 Show Posts
|
|
Pages: [1] 2 »
|
I created a user there, but apparently users can't post / reply / do anything there (?) New signups need approval before they can post -- approval which we are aiming to do within 24 hours. |
|
|
|
After 2 hours of running new firmware, I can already tell it is helping a lot.
Prior to the firmware update: 562mhz showed 360GH/s on pool and 10% error rate
After firmware update: 562mhz shows 375GH/s on pool and 0.3% error rate
I will try to push the core clock up tomorrow and see how it handles it. Before the update I could not go higher then 562mhz without it becoming unstable.
Looking forward to hearing how it goes. Most boards run poorly at certain frequencies, which can vary from board to board. So if you go up by 1 or 2 MHz at a time, you might run into a poorly performing clock speed far lower than the optimal speed. |
|
|
|
Does the RPi auto update to it if we have auto updates checked? I noticed my RPi reboot and firmware is showing as blank now. Is it safe to assume it updated to 0.4 or is there a way to tell somehow with some terminal command?  An RPi image that both auto updates itself and the firmware is planned, but not yet ready. But people have waited long enough for this firmware update. Again the image linked to in the forum post is *only* an updater; there is no MinePeon or web interface supported. Do NOT overwrite your HashFast / minepeon SD card; (unless you back it up first!). |
|
|
|
HFA: Failed to receive OP_NAME response over and over. Any ideas how to fix? (cgminer version 4.0.1) cgminer 4.0.1 basically needs new firmware flashed as 0.2 won't work with it. Firmware 0.4 has been released. It is backwards compatible, but should work best with cgminer 4.0.1. (And vice versa.) Primary Download: http://setup.hashfast.com/files/hashfast-20140224-0-multiarch.tar.gzBackup Download: https://mega.co.nz/#!bkBjwboL!cbkZgGh7EqIRviHDwJc6l0IOUN2IWLjQPcLFitejxu8 The above are for Ubuntu 32/64 and Raspberry Pi (ArchLinux) only. For users on Windows or otherwise less familiar with Linux, the HashFast firmware update version 0.4 is also available via a one-click upgrade through our custom Raspberry Pi image. This image is *only* an updater; there is no MinePeon or web interface supported. *DO NOT* write this image over the HashFast MinePeon SD card that came with your system (unless you back it up first). Use a different SD card instead. More details and full instructions on the raspberry pi option available at: http://forums.hashfast.com/forum/software/firmware-updates/34-firmware-0-4Cheers, Amy |
|
|
|
I like the idea of Dual PSU, especially for expansion and failover
A Seasonic PSU would be the last thing to fail in this or any miner. I can almost guarantee you that HF aren't using two for failover purposes. Correct, we're not using two for failover purposes, we're using two because we chose to provide a 20% safety margin on top of what the Sierra might need. For our mining rigs we have chosen to specify power supplies that are rated 20% higher than the highest wattage we expect even including significant overclocking. The GN chip uses 250W at nominal, and we have designed all the systems to handle upto 350W. When the mini motherboard is supplying 350W to the chip (this level of overclock should result in up to 540GH/s) the mini motherboard may consume up to 411W of 12v, due to losses in the VRM. Add 25W for fans, and you get to 436W per mini-motherboard. Multiply by 3 chips in the Sierra, and you get about 1300W. We've added a safety margin of 20% extra on top of this - so we needed power supplies that were rated to a minimum of 1560W @ 12v to meet these specifications, and maintain the headroom and margin we desired to ensure reliable, efficient long term operation.
There are very few manufacturers who make single 1600W supplies, and we didn't find a single unit from a high quality vendor.
|
|
|
|
For our mining rigs we have chosen to specify power supplies that are rated 20% higher than the highest wattage we expect even including significant overclocking. The GN chip uses 250W at nominal, and we have designed all the systems to handle upto 350W. When the mini motherboard is supplying 350W to the chip (this level of overclock should result in up to 540GH/s) the mini motherboard may consume up to 411W of 12v, due to losses in the VRM. Add 25W for fans, and you get to 436W per mini-motherboard. Multiply by 3 chips in the Sierra, and you get about 1300W. We've added a safety margin of 20% extra on top of this - so we needed power supplies that were rated to a minimum of 1560W @ 12v to meet these specifications, and maintain the headroom and margin we desired to ensure reliable, efficient long term operation. There are very few manufacturers who make single 1600W supplies, and we didn't find a single unit from a high quality vendor. In addition this size of supply is more expensive than 2 smaller supplies. The Sea Sonic supplies are really well built, and even significantly overbuilt. In all tests they are capable of supplying power well above their ratings, and the output regulation is very accurate and smooth. They handle hot operating environments very well. Please go read the reviews of the supplies in the Sierra - the Sea Sonic X-850FM3. http://hardocp.com/article/2013/04/04/seasonic_xseries_x850_power_supply_review/9#.UmDCKVCcc5g"... the X-850 is simply another excellent member of the X-Series of power supplies from Seasonic and users won't go wrong picking it up as it is generally better overall than any of the closest competition (Corsair AX850, Corsair HX850 Gold, and SilverStone ST-85F-G Evolution). At the end of the day, the decision is probably going to go in favor of the X-850 above those competitors if the prices are within ~$15-20, or so, of each other." Seeing as HashFast is picking up the tab for the power supply, I'd say X-850 is good decision for the customers.  |
|
|
|
wow, that's a very cool setup that i hadn't even seen.
