CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 1042.

Epsylon3

legendary

Activity: 1484

Merit: 1082

ccminer/cpuminer developer

Quote from: scryptr on July 10, 2015, 07:05:17 AM

BITCOIN--

Bitcoin is up over $285 on several exchanges. WooHoo!

I'm looking forward to moving to CUDA 7.5 sometime soon. As soon as Los Jefes sort it out, that is!

--scryptr

reminder: you can safely install cuda 7.5 on linux if you answer no to the symlink, the only thing to change is in configure.sh (to build)

--with-cuda=/usr/local/cuda-7.5

and to execute, add the line /usr/local/cuda-7.5/lib64 in a file named /etc/ld.so.conf.d/cuda.conf and type ldconfig

Epsylon3

legendary

Activity: 1484

Merit: 1082

ccminer/cpuminer developer

linux doesnt have the heavycoin hefty problem

scryptr

legendary

Activity: 1797

Merit: 1028

BITCOIN--

Bitcoin is up over $285 on several exchanges. WooHoo!

I'm looking forward to moving to CUDA 7.5 sometime soon. As soon as Los Jefes sort it out, that is!

--scryptr

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: myagui on July 10, 2015, 06:23:53 AM

@sp_:

Your latest git does not compile with cuda 7.5 on Windows (and likely not on *nix either). There's a problem with Hefty causing the following error message:

Code:

Error	6	error : 'cicc' died with status 0xC0000005 (ACCESS_VIOLATION)

The top men on IRC had also found this error already, and I understand there's at least 2 solutions, cutting out Hefty entirely, or modifying it so that the compile can succeed. I don't think either solution has made it into anyone's git yet. You should stop by IRC sometime Cool

For those with working versions, so far it looks that cuda 7.5 speeds up some things, and slows down other things.
Mixed results all around, so really just for the adventurous for now...

need to remove the shitload of "#pragma unroll" in cuda_hefty1.cu (especially the double loop) and it works fine

compiling with "-Xptxas --def-load-cache=cs -Xptxas --def-store-cache=cs " might help also a bit to lower the difference

I just committed the change to my lyra release github/djm34/ccminer-lyra

in principle cuda 7.5 give a +100kh/s on gtx980 (getting up to 2630kh/s on my card OC)
+60kh/s on gtx750ti (from 1140 to 1200kh/s on my card)
and using the compilation option allows the 780ti to run at 2.8MH/s instead of 2.1MH/s (however must keep in mind that with cuda 6.5, the 780ti is faster and does 2.9MH/s)

myagui

legendary

Activity: 1154

Merit: 1001

@sp_:

Your latest git does not compile with cuda 7.5 on Windows (and likely not on *nix either). There's a problem with Hefty causing the following error message:

Code:

Error	6	error : 'cicc' died with status 0xC0000005 (ACCESS_VIOLATION)

The top men on IRC had also found this error already, and I understand there's at least 2 solutions, cutting out Hefty entirely, or modifying it so that the compile can succeed. I don't think either solution has made it into anyone's git yet. You should stop by IRC sometime Cool

For those with working versions, so far it looks that cuda 7.5 speeds up some things, and slows down other things.
Mixed results all around, so really just for the adventurous for now...

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: scryptr on July 10, 2015, 06:02:46 AM

Quote from: sp_ on July 10, 2015, 03:04:05 AM

Quote from: AliMan on July 10, 2015, 12:14:03 AM

Quote from: sp_ on July 09, 2015, 11:55:04 AM

If you have time please run tvprovot's cuda 7.0 version of quark and compare the hashrate with my 6.5 release 54-git and post your findings.

Getting 71 MH/s using that miner compared to yours with 93 MH/s.

Could anyone try to compile and run my latest git with cuda 7.5? Is the hash still broken? what is the current speed compared to 6.5

Epsylon3--

I hope that Epsylon3 can take the time to compile and comb through your code! He's really the current goto guy for CUDA 7.5 and CCminer.

If I upgrade my Linux boxes, it will be with CUDA 7.5. Why wait for the future? --scryptr

because right now you will lose in performance unless you mine lyra2re (that's pretty much the actual picture...)
So until there is some undertsanding of what is happening and how to get our perf back, it isn't a good idea

scryptr

legendary

Activity: 1797

Merit: 1028

Quote from: sp_ on July 10, 2015, 03:04:05 AM

Quote from: AliMan on July 10, 2015, 12:14:03 AM

Quote from: sp_ on July 09, 2015, 11:55:04 AM

If you have time please run tvprovot's cuda 7.0 version of quark and compare the hashrate with my 6.5 release 54-git and post your findings.

Getting 71 MH/s using that miner compared to yours with 93 MH/s.

