FPGA development board "Lancelot" - accept bitsteam developer's orders.

[loshia]

Thank You!

[1714029729]

1714029729

[1714029729]

1714029729

[1714029729]

1714029729

[loshia]

Kano,

I have tried --icarus-timing 2:70. I was running it for two days. I have saved the results.

Today i changed the resistors on a couple of my boards. With 2:70 i see a lot of times two yellow leds (on ones with VCC core increased -changed resistors) - The sign on top of the leds is IDLE. When i run cgminer with icarus-timing long No yellow leds - these broads (changed resistors) are blinking like the rest

Can you comment that pls.
Thank you

I can compare the results after two days or so...

[kano]

Kano,

I have tried --icarus-timing 2:70. I was running it for two days. I have saved the results.

Today i changed the resistors on a couple of my boards. With 2:70 i see a lot of times two yellow leds (on ones with VCC core increased -changed resistors) - The sign on top of the leds is IDLE. When i run cgminer with icarus-timing long No yellow leds - these broads (changed resistors) are blinking like the rest

Can you comment that pls.
Thank you

I can compare the results after two days or so...

What does the API say for 'stats' about the ICA devices? (both for the ones running 'long' and the ones running '2:70')
(There are 16 Icarus specific stats in there)

You of course have to enable the API with --api-listen and can then request API 'stats' from the computer running cgminer at 127.0.0.1 ... or read the API-README

[loshia]

Kano,
Unfortunately i did not saved api-stats with timing 2:70. I am posting results with timing long. If they are not enough i can run some boards with 2:70 and some with long and post back the results tomorrow


Full nonce for moded boards with U ranging between 6.3-6.5 is:
4.413108
4.329242
4.273607
4.347153

HS:

0.000000000999920
0.000000000981032
0.000000000965784
0.000000000984132


Full nonce for rest is with U ranging between 5.5 -5.7 is:
4.989389
4.936110
4.984062
4.896753

HS:
0.000000001134453
0.000000001115143
0.000000001132101
0.000000001110573

Is that info enough?

Best

[kano]

That's suggesting they are running at around 1GH/s so I guess there must be a problem (since the U isn't 1GH/s more like ~460MH/s)
Maybe the new bitstream divides the work up differently?

All 16 values for a device would be more useful ...

Anyway - if those numbers were correct then it would be something like: 1:40

Also, the Icarus code defaults to 2 devices (which you can change with --icarus-options) that divide the nonce-range in half (however the division 'method' based on the number of devices is fixed) - so I'm not sure exactly what the bitstream is doing different to give those numbers you are getting.

... however ... if the bitstream is adjusting the clock ... then yes timing long/short would not give correct results.

long/short timing assumes the clock is fixed - if it isn't it wont work without knowing the clock for each result (and me rewriting the timing code to handle that)

Try 1:40 and see what happens ...

[loshia]

Thanks Kano!
I am trying it and i will post the results
Best

[loshia]

Kano,

Here are results with timing=long for 11 hours

Please note ICA0-ICA3 and ICA26-ICA29 are with changed resistors for sure. There are at lest 3-4 boards from the rest with changed resistors, but unfortunately i can not remember their numbers

