CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 935.

qqqq

legendary

Activity: 1596

Merit: 1011

Quote from: t-nelson on September 21, 2015, 01:08:36 PM

Quote from: t-nelson on September 21, 2015, 11:51:23 AM

Quote from: qqqq on September 21, 2015, 11:44:30 AM

Quote from: t-nelson on September 21, 2015, 11:31:21 AM

Quote from: qqqq on September 21, 2015, 11:11:03 AM

Quote from: t-nelson on September 21, 2015, 10:21:02 AM

If anyone waiting on the solo work submission crash can compile, please test https://github.com/t-nelson/ccminer/tree/solo_stack_smash and let me know if it works. Otherwise it will probably be another three days before I find out myself :/

If someones can compile binaries i will testing quark with 45 Mhs.

What platform?

Windows

Bleh... I'll try to fix my VM.

http://ge.tt/5IlhbIO2/v/0
SHA256: 00cf5e3f815107d4bbf185153393c37b8b16b4bdbc8fb8e36df4e3679e8e025a ccminer.zip
SHA256: 2c179b8196006a29022be6a6db060b93b4d9ad40f499cb8a3d042371c0fea993 ccminer.exe

Could you please upload somewhere else, mega for ex. Cause i can't download it. Very strange.

scryptr, maybe you can upload you compiled binaries ?

scryptr

legendary

Activity: 1797

Merit: 1028

Quote from: t-nelson on September 21, 2015, 12:43:37 PM

Quote from: scryptr on September 21, 2015, 12:27:38 PM

Quote from: djm34 on September 21, 2015, 12:09:39 PM

Quote from: t-nelson on September 21, 2015, 10:21:02 AM

If anyone waiting on the solo work submission crash can compile, please test https://github.com/t-nelson/ccminer/tree/solo_stack_smash and let me know if it works. Otherwise it will probably be another three days before I find out myself :/

don't think it is related...ccminer has been sending back much bigger messages (creditcoin for example) without any problem (and without changing that part).
However I know that the message length has to be the one expected by the wallet otherwise it will get refused (like for neoscrypt btw and a few others. This assertion isn't present in all wallets... but it is present for sure for lyra2rev2 and neoscrypt, but usually the problem is more with getting the message than sending it back)

edit: and again, ccminer (at least my version) was tested on the wallet in order to correct some problem which was occurring in the previous version, and both wallet and miner were released once we (james for the wallet, myself for the miner) were sure it was finding blocks properly.
(then it was tested on pool...)

GIT CLONE--

I attempted to clone the directory above with "git clone https://github.com/t-nelson/ccminer/tree/solo_stack_smash ." on the Linux command line, and received the message "fatal: repository 'https://github.com/t-nelson/ccminer/tree/solo_stack_smash/' not found". This is the same command that I use to clone tpruvot's and sp_'s CCminer.

--scryptr

Clone URL is https://github.com/t-nelson/ccminer.git
Once cloned you can checkout the branch with

Code:

git checkout -b test origin/solo_stack_smash

CHECKOUT--

I cloned ccminer.git as instructed, changed into the ccminer directory, and issued the above checkout command. After compiling, I launched your test build of ccminer with my VTC wallet running. There was the pause on launch after the thread initializaion, about 2 minutes, then the cards began posting hash results. My 6x 750ti rig is now solo-mining across my LAN to my Win 7 x64 work computer.

According to the CoinWarz calculator, My 36Mh/s VTC rig should find a block every 0.72 days at the current difficulty (~520). I do not have any error-catching or debugging software, but VertCoins are worth about a nickle each. If I hit a block and get 50 coins, I could convert them to nickles, find a slot machine, and make myself a wealthy man.

It is all in the probability. --scryptr

EDIT: I also downloaded your Windows build and have launched it on my Win 7 x64 computer. It is mining with the same VTC wallet, adding another 6Mh/s and reducing the average time to find a block to about 1 block every 2/3 day. --scryptr

t-nelson

member

Activity: 70

Merit: 10

Quote from: t-nelson on September 21, 2015, 11:51:23 AM

Quote from: qqqq on September 21, 2015, 11:44:30 AM

Quote from: t-nelson on September 21, 2015, 11:31:21 AM

Quote from: qqqq on September 21, 2015, 11:11:03 AM

Quote from: t-nelson on September 21, 2015, 10:21:02 AM

If anyone waiting on the solo work submission crash can compile, please test https://github.com/t-nelson/ccminer/tree/solo_stack_smash and let me know if it works. Otherwise it will probably be another three days before I find out myself :/

If someones can compile binaries i will testing quark with 45 Mhs.

