benchmarks/tinymembench: Add new port

This is a simple memory benchmark program, which tries to measure the
peak bandwidth of sequential memory accesses and the latency of random
memory accesses. Bandwidth is measured by running different assembly
code for the aligned memory blocks and attempting different prefetch
strategies.

benchmarks/Makefile

@@ -104,6 +104,7 @@
     SUBDIR += sysbench
     SUBDIR += tcpblast
     SUBDIR += thrulay
+    SUBDIR += tinymembench
     SUBDIR += tsung
     SUBDIR += ttcp
     SUBDIR += typometer

benchmarks/tinymembench/Makefile

@@ -0,0 +1,23 @@
+PORTNAME=	tinymembench
+DISTVERSIONPREFIX=	v
+DISTVERSION=	0.4
+CATEGORIES=	benchmarks
+
+MAINTAINER=	ehaupt@FreeBSD.org
+COMMENT=	Simple benchmark for memory throughput and latency
+WWW=		https://github.com/ssvb/tinymembench
+
+LICENSE=	MIT
+LICENSE_FILE=	${WRKSRC}/LICENSE
+
+USES=		gmake
+USE_GCC=	yes
+USE_GITHUB=	yes
+GH_ACCOUNT=	ssvb
+
+PLIST_FILES=	bin/tinymembench
+
+do-install:
+	${INSTALL_PROGRAM} ${WRKSRC}/${PORTNAME} ${STAGEDIR}${PREFIX}/bin
+
+.include <bsd.port.mk>

benchmarks/tinymembench/distinfo

@@ -0,0 +1,3 @@
+TIMESTAMP = 1703098360
+SHA256 (ssvb-tinymembench-v0.4_GH0.tar.gz) = e37a2682065b2778f2fa7f22bd59472f5ca2191be48503952ae020b8f023f23d
+SIZE (ssvb-tinymembench-v0.4_GH0.tar.gz) = 13564

benchmarks/tinymembench/files/patch-Makefile

@@ -0,0 +1,40 @@
+--- Makefile.orig	2016-03-30 04:42:59 UTC
++++ Makefile
+@@ -1,29 +1,27 @@ all: tinymembench
+ all: tinymembench
+ 
+-ifdef WINDIR
+-	CC = gcc
+-endif
++CC ?= gcc
+ 
+ tinymembench: main.c util.o util.h asm-opt.h version.h asm-opt.o x86-sse2.o arm-neon.o mips-32.o aarch64-asm.o
+-	${CC} -O2 ${CFLAGS} -o tinymembench main.c util.o asm-opt.o x86-sse2.o arm-neon.o mips-32.o aarch64-asm.o -lm
++	${CC} ${CFLAGS} -o tinymembench main.c util.o asm-opt.o x86-sse2.o arm-neon.o mips-32.o aarch64-asm.o -lm
+ 
+ util.o: util.c util.h
+-	${CC} -O2 ${CFLAGS} -c util.c
++	${CC} ${CFLAGS} -c util.c
+ 
+ asm-opt.o: asm-opt.c asm-opt.h x86-sse2.h arm-neon.h mips-32.h
+-	${CC} -O2 ${CFLAGS} -c asm-opt.c
++	${CC} ${CFLAGS} -c asm-opt.c
+ 
+ x86-sse2.o: x86-sse2.S
+-	${CC} -O2 ${CFLAGS} -c x86-sse2.S
++	${CC} ${CFLAGS} -c x86-sse2.S
+ 
+ arm-neon.o: arm-neon.S
+-	${CC} -O2 ${CFLAGS} -c arm-neon.S
++	${CC} ${CFLAGS} -c arm-neon.S
+ 
+ aarch64-asm.o: aarch64-asm.S
+-	${CC} -O2 ${CFLAGS} -c aarch64-asm.S
++	${CC} ${CFLAGS} -c aarch64-asm.S
+ 
+ mips-32.o: mips-32.S
+-	${CC} -O2 ${CFLAGS} -c mips-32.S
++	${CC} ${CFLAGS} -c mips-32.S
+ 
+ clean:
+ 	-rm -f tinymembench

benchmarks/tinymembench/files/patch-aarch64-asm.S

@@ -0,0 +1,8 @@
+--- aarch64-asm.S.orig	2016-03-30 04:42:59 UTC
++++ aarch64-asm.S
+@@ -123,3 +123,5 @@ asm_function aligned_block_copy_ld1st1_aarch64
+ .endfunc
+ 
+ #endif
++
++.section .note.GNU-stack,"",@progbits

benchmarks/tinymembench/files/patch-arm-neon.S

@@ -0,0 +1,8 @@
+--- arm-neon.S.orig	2016-03-30 04:42:59 UTC
++++ arm-neon.S
+@@ -473,3 +473,5 @@ asm_function aligned_block_copy_vfp
+ .endfunc
+ 
+ #endif
++
++.section .note.GNU-stack,"",@progbits

benchmarks/tinymembench/files/patch-mips-32.S

@@ -0,0 +1,8 @@
+--- mips-32.S.orig	2016-03-30 04:42:59 UTC
++++ mips-32.S
+@@ -181,3 +181,5 @@ asm_function aligned_block_copy_pf32_mips32
+ .endfunc
+ 
+ #endif
++
++.section .note.GNU-stack,"",@progbits

benchmarks/tinymembench/files/patch-x86-sse2.S

@@ -0,0 +1,8 @@
+--- x86-sse2.S.orig	2016-03-30 04:42:59 UTC
++++ x86-sse2.S
+@@ -243,3 +243,5 @@ asm_function aligned_block_fill_nt_sse2
+ /*****************************************************************************/
+ 
+ #endif
++
++.section .note.GNU-stack,"",@progbits

benchmarks/tinymembench/pkg-descr

@@ -0,0 +1,5 @@
+This is a simple memory benchmark program, which tries to measure the
+peak bandwidth of sequential memory accesses and the latency of random
+memory accesses. Bandwidth is measured by running different assembly
+code for the aligned memory blocks and attempting different prefetch
+strategies.