Best

	0	GPU0	41672	971221	363.593085	1.400034	0.000000
	1	ICA0	41672	12336	2.510747	0.048282	0.000002	43	4.413108	10	0.000000000999920	0.118484	1600	4274231664	4605	0.008396	160	119	long	true	115200	2	2
	2	ICA1	41672	12473	2.961499	0.214484	0.000002	42	4.329242	10	0.000000000981032	0.115743	1600	4242677362	4541	0.008696	160	62	long	true	115200	2	2
	3	ICA2	41672	12649	5.315056	1.476899	0.000002	41	4.273607	10	0.000000000965784	0.125597	1600	4288070182	4707	0.009180	160	71	long	true	115200	2	2
	4	ICA3	41672	12569	3.502263	0.262402	0.000002	42	4.347153	10	0.000000000984132	0.120340	1600	4292662870	4606	0.008492	160	129	long	true	115200	2	2
	5	ICA4	41672	11004	3.566485	0.946676	0.000002	48	4.958089	10	0.000000001126098	0.121535	1600	4278299352	4115	0.007690	160	115	long	true	115200	2	2
	6	ICA5	41672	10893	3.635474	0.945021	0.000002	48	4.978525	10	0.000000001125178	0.145921	1600	4261545708	4112	0.007901	160	110	long	true	115200	2	2
	7	ICA6	41672	10946	3.711691	0.998559	0.000002	48	4.961865	10	0.000000001124789	0.130935	1600	4272433342	3975	0.007576	160	137	long	true	115200	2	2
	8	ICA7	41672	10928	2.661877	0.121210	0.000002	48	4.967016	10	0.000000001125969	0.131015	1600	4291956988	4096	0.007345	160	97	long	true	115200	2	2
	9	ICA8	41672	10917	2.841205	0.119703	0.000002	48	4.989389	10	0.000000001134453	0.116952	1600	4277281596	4072	0.008135	160	71	long	true	115200	2	2
	10	ICA9	41672	10971	5.718872	1.240576	0.000000	48	4.936110	10	0.000000001115143	0.146609	1600	4290746994	4003	0.007442	160	6	long	true	115200	2	2
	11	ICA10	41672	10752	4.543638	1.384864	0.000002	48	4.984062	10	0.000000001132101	0.121727	1600	4281984372	3965	0.008019	160	130	long	true	115200	2	2
	12	ICA11	41672	11059	2.743326	0.119330	0.000001	47	4.896753	10	0.000000001110573	0.126880	1600	4266670588	4025	0.007508	160	29	long	true	115200	2	2
	13	ICA12	41672	10615	4.528384	1.233007	0.000000	49	5.045424	10	0.000000001130485	0.190026	1600	4278466792	3931	0.007593	160	94	long	true	115200	2	2
	14	ICA13	41672	10748	4.613970	1.352324	0.000002	49	5.037712	10	0.000000001133836	0.167924	1600	4284039906	4016	0.007426	160	18	long	true	115200	2	2
	15	ICA14	41672	10900	2.733590	0.117377	0.000002	49	5.037367	10	0.000000001149880	0.098668	1600	4277803568	4073	0.007798	160	66	long	true	115200	2	2
	16	ICA15	41672	10953	2.718361	0.119965	0.000002	48	4.923011	10	0.000000001111331	0.149882	1600	4260984088	3991	0.007511	160	151	long	true	115200	2	2
	17	ICA16	41672	10950	2.537014	0.141895	0.000002	48	4.981943	10	0.000000001137081	0.098218	1600	4275359084	4045	0.008275	160	43	long	true	115200	2	2
	18	ICA17	41672	12350	3.726626	0.664459	0.000002	42	4.392341	10	0.000000000990362	0.138770	1600	4251259860	4497	0.008671	160	20	long	true	115200	2	2
	19	ICA18	41672	12250	2.975354	0.144306	0.000002	43	4.405313	10	0.000000000996196	0.126682	1600	4288208408	4492	0.008669	160	11	long	true	115200	2	2
	20	ICA19	41672	10650	2.721982	0.144693	0.000002	49	5.093104	10	0.000000001155094	0.132013	1600	4291332084	3830	0.007291	160	152	long	true	115200	2	2
	21	ICA20	41672	10913	2.605371	0.174266	0.000002	48	4.966502	10	0.000000001118462	0.162746	1600	4277245552	3974	0.007577	160	138	long	true	115200	2	2
	22	ICA21	41672	10764	2.726643	0.147668	0.000001	49	5.004606	10	0.000000001116213	0.210508	1600	4277395872	4042	0.007937	160	45	long	true	115200	2	2
	23	ICA22	41672	12375	2.626903	0.145837	0.000001	42	4.360790	10	0.000000000991422	0.102663	1600	4288569702	4461	0.008659	160	140	long	true	115200	2	2
	24	ICA23	41672	10710	3.086680	0.144996	0.000002	49	5.093802	10	0.000000001160190	0.110824	1600	4291315650	3985	0.007460	160	143	long	true	115200	2	2
	25	ICA24	41672	10922	2.751743	0.142671	0.000002	48	4.997611	10	0.000000001127309	0.155857	1600	4246874872	3966	0.007127	160	126	long	true	115200	2	2
	26	ICA25	41672	10864	3.041689	0.141176	0.000002	49	5.016164	10	0.000000001133526	0.147707	1600	4255465354	4046	0.008038	160	50	long	true	115200	2	2
	27	ICA26	41672	12312	3.044015	0.324907	0.000002	43	4.447452	10	0.000000001015595	0.085504	1600	4284001448	4519	0.008444	160	40	long	true	115200	2	2
	28	ICA27	41672	12472	3.749297	0.924476	0.000002	42	4.376254	10	0.000000000989404	0.126796	1600	4284850334	4452	0.008082	160	135	long	true	115200	2	2
	29	ICA28	41672	12406	2.959108	0.141556	0.000002	43	4.419099	10	0.000000000999569	0.125985	1600	4292420886	4649	0.009071	160	9	long	true	115200	2	2
	30	ICA29	41672	12646	3.233519	0.663727	0.000001	42	4.364028	10	0.000000000988383	0.118956	1600	4263208878	4791	0.008888	160	151	long	true	115200	2	2

[luffy]

i think i may have damaged the board while trying to solder the parallel resistors. now the Vin of every fpga
is stuck to 0.59v no matter how many times i try to solder the original smt registors back. the white leds are off,
all the other leds are on. i hope it is a matter of bad soldering and nothing else 

