Seems people are fine with the 50% neutered miner for a week, and then full open-source release of the miner, and that's my favorite option since people can start mining today, so that's what we're going to do!

First thing's first, the download link: http://www.filedropper.com/pascalcuda

There are some incredibly important instructions though, since it's not obvious how to use, and there are a lot of caveats to the miner:
1. You must already have PascalCoin installed. If you don't have it, download it from sourceforge here: https://sourceforge.net/projects/pascalcoin/. Once it is installed, run the PascalCoinWallet.exe provided in the download.
2. You must be using a 256-bit secp256k1 key. This is the default behavior of the PascalCoin wallet.
3. Your miner name must be exactly 10 characters long. The miner expects that the input is exactly 176 total bytes (which is achieved by using a secp256k1 key and a 10-character name)
4. You must have one (and only one) active mining thread in the PascalCoin wallet. This mining thread is constantly loading solutions from datain.txt and attempting to mine a block with them.
5. You must run the GPU miner from the same directory as PascalCoinWallet.exe. If you run PascalCoinWallet.exe from the extracted PascalCuda.zip file and the PascalCoinCUDA_smXX.exe miner from the same location, it will work.

You can mine with multiple GPUs, however it might take a bit of cleverness. The miner accepts a single argument, a number, which determines which device the miner mines on. However, mining with multiple GPUs to one instance of a PascalCoin wallet will create duplicate work. A future version of the miner could avoid this by allowing each miner to use a different miner name, or something. Work in progress. Probably the easiest way to mine with multiple GPUs is to run separate copies of the PascalCoin wallet under different Windows users (and have each with either a different miner name, or a different private key being used for mining), and use a different GPU for each one.

There are a lot of different versions of the miner, compiled for different compute architectures (3.0 through 6.1). To find out which one your NVidia card supports, see: https://en.wikipedia.org/wiki/CUDA#Version_features_and_specifications

You may find that a version with a lower compute version than your card's compute version performs better for whatever reason. The only restriction is that you can't run a version with a higher compute capability than your card supports.

You will see a folder called 'Benchmarks' with a bunch of .exe files in it. This contains a standard headerout.txt file that will never change (unless you change it) that you can run all of the different smXX versions against to find out which is the fastest on your system. The benchmarker doesn't do full mining, but the performance of the benchmarker should be the same as the regular program. The benchmarker basically uses 'low-difficulty' shares to get a more accurate picture of hashrate. The benchmarker and standard miner are both neutered in the exact same fashion, so the benchmarkers should be accurate indicators of actual mining performance.

You will see that the benchmarker (and occasionally regular miner) will report the same nonce twice or more. This means there is duplicate work, an issue I'm still tracking down in my code.

Each time the actual miner (not the benchmarker) prints out a hash and a "Found nonce: ..." line, it found a solution that would work for a difficulty-20 block (AKA target = 20000000). On average, it would take 16 of these 'shares' to crack a block at a target of 25000000, or slightly less than 32 of these 'shares' to crack a block at the current target (256E5937).

If you find a way to compromise the neutering of the miner and make it run at full speed, congrats! It's probably not too difficult, and you're more than free to try.

Current forum time is October 09, 2016 8-ish PM. I will release the unneutered version and complete source-code sometime before October 16, 2016 8-ish PM.

Also since some people are complaining about this reducing the value of the coin: GPU miners are a natural progression of virtually every PoW cryptocurrency. By not making one, we're leaving the coin completely vulnerable to someone who does (and who could then freeze the network, strand it on a crazy difficulty, constantly fork it, etc.) By having someone like me make a GPU miner and releasing a 50% speed miner within about 5 hours of initial GPU mining and a full-speed miner a week later, the network is protected against a bad actor making a private GPU miner and harming the network. If anything, having a GPU miner should increase the value of the coin since it will soon be ~20x more secure (the GPU miner is roughly 20x more cost effective than equivalent CPUs). Also, this opens up the potential for a pool to be developed, and for an even faster OpenCL version to be created.

Very important!
I've detected a BUG when trying to download blockchain caused by blocks 17729,17731 and 17763, because they have a lot of operations included (more than 15000 operations!) and downloading algorithm does not work properly.

Today I'll publish a new Build correcting this issue, but meanwhile, if you cannot connect, try downloading BlockChain from WeTransfer and copy it in PascalCoin database folder (C:\Users\(My User)\AppData\Roaming\PascalCoin)

Download BlockChain:  https://we.tl/zdFspbXDUk

Note: This issue only affects users trying to download blockchain when starting Application after installation or after some hours off, if your blockchain is over Block 17763 you don't need to download from link.

Same here, I did not find any block since the last 12 hours.

1.) 40k G what is that figure?

2.) Each block gives you 100 coins.

3.) You need to wait for them to mature. It can be a while for this to happen, they have to turn green, you should see a balance when you select my wallet keys in accounts explorer.

4.) Mining stops sometimes and it goes "alone in the world" blockchain has not been updated and cannot connect to peers, it will fix itself in the 1.0.3 version.

5.) See answer 4.