DIY FPGA Mining rig for any algorithm with fast ROI - page 79.

DigitalCruncher

jr. member

Activity: 59

Merit: 1

Quote from: whitefire990 on May 06, 2018, 11:01:27 PM

Quote from: DigitalCruncher on May 06, 2018, 10:41:24 PM

Quote from: whitefire990 on May 06, 2018, 10:09:04 PM

Regarding the question about bitstream compatibility, the answer is no, a bitstream built for VCU1525 will not work on the XUPP3R. The clock & USB UART are on different pins, so I must 'build' two versions of the bitstream, one for each card. Although the change in the code is tiny, it still takes the tools 5+ hours to re-run place & route, with multiple runs needed to get a success.

Can you inplement a double set of interfaces to support both boards by the single firmware?
BUFGMUX_CTRL can be used to select the clock signal, if clocking assignment is different.
It is probably possible to build one firmware for both board.

I get the general idea but I still have a hard time seeing how to do that with the USB-UART pins.

As far as 'resale' of these FPGA cards, there is also a tremendous future for AI mining. In fact, if tens of thousands of people were mining crypto with these boards, and an AI mining server became available, they are so powerful that you might end up with a 'Skynet' type superintelligence, or the Kurzweil 'singularity' event.

My point is that if, for some reason, mining crypto became unprofitable, someone could re-sale access to the FPGA hardware for AI, weather prediction or any other compute-heavy tasks.

Suppose you have 2 different sets of pins.
RXD_BITTWARE and TXD_BITTWARE is one set, RXD_XILINX and TXD_XILINX is another set.
The uart module inside FPGA has RXD input and TXD output also.
Then make connections as follows:
TXD_BITTWARE = TXD
TXD_XILINX = TXD

RXD = RXD_BITTWARE & RXD_XILINX
The merging function varies between "or, xor, and" and it is depending on internal or external pulling unconnected pins up or down.

senseless

hero member

Activity: 1118

Merit: 541

Quote from: senseless on May 06, 2018, 06:18:03 PM

Quote from: whitefire990 on May 06, 2018, 04:33:24 PM

Both the VCU1525 and XUPP3R have a circuit that will shut down the FPGA power supply if the temperature exceeds limits. This doesn't prevent you from trying to pull too much current for a short time though.

Another important note for people considering this hardware. The VCU1525 has a fixed core voltage of 0.85V. The XUPP3R from Bittware is more flexible and you can change the core voltage from 0.74V to 1.11V, which allows a much greater flexibility in terms of over-clocking the FPGA. For memory intensive algorithms, a higher voltage will produce faster hash rates, vs. for power limited algorithms, a lower voltage would likely produce better hash rates.

The VCU1525 is also only 160A vccint... Increasing the voltage lowers amperage and allows you to use more power... But, I'm not sure the amperage of supply capacity the bittware board has... Happen to have a product manual on you? If the board is only designed for 100A, wouldn't do you much good even operating at 1V. I've got a couple VCU118's here that I picked up long before the VCU1525 was announced or released. Those are limited to like 80A vccint Cry

and they're damn near twice the price.

Went and tried to find the max amp on the XUPP3R and no data was available. Best I've been able to figure out is that you're limited to 75W on the 6 PIN aux and 75W on the PCI-E. This would be less than the VCU1525 which has 8 pin (150W) plus PCI-E (75W). Even if you're running at 1.1V, the max amps you could pull for the XUPP3R is 136A. Do NOT buy the XUPP3R for mining. Even with variable core voltage it's power supply is shite compared to the VCU1525. Even the VCU1525 at 160A, is quite limited in the maximum amount of 'things' you can have on in the FPGA at one time. You'll hit over 150A 0.85v using only 70% of the logic with all memories powered down.

whitefire990

copper member

Activity: 166

Merit: 84

Quote from: DigitalCruncher on May 06, 2018, 10:41:24 PM

Quote from: whitefire990 on May 06, 2018, 10:09:04 PM

Regarding the question about bitstream compatibility, the answer is no, a bitstream built for VCU1525 will not work on the XUPP3R. The clock & USB UART are on different pins, so I must 'build' two versions of the bitstream, one for each card. Although the change in the code is tiny, it still takes the tools 5+ hours to re-run place & route, with multiple runs needed to get a success.

Can you inplement a double set of interfaces to support both boards by the single firmware?
BUFGMUX_CTRL can be used to select the clock signal, if clocking assignment is different.
It is probably possible to build one firmware for both board.

I get the general idea but I still have a hard time seeing how to do that with the USB-UART pins.

As far as 'resale' of these FPGA cards, there is also a tremendous future for AI mining. In fact, if tens of thousands of people were mining crypto with these boards, and an AI mining server became available, they are so powerful that you might end up with a 'Skynet' type superintelligence, or the Kurzweil 'singularity' event.

