Krnlx Nvidia xevan miner - 3.3+ mh on 1070, ~6mh on 80ti FREE, OPENSOURCE

[abudfv2008]

Hi all! This is prealpha version of my xevan miner. Tested only on linux, cuda 7.5 and 1070 & 1080ti.
https://github.com/krnlx/ccminer-xevan

May be you can make a win/linux binary. Otherwise there are big chances someone very soon will share a trojan binary of your miner.
It would be much better if that binary is made by you.

[1713576961]

1713576961

[FortuneNVirtue]

Just compiled successfully (as you advised), but
I got something like :

GPU#0: NONCE FOUND
GPU#0: result for 00000128 does not validate on CPU!

Any suggestion, please ?

[krnlx]

Just compiled successfully (as you advised), but
I got something like :

GPU#0: NONCE FOUND
GPU#0: result for 00000128 does not validate on CPU!

Any suggestion, please ?

This is not good  What changes did you made to source code ?

[stas260385]

Just compiled successfully (as you advised), but
I got something like :

GPU#0: NONCE FOUND
GPU#0: result for 00000128 does not validate on CPU!

Any suggestion, please ?

Too high intensity in your case. Set -i 20
It's might help you.

Can you share your binary please?

[Gaglam]

a win binary would be highly appreciated

[rusm35]

I got some errors compiling "identifier ulong is undefined" (file: cuda_bmw512.cu)
Could you help me please?

PS. I am not a programmer, I tried to compile it with vs2013, cuda 7.5, windows 10.

try adding

#define ulong uint64_t

or just replace ulong with uint64_t

I'm not a programmer. Tell me please where to add it, I have the same mistake.

[ioglnx]

There are many issues in the windows build - I fixed already 6 and the next coming up.
Like not compiling the deps (curl etc.) krnlx may you merge the setting of your skunk version since this properly works.

[FortuneNVirtue]

Just compiled successfully (as you advised), but
I got something like :

GPU#0: NONCE FOUND
GPU#0: result for 00000128 does not validate on CPU!

Any suggestion, please ?

This is not good  What changes did you made to source code ?

1. added #define ulong uint64_t (I also tried to replace ulong with uint64_t, both result the same.)
2. changed uint ord to int ord (cuda_sha512_lbry.cu) because it show error.
3. added files cuda_x13_fugue512.cu and cuda_x13_hamsi512.cu to the project
4. copied folder "compat/curl-for-windows/out/ from somewhere else to "compat/curl-for-windows" (it asked for libcurl.x86.lib as states in linker)

I may do something wrong, but please suggest me, Thank you.

[palgin]

Stick to Linux build. Even after removing part of CPU validation from Win build yiimp rejects shares, which means hashing go wrong on Windows. Think it may be lib-related issue, maybe someone will fix

[CryptoGeneral]

There is a 3000 XLR bounty for a Windows xevan miner. I'll hope that you can make it happen.

[nax]

Just compiled successfully (as you advised), but
I got something like :

GPU#0: NONCE FOUND
GPU#0: result for 00000128 does not validate on CPU!

Any suggestion, please ?

This is not good  What changes did you made to source code ?

1. added #define ulong uint64_t (I also tried to replace ulong with uint64_t, both result the same.)
2. changed uint ord to int ord (cuda_sha512_lbry.cu) because it show error.
3. added files cuda_x13_fugue512.cu and cuda_x13_hamsi512.cu to the project
4. copied folder "compat/curl-for-windows/out/ from somewhere else to "compat/curl-for-windows" (it asked for libcurl.x86.lib as states in linker)

I may do something wrong, but please suggest me, Thank you.

same error here

[palgin]

krnlx deserves every existing bounty for this algo because he was first to release, all others will be copy-paste guys. Win binary is not obligatory, code is open and it's just a matter of time when it will become avaliable.

[}{uNdd]

Just compiled successfully (as you advised), but
I got something like :

GPU#0: NONCE FOUND
GPU#0: result for 00000128 does not validate on CPU!

Any suggestion, please ?

This is not good  What changes did you made to source code ?

1. added #define ulong uint64_t (I also tried to replace ulong with uint64_t, both result the same.)
2. changed uint ord to int ord (cuda_sha512_lbry.cu) because it show error.
3. added files cuda_x13_fugue512.cu and cuda_x13_hamsi512.cu to the project
4. copied folder "compat/curl-for-windows/out/ from somewhere else to "compat/curl-for-windows" (it asked for libcurl.x86.lib as states in linker)

I may do something wrong, but please suggest me, Thank you.

the last point error , i still finding where to copy ...

[sp_]

I have a windows build. But Xevan is not profitable. any more.

[startsts]

Can you tell me what multiplier has xevan algo?   I set to my pool multiplier 24  and my pool show more 20% hashrate than the ccminer.

[sp_]

4. copied folder "compat/curl-for-windows/out/ from somewhere else to "compat/curl-for-windows" (it asked for libcurl.x86.lib as states in linker)

Copy and replace the folder compat/curl-for-windows from the alexis 1.0 rep on github. build with cuda 7.5 and visual studio 2013

[abudfv2008]

krnlx deserves every existing bounty for this algo because he was first to release, all others will be copy-paste guys. Win binary is not obligatory, code is open and it's just a matter of time when it will become avaliable.

It is not obligatory. But do you think that coming sooner or later malware binaries are better?

[FortuneNVirtue]

Just compiled successfully (as you advised), but
I got something like :

GPU#0: NONCE FOUND
GPU#0: result for 00000128 does not validate on CPU!

Any suggestion, please ?

This is not good  What changes did you made to source code ?

1. added #define ulong uint64_t (I also tried to replace ulong with uint64_t, both result the same.)
2. changed uint ord to int ord (cuda_sha512_lbry.cu) because it show error.
3. added files cuda_x13_fugue512.cu and cuda_x13_hamsi512.cu to the project
4. copied folder "compat/curl-for-windows/out/ from somewhere else to "compat/curl-for-windows" (it asked for libcurl.x86.lib as states in linker)

I may do something wrong, but please suggest me, Thank you.

the last point error , i still finding where to copy ...

I just copied from alexis78's ccminer. (I just guessed, because it's vs2013 and cuda7.5, other versions of ccminer may work as well, I don't know)

And what palgin said is alsolutely right, krnlx deserves every bounty for opensource xevan ccminer/nvidia gpu miner.
(Solaris and Bitsend as I know)

[krnlx]

Stick to Linux build. Even after removing part of CPU validation from Win build yiimp rejects shares, which means hashing go wrong on Windows. Think it may be lib-related issue, maybe someone will fix

Can you please printout hashing result on every algo with zero input data(x17/x17.cu, commented out code) both on linux and windows to determine where is the bug related to windows version?
My rigs are headless and far away from me, I can't test it on windows ;(

[nax]

4. copied folder "compat/curl-for-windows/out/ from somewhere else to "compat/curl-for-windows" (it asked for libcurl.x86.lib as states in linker)

Copy and replace the folder compat/curl-for-windows from the alexis 1.0 rep on github. build with cuda 7.5 and visual studio 2013

How did you solve this problem?
https://ibb.co/bHPGow