EWBF's CUDA Zcash miner

[scryptr]

"SOME THEREAD IS DEAD RESTART" --

I received the above message in the console window after 2-3 hours.  The mispelling of "thread" was quoted from the message; the miner hung and required a restart.  Since there was no timestamp on the console window, I don't know how long the miner was idle.

I went back to mining with EQM.    The miner outperforms NiceHash EQM, but I notice that when I mine CryptoNight on the CPU, and Equihash with this new miner, this new miner gets 220-230Sols/s rather than the 210-220Sols/s with the EQM miner (while also mining CryptoNight on the CPU).  However, if I do not mine CryptoNight on the CPU, the new ZEC miner gets 260Sols/s mining ZEC.  The GPU involved is a GTX 980ti.

Still a few bugs in v1.0.2b.  Hope they get bashed!       --scryptr

[1714207557]

1714207557

[1714207557]

1714207557

[zunomiljan]

At this moment it is better to mine ETH, but i wish to support your work, so i will mine with 2-rigs 1070 for couple days.
https://i.imgur.com/ZOb8UaD.png

[laik2]

"SOME THEREAD IS DEAD RESTART" --

I received the above message in the console window after 2-3 hours.  The mispelling of "thread" was quoted from the message; the miner hung and required a restart.  Since there was no timestamp on the console window, I don't know how long the miner was idle.

I went back to mining with EQM.    The miner outperforms NiceHash EQM, but I notice that when I mine CryptoNight on the CPU, and Equihash with this new miner, this new miner gets 220-230Sols/s rather than the 210-220Sols/s with the EQM miner (while also mining CryptoNight on the CPU).  However, if I do not mine CryptoNight on the CPU, the new ZEC miner gets 260Sols/s mining ZEC.  The GPU involved is a GTX 980ti.

Still a few bugs in v1.0.2b.  Hope they get bashed!       --scryptr

The CPU is probably used to solve the final round. The author is not responding on my Linux query so assume sporadic support.

[antantti]

P0 state enabled?

Have you enabled power state 0 (P0)? If your memory speed is 3005 you are in P2 state which is slower in some memory hard algos.

P0 State on GTX980 gave me 210~215 Sol/s 1404 gpu / 3814 mem 74C on W8 (W10 had me some hash trouble on miner). using 376.09 nvidia driver

NOTE: up to 215~218 on v0.02... Nice work!!!!

Ok, try to limit your tdp + overclock your memory to 4000 or higher, you will find your sweetspot.

[topgeek]

1080 - 319 sols
1050ti - 114 sols

Thanks EWBF!

[Amph]

EWBF behave totaly different
ewbf use a lot GPU power, depending on card near 90% tdp, Memory controller load is ~60%, memory use ~1.5Gb
nicehash use same 10-15% less power, memory load in 2x exe go up to 71%, memory use ~250mb per instance

I presume both have one or other parts done better, so in time i bet we can have more then 300sol (maybe even 400 with OC as 1070 is a lot bettern then AMD in blace algo calculations) on default 1070 or even more, at least more then AMD HW have atm

i managed to do 250 sol at 100 watt like nicehash, but ncicehash was only 220 at the same wattage, so this is actually coded better

Your setting ? .

120 core 55 tdp 400 mem

on pool side the earning is not right, are e sure the dev fee is 2% only?

I see hashrate in miner 1:1 on suprnova

yeah the hash is right but not the final earning

On Nanopool the hashrate and profit is also in line with the miner for me - using 6 hour averages. This is great work! Thanks! EWBF!

what's your exact result, mine is 1500 sol only 0.017 zcash(0.0155 + 0.0015 in round earnings) in 3.5 hours...should have been 0.02 already

[ocminer]

EWBF behave totaly different
ewbf use a lot GPU power, depending on card near 90% tdp, Memory controller load is ~60%, memory use ~1.5Gb
nicehash use same 10-15% less power, memory load in 2x exe go up to 71%, memory use ~250mb per instance

I presume both have one or other parts done better, so in time i bet we can have more then 300sol (maybe even 400 with OC as 1070 is a lot bettern then AMD in blace algo calculations) on default 1070 or even more, at least more then AMD HW have atm

i managed to do 250 sol at 100 watt like nicehash, but ncicehash was only 220 at the same wattage, so this is actually coded better

Your setting ? .

120 core 55 tdp 400 mem

on pool side the earning is not right, are e sure the dev fee is 2% only?

I see hashrate in miner 1:1 on suprnova

yeah the hash is right but not the final earning

On Nanopool the hashrate and profit is also in line with the miner for me - using 6 hour averages. This is great work! Thanks! EWBF!

what's your exact result, mine is 1500 sol only 0.017 zcash(0.0155 + 0.0015 in round earnings) in 3.5 hours...should have been 0.02 already

You cannot reliably count that over 3 hours on prop pools, wait at least 24h to get luck variance out.

