Beam Hash III !Why is it the best and last algo change for beam?
Explanations Optimizations and Mining Setup
Beam Hash III has finally hit the main-net! This change brought a lot of optimizations for both Nvidia and AMD gpu’s. Thanks @LOLIDIEB! These changes include optimizations to saturate the memory controllers for all hardware that would mine beam-hash III.
For FPGA’s and ASICS this means that the memory requirement saturates to the point that its unfeasible and not profitable to produce an ASIC currently in a single chip design. For FPGA’s someone could take the steps to produce a bit stream; however, it would saturate the controllers to the point that its not any more profitable then a GPU. It would be cheaper to just buy gpu’s instead. Asics will inevitably be created for any network; however, the best method is to secure a network in such a way that introduction strengthens a coin instead of hurting it. This is exactly the appropriate measures taken with Beam Hash III
Currently there are four types of mining hardware!
Some of these are looked at by the mining community with disgust. They see FPGA’s and ASICs as plagues on crypto networks. While this is true under certain circumstances, in other situations its the complete opposite. There are certain network conditions for which direction this might go. There will be more on this sujbect later but for now lets talk about the types of mining hardware!
There are four types of commonly used hardware in crypto algo proof of work. Cpu’s, Fpga’s, GPU’s, Asics!
CPU
Cpu’s are the most decentralized piece of hardware. However mining on them leaves the computer unable to be utilized and from this perspective its not good to produce a CPU efficient algo. This creates a situation where the only hash available is from data centers .This hash rate ultimately gets sold on nice hash. People cant use their computers while they are mining and this results in the entire computer becoming inefficient for everyday use. For network stability a network needs to not have hash-rate cut on and off. On cpu networks,sadly someone ultimately produces a GPU algo in secret.
GPU
Graphic cards are the defacto secondary decentralized piece of hardware. Most crypto algorithms start out on graphic cards and with good reason. You can continue to use your computer while the computer is mining. Floating point performance is vastly superior per watt to cpu’s and much cheaper per diesize to produce then an FPGA.
For example the upcoming 3080 TI is approximately 21 teraflops of (fp32 performance) The upcoming 6900xt is 21 teraflops of (fp32 performance)and both are to be set to be produced on 7 nanometers. One of the current FPGA’s that are used for mining is the Squirrel Research FPGA and it is 16.5 teraflops of FP32 performance if use of the full Gate array for FP32 Math. Gpu’s are used in Large Data centers for Computation’s They are cheaper then FPGA when all is said and done. On a bad algo graphics cards become exploitable as someone produces an FPGA bit stream in secret
FPGA
(FPGAs) are semiconductor devices that are based around a matrix of configurable logic blocks (CLBs) connected via programmable interconnects. They contain an extremely configurable circuit that can be programmed at the hardware level . This leads to extremely efficient circuit design. There are even some FPGA’s that can compile OPEN CL code the same as a GPU. However the resulting performance would all depend upon certain criteria. If the algo can be streamlined you can create an FPGA bit stream that is twenty times as fast per watt. This does not make FPGA’s bad. It just makes them exploit the algo! More on this and why Fpga’s and Asic’s are good and bad as well Later. Some algo’s can not be streamlined for example CN-GPU is full NIST compatible Floating point math. The resulting bitstream is 16.5 teraflops. The same as a highend GPU. Thus the hashvalue is the same.
ASICS
An application-specific integrated circuit (ASIC /ˈeɪsɪk/) is an integrated circuit (IC) chip customized for a particular use, rather than intended for general-purpose use. Roughly speaking they use a laser and etch teh circuit needed to produce a chip capable of pow for the algo directly on a silicon wafer. This results in the most optimized and efficient piece of tech at a given nano meter!
If you have a CPU, GPU, FPGA, and ASIC produced on the same nanometer and programmed for the same algo and it was exploitable the cpu would perform at a specific rate. The gpu would perform at a ratio a multitude faster then that specific rate. The FPGA would demolish the CPU and GPU by a factor of a tremendous number by hundreds of times. . The Asic would be three times to ten times faster then the FPGA.
