The Habanero - 650GH/s - OOS

[SVK]

It's going well. I haven't done as much testing as I'd like since I sent out the boards I made to ChipGeek and Gateway for testing/coding purposes. Firmware work is coming along well, and will be done by the time we're ready to assemble them.

The current plan is to have the assembly house accept all the parts by May 6th for kitting in order to have assembly start May 9th, with surface mount, through hole and test occurring concurrently the following week. The production revision of the PCB has been sent to the board house, and they will be ready in time to be delivered on the 6th.

I'm currently bench testing a board to get a better handle on the numbers to give us some baselines that all the boards should hit and that we feel comfortable advertising the board at, as well as testing more extreme situations in order to push the limits. Once that's done we'll open the store here right away to orders.

Many thanks for update. It looks like it's coming well together
Rough math on dates it looks like another 4-5 weeks ? 

Thank you again and hurry up 

[1714853554]

1714853554

[1714853554]

1714853554

[1714853554]

1714853554

[MrTeal]

It's going well. I haven't done as much testing as I'd like since I sent out the boards I made to ChipGeek and Gateway for testing/coding purposes. Firmware work is coming along well, and will be done by the time we're ready to assemble them.

The current plan is to have the assembly house accept all the parts by May 6th for kitting in order to have assembly start May 9th, with surface mount, through hole and test occurring concurrently the following week. The production revision of the PCB has been sent to the board house, and they will be ready in time to be delivered on the 6th.

I'm currently bench testing a board to get a better handle on the numbers to give us some baselines that all the boards should hit and that we feel comfortable advertising the board at, as well as testing more extreme situations in order to push the limits. Once that's done we'll open the store here right away to orders.

Many thanks for update. It looks like it's coming well together
Rough math on dates it looks like another 4-5 weeks ? 

Thank you again and hurry up 

I am going to guess we'll call this a May batch, and any shipping before the end of May will be a bonus. I would guess that it will be sooner, but I'd rather have pleasantly surprised customers than disappointed ones.

As a point of reference, for the first batch of Chilis we got parts to the assembly house on Oct 7th, assembly started on Oct 9th and first units shipped Oct 14th, with the last ones going out the 18th. I imagine the Habanero will take a little longer to process though.

[SVK]

It's going well. I haven't done as much testing as I'd like since I sent out the boards I made to ChipGeek and Gateway for testing/coding purposes. Firmware work is coming along well, and will be done by the time we're ready to assemble them.

The current plan is to have the assembly house accept all the parts by May 6th for kitting in order to have assembly start May 9th, with surface mount, through hole and test occurring concurrently the following week. The production revision of the PCB has been sent to the board house, and they will be ready in time to be delivered on the 6th.

I'm currently bench testing a board to get a better handle on the numbers to give us some baselines that all the boards should hit and that we feel comfortable advertising the board at, as well as testing more extreme situations in order to push the limits. Once that's done we'll open the store here right away to orders.

Many thanks for update. It looks like it's coming well together
Rough math on dates it looks like another 4-5 weeks ? 

Thank you again and hurry up 

I am going to guess we'll call this a May batch, and any shipping before the end of May will be a bonus. I would guess that it will be sooner, but I'd rather have pleasantly surprised customers than disappointed ones.

As a point of reference, for the first batch of Chilis we got parts to the assembly house on Oct 7th, assembly started on Oct 9th and first units shipped Oct 14th, with the last ones going out the 18th. I imagine the Habanero will take a little longer to process though.

You are absolutely right with the fact that rather say later than sooner. People here go ballistic even if you are 1 day late. SCAM SCAM lol.
I'm resisting from buying Ants S1's since I have heard about your project. 1 chip with 4 cores is true advancement at least in my eyes and not 640 chips on 5 boards

Keep up the good work and please be competitive with the pricing I would not like to bang my head that I didn't buy them Ants at the end  

[MrTeal]

I've talked with Gateway and it sound like the store will likely go live either this weekend or Monday.

Testing has gone well so far, though it has been limited because I have it set up in my office. Besides the fact that I can really only run it in the evenings/night, it's still cold enough here (high of 5C today) that our building is running the heat. Combine that with 700W from the Habanero and my small office was a sauna this morning. It did spend the night running comfortably at 875MHz (640GH/s poolside). We really need to work with the mining software providers to get realtime voltage and frequency adjustments back into the software though, similar to the GPU days.

I'll be bringing it home tonight, so I should be able to run it longer and harder going forward.

[nexus99]

I'm crossing my fingers for a Saturday store launch. Other wise I'm gone for two weeks and will probably miss out. 

[Abraxas601]

It's going well. I haven't done as much testing as I'd like since I sent out the boards I made to ChipGeek and Gateway for testing/coding purposes. Firmware work is coming along well, and will be done by the time we're ready to assemble them.

The current plan is to have the assembly house accept all the parts by May 6th for kitting in order to have assembly start May 9th, with surface mount, through hole and test occurring concurrently the following week. The production revision of the PCB has been sent to the board house, and they will be ready in time to be delivered on the 6th.

I'm currently bench testing a board to get a better handle on the numbers to give us some baselines that all the boards should hit and that we feel comfortable advertising the board at, as well as testing more extreme situations in order to push the limits. Once that's done we'll open the store here right away to orders.

Many thanks for update. It looks like it's coming well together
Rough math on dates it looks like another 4-5 weeks ? 

Thank you again and hurry up 

I am going to guess we'll call this a May batch, and any shipping before the end of May will be a bonus. I would guess that it will be sooner, but I'd rather have pleasantly surprised customers than disappointed ones.

As a point of reference, for the first batch of Chilis we got parts to the assembly house on Oct 7th, assembly started on Oct 9th and first units shipped Oct 14th, with the last ones going out the 18th. I imagine the Habanero will take a little longer to process though.

