The Bitmain Antminer S3 Bitcoin ASIC Miner is their next step in the evolution of their ASIC chips as well as going back to the basics with the blade design that started it all. There are some changes that have been made along the way though. First is the cowled dual fan enclosure that the S3 has is a new twist on the first-blade design. They also made overclocking easier as well. The Antminer S3 was initially announced to run at a 478 gh/s and then revised down to 440 gh/s.
Chips 32x Bitmain BM1382Gen2
Hash Rate: 440 GH/s±5%
Power Consumption: 366 Watt at the wall
Power Efficiency: 0.77 J/GH on wall
Power Supply: 12V DC
Size: 331 mm x 137 mm x 160 mm
Fans: Two 14038 fans mounted on both front and back
Operating Temperature: 0 °C to 35-°C
Complied with: FCC / CE
Network Connection: Ethernet
The shipping gods were good to me yet again. The S3 arrived safe and secure thanks to simple yet effective foam and bubble wrap packing. For those familiar with the S1 and S2s the interface is the same with a few tweaks. You log in as usual to 192.168.1.99 through your web browser and the default password is “root.” Change it immediately. If you are going to have more than one S3, i suggest you change it to something else as well. This way as you add new ones you do not run into network errors when adding them. Once done, you can go to the miner config page and set your pools. You can also set the S3 to failover or to perform load balancing for pool management. The setting is convenient as you can use multiple pools at once. I tried it on NastyPool and BTCGuild. Both sites reported 200 gh/s to 230 gh/s. My S3 has been great running at 260 gh/s with no overclocking and pulling in 340 watts at the wall. If you want to OC, you can plug in the extra set of PCIe cables to provide extra power. These extra power connections make overclocking much easier as you have the power on hand to OC.
Bitmain listened to many in the mining community that mention the S1 could be better cooled with a cowl on it. The S3 has a cowl over the blades and dual fans. The cowl allows for very efficient cooling. The fans, despite there being two of them, were actually not as loud as my S1. My S3 ran at 42 to 44-degree Celsius. In contrast, my S1 runs between 47 and 50 Celsius. Bitmain has focused on making the S3 as efficient as possible. Accuracy has been tweaked so hardware error rates are as low as .01%. On my units, the hardware errors hovered between .01 and .02. The low-error rate can be hundreds of times better than many of the competition’s miners currently.
The S3 is super easy to use as usual. Bitmain did not cut any corners. The new design helps show off the new chips. I have overclocked it only mildly, I was able to hit 502 gh/s and 374 watts at the wall and the temp was 48 degrees. I am sure that the OC masters in the community will be able to push it even further. I have to stress the build quality again is amazing. It is light yet built solid. Each blade has an aluminum heatsink on the front that is large also one on the backside of the board this is all set in the cowl in such a way as to promote the best heat exchange.
I heartily recommend the Antminer S3. I was able to get batch one S3s from Juan at https://www.112bit.com. Juan had this to say about the S3 and how fast the sales have been as well as how far it has evolved.
Juan: The S3 is a huge step forward in both power consumption and performance. We also have to remember that the S1 came out and was $3000, and Bitmain said “here we are, and we have miners in hand.” There were no presales just miners shipped and lots of them. The S3 sold out very very fast. People know and trust Bitmain to deliver, and they do.
With sales of the S3 already into batch 3 and sold out, Bitmain is looking again to capture a huge portion of the global hash rate being run on their equipment. While their competitors like Butterfly Labs cannot ship anything and KNC ran out of chips building their personal mining farm Bitmain delivers. The S3 is one of my favorite miners along with the Spondoolies-Tech SP10Dawson.
Last modified: October 6, 2014 11:15 UTC