This results in an ASIC that is 1000 times faster then a CPU.
This might seem like a problem and actually is on algo’s that were not planned out are once that were created to cater to specific hardware. You cant try to cater to them completely only perfer certain hardware are you create an issue where it could become centralized . If you cater to CPU’s only for example the above mentioned method would happen. The most decentralized piece of hardware is the GPU and the most efficient is the ASIC. It makes since to transition from graphics cardsto ASIC. But their has to be a situation created in such a way that the transition is smooth because its is inevitable.
The reason people use different equipment is because of different profitability. You can calculate this value via hash value to determine if its profitable are not.
Hashvalue
Current_Hashvalue =(blockreward*exchange_price)/(Network_hashrate*block_target_time)
What this does is take the coins emission curve, the exchange price, the network hashrate, and block target time. This tells you how much each solution a second is worth on the network based on current exchange price. For every hash per second your getting a piece of emission per second.
Roughly speaking you want all hardware to have as close of a hashvalue as possible on a good algo.
If a piece of mining equipment is ten times as profitable then what happens is the hash value is ten times more then the hash value for another given piece of equipment
With this knowledge lets explain the purpose of beamhash III and why beam has done an amazing thing. One that is not understood by the majority of crypto.
What makes an algo exploitable is the centralization of different hardware. Its okay to have a higher profit and different hash value . Its not okay to be centralized.
For example.
ONE GPU MINER with a large percentage of hash on a CPU NETWORK…
ONE FPGA MINER with a large percentage of hash on a GPU NETWORK
ONE ASIC MINER with a large percentage of hash on a GPU NETWORK.
This would be really bad and unhealthy. Do not let this be confused with Nice hash as most rigs renting are indeed GPU’S and their resulting hashvalue is the same as others on BEAMHASHIII
However from this aspect nicehash is fine. However Nicehash does hurt other aspects of network security from a 51 percent hash rate as it can put control of the network in the hands of one miner who might have nefarious purposes. But this can happen on any network and as the network grows the likely hood of this happening is less.
What are the good aspects of FPGA’s and ASICS?
Well roughly speaking they are more efficient and use less power. Their are a lot of POS networks starting up because of energy efficiency. The whole planet is going green.
However POW works by having a given cost on a specific coin. For example Bit mains current most efficient 110TH bitcoin miner cost 3396.62 a year to run and makes 3369.60 a year in BTC. The minimum price new emission will sell for would be whatever a miner can get in profit in between this price .9,000 and a few percent… 9500 USD. This is one of the reasons the price of BTC is staying around this price. This is the minimum cost of emission anything dumping below that was previously cheaper minted emission getting dumped. But the price will rise back up the stock flow and current cost to produce. This is also why asic’s are good for a network. People keep mining with them because that is all they can mine with them and they know the price will rise back up to that cost if all new emission costs the minimum cost in electricity it takes to mint a coin. Read this paragraph twice if you didn't catch what I said the first time!
Because of this reason asics can either destroy a coin or they can make the price rise much higher then any other type of coin that uses POW. If there were all CPU’s on a network the minimum cost to mint a coin would be the emission curve and total CPU watt dissipation.
ASICS are inevitable! If embraced correctly and an algo created that streamlines hashvalue on GPU’s and holds off ASICs until a time when it is inevitable a coin can give a decentralized emission to everyone who joins. This can happen at the same time allowing for asics to be made when the tech is right.
Beamhash III
A good algo will be designed in such a way that CPU’s are relevant but then quickly irrelevant once a GPU pow has been made. This allows for people to use their computer while letting their GPU mine in the background.
Also Either resist and make unprofitable FPGA’s are Embrace them and give them the Same hash value as a GPU where each watt dissipated results in a similar hash-rate!
