Merge pull request #2 from mumax/master

update my fork to match the mumax git

.gitignore

@@ -19,3 +19,4 @@ mumax3.*linux*cuda*
 *.*~
 tmp/
 *.out
+.idea/

.travis.yml

@@ -0,0 +1,16 @@
+language: go
+dist: trusty
+sudo: required
+install: true
+env:
+  global:
+    - GOARCH=amd64
+before_install:
+  - wget http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1404/x86_64/cuda-repo-ubuntu1404_8.0.44-1_amd64.deb
+  - sudo dpkg -i cuda-repo-ubuntu1404_8.0.44-1_amd64.deb
+  - sudo apt-get -qq update
+  - sudo apt-get install cuda -y
+script:
+  - go get -u github.com/barnex/fftw 
+  - go get -u github.com/barnex/matrix
+  - go build ./...

README.md

@@ -1,5 +1,6 @@
 mumax3 
 ======
+[![Build Status](https://travis-ci.org/mumax/3.svg?branch=master)](https://travis-ci.org/mumax/3)
 
 GPU accelerated micromagnetic simulator.
 
@@ -58,7 +59,7 @@ Your binary is now at `$GOPATH/bin/mumax3`
 To do all at once on Ubuntu:
 ```
 sudo apt-get install git golang-go gcc nvidia-cuda-toolkit nvidia-cuda-dev nvidia-340 gnuplot
-export GOPATH=$HOME go get -u -v github.com/mumax/3/cmd/mumax3
+GOPATH=$HOME go get -u -v github.com/mumax/3/cmd/mumax3
 ```
 
 Contributing

bench/gpus.gplot

@@ -1,16 +1,16 @@
 #! /usr/bin/gnuplot
 
-set term pdf size 4in, 2in;
+set term pdf size 4in, 3in;
 set output "gpus.pdf"
 set boxwidth 0.5
 set style fill solid
 set key off
 set ylabel "throughput (M cells/s)"
-set xtics rotate by -45
+set xtics rotate by -90
 #set xtics out offset 0,-1.2
 
-set yrange[0:200]
-plot "gpus.txt" u ($0+1):($2/1e6):xtic(4) w boxes, "oommf4M.txt" u (0):(4*$1**2 * $2 /$3/1e6):xtic("OOMMF(CPU)") w boxes 
+set yrange[0:350]
+plot "gpus.txt" u ($0+1):($2/1e6):xtic(4) w boxes, "oommf4M.txt" u (0):(4*$1**2 * $2 /$3/1e6):xtic("OOMMF(CPU)") w boxes
 
 set output
 

bench/gpus.txt

@@ -1,12 +1,23 @@
 4.194304e+06 2.0895753045290217e+07 1.6945210251645377e-14  "GTX 650M"
-4.194304e+06 6.696363509497398e+07 1.6945210251645377e-14	"M2070"
+1.048576e+06 4.033703669606458e+07 1.6946205941339026e-14 "GT 755M (iMac 2013)"
+4.194304e+06 5.527840160261479e+07 1.6945213680693907e-14   "GTX 860M"
+4.194304e+06 6.696363509497398e+07 1.6945210251645377e-14       "TESLA M2070"
+4.194304e+06 6.945448262641394e+07 1.6945209171548022e-14   "TESLA 2050"
 4.194304e+06 8.771628448955536e+07 1.5722384085696076e-14   "GTX 660"
-4.194304e+06 1.1436826272529118e+08 1.5722384085696076e-14  "GTX 680"
+4.194304e+06 9.70511185846317e+07 1.6945209171548022e-14 "Quadro K4200"
 4.194304e+06 1.1248743156367055e+08 1.5722384085696076e-14  "GTX 480"
-4.194304e+06 1.2765247865136869e+08 1.5722384085696076e-14  "GTX 580"
+4.194304e+06 1.1436826272529118e+08 1.5722384085696076e-14  "GTX 680"
 4.194304e+06 1.205878382259874e+08 1.6945213680693907e-14   "GTX 970"
-4.194304e+06 1.3222269984079185e+08 1.6945213680693907e-14  "GTX 980"  
-4.194304e+06 1.3064462210481596e+08 1.6945210251645377e-14  "K20XM"
+4.194304e+06 1.2765247865136869e+08 1.5722384085696076e-14  "GTX 580"
+4.194304e+06 1.282698958240901e+08 1.6945224358062508e-14  "GTX 1060 (mobile)"
+4.194304e+06 1.3064462210481596e+08 1.6945210251645377e-14  "TESLA K20XM"
+4.194304e+06 1.3222269984079185e+08 1.6945213680693907e-14  "GTX 980"
+4.194304e+06 1.702974441584964e+08 1.6945224358062508e-14   "GTX 1070"
+4.194304e+06 1.752769108651334e+08 1.6945209171548022e-14   "GTX TITAN BLACK FE"
 4.194304e+06 1.7967269114941204e+08 1.6945210251645377e-14  "GTX TITAN"
 4.194304e+06 1.8968187897582138e+08 1.6945207552204512e-14  "GTX 1080"
-4.194304e+06 1.9961744689897743e+08 1.6945207552204512e-14	"GTX 980 Ti"
+4.194304e+06 1.9961744689897743e+08 1.6945207552204512e-14      "GTX 980 Ti"
+4.194304e+06 2.7516254149838316e+08 1.6945224358062508e-14  "GTX 1080 Ti"
+4.194304e+06 2.747775864824991e+08 1.694521108780564e-14        "GTX TITAN X (Pascal)"
+4.194304e+06 3.413751301174529e+08 1.6945224358062508e-14       "GTX TITAN Xp"
+4.194304e+06 3.6200963642011607e+08 1.6945207552204512e-14	"TESLA P100"