My point is that if, for some reason, mining crypto became unprofitable, someone could re-sale access to the FPGA hardware for AI, weather prediction or any other compute-heavy tasks.

DigitalCruncher

jr. member

Activity: 59

Merit: 1

Quote from: whitefire990 on May 06, 2018, 10:09:04 PM

Regarding the question about bitstream compatibility, the answer is no, a bitstream built for VCU1525 will not work on the XUPP3R. The clock & USB UART are on different pins, so I must 'build' two versions of the bitstream, one for each card. Although the change in the code is tiny, it still takes the tools 5+ hours to re-run place & route, with multiple runs needed to get a success.

Can you inplement a double set of interfaces to support both boards by the single firmware?
BUFGMUX_CTRL can be used to select the clock signal, if clocking assignment is different.
It is probably possible to build one firmware for both board.

arielbit

legendary

Activity: 3444

Merit: 1061

great! let the FPGA manufacturers take a nice slice of the pie...the more the merrier Cheesy

so where can i buy?...looking forward in making a rig out of these..

you make the firmware for different algos that can be loaded to FPGA..and we mine with a fee..

anything i missed?

...

do people who know c language and assembly language, who have experience in programming micro-controllers like zilog, atmel and pic ....can easily adapt to using this? like doing the stuff that you are doing? create firmware for this FPGAs

the lack of developers in this arena and the lesser utility/other uses of FPGA (in general population) makes me think twice in investing more in this kind of stuff...unless i knew a guy in person that i can work with, not just some random guy in the internet..

whitefire990

copper member

Activity: 166

Merit: 84

Quote from: ZaebaloRegitsa4 on May 06, 2018, 09:21:29 PM

VCU1525 card is the Development Kit! This is not mass produced. And it can not ramp up production.

I had a conference call with Xilinx before announcing this project and they agreed to transition the VCU1525 into a production product for a slightly higher price; the production version will be released in June/July, in much higher quantities.

Regarding the question about bitstream compatibility, the answer is no, a bitstream built for VCU1525 will not work on the XUPP3R. The clock & USB UART are on different pins, so I must 'build' two versions of the bitstream, one for each card. Although the change in the code is tiny, it still takes the tools 5+ hours to re-run place & route, with multiple runs needed to get a success.

ZaebaloRegitsa4

jr. member

Activity: 64

Merit: 1

VCU1525 card is the Development Kit! This is not mass produced. And it can not ramp up production.

jasemoney

legendary

Activity: 1610

Merit: 1008

Forget-about-it

Hey nice writeup, really interested.
Wondering if Bittware or Avnet taking bitcoin for purchases?

Also if you make a bitstream for the VCU1525 is it going to be working for the XUPP3R-VU9P?

I like to do spec mining on many coins, so I'd like to have the biggest option set available (within reason).

what speed/type DDR4 in these fpga?

have you tried porting sha244 into an asic built for sha256?

HardwareCollector

member

Activity: 144

Merit: 10

If you are data center guy such as myself, then you may prefer these server cards.

Lead time is 2 weeks:
https://www.xilinx.com/products/boards-and-kits/vcu1525-p.html#hardware

I do not trust PCIe risers with $4K hardware, there are some brave souls out here.

Reggie0

member

Activity: 107

Merit: 13

Quote from: 2112 on May 06, 2018, 04:57:59 PM

Quote from: Reggie0 on May 06, 2018, 04:22:52 PM

Quote from: GPUHoarder on May 06, 2018, 10:21:25 AM

Quote from: DigitalCruncher on May 06, 2018, 10:11:59 AM

Turn on the overtemperature shutdown in the bitstream, and control the temperature in the mining software. It is easy.

“Turn on overtemp shutdown in the bitstream”

AFAIK there’s no such magic button. If you want temp monitoring and thermal limits with an FPGA you have to include them in your design logic.

False. The better fpgas(like virtex 5/6, complete series 7 and newer xilinx) has pin terminals to an separated internal die temperature sensor transistor. A properly designed board should use it for temperature measurement and shutdown. It works even when FPGA isn't configured. An alternative is using jtag to read XADC (7 series and newer).

I actually went ahead and RTFM'd the "UltraScale Architecture SYStem MONitor User's Guide". It looks like it is very close to being a "magic button", it takes only one line of code:

Code:

set_property BITSTREAM.CONFIG.OVERTEMPSHUTDOWN ENABLE

On the other hand the whole guide to the SYSMON block has 113 pages, so it must be also relatively easy to accidentally mis-configure it.