Could anyone try to compile and run my latest git with cuda 7.5? Is the hash still broken? what is the current speed compared to 6.5

Epsylon3--

I hope that Epsylon3 can take the time to compile and comb through your code! He's really the current goto guy for CUDA 7.5 and CCminer.

If I upgrade my Linux boxes, it will be with CUDA 7.5. Why wait for the future? --scryptr

sp_

legendary

Activity: 2926

Merit: 1087

Team Black developer

Quote from: AliMan on July 10, 2015, 12:14:03 AM

Quote from: sp_ on July 09, 2015, 11:55:04 AM

If you have time please run tvprovot's cuda 7.0 version of quark and compare the hashrate with my 6.5 release 54-git and post your findings.

Getting 71 MH/s using that miner compared to yours with 93 MH/s.

Could anyone try to compile and run my latest git with cuda 7.5? Is the hash still broken? what is the current speed compared to 6.5

sp_

legendary

Activity: 2926

Merit: 1087

Team Black developer

Quote from: cyberspacemonkey on July 09, 2015, 10:29:37 PM

So you mean CUDA's toolkit 6.5 is better for compiling than 7.0? I'm currently using 7.0 not sure if I should downgrade.

If you want to use my fork you need to downgrade. Cuda toolkit 7.0 is producing an exe file wich gives error in the hash. (doesn't work) and it's much slower.

sp_

legendary

Activity: 2926

Merit: 1087

Team Black developer

The yaamp clone http://hashpower.co/ is now paying out in DASH. I tired to testmine yesterday and receieved payments.

here is their thread:

https://bitcointalksearch.org/topic/pool-hashpowerco-yaamp-clone-x11-x13-x15-quark-neoscrypt-1108818

Epsylon3

legendary

Activity: 1484

Merit: 1082

ccminer/cpuminer developer

i will fully skip the cuda 7.0... and never made any release with it for that, 7.5 is different

I just fixed a few problems with it...

AliMan

hero member

Activity: 1974

Merit: 502

Vave.com - Crypto Casino

Quote from: sp_ on July 09, 2015, 11:55:04 AM

If you want a working miner for cuda7.0 this is the correct branch:

https://github.com/tpruvot/ccminer

I tried to compile my fork for the cuda 7.0 and started modding a bit. But the compiler wasn't good enough. Not worth the effort..
Around 10% drop in hashrate in all algos..

If you have time please run tvprovot's cuda 7.0 version of quark and compare the hashrate with my 6.5 release 54-git and post your findings.

Getting 71 MH/s using that miner compared to yours with 93 MH/s.

cyberspacemonkey

legendary

Activity: 1288

Merit: 1002

Quote from: sp_ on July 09, 2015, 11:55:04 AM

If you want a working miner for cuda7.0 this is the correct branch:

https://github.com/tpruvot/ccminer

I tried to compile my fork for the cuda 7.0 and started modding a bit. But the compiler wasn't good enough. Not worth the effort..
Around 10% drop in hashrate in all algos..

If you have time please run tvprovot's cuda 7.0 version of quark and compare the hashrate with my 6.5 release 54-git and post your findings.

So you mean CUDA's toolkit 6.5 is better for compiling than 7.0? I'm currently using 7.0 not sure if I should downgrade.

flipclip

member

Activity: 111

Merit: 10

Quote from: scryptr on July 08, 2015, 10:49:30 AM

Quote from: flipclip on July 08, 2015, 10:31:21 AM

How do you do checkouts then? I'm used to the command line and using the sha for checkouts, so that is why I am wondering.

COMMAND LINE--

I use the command line, and refer to the commit number when posting about performance. The sha will verify checksum, and is very precise for that purpose. Commit numbers are sequential.

The line, "git clone https://github.com/sp-hash/ccminer", should clone the latest commit. If I am wrong, please tell me!

--scryptr

Yes that is the correct command to get the latest commit. I was just wondering if you used some type of command to get a previous commit via "COMMIT #". An example would be someone says "COMMIT #820 is doing faster lyra still" (completely made up example by the way), but you are already on COMMIT #843, so you use a git command to go back to COMMIT #820 to try compiling from that point in the past. I've always used "git checkout sha" to go "back in time" (hence the reason I always find the sha information more informative then "COMMIT #") but I was thinking maybe I was missing a command.

sp_

legendary

Activity: 2926

Merit: 1087

Team Black developer

If you want a working miner for cuda7.0 this is the correct branch:

https://github.com/tpruvot/ccminer

I tried to compile my fork for the cuda 7.0 and started modding a bit. But the compiler wasn't good enough. Not worth the effort..
Around 10% drop in hashrate in all algos..