Beam hash II transitioned to Beam hash III in an attempt to create a situation where the hash value is as close to possible (excluding mining software optimizations) This creates the situation where each watt dissipated is as close as possible per given die size at a given nano-meter of lithography when each piece of hardware is produced.
Beam hash III attempts to do both methods of evasion to FPGA’s. This makes them unprofitable because a bit stream was made the hash value would be close as possible to a GPU then most other algo’s. This would put negative stipulations on even attempting it.
The good thing about beam hash III is that it was made in such a way that its simple to make on an ASIC but its also put a memory saturation on it to the point that it is unfeasible to make at this time and gives a good 24 months at least IMHO before possible.
This creates the situation where FPGA’s are not feasible. GPU’s are practical and have the same hash value. Asic’s are pushed out but once practical to make will transition over nicely as they are all but inevitable
Some of these changes included a memory allocation by default of around six gigs! However “roughly speaking” by shuffling calculations around its possible to squeeze a smaller kernel into some graphic processing units.
The Kernel is really just the compiled Open Cl code and resulting scratchpad loaded onto a GPU for AMD. For NVIDIA it is also the resulting CUDA output and resulting Scratchpad!
Most notably at Launch Mini Z was the well known NVIDIA miner was able to make the scratchpad fit in 2 gigs. Now this doesn't stop the GPU from being saturated and a resulting better hash value. it just allows the GPU to mine. AMD openCL miner LOLIMINER managed to squeeze it down to 3gigs/4gigs for certain GPUS.
The resulting hashvalue of the current network is 4,588 satoshis cost and they are selling for 4,531 each.
Bare in mind not all of the network has optimized their miners yet.
Below is some steps to help you optimize your miner for AMD. I am not familiar with NVIDIA software. So I am only going to comment on what i feel I am adequate at!
If you have questions are need help please feel free!
Join Beam Community Mining.
https://t.me/BeamMiners
For AMD to Optimize your rig sometimes its necessary to start over. In Linux its straight forward. But on Windows its sometimes difficult due to driver updates. Here are the steps necessary to keep windows from messing up your mining rig.
STEP 1
System>Advance system settings>Hardware> Device Installation Settings
Once Disabling this allow windows to update 100 percent completely. There are many situations where people feel windows updates break your mining rig. The truth is 99 percent of people do not uncheck this box!
Step 2
Delete C:/AMD folder
When reinstalling drivers sometimes windows will keep a copy of this and any bluescreen you have for a complete unrelated reason will cause a rollback to the most “Microsoft compatible” driver in that folder. Sometimes that driver contains an equivalent older OPENCL version and your system path is updated to the older OPENCL version Reluctantly.
Step 3
Reboot in to safe mode and “DDU”
DDU Uninstalls ALL AMD DRIVERS
https://www.guru3d.com/files-details/display-driver-uninstaller-download.html
At this point please install latest drivers from AMD. DO NOT INSTALL THE OPTIONAL. ONLY INSTALL THE WHQL VERSION, otherwise windows will default back to C:/AMD and look for the most compatible version!
https://www.amd.com/en/support
If you have modified bios make sure that you run a pixel patcher on it before resetting
Once Reset Change over to Compute mode.
Please download the best AMD miner I have found from here.
https://github.com/Lolliedieb/lolMiner-releases/releases/download/1.01/lolMiner_v1.01_Lin64.tar.gz
https://github.com/Lolliedieb/lolMiner-releases/releases/download/1.01/lolMiner_v1.01_Win64.zip
In the future when these releases are older you can find his default github here
To run the miner its very simple.
lolMiner.exe — coin BEAM-III — tls 1 — pool stratum+tcp://pool.raskul.com:1690 — user YOUR_PERMANENT_ADDRESS
Get your address from within the Beam Wallet. Set it as never expiring
You can get the latest version from the official website
Set cards to 1100 mhz and ram as high as possible.