You are going too far to look up some useful information, but it is on the top. Read the "VCU1525 Acceleration Platform User Guide", on page 47. It says fan controller IC has a die temperature sensor connected to FPGA-s DXP/DXN pin, and has an TCRIT#output which connected to the supervisory IC and it will shut down the FPGA, when overheated. It is independent from the FPGA and FPGA's SYSMON.

It looks like the Xilinx's card is a properly designed FPGA board

senseless

hero member

Activity: 1118

Merit: 541

Quote from: whitefire990 on May 06, 2018, 04:33:24 PM

Both the VCU1525 and XUPP3R have a circuit that will shut down the FPGA power supply if the temperature exceeds limits. This doesn't prevent you from trying to pull too much current for a short time though.

Another important note for people considering this hardware. The VCU1525 has a fixed core voltage of 0.85V. The XUPP3R from Bittware is more flexible and you can change the core voltage from 0.74V to 1.11V, which allows a much greater flexibility in terms of over-clocking the FPGA. For memory intensive algorithms, a higher voltage will produce faster hash rates, vs. for power limited algorithms, a lower voltage would likely produce better hash rates.

The VCU1525 is also only 160A vccint... Increasing the voltage lowers amperage and allows you to use more power... But, I'm not sure the amperage of supply capacity the bittware board has... Happen to have a product manual on you? If the board is only designed for 100A, wouldn't do you much good even operating at 1V. I've got a couple VCU118's here that I picked up long before the VCU1525 was announced or released. Those are limited to like 80A vccint Cry

and they're damn near twice the price.

GPUHoarder

member

Activity: 154

Merit: 37

Quote from: 2112 on May 06, 2018, 04:57:59 PM

Quote from: Reggie0 on May 06, 2018, 04:22:52 PM

Quote from: GPUHoarder on May 06, 2018, 10:21:25 AM

Quote from: DigitalCruncher on May 06, 2018, 10:11:59 AM

Turn on the overtemperature shutdown in the bitstream, and control the temperature in the mining software. It is easy.

“Turn on overtemp shutdown in the bitstream”

AFAIK there’s no such magic button. If you want temp monitoring and thermal limits with an FPGA you have to include them in your design logic.

False. The better fpgas(like virtex 5/6, complete series 7 and newer xilinx) has pin terminals to an separated internal die temperature sensor transistor. A properly designed board should use it for temperature measurement and shutdown. It works even when FPGA isn't configured. An alternative is using jtag to read XADC (7 series and newer).

I actually went ahead and RTFM'd the "UltraScale Architecture SYStem MONitor User's Guide". It looks like it is very close to being a "magic button", it takes only one line of code:

Code:

set_property BITSTREAM.CONFIG.OVERTEMPSHUTDOWN ENABLE

On the other hand the whole guide to the SYSMON block has 113 pages, so it must be also relatively easy to accidentally mis-configure it.

I stand corrected, I was not aware of the existence of that flag! I knew you could read the XADC, but did not realize there was a bitstream flag. Good to know, although I don’t know if I would rely on that as your only protection.

I appreciate the wealth of knowledge and skill coming out on this forum regarding the FPGA work.

Regarding workstation GPUs, I said they could be useful for mining, and can work alongside FPGAs given the direct access to their memory, not that they were better than normal GPUs for mining or that they were like FPGAs. You can however very often find large batches of the older generation passive cards for cheap as most people don’t have the ability to use them. I last paid $130/ea about a month ago for a bunch of older FirePro workstation cards that are now hashing at 24MH ea.

2112

legendary

Activity: 2128

Merit: 1073

Quote from: Reggie0 on May 06, 2018, 04:22:52 PM

Quote from: GPUHoarder on May 06, 2018, 10:21:25 AM

Quote from: DigitalCruncher on May 06, 2018, 10:11:59 AM

Turn on the overtemperature shutdown in the bitstream, and control the temperature in the mining software. It is easy.

“Turn on overtemp shutdown in the bitstream”

AFAIK there’s no such magic button. If you want temp monitoring and thermal limits with an FPGA you have to include them in your design logic.

False. The better fpgas(like virtex 5/6, complete series 7 and newer xilinx) has pin terminals to an separated internal die temperature sensor transistor. A properly designed board should use it for temperature measurement and shutdown. It works even when FPGA isn't configured. An alternative is using jtag to read XADC (7 series and newer).

I actually went ahead and RTFM'd the "UltraScale Architecture SYStem MONitor User's Guide". It looks like it is very close to being a "magic button", it takes only one line of code:

Code:

set_property BITSTREAM.CONFIG.OVERTEMPSHUTDOWN ENABLE

On the other hand the whole guide to the SYSMON block has 113 pages, so it must be also relatively easy to accidentally mis-configure it.

whitefire990

copper member

Activity: 166

Merit: 84

Both the VCU1525 and XUPP3R have a circuit that will shut down the FPGA power supply if the temperature exceeds limits. This doesn't prevent you from trying to pull too much current for a short time though.