You are absolutely right with the fact that rather say later than sooner. People here go ballistic even if you are 1 day late. SCAM SCAM lol.
I'm resisting from buying Ants S1's since I have heard about your project. 1 chip with 4 cores is true advancement at least in my eyes and not 640 chips on 5 boards

Keep up the good work and please be competitive with the pricing I would not like to bang my head that I didn't buy them Ants at the end  

Ya, I was holding off on another S1 purchase but since it looks like at least a good month before Habanero availability I think I'll purchase 4 more Ants at the new BTC price they just posted.

I'm definitely not buying an S2 though since I'm enjoying what I see from the Habanero.  I'm hoping your pricing can be competitive with the Chinese offerings.

[CurcO]

The chip has been seen to go up to 1 Ghz  , however 1200 watts PSU is needed ... good luck

[MrTeal]

The chip has been seen to go up to 1 Ghz  , however 1200 watts PSU is needed ... good luck

Yeah, we're not quite there yet. Given some of the rumblings I'm hearing 1GHz might be a case by case basis, like hitting 5GHz on an i5-2500k. Some can get there, and some can't. Part of our profiling process is getting the exact mix right. For instance right now I'm running at 912MHz, and one die is marginally stable at my current frequency/voltage settings. The other dies can do better. Once FW and SW support gets to the point where we can decouple the dies and run them each with a separate frequency and voltage, we should be able to extract more from the chips.

[aerobatic]

The chip has been seen to go up to 1 Ghz  , however 1200 watts PSU is needed ... good luck

Yeah, we're not quite there yet. Given some of the rumblings I'm hearing 1GHz might be a case by case basis, like hitting 5GHz on an i5-2500k. Some can get there, and some can't. Part of our profiling process is getting the exact mix right. For instance right now I'm running at 912MHz, and one die is marginally stable at my current frequency/voltage settings. The other dies can do better. Once FW and SW support gets to the point where we can decouple the dies and run them each with a separate frequency and voltage, we should be able to extract more from the chips.

What hash rate does 912 Mhz clock equate to...?  650 GH/s?

[cedivad]

What hash rate does 912 Mhz clock equate to...?  650 GH/s?

if
875MHz = 640GH/s

then
912MHz = 667GH/s?

[aerobatic]

if
875MHz = 640GH/s  then  912MHz = 667GH/s?

how does that reconcile with the yoli boards on hashfast's web site and the sierra evo's that are claimed to hash at 750 GH/s (per board)... which would presumably be a clock speed of 1,025 Mhz ?

[cedivad]

how does that reconcile with the yoli boards on hashfast's web site and the sierra evo's that are claimed to hash at 750 GH/s (per board)... which would presumably be a clock speed of 1,025 Mhz ?

He knows it:

The chip has been seen to go up to 1 Ghz  , however 1200 watts PSU is needed ... good luck

Posted from Bitcointa.lk - #Q5AesHI8ihy45RgA

[MrTeal]

A 1200W PSU seems excessive, but you might want one for overhead. I haven't pushed the Habanero up to 750GH/s yet, but the one test hashing right now at
http://eligius.st/~wizkid057/newstats/userstats.php/1CAqwaYmNn6SZbk4G3e9cXxsJ3tJVfGuzt is pulling 54.1A at 11.92V, or 645W. That last 10% might really push the power usage if you need lots of voltage, but I'd be surprised if you couldn't get away with a good 1000W supply.
Edit: I just went and measured it again as that seemed a little low, and the number is 59.1A. Sorry for the confusion.

[qumqats]

Does it need to be a PC power supply?
Or will a simple 12V 100A supply work with the right cable/connectors?

ADVthaAnksNCE

[Abraxas601]

Does it need to be a PC power supply?
Or will a simple 12V 100A supply work with the right cable/connectors?

ADVthaAnksNCE

Shouldn't matter as long as the source can provide the required DC 12V and Current.  Source could be a generator, battery, solar cells, warp core...

[gateway]

Bit of a tiny update can be found here. https://peppermining.com/the-habanero/

Cheers..

[CurcO]

The Chip when run at 1ghz will consume 800 watts, yes you can get away running on a 1000w PSU however for continuous run time (24x7) you will be putting considerable amount of load on the 1000w PSU (80%) this will impact the efficiency, and will increment the risk of PSU burnout.  When you compare it to running on a 1200 PSU you will be at 66.7% load which gives you efficiency and stable current . Which IMO if allows the unit to be run 24x7 with little risk. 

A 1200W PSU seems excessive, but you might want one for overhead. I haven't pushed the Habanero up to 750GH/s yet, but the one test hashing right now at
http://eligius.st/~wizkid057/newstats/userstats.php/1CAqwaYmNn6SZbk4G3e9cXxsJ3tJVfGuzt is pulling 54.1A at 11.92V, or 645W. That last 10% might really push the power usage if you need lots of voltage, but I'd be surprised if you couldn't get away with a good 1000W supply.
Edit: I just went and measured it again as that seemed a little low, and the number is 59.1A. Sorry for the confusion.

[rob143]

One solution is to check out sidehacks load balancing server power supply control board. Server power supplies can run up to 90% and are dirt cheap in comparison to a desktop power supply.

[qumqats]

One solution is to check out sidehacks load balancing server power supply control board. Server power supplies can run up to 90% and are dirt cheap in comparison to a desktop power supply.

Interesting, but @750W 62A it seems to be just shy of having enough current/watts with a reasonable safety factor, unless you do a redundant 3 per 2 Habs.

[MrTeal]

Each die has an independent voltage controller and 12V input, so you can run whatever power supply configuration you want. Even if you have a server power supply with a higher 13V input voltage and an ATX one with a 12V output, you could run them together without contention.