Add cwager_x64ff_mt multithreaded amd64 assembly solution (#1056)

Author: Kexisx

PrimeAssembly/solution_4/.gitignore

@@ -0,0 +1,3 @@
+cwager_x64ff_mt
+cwager_x64ff_mt.o
+core

PrimeAssembly/solution_4/Dockerfile

@@ -0,0 +1,11 @@
+FROM alpine:3.13
+
+RUN apk add --no-cache build-base nasm
+
+WORKDIR /opt/app
+
+COPY *.sh *.asm README.md ./
+
+RUN ./build.sh
+
+ENTRYPOINT ["./run.sh"]

PrimeAssembly/solution_4/README.md

@@ -0,0 +1,62 @@
+# cwager_x64ff_mt
+
+![Algorithm](https://img.shields.io/badge/Algorithm-base-green)
+![Faithfulness](https://img.shields.io/badge/Faithful-yes-green)
+![Parallelism](https://img.shields.io/badge/Parallel-yes-green)
+![Bit count](https://img.shields.io/badge/Bits-1-green)
+
+`cwager_x64ff_mt` is a multithreaded x86-64 NASM implementation of the base Sieve of Eratosthenes.
+
+The main source file is `cwager_x64ff_mt.asm`.
+
+Each worker thread repeatedly:
+
+- allocates a fresh sieve buffer dynamically at runtime
+- initializes that buffer from scratch
+- runs a faithful base-algorithm sieve over odd candidates only
+- discards the sieve unless it is the final completed pass for that worker
+
+The implementation keeps the benchmark state in a dynamically allocated benchmark structure and gives each worker its own dynamically allocated worker state structure.
+
+Each worker state structure contains the runtime sieve metadata and the sieve buffer pointer:
+
+- sieve size
+- derived bit count
+- derived word count
+- pass count
+- sieve buffer pointer
+
+This is intended to be the assembly equivalent of the "class containing the full state of the sieve" required by the current drag-race faithfulness rules. The sieve size and corresponding buffer are established dynamically at runtime for each worker, and each timed pass recreates a fresh sieve buffer from scratch before running the base algorithm.
+
+## Run instructions
+
+### NASM/GCC
+
+From this directory:
+
+```bash
+./build.sh
+./run.sh
+```
+
+This produces the executable `cwager_x64ff_mt` from `cwager_x64ff_mt.asm`.
+
+### Docker
+
+```bash
+docker build -t cwager-x64ff-mt .
+docker run --rm cwager-x64ff-mt
+```
+
+## Output
+
+Example output from this machine:
+
+```text
+cwager_x64ff_mt;128068;5.001;16;algorithm=base,faithful=yes,bits=1
+cwager_x64ff_mt;128397;5.000;16;algorithm=base,faithful=yes,bits=1
+```
+
+## Notes
+
+This implementation targets amd64/x86-64 Linux, links against pthreads, and reports using the label `cwager_x64ff_mt`.

PrimeAssembly/solution_4/arch-amd64

@@ -0,0 +1 @@
+

PrimeAssembly/solution_4/build.sh

@@ -0,0 +1,4 @@
+#!/bin/sh
+
+nasm -felf64 cwager_x64ff_mt.asm -o cwager_x64ff_mt.o
+gcc -no-pie -pthread cwager_x64ff_mt.o -o cwager_x64ff_mt