What platform?

Windows

Bleh... I'll try to fix my VM.

http://ge.tt/5IlhbIO2/v/0
SHA256: 00cf5e3f815107d4bbf185153393c37b8b16b4bdbc8fb8e36df4e3679e8e025a ccminer.zip
SHA256: 2c179b8196006a29022be6a6db060b93b4d9ad40f499cb8a3d042371c0fea993 ccminer.exe

t-nelson

member

Activity: 70

Merit: 10

Quote from: scryptr on September 21, 2015, 12:27:38 PM

Quote from: djm34 on September 21, 2015, 12:09:39 PM

Quote from: t-nelson on September 21, 2015, 10:21:02 AM

If anyone waiting on the solo work submission crash can compile, please test https://github.com/t-nelson/ccminer/tree/solo_stack_smash and let me know if it works. Otherwise it will probably be another three days before I find out myself :/

don't think it is related...ccminer has been sending back much bigger messages (creditcoin for example) without any problem (and without changing that part).
However I know that the message length has to be the one expected by the wallet otherwise it will get refused (like for neoscrypt btw and a few others. This assertion isn't present in all wallets... but it is present for sure for lyra2rev2 and neoscrypt, but usually the problem is more with getting the message than sending it back)

edit: and again, ccminer (at least my version) was tested on the wallet in order to correct some problem which was occurring in the previous version, and both wallet and miner were released once we (james for the wallet, myself for the miner) were sure it was finding blocks properly.
(then it was tested on pool...)

GIT CLONE--

I attempted to clone the directory above with "git clone https://github.com/t-nelson/ccminer/tree/solo_stack_smash ." on the Linux command line, and received the message "fatal: repository 'https://github.com/t-nelson/ccminer/tree/solo_stack_smash/' not found". This is the same command that I use to clone tpruvot's and sp_'s CCminer.

--scryptr

Clone URL is https://github.com/t-nelson/ccminer.git
Once cloned you can checkout the branch with

Code:

git checkout -b test origin/solo_stack_smash

qqqq

legendary

Activity: 1596

Merit: 1011

Quote from: djm34 on September 21, 2015, 12:39:52 PM

Quote from: t-nelson on September 21, 2015, 12:26:33 PM

Quote from: djm34 on September 21, 2015, 12:09:39 PM

Quote from: t-nelson on September 21, 2015, 10:21:02 AM

If anyone waiting on the solo work submission crash can compile, please test https://github.com/t-nelson/ccminer/tree/solo_stack_smash and let me know if it works. Otherwise it will probably be another three days before I find out myself :/

don't think it is related...ccminer has been sending back much bigger messages (creditcoin for example) without any problem (and without changing that part).
However I know that the message length has to be the one expected by the wallet otherwise it will get refused (like for neoscrypt btw and a few others. This assertion isn't present in all wallets... but it is present for sure for lyra2rev2 and neoscrypt, but usually the problem is more with getting the message than sending it back)

The sprintf() call was the only obvious path to a smashed stack that I saw in submit_upstream_work(). Being as the stack was utterly destroyed, I couldn't examine the parameters passed in to see if they were valid or not. There could be a scribbler elsewhere, but it's going to take me a few more days to solve another block and hit my breakpoint.

if you are in luck you can try testnet (I tried, however, I haven't been able to sync... )

Testnet is a not good thing for testing like finding real blocks. I know the guys who testing via testnet and thinking that eveything is ok, but in real mining is not.

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: t-nelson on September 21, 2015, 12:26:33 PM

Quote from: djm34 on September 21, 2015, 12:09:39 PM

Quote from: t-nelson on September 21, 2015, 10:21:02 AM

If anyone waiting on the solo work submission crash can compile, please test https://github.com/t-nelson/ccminer/tree/solo_stack_smash and let me know if it works. Otherwise it will probably be another three days before I find out myself :/

don't think it is related...ccminer has been sending back much bigger messages (creditcoin for example) without any problem (and without changing that part).
However I know that the message length has to be the one expected by the wallet otherwise it will get refused (like for neoscrypt btw and a few others. This assertion isn't present in all wallets... but it is present for sure for lyra2rev2 and neoscrypt, but usually the problem is more with getting the message than sending it back)

