CCminer(SP-MOD) Modded GPU kernels.

[tbearhere]

here is the build

Zcash sp-mod #1

https://github.com/sp-hash/ccminer/releases/tag/zcash-spmod1

sp_, you're the man!! Thanks!!!

What rates are you getting?

After 12+ hours, looks like the software is reporting stable 114-121 Sols with 4 -Asus GTX 750ti GPU's standard clocked on this rig. Even with this, it's still more than double I was getting before. No support for AVX or AVX2 on the CPU. I think 8GB of RAM on this system. Pleased so far. Thank you!!

Try downing your memory .. you should get the same s/s with less power. I did this on my 750ti rig.

[1715648507]

1715648507

[1715648507]

1715648507

[1715648507]

1715648507

[1715648507]

1715648507

[antantti]

thebearhere,

You are doing it wrong.

I can tell you how to fix those issues if you get rid of that astalavista.

Upgrade to win7, there is nothing new to learn, most of the time it is the best OS (de facto) when mining.

[fenomenhaa]

Sp1 vers.
1070  +850mem +30 core 115 sol
970  +350mem  72 Sol

Did you set your clocks to the p0 state?  Thx
EDIT: I have my 970 set to 0 on the core and -501 mem getting ~75 sols/s. Less wattage too. Normal p2 state.

yeah it's on p0 state ,I'll try p2 state when at work.

[sp_]

Zcash sp-mod #1 has a memory leak. Klaust's fork has the same. Seems to be an error in the Silent Army cuda port

[ioglnx]

For me it runs stable for days also your build :-)

[sp_]

For me it runs stable for days also your build :-)

Yes it runs for days, but the process is eating memory and eventually your rig will crash. You can observe it easily in a 6card rig if you open taskmanager.

[ioglnx]

Have 5 cards but memory usage is just 4GB of all processes..since i use it..i monitor this too but no leak so far.

[Schleicher]

Zcash sp-mod #1 has a memory leak. Klaust's fork has the same. Seems to be an error in the Silent Army cuda port

There's this malloc:

Code:

eq->sols = (sols_t *)malloc(sizeof(*eq->sols));

The free() is missing.
Also there's this:

Code:

checkCudaErrors(cudaMalloc((void**)&buf_blake_st, sizeof(blake2b_state_s)));

The cudaFree() is missing.

[ioglnx]

Zcash sp-mod #1 has a memory leak. Klaust's fork has the same. Seems to be an error in the Silent Army cuda port

There's this malloc:

Code:

eq->sols = (sols_t *)malloc(sizeof(*eq->sols));

The free() is missing.
Also there's this:

Code:

checkCudaErrors(cudaMalloc((void**)&buf_blake_st, sizeof(blake2b_state_s)));

The cudaFree() is missing.

Klaus :-D du Haudegen..danke ..Thanks is it fixed in your repo?
Ah seh schon ist gefixt.

[Amph]

Zcash sp-mod #1 has a memory leak. Klaust's fork has the same. Seems to be an error in the Silent Army cuda port

so it happened to you also?

Have 5 cards but memory usage is just 4GB of all processes..since i use it..i monitor this too but no leak so far.

maybe don't happen with less than 6 cards

[tbearhere]

thebearhere,

You are doing it wrong.

I can tell you how to fix those issues if you get rid of that astalavista.

Upgrade to win7, there is nothing new to learn, most of the time it is the best OS (de facto) when mining.

I wish someone could help me with that.
Why Ive been mining on it fine for years. I cloned the going bad hdd and when i did that im missing a file that wont let me update in order to install  msvs 2015. Im missing one file and it can be updated but I don't know how to install it.  If i could burn msvs 2015 on a cd, i would then put it on top of the new hdd and hammer it in.
No thx i dont need to Max allowcate 100% stuff  

ps curious is it free win 7 ?

[tbearhere]

Sp1 vers.
1070  +850mem +30 core 115 sol
970  +350mem  72 Sol

Did you set your clocks to the p0 state?  Thx
EDIT: I have my 970 set to 0 on the core and -501 mem getting ~75 sols/s. Less wattage too. Normal p2 state.

yeah it's on p0 state ,I'll try p2 state when at work.

If your mining in the p0 state great.. don't touch it.

[tbearhere]

Zcash sp-mod #1 has a memory leak. Klaust's fork has the same. Seems to be an error in the Silent Army cuda port

so it happened to you also?

Have 5 cards but memory usage is just 4GB of all processes..since i use it..i monitor this too but no leak so far.

maybe don't happen with less than 6 cards

What is a memory leak.....a question I never asked. thx
ps my 6 card 750ti hasn't crashed.