[jiggytom]

Any comments on cpu usage?  Most of my rigs run celeron, will this underperform?  I won't be able to test until later tonight.

[reb0rn21]

Any comments on cpu usage?  Most of my rigs run celeron, will this underperform?  I won't be able to test until later tonight.

there is almost none CPU usage 5x 0170 on G1840

[bensam1231]

Looks good despite the miner not really putting out errors when you input a wrong syntax. You kinda have to guess and check to figure out what exactly wrong with it.

The miner is also not compatible with miningrigrentals.com. Please fix this. I actually don't think it's compatible with any pools that don't have bat files (the miner must mine on that specific pool for the dev and can't redirect to a new pool when it needs to).

The miner does use roughly ~13% more power then Nicehash measured at the wall.

The miner gets ~235sols on STOCK settings, 250 is with OCs.

So far have not gotten any crashes.

EWBF behave totaly different
ewbf use a lot GPU power, depending on card near 90% tdp, Memory controller load is ~60%, memory use ~1.5Gb
nicehash use same 10-15% less power, memory load in 2x exe go up to 71%, memory use ~250mb per instance

I presume both have one or other parts done better, so in time i bet we can have more then 300sol (maybe even 400 with OC as 1070 is a lot bettern then AMD in blace algo calculations) on default 1070 or even more, at least more then AMD HW have atm

i managed to do 250 sol at 100 watt like nicehash, but ncicehash was only 220 at the same wattage, so this is actually coded better

Your setting ? .

120 core 55 tdp 400 mem

on pool side the earning is not right, are we sure the dev fee is 2% only?

When you tell someone your core speed the +120 DOES NOT HELP. All GPUs have different core frequencies. Where memory is the same, so +400 memory is the same on all cards. +120 core can be the difference between 1900mhz and 2190 depending on what the stock settings are for the card.

[Amph]

EWBF behave totaly different
ewbf use a lot GPU power, depending on card near 90% tdp, Memory controller load is ~60%, memory use ~1.5Gb
nicehash use same 10-15% less power, memory load in 2x exe go up to 71%, memory use ~250mb per instance

I presume both have one or other parts done better, so in time i bet we can have more then 300sol (maybe even 400 with OC as 1070 is a lot bettern then AMD in blace algo calculations) on default 1070 or even more, at least more then AMD HW have atm

i managed to do 250 sol at 100 watt like nicehash, but ncicehash was only 220 at the same wattage, so this is actually coded better

Your setting ? .

120 core 55 tdp 400 mem

on pool side the earning is not right, are e sure the dev fee is 2% only?

I see hashrate in miner 1:1 on suprnova

yeah the hash is right but not the final earning

On Nanopool the hashrate and profit is also in line with the miner for me - using 6 hour averages. This is great work! Thanks! EWBF!

what's your exact result, mine is 1500 sol only 0.017 zcash(0.0155 + 0.0015 in round earnings) in 3.5 hours...should have been 0.02 already

You cannot reliably count that over 3 hours on prop pools, wait at least 24h to get luck variance out.

right in fact now it's on par, 0.026 on 4 hours, i've counted at least 0.007 every hour

120 core 55 tdp 400 mem

on pool side the earning is not right, are we sure the dev fee is 2% only?

When you tell someone your core speed the +120 DOES NOT HELP. All GPUs have different core frequencies. Where memory is the same, so +400 memory is the same on all cards. +120 core can be the difference between 1900mhz and 2190 depending on what the stock settings are for the card.

it's limited by the tdp anyway, which is 55%, final is 1720 core

just tried with 500 mem, give slightly better result

[jiggytom]

I'm curious how the Dev was able to get the 1080 to run faster than the 1070. This means something similar should be doable for ETH mining.  Dev, have you worked on an eth miner before?

[ZenFr]

linux version?

Asking for linux version too :-).
(an binary file if you don't want publish the sources)

[topgeek]

I'm curious how the Dev was able to get the 1080 to run faster than the 1070. This means something similar should be doable for ETH mining.  Dev, have you worked on an eth miner before?

What he said!   

Please do an ETH miner that takes advantage of the 1080 as well :popcorn:

[reb0rn21]

Yeah it looks like EWBF is GPU limited, that just mean some/lot optimization can be done on blace GPU code to make it faster, if 1080 is fater then 1070 with memory that have slow latency

[tbearhere]

I know you do not have to believe me but it is clean

Thank you........ how do you call out a card if you have more then 1 card and need to split the cards up? Thx

[tbearhere]

And what is the difference between version a and version b please. Thank you.

[kondiomir]

Getting about 190 Sol/s @EVGA GTX 970...
Not bad at all.


Is there any readme file to see the options ?

[jiggytom]

Getting thread exited with code 30 on one rig...memory issues?

[tbearhere]

Whats the power usage? Did you try to underclock / reduce tdp. (-500 on the memory)

If I did my math correctly ... +500 is better with 77% tdp.