Another important note for people considering this hardware. The VCU1525 has a fixed core voltage of 0.85V. The XUPP3R from Bittware is more flexible and you can change the core voltage from 0.74V to 1.11V, which allows a much greater flexibility in terms of over-clocking the FPGA. For memory intensive algorithms, a higher voltage will produce faster hash rates, vs. for power limited algorithms, a lower voltage would likely produce better hash rates.

Reggie0

member

Activity: 107

Merit: 13

Quote from: GPUHoarder on May 06, 2018, 10:21:25 AM

Quote from: DigitalCruncher on May 06, 2018, 10:11:59 AM

From my practice there are simple measures to avoid failures.
Do not place these boards in the closed case, setup the fan to cool the bottom layer of PCB.
Do not allow the local overheat of any component like DC/DC or external power connector.
Turn on the overtemperature shutdown in the bitstream, and control the temperature in the mining software. It is easy.
Then the FPGA board will work for years.

“Turn on overtemp shutdown in the bitstream”

AFAIK there’s no such magic button. If you want temp monitoring and thermal limits with an FPGA you have to include them in your design logic. Same as an ASIC. Otherwise you very much can exceed junction temps and damage the hardware if you have enough power going into the board to begin with.

This is also possible on some GPUs with poor drivers. A few of my old Titan Blacks from HPC work had power stages for memory that would overheat if the memory was trashed too much for too long, and the mosfet would slip down on the boards till they caused a short. Actually caught a entire server on fire that way once...

False. The better fpgas(like virtex 5/6, complete series 7 and newer xilinx) has pin terminals to an separated internal die temperature sensor transistor. A properly designed board should use it for temperature measurement and shutdown. It works even when FPGA isn't configured. An alternative is using jtag to read XADC (7 series and newer).

Xeonus

newbie

Activity: 52

Merit: 0

This thread is super interesting to read. As others have stated, it is interesting to see these ASIC, FPGA, GPU cycles and discussions happen over the years.

Are now modern FPGAs (and the ones the OP uses) more suitable than the ones that were used in the previous attempts? Are FPGAs now in a sense more flexible in terms of mining so it makes sense to focus development on those?

gameboy366

jr. member

Activity: 252

Merit: 8

Quote from: HardwareCollector on May 06, 2018, 01:11:12 PM

Quote from: gameboy366 on May 06, 2018, 01:00:24 PM

Yo ! Can Workstation gpus outperform gaming gpus in mining. One guy here said that they are like fpga (or I didn't understand correctly). Can they be modified and made into a better mining machine ?

No, a Quadro P6000 ($4,999) does not out perform a GTX 1080 Ti ($699) for mining, they are basically the same cards compute-wise.

But, but this guy said something like this :

Quote from: GPUHoarder on May 05, 2018, 02:39:39 PM

Quote from: gameboy366 on May 05, 2018, 02:29:09 PM

Can workstation cards become useful for mining ? Damn ! They're gonna teach us FPGA next year in college.

You mean the FirePro type GPUs? Definitely. I have many many older FirePro cards that had plenty of memory bandwidth but underpowered (or over powered depending how you look at it) cores compared to the current gen. This cards got large page table support and some other niceties because of who the customers were. The also support an FPGA on the PCIe bus directly writing into / readying from their GPU memory (albeit at PCIe speeds). This works great with something like ETH, because you can do all the compute heavy Keccak off GPU but still use all the memory bandwidth.

HardwareCollector

member

Activity: 144

Merit: 10

Quote from: gameboy366 on May 06, 2018, 01:00:24 PM

Yo ! Can Workstation gpus outperform gaming gpus in mining. One guy here said that they are like fpga (or I didn't understand correctly). Can they be modified and made into a better mining machine ?

No, a Quadro P6000 ($4,999) does not out perform a GTX 1080 Ti ($699) for mining, they are basically the same cards compute-wise.

HardwareCollector

member

Activity: 144

Merit: 10

Quote from: mrb on May 06, 2018, 12:52:44 PM

It would be nice to have a complete database documenting all PoW algos (including pseudo-code). If someone feels motivated to do it, I'd be glad to sponsor them.

Can you please elaborate on the sponsorship. I understand the equihash algorithm very well, by the way.

gameboy366

jr. member

Activity: 252

Merit: 8

Yo ! Can Workstation gpus outperform gaming gpus in mining. One guy here said that they are like fpga (or I didn't understand correctly). Can they be modified and made into a better mining machine ?

Topic: DIY FPGA Mining rig for any algorithm with fast ROI - page 79. (Read 99472 times)