The sprintf() call was the only obvious path to a smashed stack that I saw in submit_upstream_work(). Being as the stack was utterly destroyed, I couldn't examine the parameters passed in to see if they were valid or not. There could be a scribbler elsewhere, but it's going to take me a few more days to solve another block and hit my breakpoint.

if you are in luck you can try testnet (I tried, however, I haven't been able to sync... )

scryptr

legendary

Activity: 1797

Merit: 1028

Quote from: djm34 on September 21, 2015, 12:09:39 PM

Quote from: t-nelson on September 21, 2015, 10:21:02 AM

If anyone waiting on the solo work submission crash can compile, please test https://github.com/t-nelson/ccminer/tree/solo_stack_smash and let me know if it works. Otherwise it will probably be another three days before I find out myself :/

don't think it is related...ccminer has been sending back much bigger messages (creditcoin for example) without any problem (and without changing that part).
However I know that the message length has to be the one expected by the wallet otherwise it will get refused (like for neoscrypt btw and a few others. This assertion isn't present in all wallets... but it is present for sure for lyra2rev2 and neoscrypt, but usually the problem is more with getting the message than sending it back)

edit: and again, ccminer (at least my version) was tested on the wallet in order to correct some problem which was occurring in the previous version, and both wallet and miner were released once we (james for the wallet, myself for the miner) were sure it was finding blocks properly.
(then it was tested on pool...)

GIT CLONE--

I attempted to clone the directory above with "git clone https://github.com/t-nelson/ccminer/tree/solo_stack_smash ." on the Linux command line, and received the message "fatal: repository 'https://github.com/t-nelson/ccminer/tree/solo_stack_smash/' not found". This is the same command that I use to clone tpruvot's and sp_'s CCminer.

--scryptr

t-nelson

member

Activity: 70

Merit: 10

Quote from: djm34 on September 21, 2015, 12:09:39 PM

Quote from: t-nelson on September 21, 2015, 10:21:02 AM

If anyone waiting on the solo work submission crash can compile, please test https://github.com/t-nelson/ccminer/tree/solo_stack_smash and let me know if it works. Otherwise it will probably be another three days before I find out myself :/

don't think it is related...ccminer has been sending back much bigger messages (creditcoin for example) without any problem (and without changing that part).
However I know that the message length has to be the one expected by the wallet otherwise it will get refused (like for neoscrypt btw and a few others. This assertion isn't present in all wallets... but it is present for sure for lyra2rev2 and neoscrypt, but usually the problem is more with getting the message than sending it back)

The sprintf() call was the only obvious path to a smashed stack that I saw in submit_upstream_work(). Being as the stack was utterly destroyed, I couldn't examine the parameters passed in to see if they were valid or not. There could be a scribbler elsewhere, but it's going to take me a few more days to solve another block and hit my breakpoint.

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: t-nelson on September 21, 2015, 10:21:02 AM

If anyone waiting on the solo work submission crash can compile, please test https://github.com/t-nelson/ccminer/tree/solo_stack_smash and let me know if it works. Otherwise it will probably be another three days before I find out myself :/

don't think it is related...ccminer has been sending back much bigger messages (creditcoin for example) without any problem (and without changing that part).
However I know that the message length has to be the one expected by the wallet otherwise it will get refused (like for neoscrypt btw and a few others. This assertion isn't present in all wallets... but it is present for sure for lyra2rev2 and neoscrypt, but usually the problem is more with getting the message than sending it back)

edit: and again, ccminer (at least my version) was tested on the wallet in order to correct some problem which was occurring in the previous version, and both wallet and miner were released once we (james for the wallet, myself for the miner) were sure it was finding blocks properly.
(then it was tested on pool...)

edit2: may-be something to try instead of wasting 3 days: pick up an already found block (in the wallet) and try to submit it back with the correct nonce. it should be rejected, however it shouldn't generate a json error.

t-nelson

member

Activity: 70

Merit: 10

Quote from: qqqq on September 21, 2015, 11:44:30 AM

Quote from: t-nelson on September 21, 2015, 11:31:21 AM

Quote from: qqqq on September 21, 2015, 11:11:03 AM

Quote from: t-nelson on September 21, 2015, 10:21:02 AM

If anyone waiting on the solo work submission crash can compile, please test https://github.com/t-nelson/ccminer/tree/solo_stack_smash and let me know if it works. Otherwise it will probably be another three days before I find out myself :/

If someones can compile binaries i will testing quark with 45 Mhs.

