Topic: Algorithmically placed FPGA miner: 255MH/s/chip, supports all known boards - page 28. (Read 119440 times)

Intriguing.  But how am I supposed to tell this apart from a chip damaged by failure to bother installing a heatsink and fan?  Unless I can do that, I'll simply end up taking over your warranty liabilities...

I think you missed the part of my post where I showed that it takes 26 Amps to get the junction up to 125C with a heatsink+fan.

No.  There is an option to quit the miner if it is unable to contact the signcryption server.  So you launch it from a three-line shell script:


#!/bin/bash
run-tml-miner
run-old-miner


problem solved.


I hereby open-source the above three-line shell script.

On ZTEX boards, the FPGA's JTAG signals are not even connected to the Cypress FX2 microcontroller.

If you're sending commands to urjtag over a pipe, don't forget to disable urjtag's command history. Otherwise it seems to store every single command sent in RAM for the duration of the session, so its memory use slowly bloats.

Also, I seem to recall there's code out there to emulate the Altera USB Blaster JTAG adapter using a Cypress FX2 like the one on the Ztex board, e.g. http://fpga4u.epfl.ch/wiki/FX2. No idea how practical it'd be to port it though.

Are the Xilinx-brand JTAG cables supported? None? All? Some? If the latter: Which ones?

I have no experience with urjtag, even though (or maybe because) I sold JTAG hardware myself about a decade ago.

Necessary patch to fix urjtag omission: http://article.gmane.org/gmane.comp.embedded.jtag.urjtag.devel/1288

Just FYI, and especially for Dr. Tyrell, the X6500 and ModMiner are fully compatible in terms of bitstreams. Only the host-side protocols are different.

Oh mighty mighty Tricone Master.
May I humbly request you talking to enterpoint to get the boards specs of their cairnsmore quad fpga board to run your firmware.

Would be awesome!!!

The guy who sells it announced that already

I'd like to know when this bitstream will be available for the ModMiner - http://www.btcfpga.com/

What about Icarus boards - any chance we'll see a version of the bitstream for them?

That quote's from a thread about a Virtex-5 device. As I recall those have metal heatspreaders. So a different chip, different packaging, and built on a different process (65nm rather than 45nm).

My concern is not that your bitstream will kill the board immediately. My concern is the reliability over the warranty period of 2 years.  I do not want to get trouble because customers want me to replace 5% to 10% of their boards after two years. This assumption is IMHO not unrealistic at permanent operation at 80° to 90°C junction temperature.

Of course, every one who accepts that risk can use your bitstream.

The second issue I pointed out is the lack of direct or indirect core temperature measurement. There is no way to detect improperly installed heat sinks or fan failure. The algorithm of the BTCMiner which stops the FPGA if the error rate increases would not work with your bitstream because the frequency has to be limited. (And so far this feature saved several FPGA's). An external temperature sensor would not solve this problem because it can only detect fan failure

If you think my concerns are unfounded you can offer to replace/refund FPGA's that fail during usage of your software.


Why they use this odd plastic packages if these FPGA's are also intended for extreme applications (like bitcoin mining)? At 12W the temperature difference between die and heat sink would be about 26°C in CSG484 packages (I use) and 44°C in FGG484 packages (used by other board vendors). IMHO this is not optimal.

Well ... with GPU's there is an advantage if they die in less than 12 months ...

You mine **% faster only to kill your equipment **% sooner?

Same goes for GPUs. AMD says their chips will happily run at 100c+, but me and everyone else has been using 85c as the maximum safe 24/7 mining temp.

That's the Absolute Maximum Rating. If you look at the footnote it warns: "Exposure to Absolute Maximum Ratings conditions for extended periods of time might affect device reliability." So they guarantee it won't die immediately, but not that it won't eventually fail in a few weeks or months as a result of running it out of spec which is what ztex is worried about.

I don't understand why people are getting their panties in a bunch about this.

You have the choice to either USE it or NOT use it.

Considering this improvement is going to increase mining income for everyone who chooses to use it, why can't the developer also increase HIS income ?

As an aside....
Pirate@40 currently generates approx 10% income directly off the backs of everyone who CHOOSES to mine at GPUMAX (when mining shares are sold and public work becomes available, that is).
Using your (skewed at best) reasoning, shouldn't you be praying for HIS downfall ? or perhaps all of the miners that are currently making 20% or greater profits mining, while CHOOSING to use GPUMAX ?
Your 'free software' soap-box spiel applies far more to something such as GPUMAX than it does to the licensing/income model that will be implimented for the OPTIONAL USE of this Bitstream.

Step 1.
CHOOSE

Step 2.
STFU and collect your mining rewards.

Also, as far as a per-FPGA licensing model, for direct sale to end users ?
You can be sure that everyone who decides to purchase (if this retarded pricing model WAS available) would suddenly find themselves only running a FRACTION of the hardware they were prior to purchasing....LMAO
Suddenly, EVERYONE only has but a SINGLE FPGA to purchase the Bitstream license for....LOL

This licensing model is FAIR. As far as another dependency or point of failure (LMAO) The Developer has EVERY incentive to keep his side of the 'operation' running, otherwise, he makes NOTHING.

EVERYONE is here to increase their financial holdings in 1 way or another, so calling the Developer of this bitstream GREEDY is nothing more than the Pot calling the Kettle...well, you know.

Again, CHOICE is a great thing. FTW !

Anyone who cries about this is clearly retarded.