From aaf77c78f66fae093a684cb4a9e22b368399d8f6 Mon Sep 17 00:00:00 2001
From: Armin Friedl <friedl.armin@gmail.com>
Date: Fri, 24 Jun 2016 03:45:10 +0200
Subject: [PATCH] added logging, added fastmath undermining for simple8

---
 plot/plot.py                               |   1 +
 roofline/src/Makefile                      |  17 +++-
 roofline/src/aikern.c                      |  36 ++++++--
 roofline/src/aikern.h                      |  49 +++++++++--
 roofline/src/log/fma16                     |   5 ++
 roofline/src/log/fma8                      |   5 ++
 roofline/src/log/simple16                  |   5 ++
 roofline/src/log/simple8                   |   5 ++
 roofline/src/log/simple8fastmath           |   5 ++
 roofline/src/roofline                      | Bin 31088 -> 0 bytes
 roofline/src/roofline.c                    |  92 +++++++++++++++++++--
 roofline/src/roofline_fma                  | Bin 18800 -> 0 bytes
 roofline/src/roofline_fma_fast_fastmath_o3 | Bin 22872 -> 0 bytes
 roofline/src/roofline_fma_fast_o3          | Bin 22872 -> 0 bytes
 roofline/src/roofline_fma_o3               | Bin 22872 -> 0 bytes
 roofline/src/roofline_o3                   | Bin 26968 -> 0 bytes
 16 files changed, 202 insertions(+), 18 deletions(-)
 create mode 100644 roofline/src/log/fma16
 create mode 100644 roofline/src/log/fma8
 create mode 100644 roofline/src/log/simple16
 create mode 100644 roofline/src/log/simple8
 create mode 100644 roofline/src/log/simple8fastmath
 delete mode 100755 roofline/src/roofline
 delete mode 100755 roofline/src/roofline_fma
 delete mode 100755 roofline/src/roofline_fma_fast_fastmath_o3
 delete mode 100755 roofline/src/roofline_fma_fast_o3
 delete mode 100755 roofline/src/roofline_fma_o3
 delete mode 100755 roofline/src/roofline_o3

diff --git a/plot/plot.py b/plot/plot.py
index 14a5fd8..198739c 100644
--- a/plot/plot.py
+++ b/plot/plot.py
@@ -21,6 +21,7 @@ while i<=64:
     xlbl.append(repr(i))
     i *= 2
 
+print(xlbl)    
 # memory
 values = []
 bandwidth = 10.6
diff --git a/roofline/src/Makefile b/roofline/src/Makefile
index adb4fb0..845fe45 100644
--- a/roofline/src/Makefile
+++ b/roofline/src/Makefile
@@ -1,7 +1,7 @@
-all: bin lib
+all: clean bin lib
 
 # Roofline Binary
-bin: roofline roofline_o3 roofline_fma roofline_fma_o3 roofline_fma_fast_o3 roofline_fma_fast_fastmath_o3
+bin: roofline roofline_o3 roofline_fma roofline_fma_o3 roofline_fma_fast_o3 roofline_fma_fast_o2 roofline_fma_fast_fastmath_o3
 	mkdir bin
 	mv $^ bin
 
@@ -20,12 +20,15 @@ roofline_fma_o3: roofline.c aikern_fma_o3.a
 roofline_fma_fast_o3: roofline.c aikern_fma_fast_o3.a
 	gcc -Wall -Wextra -std=c99 -fopenmp $^ -o $@
 
+roofline_fma_fast_o2: roofline.c aikern_fma_fast_o2.a
+	gcc -Wall -Wextra -std=c99 -fopenmp $^ -o $@
+
 roofline_fma_fast_fastmath_o3: roofline.c aikern_fma_fast_fastmath_o3.a
 	gcc -Wall -Wextra -std=c99 -fopenmp $^ -o $@
 
 
 # Static Libraries
-lib: aikern.a aikern_o3.a aikern_fma.a aikern_fma_o3.a aikern_fma_fast_o3.a aikern_fma_fast_fastmath_o3.a
+lib: aikern.a aikern_o3.a aikern_fma.a aikern_fma_o3.a aikern_fma_fast_o2.a aikern_fma_fast_o3.a aikern_fma_fast_fastmath_o3.a
 	mkdir lib
 	mv $^ lib
 
@@ -49,6 +52,11 @@ aikern_fma_o3.a: aikern.c aikern.h
 	ar rcs $@ aikern_fma_o3.o
 	rm aikern_fma_o3.o
 