cmd/mumax3-convert/main.go

@@ -39,6 +39,7 @@ import (
 	"log"
 	"os"
 	"path"
+	"path/filepath"
 	"strconv"
 	"strings"
 
@@ -125,8 +126,16 @@ func main() {
 		log.Fatal("no output format specified (e.g.: -png)")
 	}
 
+	// expand wildcards which are not expanded by the shell
+	// (pointing a finger at cmd.exe)
+	var fnames []string
+	for _, input := range flag.Args() {
+		fmt.Println(input)
+		expanded, _ := filepath.Glob(input)
+		fnames = append(fnames, expanded...)
+	}
 	// read all input files and put them in the task que
-	for _, fname := range flag.Args() {
+	for _, fname := range fnames {
 		for _, outp := range wantOut {
 			fname := fname // closure caveats
 			outp := outp

cmd/mumax3-server/main.go

@@ -72,15 +72,20 @@ func main() {
 	go func() {
 		log.Println("serving at", thisAddr)
 
+		// Resolve the IPs for thisHost
+		thisIP, err := net.LookupHost(thisHost)
+		Fatal(err)
+
 		// try to listen and serve on all interfaces other than thisAddr
 		// this is for convenience, errors are not fatal.
 		_, p, err := net.SplitHostPort(thisAddr)
 		Fatal(err)
 		ips := util.InterfaceAddrs()
 		for _, ip := range ips {
 			addr := net.JoinHostPort(ip, p)
-			if addr != thisAddr { // skip thisAddr, will start later and is fatal on error
+			if !contains(thisIP, ip) { // skip thisIP, will start later and is fatal on error
 				go func() {
+					log.Println("serving at", addr)
 					err := http.ListenAndServe(addr, nil)
 					if err != nil {
 						log.Println("info:", err, "(but still serving other interfaces)")
@@ -178,7 +183,7 @@ func parseIPs() []string {
 		if len(split) != 4 {
 			log.Fatal("invalid IP address range:", s)
 		}
-		var start, stop [4]byte
+		var start, stop [4]uint
 		for i, s := range split {
 			split := strings.Split(s, "-")
 			first := atobyte(split[0])
@@ -204,13 +209,13 @@ func parseIPs() []string {
 	return IPs
 }
 
-func atobyte(a string) byte {
+func atobyte(a string) uint {
 	i, err := strconv.Atoi(a)
 	if err != nil {
 		panic(err)
 	}
 	if int(byte(i)) != i {
 		panic("too large")
 	}
-	return byte(i)
+	return uint(i)
 }

cmd/mumax3/main.go

@@ -164,4 +164,6 @@ func printVersion() {
 	fmt.Print("//", cuda.GPUInfo, ", using CC", cuda.UseCC, " PTX \n")
 	fmt.Print("//(c) Arne Vansteenkiste, Dynamat LAB, Ghent University, Belgium", "\n")
 	fmt.Print("//This is free software without any warranty. See license.txt", "\n")
+	fmt.Print("//If you use mumax in any work or publication,", "\n")
+	fmt.Print("//we kindly ask you to cite the references in references.bib", "\n")
 }

cmd/mumax3/queue.go

@@ -123,9 +123,10 @@ func run(inFile string, gpu int, webAddr string) {
 
 	cmd := exec.Command(os.Args[0], flags...)
 	log.Println(os.Args[0], flags)
-	err := cmd.Run()
+	output, err := cmd.CombinedOutput()
 	if err != nil {
 		log.Println(inFile, err)
+		log.Printf("%s\n", output)
 		exitStatus.set(1)
 		numFailed.inc()
 		if *flag_failfast {