notice 2 of the modules are stacked. 1 of the radiators is flipped upside down to accommodate this stacked setup. also, i'm glad to see that 2 of the fans have been moved to the back to provide exhaust. also note that we have 2 Seasonic PSU's in place to drive the 3 modules.
Yes, I got two questions on this already so let me clarify: there are three modules in that image of the Sierra. Two are on the bottom layer (side by side), and one module seated above. This let us have better airflow in the case (and more room for the SeaSonic PSUs). |
|
|
|
P.S. If people DO have other questions about HashFast, please feel free to send them my way.
Maybe I'll post a bit about the mysterious Sierra Controller. (drum roll please.....) ... ... "The Controller" is simply a computer that can run cgminer, and has a USB port. That's it. We're recommending it be a linux box, but other than that it could be most anything: Your desktop. That old laptop that's been sitting in your closet for the last few months. A 1U (or 2U, if you prefer) linux server that is rack-mountable. Or even a Raspberry Pi. Though until we can do more testing, we don't know how many Sierras a single Raspberry Pi could control. A single Sierra is a 1.2THash/sec machine, remember. So the difficulty setting you choose has a lot to do with how chatty that Sierra will be. (We obviously do not recommend using a difficulty level of 1 or 2 on a Sierra. It'd be a waste.) So, if you want, you can BUY a low-end linux server to act as the controller for the Sierra(s) that you have bought. (And of course, with every order of 10 or more we provide that controller.) But it's quite probable (if you're a geek) that you already have a computer around that you can plug your 1.2 THash/sec Sierras into. |
|
|
|
I do wonder - what else is going on today, that these repeated 'embarrassing' posts are supposed to distract us from?
Of course, maybe there is nothing else going on today. Just another day on the forums. Cheers, Amy |
|
|
|
I'm at a BTC meetup in Sunnyvale, where the Hashfast VP of Engineering is giving a talk right now.
This is admittedly rather time-sensitive, but if anyone has any questions, I'll be happy to ask 'em!
Ask if they're still on schedule for shipping next week. Not surprisingly, this was indeed one of the questions that came up last night. And, as of a few days ago, ... ... it looks like we're probably going to slip, by about a week. Ironically, not because of the silicon, even though that is the part of the system that takes the longest to manufacture. That's going through the foundry smoothly. But we ran into a delay sourcing one of our other components. ... It's the 17th, your projected date for getting the chips from TSMC? Am i misremembering? Trolls are funny. This is what, the eighth time you've posted this hour? The eighth time in one hour you repeat what you hope is an embarrassing question? The process at the foundry is going well. Someone posted an optimistic date, one I personally did not approve of nor condone. (Sorry, cypherdoc.) And I cannot give any other dates for the foundry. I can tell you when they pop out of the oven, nothing sooner. But honestly the foundry is not important, because that is not what will ultimately delay us. That delayed component is what is pushing our date out. Funny that you harp on the foundry when you could be harping on delayed shipping instead. I do wonder - what else is going on today, that these repeated 'embarrassing' posts are supposed to distract us from? Amy .... Who is somewhat regretting her decision to hang out on the forums this afternoon. P.S. If people DO have other questions about HashFast, please feel free to send them my way. |
|
|
|
Ask if they're still on schedule for shipping next week.
Not surprisingly, this was indeed one of the questions that came up last night. And, as of a few days ago, ... ... it looks like we're probably going to slip, by about a week. Ironically, not because of the silicon, even though that is the part of the system that takes the longest to manufacture. That's going through the foundry smoothly. But we ran into a delay sourcing one of our other components. We've now found an alternative source. But that delay means it'll be challenging for us to ship finished systems before the end of October. It's disappointing, but unfortunately these things happen. I'm glad we've made it this far through the process, and are still so close to our target date. (I'm watching every day to see how close Cointerra will come to their targeted tapeout date.) Amy Woodward VP Engineering HashFast Technologies Amy, Thank you for directly addressing these questions in this thread. I really appreciate you giving an honest reply as to shipping dates. With that being said, as we get closer to that date, can we expect more frequent updates regarding shipping? Thanks DHJ/WH My available time ebbs and flows with the demands of each day. But Yes. I can promise you more frequent updates. Those I'll probably offload to Sam, Erin or Cara. But I wanted to make sure the post about the slippage in dates came straight from me. Amy Woodward |
|
|
|
as Order #1, i can assure you i will be in constant contact with them regarding expected shipping dates and will keep all of you posted.