What platform?

Windows

Bleh... I'll try to fix my VM.

qqqq

legendary

Activity: 1596

Merit: 1011

Quote from: t-nelson on September 21, 2015, 11:31:21 AM

Quote from: qqqq on September 21, 2015, 11:11:03 AM

Quote from: t-nelson on September 21, 2015, 10:21:02 AM

If anyone waiting on the solo work submission crash can compile, please test https://github.com/t-nelson/ccminer/tree/solo_stack_smash and let me know if it works. Otherwise it will probably be another three days before I find out myself :/

If someones can compile binaries i will testing quark with 45 Mhs.

What platform?

Windows

t-nelson

member

Activity: 70

Merit: 10

Quote from: qqqq on September 21, 2015, 11:11:03 AM

Quote from: t-nelson on September 21, 2015, 10:21:02 AM

If anyone waiting on the solo work submission crash can compile, please test https://github.com/t-nelson/ccminer/tree/solo_stack_smash and let me know if it works. Otherwise it will probably be another three days before I find out myself :/

If someones can compile binaries i will testing quark with 45 Mhs.

What platform?

qqqq

legendary

Activity: 1596

Merit: 1011

Quote from: t-nelson on September 21, 2015, 10:21:02 AM

If anyone waiting on the solo work submission crash can compile, please test https://github.com/t-nelson/ccminer/tree/solo_stack_smash and let me know if it works. Otherwise it will probably be another three days before I find out myself :/

If someones can compile binaries i will testing quark with 45 Mhs.

rednoW

legendary

Activity: 1510

Merit: 1003

Fired my poor gtx750 1gb again, 1500/1600 = 6300kh/s (release 68, quark)

t-nelson

member

Activity: 70

Merit: 10

If anyone waiting on the solo work submission crash can compile, please test https://github.com/t-nelson/ccminer/tree/solo_stack_smash and let me know if it works. Otherwise it will probably be another three days before I find out myself :/

t-nelson

member

Activity: 70

Merit: 10

Quote from: scryptr on September 21, 2015, 09:51:07 AM

Quote from: qqqq on September 21, 2015, 06:33:17 AM

Btw you can test ccminer with even SRC it has same solo bug (crashing miner with json_rpc_call execution or something) but quark algo. So you waste much lower time for block.

A CLUE--

Thanks for the information. I suppose there may be a portion of the code that is not algorithm specific that is broken. However, I completely failed to get CCminer to solo-mine FeatherCoin (FTC), it would never communicate properly with the wallet. SecureCoin (SRC) may be as good a candidate as VertCoin (VTC) for the purpose of debugging.

I am not familiar enough with the code to make more than a WildAssGuess (WAG). So far, I've solo-mined Ether Coin (ETH) and WertCoin. I have a copy of the SpreadCoin (SPR) solo-miner, and I assume that it works properly, but I haven't tried it yet. Maybe the code could be compared where the miners (CCminer, and SPR solo-miner) communicate with the wallet.

--scryptr

Code:

[2015-09-21 03:01:10] GPU #0: GeForce GTX 960, 5726 Temp= 56C Fan= 60%
[2015-09-21 03:01:10] GPU #1: GeForce GTX 750 Ti, 4178 Temp= 48C Fan= 50%
[2015-09-21 03:01:35] GPU #0: GeForce GTX 960, 5726 Temp= 56C Fan= 60%
[2015-09-21 03:01:35] GPU #1: GeForce GTX 750 Ti, 4178 Temp= 48C Fan= 50%
[2015-09-21 03:01:58] GPU #0: GeForce GTX 960, 5726 Temp= 57C Fan= 60%
[2015-09-21 03:01:58] JSON-RPC call failed: Invalid parameter
[2015-09-21 03:01:58] submit_upstream_work json_rpc_call failed
*** stack smashing detected ***: /dist/ccminer terminated

Program received signal SIGABRT, Aborted.

Finally...

I think I know what the problem is.

scryptr

legendary

Activity: 1797

Merit: 1028

Quote from: qqqq on September 21, 2015, 06:33:17 AM

Btw you can test ccminer with even SRC it has same solo bug (crashing miner with json_rpc_call execution or something) but quark algo. So you waste much lower time for block.