[loshia]

luffy,

Same happened to me also. Although original resistors look OK they were damaged. I suggest to find some 0603 Resistors and replace them. All will be fine. Another approach will be to unsloder them and cheek them with multimeter and  you will see that they are damaged (no resistance). Or you can measure the resistance between two ends of the resistors soldered on PCB. if they are damaged the resistance will be 10 KOhms if they are ok it should be about 5 Kohms

I was trying parallel soldering also. It turned out to be a bad idea because of easy "blowing out"
original ones. Better approach is just to replace them

[hardcore-fs]

DON'T try to continually solder resistors to your PCB.....

Get some VERY THIN FLEXIBLE linking wire, then solder that to the PCB, THEN do your testing by soldering the resistors to the ends of the wire.
That way you only solder your VALUABLE PCB once or twice, but you can solder and abuse your shitty connection wire multiple times.

[libertybuck]

DON'T try to continually solder resistors to your PCB.....

Get some VERY THIN FLEXIBLE linking wire, then solder that to the PCB, THEN do your testing by soldering the resistors to the ends of the wire.
That way you only solder your VALUABLE PCB once or twice, but you can solder and abuse your shitty connection wire multiple times.

Good idea.

[luffy]

thanks guys. i measure them around 3Kohm each. i wonder if i have damaged any neighbour resistor!

[loshia]

Probably you did. I can not be 100% sure but my core voltage was 0.5 like yours when they were 0 kohm each...being 3 kohm each it means that you should get very high VCC core (my oppinion) voltage or something else around is damaged which shall be the case

Try the wire thing as suggested. As stated PCB is easy to damage. If you happen to fix at least one of the cores you will able to measure correct core voltage

[fpf]

Dear N.G.Zhang,

Will the source code for that bitstream be published?
How does the FPGA determine the max. clock rate / hash rate?
Over the I2C temperature sensor that is available on the lancelot boards but not on the icarus boards?
And what does that mean in case of icarus boards.
What functions do the 4 dip switches have now? (Played a bit with them but it seems they lost their old functions and do some weird stuff now)

Regards

FPF

[hardcore-fs]

Probably you did. I can not be 100% sure but my core voltage was 0.5 like yours when they were 0 kohm each...being 3 kohm each it means that you should get very high VCC core (my oppinion) voltage or something else around is damaged which shall be the case

Try the wire thing as suggested. As stated PCB is easy to damage. If you happen to fix at least one of the cores you will able to measure correct core voltage

There are always better solutions (routing/VHDL/Verilog/partitioning optimizations)

Really no one should be playing with the core voltage, UNLESS they have enabled both in-chip/external temp  and external  current  measurement.

Problem is that such things that allow you to fiddle with the core voltage vary significantly from batch to batch & the age of the Chip.

You will find that current and internal die temp start to show signs that you are getting close to the danger zone, whilst you still have time to back them off.
Certainly making mods to a small group of FPGA then retroactively applying the 'fix' to masses of production out in the field, really is only going to end one way, plus such damage is really going to F*** any chances of returning the units for trade in.

The question should be:
Is the mod going to make more bit coins between now and trade in, than the trade in value of the device?

If not then leave it the F***  ALONE!!!

[burnin]

Dear N.G.Zhang,

Will the source code for that bitstream be published?
How does the FPGA determine the max. clock rate / hash rate?
Over the I2C temperature sensor that is available on the lancelot boards but not on the icarus boards?
And what does that mean in case of icarus boards.
What functions do the 4 dip switches have now? (Played a bit with them but it seems they lost their old functions and do some weird stuff now)

Regards

FPF

From what he said before i don't think so.
Would be awesome though.
The Source won't be the only key, i think he did a lot of manual routing stuff.

[blackarrow]

Hi!

We've tried v42 in -3C Xilinx FPGA and it is actually slower than the 200Mhz firmware.

N.G.Zhang, do you consider posting the code for the v42 firmware so we can fiddle with it?

Thank you.

[Dexter770221]

Hi!

We've tried v42 in -3C Xilinx FPGA and it is actually slower than the 200Mhz firmware.

N.G.Zhang, do you consider posting the code for the v42 firmware so we can fiddle with it?

Thank you.

+1!
There's always room for improvements

[induktor]

Hello
What happened to this project? is still alive? can I get dev-boards?, is some firmware already working with cgminer?

I just jumped into this and wanted to start playing with FPGA.
thanks

[wsoei]

Hello
What happened to this project? is still alive? can I get dev-boards?, is some firmware already working with cgminer?

I just jumped into this and wanted to start playing with FPGA.
thanks

you can buy them in this thread:
https://bitcointalk.org/index.php?topic=187549.0