We will be adding a new PoW algo RamHash (also known as LXRHash) from https://gitlab.com/accumulatenetwork/ecosystem/LXRMining/-/tree/main/c and documenting the process of adding a new PoW algo in #317 . It will be able to be used in HSC's by using the -ac_algo=ramhash option . This issue will keep track of any issues we run into in doing this.
Add ramhash option to -ac_algo
Create new RamHashMiner class
Import RamHash code into our codebase at src/Ramhash
Add the buiding of RamHash to our build system via it's Makefile
Use the RamHashMiner class if -ac_algo=ramhash is used
Decide if RamHash constants will depend on -ac_name
Decide if RamHash will have an optional CLI arguments
Call LXRHash::init() from miner.cpp
Call LXRHash::LxrPOW() from miner.cpp
Create an HSC that uses RamHash and verify mining works on
Linux
Mac
Windows
We will be adding a new PoW algo RamHash (also known as LXRHash) from https://gitlab.com/accumulatenetwork/ecosystem/LXRMining/-/tree/main/c and documenting the process of adding a new PoW algo in #317 . It will be able to be used in HSC's by using the `-ac_algo=ramhash` option . This issue will keep track of any issues we run into in doing this.
* [x] Add `ramhash` option to `-ac_algo`
* [x] Create new `RamHashMiner` class
* [x] Import RamHash code into our codebase at src/Ramhash
* [x] Add the buiding of RamHash to our build system via it's Makefile
* [x] Use the `RamHashMiner` class if `-ac_algo=ramhash` is used
* [x] Decide if RamHash constants will depend on `-ac_name`
* [ ] Decide if RamHash will have an optional CLI arguments
* [x] Call LXRHash::init() from miner.cpp
* [x] Call LXRHash::LxrPOW() from miner.cpp
* [ ] Create an HSC that uses RamHash and verify mining works on
* [ ] Linux
* [ ] Mac
* [ ] Windows
One thing we need to look into is if the ramhash library code we are going to use can support multiple threads. If not, we can just support mining with a single thread at first and figure that out later.
One thing we need to look into is if the ramhash library code we are going to use can support multiple threads. If not, we can just support mining with a single thread at first and figure that out later.
ramhash branch now compiles ramhash in build.sh and links it when compiling hushd .
Currently the code coredumps when the RamhashMiner function is called.
(gdb) bt
#0 0x000055fab002d1c0 in RamhashMiner (pwallet=0x55fab218ce70) at miner.cpp:1504
#1 0x000055fab04c702b in thread_proxy ()
#2 0x00007efc9cffa609 in start_thread (arg=<optimized out>) at pthread_create.c:477
#3 0x00007efc9cf1f293 in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:95
It's not clear yet exactly what is causing this coredump, but some of the code does still need to change. None of the other mining algorithms does anything relating to the disk but Ramhash has a 1GB bytemap file that it needs to work with.
`ramhash` branch now compiles ramhash in build.sh and links it when compiling hushd .
Currently the code coredumps when the `RamhashMiner` function is called.
```
(gdb) bt
#0 0x000055fab002d1c0 in RamhashMiner (pwallet=0x55fab218ce70) at miner.cpp:1504
#1 0x000055fab04c702b in thread_proxy ()
#2 0x00007efc9cffa609 in start_thread (arg=<optimized out>) at pthread_create.c:477
#3 0x00007efc9cf1f293 in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:95
```
It's not clear yet exactly what is causing this coredump, but some of the code does still need to change. None of the other mining algorithms does anything relating to the disk but Ramhash has a 1GB bytemap file that it needs to work with.
duke
We will be adding a new PoW algo RamHash (also known as LXRHash) from https://gitlab.com/accumulatenetwork/ecosystem/LXRMining/-/tree/main/c and documenting the process of adding a new PoW algo in #317 . It will be able to be used in HSC's by using the
-ac_algo=ramhashoption . This issue will keep track of any issues we run into in doing this.ramhashoption to-ac_algoRamHashMinerclassRamHashMinerclass if-ac_algo=ramhashis used-ac_nameRamHash uses a psuedorandom number generator called Xorshift https://en.wikipedia.org/wiki/Xorshift
This is being worked on in the
ramhashbranchOne thing we need to look into is if the ramhash library code we are going to use can support multiple threads. If not, we can just support mining with a single thread at first and figure that out later.
Ramhash C++ implementation is now added on the
ramhashbranch atsrc/Ramhashramhashbranch now compiles ramhash in build.sh and links it when compiling hushd .Currently the code coredumps when the
RamhashMinerfunction is called.It's not clear yet exactly what is causing this coredump, but some of the code does still need to change. None of the other mining algorithms does anything relating to the disk but Ramhash has a 1GB bytemap file that it needs to work with.