Xaltotun
|
 |
July 23, 2015, 11:51:07 PM |
|
Interesting coin, looking forward to the future. Still confused like others about total coin supply. Alpha explorer says: Total Siacoin 3.095169 GS. What does that mean? Giga-Sia, I would guess. Cool, thank you. So, just over 3 billion it is.
|
|
|
|
|
vince232
|
 |
July 24, 2015, 05:03:38 AM |
|
where can i buy this coin?"
|
|
|
|
|
|
pallas
Legendary
Offline
Activity: 2716
Merit: 1094
Black Belt Developer
|
 |
July 25, 2015, 07:43:30 AM |
|
I'm trying the gpu miner on nvidia and noticing it takes 100% cpu. It's compiled from git on ubuntu 15.04. On amd it takes less than 1%. Anyone else?
|
|
|
|
go6ooo1212
Legendary
Offline
Activity: 1512
Merit: 1000
quarkchain.io
|
 |
July 25, 2015, 09:51:39 AM |
|
I'm trying the gpu miner on nvidia and noticing it takes 100% cpu. It's compiled from git on ubuntu 15.04. On amd it takes less than 1%. Anyone else?
Indeed , that happened when the 0.3.3.3 were released. Then I needed to stop mining with my nvidias, because of the huge CPU-load...
|
|
|
|
Schleicher
|
 |
July 25, 2015, 03:19:56 PM |
|
I'm trying the gpu miner on nvidia and noticing it takes 100% cpu. It's compiled from git on ubuntu 15.04. On amd it takes less than 1%. Anyone else?
I don't know if it also happens on Linux, but on Windows all the OpenCL miners take 100% of one CPU core. You can blame Nvidia for this. I have made a simple CUDA version of the Sia miner, but currently it only compiles on Windows. https://github.com/KlausT/Sia-CUDA-MinerSomeone should probably fix the makefile, I have no clue how to do that since I don't use Linux.
|
|
|
|
pallas
Legendary
Offline
Activity: 2716
Merit: 1094
Black Belt Developer
|
 |
July 25, 2015, 03:29:29 PM |
|
I'm trying the gpu miner on nvidia and noticing it takes 100% cpu. It's compiled from git on ubuntu 15.04. On amd it takes less than 1%. Anyone else?
I don't know if it also happens on Linux, but on Windows all the OpenCL miners take 100% of one CPU core. You can blame Nvidia for this. I have made a simple CUDA version of the Sia miner, but currently it only compiles on Windows. https://github.com/KlausT/Sia-CUDA-MinerSomeone should probably fix the makefile, I have no clue how to do that since I don't use Linux. after a couple Makefile fixes, it still doesn't compile: gpu-miner.o: In function `grindNonces(unsigned int, int)': gpu-miner.cpp:(.text+0xe8): undefined reference to `std::chrono::_V2::system_clock::now()' gpu-miner.cpp:(.text+0x336): undefined reference to `nonceGrindcuda(CUstream_st*, unsigned int, unsigned long*, unsigned long*, unsigned long*, unsigned long*)' gpu-miner.cpp:(.text+0x42e): undefined reference to `std::chrono::_V2::system_clock::now()' gpu-miner.o: In function `main': gpu-miner.cpp:(.text.startup+0x34f): undefined reference to `std::chrono::_V2::system_clock::now()' gpu-miner.cpp:(.text.startup+0x366): undefined reference to `std::chrono::_V2::system_clock::now()' network.o: In function `get_header_for_work(unsigned char*, unsigned char*)': network.cpp:(.text+0x3cb): undefined reference to `lround' some defines are needed for the clock functions, don't know about the other errors. what hashrate are you getting with it?
|
|
|
|
Taek (OP)
|
 |
July 25, 2015, 03:35:09 PM |
|
Luke has written a blog post about IPv4, nat, upnp, and port forwarding: http://blog.siacoin.com/2015/07/25/why-ipv4-inhibits-decentralization/0.4.0 will support upnp, which means many of you will no longer need to manually forward your ports  I'm trying the gpu miner on nvidia and noticing it takes 100% cpu. It's compiled from git on ubuntu 15.04. On amd it takes less than 1%. Anyone else?
Indeed , that happened when the 0.3.3.3 were released. Then I needed to stop mining with my nvidias, because of the huge CPU-load... We are not sure why that started happening, but we also haven't had a lot of time to look into it. We're more concerned about building the mining pool. I'm trying the gpu miner on nvidia and noticing it takes 100% cpu. It's compiled from git on ubuntu 15.04. On amd it takes less than 1%. Anyone else?
I don't know if it also happens on Linux, but on Windows all the OpenCL miners take 100% of one CPU core. You can blame Nvidia for this. I have made a simple CUDA version of the Sia miner, but currently it only compiles on Windows. https://github.com/KlausT/Sia-CUDA-MinerSomeone should probably fix the makefile, I have no clue how to do that since I don't use Linux. Thanks for doing that. I think that we will be able to get it compiling on Linux at some point, but we're so busy with other things that I'm not sure when we will get to it.
|
|
|
|
Schleicher
|
 |
July 25, 2015, 04:09:12 PM |
|
some defines are needed for the clock functions, don't know about the other errors.
what hashrate are you getting with it?
You need to enable C++11. On my GTX970 it's about 800 MH/s, maybe a little bit more if I had better cooling.
|
|
|
|
pallas
Legendary
Offline
Activity: 2716
Merit: 1094
Black Belt Developer
|
 |
July 25, 2015, 04:34:01 PM |
|
some defines are needed for the clock functions, don't know about the other errors.
what hashrate are you getting with it?