And as Order #1, cypherdoc has my cell phone number. (And we've already talked this afternoon, though I called him this time.) ...Fortunately there've been no midnight calls yet... |
|
|
|
I'm at a BTC meetup in Sunnyvale, where the Hashfast VP of Engineering is giving a talk right now.
This is admittedly rather time-sensitive, but if anyone has any questions, I'll be happy to ask 'em!
Ask if they're still on schedule for shipping next week. Not surprisingly, this was indeed one of the questions that came up last night. And, as of a few days ago, ... ... it looks like we're probably going to slip, by about a week. Ironically, not because of the silicon, even though that is the part of the system that takes the longest to manufacture. That's going through the foundry smoothly. But we ran into a delay sourcing one of our other components. We've now found an alternative source. But that delay means it'll be challenging for us to ship finished systems before the end of October. It's disappointing, but unfortunately these things happen. I'm glad we've made it this far through the process, and are still so close to our target date. (I'm watching every day to see how close Cointerra will come to their targeted tapeout date.) Amy Woodward VP Engineering HashFast Technologies |
|
|
|
So knite,
Was anything interesting asked or said regarding HF at the BTC meetup?
Nothing? I thought some interesting questions were asked at the meetup, but then, I am biased. I'm in the midst of a meeting at the moment (the closest I get to free time these days, hence checking the forums). But will put aside some time to put down some Q&As this afternoon. There was also video taken. Not sure what Plug and Play's schedule is to get that edited & posted, but I'll let you know once I do. Cheers, Amy Woodward VP Engineering HashFast Technologies |
|
|
|
There is a lot more great content on the HashFast blog that somehow escaped the forum's notice!
I'm pretty sure most people here have read all the blog content. Speaking of the blog doesn't the post titled, "Baby Jet Controller" directly contradict what is being said here in this forum? On the blog it says, "The Baby Jet ships with a built in Raspberry Pi model B. This acts as the controller for the one or two GN chips in the Baby Jet. The controller talks to the chips over its internal serial port." but in recent posts on this forum company representatives are saying that each "mini-board" will communicate with the controller over USB. So no the content of the blog hasn't escaped notice which is why I was surprised when it was said that the boards connected to the controller via USB vs. the earlier stated serial connection. https://hashfast.com/the-baby-jet-controller/Hi pmorici, Thanks for the question. We made the change from serial to USB connection for the Baby Jet so that we could have the same interface for Baby Jets as for the Sierras. USB also allows us to provide a much richer interface than a serial connection. And finally, the "serial"/UART interface on the R-Pi is 3.3V levels. The chain connectors on the modules are 1.8V levels. Using the serial interconnection method would have required level translators. Using the USB connection requires a simple USB cable. Cheers, Amy Amy Woodward VP Engineering HashFast Technologies |
|
|
|
Just to be clear, what is the difference between a "module" and a "motherboard" by your definition? How does the "module" receive it's power without some kind of VRM mounted on a PCB?
Hi donch, The website has been updated; the description was old - sometime in July old. Yes, each module is a PCB board, with amongst other things, a GN chip, a VRM, usb port, power connectors, mounts for the water cooler, etc. And a Baby Jet has enough room inside to fit two modules. Amy Woodward VP Engineering HashFast Technologies |
|
|
|
I got the same doubt and I have been told that is all about chips layering and that they are keeping in schedule Lets wait for the blog update  The gist of the blog entry is that everything at the foundry is going well. Unfortunately, our NDA with TSMC forbid us going into detail about where exactly in the process our wafers are. On the other hand, nothing forbids me writing a general educational article on some of the different stages in semiconductor fabrication. Which I have done, and you can check out on our blog. www.hashfast.com/blog |
|
|
|
Hello world! I am the new community liaison for HashFast. We are striving to forge close links with our customers and the mining community. To make that happen, I'll be checking the forum almost everyday, posting updates, answering questions, and gathering feedback. You all may contact me via email ( Community@Hashfast.com), PM (after I escape from Noob Island), or on the 'HashFast Community' thread I will create ASAP. Welcome, Community Liaison. Amy Woodward VP Engineering HashFast Technologies |
|
|
|
Are the boards and cases sourced already for the chips to be assembled into?
Yes. To be more verbose... Similar to what we did with Uniquify, we have been working extremely closely with both the board manufacturer / assembly house, and with the company that will be assembling the finished Baby Jets and Sierras. We are in daily, often hourly contact with them. I'll request pix. |
|
|
|
Are the boards and cases sourced already for the chips to be assembled into?
Yes. |
|
|
|
|
