X6500 Custom FPGA Miner

[arklan]

pondering picking up some of these. what kind of time frame would i be looking at for an order of ten or so?

For an order of 10 boards, we could ship the next business day. As I said above, heatsink availability may be a problem, though. Please email sales@fpgamining.com to discuss bulk prices!

good to know. thanks.

[1714008675]

1714008675

[1714008675]

1714008675

[1714008675]

1714008675

[nbtcminer]

@TheSeven/Fizzisist:

Question about power: Would using a 12v 5a 2.1mm power adapter with a splitter be alright to power two x6500?

I keep wanting to come up with things to do with my x6500 XD!!!

[fizzisist]

@TheSeven/Fizzisist:

Question about power: Would using a 12v 5a 2.1mm power adapter with a splitter be alright to power two x6500?

I keep wanting to come up with things to do with my x6500 XD!!!

Sounds perfectly fine to me. 2A per board is generally more than enough, unless you're running some crazy bitstream or powering special fans.

[TheSeven]

@TheSeven/Fizzisist:

Question about power: Would using a 12v 5a 2.1mm power adapter with a splitter be alright to power two x6500?

I keep wanting to come up with things to do with my x6500 XD!!!

Assuming the PSU can really delive 5A (and I've seen quite a lot of them which break their promises), even three boards should work with the current bitstreams.

[coretechs]

I have an unopened set of 8 heatsinks/fans for these if anyone needs them.  I used different heatsinks on mine for passive cooling.

[bulanula]

I have an unopened set of 8 heatsinks/fans for these if anyone needs them.  I used different heatsinks on mine for passive cooling.

Can you please post a pic !

[coretechs]

Well, I guess technically it's not "passive cooling", they just happen to sit directly in front of the air conditioner vent...  

I'm using the same Enzotech copper heatsinks that were used in the mineral oil submerged rig a few posts back in this thread.  Temps are usually ~25-35C depending on how hot it is in the room.

[fizzisist]

Well, I guess technically it's not "passive cooling", they just happen to sit directly in front of the air conditioner vent...  

I'm using the same Enzotech copper heatsinks that were used in the mineral oil submerged rig that someone posted the other day.

Nice, coretechs! Another option for passive would be these heatsinks: http://www.newegg.com/Product/Product.aspx?Item=N82E16835999031

I haven't tried those myself, but mike2kt said they worked great for him.

[thirdlight]

What's your take on this development?

Does the x6500 have enough power to do it justice?

Will you be submitting the details to get an x6500 bitstream?

Thanks,

[nbtcminer]

What's your take on this development?

Does the x6500 have enough power to do it justice?

Will you be submitting the details to get an x6500 bitstream?

Thanks,

Few notes on this development (for when they make the Bitstream available):

1.) Anyone using the 12v 2a 5.5x2.1mm wall worts will have to change over to either a brick power supply (i.e 12V 5a or higher) or a 200W PSU or higher.

2.) Rev 2.0 units with their stock heatsinks may not be good enough to keep the X6500 cool. An active cooling solution or heatsink replacement (i.e Logisys or Zalman "Flower" will be needed for sure. Rev 3.0 may fair a lot better since they have push-pin mountable heatsinks and active cooling already included in the solution.

3.) Note that 5% of your total hashing power will go towards paying Eldentyrell for his work (on 540 mh/s on a X6500, you'd be paying around 27mh/s).

4.) Power regulator should be ok, but I remember Fizzisist mentioning something about swapping resistors to get more power to the spartan chips. Not sure if this is needed or not for the new bitstream.


This development definitely shakes the FPGA /  ASIC market up

[shad]

i guess nobody is driving his rev2 board without active cooling since 200mhz/mhs firmware

How much is the commission?

The commission is 20% of the difference in hashrate between the TML and the leading open-source implementation. The current TML achieves 270MH/s, and the leading open-source implementation achieves 210MH/s; this means that you must devote 12MH/s worth of hashpower to commission jobs in order to keep your hardware busy. The remaining post-commission hashrate at your disposal is 258MH/s.

so you pay 24mh/s with an X6500, but you get extra 116mh/s for yourself

the question is more will it be possible to run that firmware with MPBM

some infos of supporting that firmeware would be nice

[TheSeven]

1.) Anyone using the 12v 2a 5.5x2.1mm wall worts will have to change over to either a brick power supply (i.e 12V 5a or higher) or a 200W PSU or higher.