If you have time please run tvprovot's cuda 7.0 version of quark and compare the hashrate with my 6.5 release 54-git and post your findings.

dga

hero member

Activity: 737

Merit: 511

Quote from: sp_ on July 09, 2015, 09:56:12 AM

Submittet a bugfix and a speedup in quark.

The gtx windforce 970 is now peaking at 16130 on standard clocks. (up from 15800)

Note that release 54 has a small bug in the hash that will report lower rates on the pool.

Just fyi, gtx 980:

Code:

[2015-07-09 15:24:05] GPU #0: result for nonce $0353D188 does not validate on CPU!
[2015-07-09 15:24:09] GPU #1: result for nonce $8BFDF7B0 does not validate on CPU!

compiling with cuda 7.0 on ubuntu for 980.

Does work under 6.5 on a 750ti. Not sure if it's card or cuda version.

sp_

legendary

Activity: 2926

Merit: 1087

Team Black developer

Submittet a bugfix and a speedup in quark.

The gtx windforce 970 is now peaking at 16130 on standard clocks. (up from 15800)

bensam1231

legendary

Activity: 1764

Merit: 1024

Quote from: djm34 on July 09, 2015, 04:08:22 AM

Quote from: bensam1231 on July 08, 2015, 11:08:37 PM

Quote from: bathrobehero on July 08, 2015, 06:22:50 PM

Quote from: djm34 on July 08, 2015, 05:56:16 PM

Quote from: jjjordan on July 08, 2015, 03:43:40 PM

7x970 - Lyra2 won't start (out of memory)
4GB system memory

7x 970 ?!

mobo are so expensive...
you need at least as much ram than vram here something like 28Gb to run that kind of system

Surely there has to be a workaround. I mean the memory/swap doesn't even seem to be allocated let alone used, not even for a second.
Something like initializing the cards one after the other instead of all at the same time or something? Or giving the cards different jobs instead of working together on one big job? I have no idea but I'm sure there's a way.

Also agree, the same thing was happening to me with Neoscrypt and just had to throw more memory at it even though system memory basically isn't used at all.

If it just uses it to 'load' into the vram on the miner, a asynchronous load should help with it (load each card into memory, then into vram one at a time). Right now though I don't really see any indication of memory usage on the system.

if you open msi AB and watch both ram and pagefile graphics, you'll see it gets allocated (more on the pagefile than on the memory) so may-be trying to increase pagefile could work.
There isn't really a work around on the code side, global memory variables have to be allocated from the host and cudamalloc works in mysterious way...)

So loading one card at a time, waiting for memory allocation, then loading another wouldn't help fix this? Do you guys already do this? The memory usage is increased, but there isn't anything indicating the system is anywhere close to out of memory, so when this happens, I would speculate it being a 'peak' allocation which happens right at the beginning of the mine where a lot of things are loaded into memory and instantly loaded into vram, but that instant is enough to push memory usage over the top.

Maybe I'm mistaken about that. It could just be 'holding' memory too after it moves it from system memory to vram even though it wont really ever use that much memory again.

Quote from: sp_ on July 09, 2015, 08:09:14 AM

Quote from: pallas on July 09, 2015, 07:51:22 AM

Quote from: sp_ on July 09, 2015, 07:49:33 AM

http://hashpower.co/ (yaamp clone) is currently paying 0.7BTC/GHASH for quark. Have anyone tried this pool?

I did, and didn't get my payments because they used a too low transaction fee.
Furthermore, payments have been stopped AFAICS.

I will try it out, with payouts in DASH

Neoscrypt 20BTC/GHASH

Looks as though that's about what Nicehash is currently paying.

Also keep in mind because it's such a small pool this could be finders 'luck' and not a expected payout. Basically the pool got lucky finding blocks and it's not big enough to give you a realistic representation of payouts.

sp_

legendary

Activity: 2926

Merit: 1087

Team Black developer

Quote from: pallas on July 09, 2015, 07:51:22 AM

Quote from: sp_ on July 09, 2015, 07:49:33 AM

http://hashpower.co/ (yaamp clone) is currently paying 0.7BTC/GHASH for quark. Have anyone tried this pool?

I did, and didn't get my payments because they used a too low transaction fee.
Furthermore, payments have been stopped AFAICS.

I will try it out, with payouts in DASH

Neoscrypt 20BTC/GHASH

pallas

legendary

Activity: 2716

Merit: 1094

Black Belt Developer

Quote from: sp_ on July 09, 2015, 07:49:33 AM

http://hashpower.co/ (yaamp clone) is currently paying 0.7BTC/GHASH for quark. Have anyone tried this pool?

I did, and didn't get my payments because they used a too low transaction fee.
Furthermore, payments have been stopped AFAICS.

Topic: CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 1042. (Read 2347601 times)