cuda/copypadmul2_20.ptx

@@ -0,0 +1,125 @@
+//
+// Generated by NVIDIA NVVM Compiler
+// Compiler built on Thu Jul 18 02:37:37 2013 (1374107857)
+// Cuda compilation tools, release 5.5, V5.5.0
+//
+
+.version 3.2
+.target sm_20
+.address_size 64
+
+	.file	1 "/home/ragnarok/go/src/github.com/mumax/3/cuda/copypadmul2.cu", 1508839776, 830
+
+.visible .entry copypadmul2(
+	.param .u64 copypadmul2_param_0,
+	.param .u32 copypadmul2_param_1,
+	.param .u32 copypadmul2_param_2,
+	.param .u32 copypadmul2_param_3,
+	.param .u64 copypadmul2_param_4,
+	.param .u32 copypadmul2_param_5,
+	.param .u32 copypadmul2_param_6,
+	.param .u32 copypadmul2_param_7,
+	.param .u64 copypadmul2_param_8,
+	.param .f32 copypadmul2_param_9,
+	.param .u64 copypadmul2_param_10
+)
+{
+	.reg .pred 	%p<8>;
+	.reg .s32 	%r<22>;
+	.reg .f32 	%f<15>;
+	.reg .s64 	%rd<17>;
+	.reg .f64 	%fd<3>;
+
+
+	ld.param.u64 	%rd7, [copypadmul2_param_0];
+	ld.param.u32 	%r5, [copypadmul2_param_1];
+	ld.param.u32 	%r6, [copypadmul2_param_2];
+	ld.param.u64 	%rd8, [copypadmul2_param_4];
+	ld.param.u32 	%r7, [copypadmul2_param_5];
+	ld.param.u32 	%r8, [copypadmul2_param_6];
+	ld.param.u32 	%r9, [copypadmul2_param_7];
+	ld.param.u64 	%rd5, [copypadmul2_param_8];
+	ld.param.f32 	%f13, [copypadmul2_param_9];
+	ld.param.u64 	%rd6, [copypadmul2_param_10];
+	cvta.to.global.u64 	%rd1, %rd7;
+	cvta.to.global.u64 	%rd2, %rd8;
+	cvta.to.global.u64 	%rd3, %rd6;
+	cvta.to.global.u64 	%rd4, %rd5;
+	.loc 1 14 1
+	mov.u32 	%r10, %ntid.x;
+	mov.u32 	%r11, %ctaid.x;
+	mov.u32 	%r12, %tid.x;
+	mad.lo.s32 	%r1, %r10, %r11, %r12;
+	.loc 1 15 1
+	mov.u32 	%r13, %ntid.y;
+	mov.u32 	%r14, %ctaid.y;
+	mov.u32 	%r15, %tid.y;
+	mad.lo.s32 	%r2, %r13, %r14, %r15;
+	.loc 1 16 1
+	mov.u32 	%r16, %ntid.z;
+	mov.u32 	%r17, %ctaid.z;
+	mov.u32 	%r18, %tid.z;
+	mad.lo.s32 	%r3, %r16, %r17, %r18;
+	.loc 1 18 1
+	setp.lt.s32	%p1, %r1, %r7;
+	setp.lt.s32	%p2, %r2, %r8;
+	and.pred  	%p3, %p1, %p2;
+	.loc 1 18 1
+	setp.lt.s32	%p4, %r3, %r9;
+	and.pred  	%p5, %p3, %p4;
+	.loc 1 18 1
+	@!%p5 bra 	BB0_7;
+	bra.uni 	BB0_1;
+
+BB0_1:
+	.loc 1 19 1
+	mad.lo.s32 	%r19, %r3, %r8, %r2;
+	mad.lo.s32 	%r4, %r19, %r7, %r1;
+	.loc 1 20 1
+	setp.eq.s64	%p6, %rd5, 0;
+	@%p6 bra 	BB0_3;
+
+	mul.wide.s32 	%rd9, %r4, 4;
+	add.s64 	%rd10, %rd4, %rd9;
+	.loc 1 20 1
+	ld.global.f32 	%f7, [%rd10];
+	mul.f32 	%f13, %f7, %f13;
+
+BB0_3:
+	cvt.f64.f32	%fd1, %f13;
+	mul.f64 	%fd2, %fd1, 0d3EB515370F99F6CB;
+	cvt.rn.f32.f64	%f3, %fd2;
+	.loc 1 21 1
+	setp.ne.s64	%p7, %rd6, 0;
+	@%p7 bra 	BB0_5;
+
+	mov.f32 	%f14, 0f3F800000;
+	bra.uni 	BB0_6;
+
+BB0_5:
+	mul.wide.s32 	%rd11, %r4, 4;
+	add.s64 	%rd12, %rd3, %rd11;
+	.loc 1 21 1
+	ld.global.f32 	%f8, [%rd12];
+	mul.f32 	%f14, %f8, 0f3F800000;
+
+BB0_6:
+	mul.wide.s32 	%rd13, %r4, 4;
+	add.s64 	%rd14, %rd2, %rd13;
+	.loc 1 22 1
+	ld.global.f32 	%f10, [%rd14];
+	mul.f32 	%f11, %f3, %f14;
+	mul.f32 	%f12, %f11, %f10;
+	mad.lo.s32 	%r20, %r3, %r6, %r2;
+	mad.lo.s32 	%r21, %r20, %r5, %r1;
+	mul.wide.s32 	%rd15, %r21, 4;
+	add.s64 	%rd16, %rd1, %rd15;
+	.loc 1 22 1
+	st.global.f32 	[%rd16], %f12;
+
+BB0_7:
+	.loc 1 24 2
+	ret;
+}
+
+