Those numbers are simply bullshit. 12V 3A will be sufficient for sure, even 2A might be sufficient with some luck. And that PSU wattage is also way off, and highly depends on the number of boards you want to power from it.

2.) Rev 2.0 units with their stock heatsinks may not be good enough to keep the X6500 cool. An active cooling solution or heatsink replacement (i.e Logisys or Zalman "Flower" will be needed for sure. Rev 3.0 may fair a lot better since they have push-pin mountable heatsinks and active cooling already included in the solution.

For Rev3, this all depends on your fan setup. Rev3 stock heatsinks should handle it well.

4.) Power regulator should be ok, but I remember Fizzisist mentioning something about swapping resistors to get more power to the spartan chips. Not sure if this is needed or not for the new bitstream.

The voltage regulators are actually going to be the bottleneck. I'm fairly certain that they can handle it, but they might require some airflow to prevent overheating. And swapping the resistors on a Rev2 board is definitely going to increase the achievable hashrate, at the expense of power consumption.

[TheSeven]

i guess nobody is driving his rev2 board without active cooling since 200mhz/mhs firmware

How much is the commission?

The commission is 20% of the difference in hashrate between the TML and the leading open-source implementation. The current TML achieves 270MH/s, and the leading open-source implementation achieves 210MH/s; this means that you must devote 12MH/s worth of hashpower to commission jobs in order to keep your hardware busy. The remaining post-commission hashrate at your disposal is 258MH/s.

so you pay 24mh/s with an X6500, but you get extra 116mh/s for yourself

the question is more will it be possible to run that firmware with MPBM

some infos of supporting that firmeware would be nice

I'm very willing to support this in MPBM, but it doesn't look like E.T. is. At least he didn't release any information about the protocol, thereby forcing people to stick with his bloaty Java miner thing for now. But I'm tempted to bet that it will be reverse engineered from the class files if he doesn't provide the source code rather soon, once this software is actually available.

Oh, and on another side note: This is not going to work well with p2pool. There will be two effects: High stales, and a higher-than-normal commission if you run this on a pool with a high "heart rate".

[nbtcminer]

Those numbers are simply bullshit. 12V 3A will be sufficient for sure, even 2A might be sufficient with some luck. And that PSU wattage is also way off, and highly depends on the number of boards you want to power from it.

I apologize for sharing the wrong numbers. I read that the firmwre would demand 48 watts or more, which works out to 12V 2.886a. In anycase, I'm just trying to err on the side of error / shitty power brick power supplies, but I don't know as much about power draw as an engineer as yourself.

[fizzisist]

Those numbers are simply bullshit. 12V 3A will be sufficient for sure, even 2A might be sufficient with some luck. And that PSU wattage is also way off, and highly depends on the number of boards you want to power from it.

I apologize for sharing the wrong numbers. I read that the firmwre would demand 48 watts or more, which works out to 12V 2.886a. In anycase, I'm just trying to err on the side of error / shitty power brick power supplies, but I don't know as much about power draw as an engineer as yourself.

I don't think TheSeven meant that to sound so harsh! We don't have any firm numbers from eldentyrell on the power consumption, but the most the regulators on the X6500 can supply is about 12W to each FPGA. Taking into account inefficiencies, 12V*3A=36W seems like a good power supply limit to aim for, and is definitely higher than the X6500 will ever draw.

[Qoheleth]

Does the overclocker firmware clock up to 200MHz yet? If not, has there been any work recently towards that?

[bonks]

Does the overclocker firmware clock up to 200MHz yet? If not, has there been any work recently towards that?

I'm using the revision 4 overclocker on http://fpgamining.com/documentation/firmware and seeing 182-210 MH/s with an average of 199.

When you load the firmware, set the initial clock to 200 and max to 250.

[Pipesnake]

Does the overclocker firmware clock up to 200MHz yet? If not, has there been any work recently towards that?

I'm using the revision 4 overclocker on http://fpgamining.com/documentation/firmware and seeing 182-210 MH/s with an average of 199.

When you load the firmware, set the initial clock to 200 and max to 250.

+1

[tgmarks]

Just curious, why would you bother with the overclocker firmware when the 200mhz firmware is rock solid and you don't gain anything on the overclocker?

[coretechs]

Is there a ztexmerge_190mhz firmware available?