New ccMiner 1.5.1-git by tpruvot With Lyra2RE Support Available

11 Dec
2014

ccminer-1-5-1-tpruvot

We have compiled a new windows binary from the latest ccMiner 1.5.1-git by tpruvot (source) that comes with support for the Lyra2RE algorithm – the algorithm that VertCoin (VTC) is going to be forking to. There is an official ccMiner fork with support for Lyra2RE available, but it seems that tpruvot’s fork is slightly better in terms of performance. We are getting about 1180 KHS with a GTX 980 GPU running at stock frequencies and around 1140 KHS with the official Lyra2RE ccMiner on the same hardware. Do note that this is not an official 1.5.1 release from tpruvot, we have just compiled a binary form the latest source in order to be able to test Lyra2RE mining performance on the available test pool. Our windows binary is compiled with VS2013 and supports Compute 3.0, 3.5, 5.0 and 5.2 GPUs, you can download and try it below, there is an example BAT file setup for testing hashrate at the Hashlink’s test VTC pool with Lyra2RE support.

To download the latest ccMiner version 1.5.1-git by tpruvot for Windows OS (CUDA 6.5)…



Other Similar Publications:

6 Responses to New ccMiner 1.5.1-git by tpruvot With Lyra2RE Support Available

d

December 11th, 2014 at 18:28

For those wondering. 750Ti getting about 593 kH/s and 615 kH/s (two different cards, both stock). Latest tpruvot build (12.7.14), all default settings, Linux.

Thanks to hashlink for the test pool.

[reposted comment from earlier VTC test pool]

d

December 13th, 2014 at 06:09

sp-hash (src) has merged with tpruvot’s 1.5.1 branch. The merge doesn’t seem complete b/c compiling doesn’t work without some edits (missing files [Algo256, Lyra, and Heavy if I am remembering correctly… though I was able to copy from tpruvot’s] and some code in x11/x11.cu and x13/x13.cu is erroring due to “undefined reference of cuda_jh512Keccak512_cpu_hash_64…”) [this might not happen in a windows build, I’m compiling under linux, I believe the Makefile.am isn’t correct, but I can’t figure it out… so I just reverted the problem lines to pre-“Merged jh512 and keccak512” and success]). After a succesful compile, compared to tpruvot’s 1.5.1-git (12.7.2014) compile I’m seeing:
x11 ~+25kh/s
x13 ~+15kh/s
x14 ~+100kh/s

(These numbers are with my fiddling, so they could actually be higher if the “merged jh512 and keccak512” helped increase hash too.)

Something to keep an eye on.

d

December 13th, 2014 at 16:21

Further refinement to sp-hash ccminer fork. From today’s compile (12.13.2014):

x11 ~+90kh/s
x13 ~+90kh/s

(This is over and above the increases from last night.)

For linux compiling the makefile.am has to be edited. To line 40 add:
quark/cuda_jh512keccak512.cu

It should look like:
quark/cuda_bmw512.cu quark/cuda_quark_keccak512.cu quark/cuda_jh512keccak512.cu \

Lexxus

December 15th, 2014 at 21:52

Any support for older 2.x cards?

d

December 15th, 2014 at 23:50

Unfortunately not. From the ReadMe:

“Support for Compute 2.0 has been dropped so we can more efficiently implement new algorithms using the latest hardware features.”

Fladirmacht

January 23rd, 2015 at 00:09

Here is dirty version for compute 2.x cards. Use only Lyra@re algo..
http://www.dropbox.com/s/aii95bx9ogv0lbh/ccminer_lyra2re_compute21.zip

Leave a Reply

Your email address will not be published. Required fields are marked *

top