longer period of time such as an hour or even a day and you can easily get such results from a pool monitoring and reporting not only your current hashrate again based on the shares submitted in the last minute for example. Do note that the Falcon Scrypt ASIC from GAWMiners is essentially the same in terms of hardware inside as the ZeusMiner THUNDER X3 and should produce the same hashrate. So far the recommended operating frequency of 328 MHz seems to work very well, though the standard rating of the ZeusMiner ASIC chips seems to be 300 MHz. The results you see here are made using the 328 MHz operating frequency with an HW error rate hovering around 5%, slightly higher than we would want, but still acceptable and apparently the normal one for the Zeus chips.


Here you can see a 24 hour chart with the hashrate reported by the LTC Rabbit mining pool. The minimum average hashrate per hour we got at the pool for the 24 hour test period was 25674 KHS and the maximum was 29332 KHS with an average across the 24 hour period of 27882 KHS. A very good result and with that hashrate you can currently get about 2.5-2.6 LTC mined a day, not too bad with the difficulty recently going past 10 thousand, but not that great either with the current exchange rate for Litecoin. So mining LTC now and not selling it immediately seems like the smarter choice if you have decided to invest in larger Scrypt ASIC miner such as the GAW Falcon.


Trying out the Scrypt ASIC miner at the ScryptGuild mining pool where you mine altcoins and get payed in BTC has shown us very similar performance. The average hashrate poolside was 27738 KHS and with it the expected payout for 24 hours of mining could get you around 0.05 BTC at the moment. So poolside we are actually getting a bit more than the 27 MHS promised by GAWMiners, though we are a bit short from the 28 MHS that Zeus claims for the THUNDER X3, but it is still acceptable. It is possible that with some tweaking that we can get slightly better results by lowering the HW error level. The quite high power usage of the miner however does not leave that much room for optimizing the cooling and squeezing some more MHz from the miner. We are not ready to give up, so stay tuned for more information as we continue to play around with the miner. It will be interesting to compare this miner with the A2-based 27 MHS mini miner that uses Innosilicon A2 Scrypt ASIC chips and we are hopefully soon be able to compare the two devices.