A CLUE--

Thanks for the information. I suppose there may be a portion of the code that is not algorithm specific that is broken. However, I completely failed to get CCminer to solo-mine FeatherCoin (FTC), it would never communicate properly with the wallet. SecureCoin (SRC) may be as good a candidate as VertCoin (VTC) for the purpose of debugging.

I am not familiar enough with the code to make more than a WildAssGuess (WAG). So far, I've solo-mined Ether Coin (ETH) and WertCoin. I have a copy of the SpreadCoin (SPR) solo-miner, and I assume that it works properly, but I haven't tried it yet. Maybe the code could be compared where the miners (CCminer, and SPR solo-miner) communicate with the wallet.

--scryptr

qqqq

legendary

Activity: 1596

Merit: 1011

Btw you can test ccminer with even SRC it has same solo bug (crashing miner with json_rpc_call execution or something) but quark algo. So you waste much lower time for block.

ol92

sr. member

Activity: 445

Merit: 255

Quote from: Genoil on September 21, 2015, 04:58:17 AM

Quote from: ol92 on September 21, 2015, 03:07:24 AM

Quote from: Genoil on September 21, 2015, 02:59:11 AM

Quote from: ol92 on September 21, 2015, 02:43:44 AM

Quote from: theotherme on September 21, 2015, 02:14:02 AM

Quote from: bensam1231 on September 20, 2015, 11:52:18 PM

...

I was mining Eth instead of Vert for awhile, but right now a 280x mines almost twice as fast as 970, so after all the AMD miners hoped on board it was no longer all that profitable. It's about break even with Vert right now though, still sometimes more profitable, regardless of it being slower.

Quote from: djm34 on September 19, 2015, 08:38:54 PM

Quote from: sp_ on September 19, 2015, 07:25:32 PM

Quote from: myagui on September 19, 2015, 12:47:30 PM

ETH can probably run much faster on the high end cards if some of the memory access can b replaced with computation.

The ET performance depends on the DAG file.. Without a random file, every random acces isside the cached and the hashrate explodes

duh ?! there is no way to get rid of those... otherwise,as Myaguy was explaining, it may possible to calculate some of the dag element hence reducing memory usage

Have you guys considered more compression?

thinking about it, but it isn't really obvious it would help... need to find a compression algorithm which works on it (I tried zipping the file, just to have an idea of what could be done and ended up with a larger file... Grin

) so it seems there isn't much to compress (well 1.2Gb of integers...). For info, for a 780ti to work correctly on windows 8.1, only 1.3% less dag is required (but that 1.3% is still too large to store anywhere that amount of data (too large for either register or texture/shared )

The GTX 980 ti with +50% more bandwith has similar performance than a GTX 980 : there should be some bottleneck elsewhere...
an amd 280x has better performance ...

It's not like the algo loads large chunks of contiguous blocks of memory like texture maps. Instead it loads 64 times a 128 byte random chunk, where the index of each can only be found based on the previous one. So the available bandwidth cannot be fully utilized. I've already tried running two of those loops in parallel, but haven't been succesful at that yet.

So where does the amd advantage come from ? better memory controler with tighter timings ?

I think wider memory bus and therefore higher overall bandwidth. In games, nVidia makes that up with compression, but here there nothing to compress in random chunks of 128 bytes of hashed hashes. The effciency of AMD vs NVidia is about the same. i.e for a GTX980, that hashes about 19MH, the theoretical max. hashrate is 224GBps/8KBph = 28MH. 19MH/28MH = 67%. For a 280/X.290/X that hash at about 27MH on average, the theoretical max is 320GBps/8KBph=40MH. 27MH/40MH= 67%.

This reasoning works for gtx 980 but not for gtx 980 ti far less efficient nor r9 280x which has a higher hashrate / bandwidth ratio

chrysophylax

legendary

Activity: 2940

Merit: 1091

--- ChainWorks Industries ---

hi all ...

donation by mining ... dbm ...

im very happy to announce that the donation links are now active for community use ...

please have a look and read - and help with the mining donations to these awesome devs ...

the op will be updated as soon as i collate the rest of the info required ...

the thread post is here - https://bitcointalksearch.org/topic/m.12480969 ...

join in and help hash for the devs ...

#crysx

Topic: CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 935. (Read 2347664 times)