You need to enable C++11. On my GTX970 it's about 800 MH/s, maybe a little bit more if I had better cooling. C++11 already enabled. Without that it didn't reach linking stage. 800 is more than default opencl implementation but less than 870, which I reached modding it a bit.
|
|
|
|
Schleicher
|
 |
July 25, 2015, 05:31:27 PM |
|
some defines are needed for the clock functions, don't know about the other errors.
what hashrate are you getting with it?
You need to enable C++11. On my GTX970 it's about 800 MH/s, maybe a little bit more if I had better cooling. C++11 already enabled. Without that it didn't reach linking stage. 800 is more than default opencl implementation but less than 870, which I reached modding it a bit. I have no clue about gcc. Maybe it doesn't find the right library? Or the library is too old? It's standard c++11, nothing special.
|
|
|
|
Tobo
|
 |
July 26, 2015, 01:38:26 AM Last edit: October 25, 2015, 02:37:50 AM by Tobo |
|
|
|
|
|
WilderX
|
 |
July 26, 2015, 12:12:29 PM |
|
Luke has written a blog post about IPv4, nat, upnp, and port forwarding: http://blog.siacoin.com/2015/07/25/why-ipv4-inhibits-decentralization/0.4.0 will support upnp, which means many of you will no longer need to manually forward your ports  I'm trying the gpu miner on nvidia and noticing it takes 100% cpu. It's compiled from git on ubuntu 15.04. On amd it takes less than 1%. Anyone else?
Indeed , that happened when the 0.3.3.3 were released. Then I needed to stop mining with my nvidias, because of the huge CPU-load... We are not sure why that started happening, but we also haven't had a lot of time to look into it. We're more concerned about building the mining pool. I'm trying the gpu miner on nvidia and noticing it takes 100% cpu. It's compiled from git on ubuntu 15.04. On amd it takes less than 1%. Anyone else?
I don't know if it also happens on Linux, but on Windows all the OpenCL miners take 100% of one CPU core. You can blame Nvidia for this. I have made a simple CUDA version of the Sia miner, but currently it only compiles on Windows. https://github.com/KlausT/Sia-CUDA-MinerSomeone should probably fix the makefile, I have no clue how to do that since I don't use Linux. Thanks for doing that. I think that we will be able to get it compiling on Linux at some point, but we're so busy with other things that I'm not sure when we will get to it. Damned nice progress Taek + Team! UPnP is good for mass adoption, but I don't know why that mining pool should be such a high priority? The way I see it these should be the next top priorities: 1. Password protection for the wallet (easy to do from the wallet UI) 2. Increasing file size cap (at least 4GB "DVD size", but preferably 50GB "Blu-ray" and obviously much more in near future when 4k goes viral) 3. Dropbox style easy user interface and automated file contracts renewals and not manual. (without this there will be no mass adoption). Just my thoughts, but surely you know best your own project 
|
|
|
|
pallas
Legendary
Offline
Activity: 2716
Merit: 1094
Black Belt Developer
|
 |
July 26, 2015, 01:39:02 PM |
|
some defines are needed for the clock functions, don't know about the other errors.
what hashrate are you getting with it?
You need to enable C++11. On my GTX970 it's about 800 MH/s, maybe a little bit more if I had better cooling. C++11 already enabled. Without that it didn't reach linking stage. 800 is more than default opencl implementation but less than 870, which I reached modding it a bit. I have no clue about gcc. Maybe it doesn't find the right library? Or the library is too old? It's standard c++11, nothing special. with some changes to Makefile and network.cpp I could compile it successfully on Linux: Sia-CUDA-Miner 1.06 Initializing... Device 0: GeForce GTX 970 (Compute Capability 5.2) Using device 0 new difficulty = 72164 Mining at 722.161 MH/s 0 blocks mined But I can get 870 Mh/s with the modded opencl miner, although using 100% of a cpu core (your cuda version uses almost 0%).
|
|
|
|
pallas
Legendary
Offline
Activity: 2716
Merit: 1094
Black Belt Developer
|
 |
July 27, 2015, 11:01:54 AM |
|
Since nobody asked about the cuda miner for linux or the modified opencl one, I assume nobody's mining Sia on nvidia.... Just to understand if it's worth working on it.
|
|
|
|
go6ooo1212
Legendary
Offline
Activity: 1512
Merit: 1000
quarkchain.io
|
 |
July 27, 2015, 11:32:05 AM |
|
Since nobody asked about the cuda miner for linux or the modified opencl one, I assume nobody's mining Sia on nvidia.... Just to understand if it's worth working on it.
Actually I am... I found Nvidia miners recently. In fact they are working fine to me... 320 MH/s on 750ti
|
|
|
|
pallas
Legendary
Offline
Activity: 2716
Merit: 1094
Black Belt Developer
|
 |
July 27, 2015, 11:38:52 AM |
|
Since nobody asked about the cuda miner for linux or the modified opencl one, I assume nobody's mining Sia on nvidia.... Just to understand if it's worth working on it.
Actually I am... I found Nvidia miners recently. In fact they are working fine to me... 320 MH/s on 750ti The cuda version can surely be optimised: it should at least match the opencl one, as of hashrate. Now If only the opencl version wouldn't eat a full cpu core....
|
|
|
|
go6ooo1212
Legendary
Offline
Activity: 1512
Merit: 1000
quarkchain.io
|
 |
July 27, 2015, 12:20:44 PM |
|
Yeah the CPU is cold now...
|
|
|
|
|