[thebigjdoe]

Zcash sp-mod #1 has a memory leak. Klaust's fork has the same. Seems to be an error in the Silent Army cuda port

so it happened to you also?

Have 5 cards but memory usage is just 4GB of all processes..since i use it..i monitor this too but no leak so far.

maybe don't happen with less than 6 cards

What is a memory leak.....a question I never asked. thx
ps my 6 card 750ti hasn't crashed.

Memory gets allocated for some value but never gets deallocated.  This memory is never allowed to be re-used because the OS thinks it is in use and eventually you run out of memory.  if you have a bucket full of water (memory) that has a leak, it will eventually be empty (no more memory)

[tbearhere]

Zcash sp-mod #1 has a memory leak. Klaust's fork has the same. Seems to be an error in the Silent Army cuda port

so it happened to you also?

Have 5 cards but memory usage is just 4GB of all processes..since i use it..i monitor this too but no leak so far.

maybe don't happen with less than 6 cards

What is a memory leak.....a question I never asked. thx
ps my 6 card 750ti hasn't crashed.

Memory gets allocated for some value but never gets deallocated.  This memory is never allowed to be re-used because the OS thinks it is in use and eventually you run out of memory.  if you have a bucket full of water (memory) that has a leak, it will eventually be empty (no more memory)

I see ... thx  thebigjdoe  
EDIT: So the app keeps filling up the ram till it's full and just keeps on trying to fill it up " leaking memory ".

[bensam1231]

Just want to mention I had my first EVGA failure. Shower of sparks from the PCIE VRM area, much like reports on reddit. Not sure if it was the VRMs or I just had a lemon. Rest of my EVGA cards are working fine. Mining Equihash doesn't remotely stress your cards, hopefully burn in will go smoothly with these. Stay away from MSI.

[_javi_]

Just want to mention I had my first EVGA failure. Shower of sparks from the PCIE VRM area, much like reports on reddit. Not sure if it was the VRMs or I just had a lemon. Rest of my EVGA cards are working fine. Mining Equihash doesn't remotely stress your cards, hopefully burn in will go smoothly with these. Stay away from MSI.

i had an EVGA burn last month.. fried pci riser (two of them). I got my rma waiting in Miami for me to pick it up (someday..).

Both MSi (micron) are also comming back from rma.. just got email last week.

Very bad experience with the 1070s... expensive.. profit plunged.. lack of miners... and defective cards.

My 4 years old 7950s are paying more...


ah.. forgot to say... 2 fried mobos from the 4x1070 rig, while changing algos (LBC to ETH). It seems the gpus took too much juice from the mobo while doing dagger.. and lifted tracks. (My fault, but never happened before with 7950/280x)

Tell me about buying a nvidia card again...

[Nikolaj]

Just want to mention I had my first EVGA failure. Shower of sparks from the PCIE VRM area, much like reports on reddit. Not sure if it was the VRMs or I just had a lemon. Rest of my EVGA cards are working fine. Mining Equihash doesn't remotely stress your cards, hopefully burn in will go smoothly with these. Stay away from MSI.

i had an EVGA burn last month.. fried pci riser (two of them). I got my rma waiting in Miami for me to pick it up (someday..).

Both MSi (micron) are also comming back from rma.. just got email last week.

Very bad experience with the 1070s... expensive.. profit plunged.. lack of miners... and defective cards.

My 4 years old 7950s are paying more...


ah.. forgot to say... 2 fried mobos from the 4x1070 rig, while changing algos (LBC to ETH). It seems the gpus took too much juice from the mobo while doing dagger.. and lifted tracks. (My fault, but never happened before with 7950/280x)

Tell me about buying a nvidia card again...

I've got 30 1070 and 1080, 2 of them (founders edition) needed to be sent in rma due shitty chinese risers (the only model available).

You should tell us also the motherboard type, I hadn't any problem with asrock h81 pro btc

It has nothing to do with the nvidias

[bensam1231]

Yeah, not going to sell my 1070s, but definitely going to start diversifying into AMD. The real lack of Nvidia support is killing things for these cards. No one is making miners for them. It's all shitty OCL ports for Equihash and I haven't heard peeps out of the devs as far as any new projects even though there are ripe opportunities.

There is no telling what is going to happen this January, but from the looks of it Equihash might take over for Ethereum... hopefully.


In my experience over the years, get powered risers. I personally use 25cm powered risers with a capacitor. Haven't had any problems with them. I don't use the PCB risers as the couple I ordered caused detection issues, they also interfere with my rig racks. You can order a bunch of them from China a couple months in advance.

[ioglnx]

I have split my rig into parity wise AMD:Nvidia. Sold some GTX1080s for the gamers and bought RX480.
While a friend of mine did the same but he also complaint that the AMD are not so efficient as the GTX1070..