+aikern_fma_fast_o2.a: aikern.c aikern.h
+	gcc -Wall -Wextra -Wno-unused -O2 -mavx -mfma -fopenmp -Ofast -c -o aikern_fma_fast_o2.o $<
+	ar rcs $@ aikern_fma_fast_o2.o
+	rm aikern_fma_fast_o2.o
+
 aikern_fma_fast_o3.a: aikern.c aikern.h
 	gcc -Wall -Wextra -Wno-unused -O3 -mavx -mfma -fopenmp -Ofast -c -o aikern_fma_fast_o3.o $<
 	ar rcs $@ aikern_fma_fast_o3.o
@@ -61,6 +69,9 @@ aikern_fma_fast_fastmath_o3.a: aikern.c aikern.h
 
 # Cleanup
 clean:
+	rm -f *.a
+	rm -f *.o
+	rm -f roofline roofline_o3 roofline_fma roofline_fma_o3 roofline_fma_fast_o3 roofline_fma_fast_o2 roofline_fma_fast_fastmath_o3
 	rm -fR bin
 	rm -fR lib
 
diff --git a/roofline/src/aikern.c b/roofline/src/aikern.c
index f043d01..1068a41 100644
--- a/roofline/src/aikern.c
+++ b/roofline/src/aikern.c
@@ -25,10 +25,11 @@ kern_result kernel_dispatch(kernel_t kernel,
 							size_t size, size_t runs)
 {
   
-  kern_result result;
+  kern_result result = {0};
   result.runs = runs;
   result.starts = malloc(sizeof(double)*(runs));
   result.ends = malloc(sizeof(double)*(runs));
+  result.size = size;
 
   if(result.starts==NULL || result.ends==NULL)
 	{
@@ -41,6 +42,7 @@ kern_result kernel_dispatch(kernel_t kernel,
 
 	case SIMPLE_1_16:
 	  result.flops = 1;
+	  result.kern_name = "Simple 1/16";
 	  for(size_t r=0; r<runs; r++)
 		{
 		  result.starts[r] = pin_time();
@@ -50,6 +52,7 @@ kern_result kernel_dispatch(kernel_t kernel,
 	  break;
 	case FMA_1_16:
 	  result.flops = 2;
+	  result.kern_name = "FMA aware 1/16";
 	  for(size_t r=0; r<runs; r++)
 		{
 		  result.starts[r] = pin_time();
@@ -59,6 +62,7 @@ kern_result kernel_dispatch(kernel_t kernel,
 	  break;
 	case SIMPLE_8_1:
 	  result.flops = 128;
+	  result.kern_name = "Simple 8";
 	  for(size_t r=0; r<runs; r++)
 		{
 		  result.starts[r] = pin_time();
@@ -68,6 +72,7 @@ kern_result kernel_dispatch(kernel_t kernel,
 	  break;
 	case FMA_8_1:
 	  result.flops = 128;
+	  result.kern_name = "FMA aware 8";
 	  for(size_t r=0; r<runs; r++)
 		{
 		  result.starts[r] = pin_time();
@@ -75,6 +80,16 @@ kern_result kernel_dispatch(kernel_t kernel,
 		  result.ends[r] = pin_time();
 		}
 	  break;
+	case SIMPLE_8_1_FASTMATH:
+	  result.flops = 128;
+	  result.kern_name = "Simple 8 undermining fastmath";
+	  for(size_t r=0; r<runs; r++)
+		{
+		  result.starts[r] = pin_time();
+		  kernel_8_1_simple_fastmath(a, size);
+		  result.ends[r] = pin_time();
+		}
+	  break;
 	default:
 	  bail_out("No such kernel %s", kernel);
 	}
@@ -82,7 +97,7 @@ kern_result kernel_dispatch(kernel_t kernel,
   return result;
 }
 
-void kernel_1_16_simple(double* a, size_t size)
+inline void kernel_1_16_simple(double* a, size_t size)
 {
   #pragma omp parallel for
   for(size_t i=0; i<size; i++)
@@ -91,7 +106,7 @@ void kernel_1_16_simple(double* a, size_t size)
 	}
 }
 
-void kernel_1_16_fuseaware(double* a, double* b, double* c, size_t size)
+inline void kernel_1_16_fuseaware(double* a, double* b, double* c, size_t size)
 {
   #pragma omp parallel for
   for(size_t i=0; i<size; i++)
@@ -100,7 +115,7 @@ void kernel_1_16_fuseaware(double* a, double* b, double* c, size_t size)
 	}
 }
 
-void kernel_8_1_simple(double* a, size_t size)
+inline void kernel_8_1_simple(double* a, size_t size)
 {
   #pragma omp parallel for
   for(size_t i=0; i<size; i++)
@@ -112,7 +127,18 @@ void kernel_8_1_simple(double* a, size_t size)
   }
 }
 
-void kernel_8_1_fuseaware(double* a, size_t size)
+inline void kernel_8_1_simple_fastmath(double* a, size_t size)
+{
+  #pragma omp parallel for
+  for(size_t i=0; i<size; i++)
+	{
+	  REP100(a[i]=a[i]*a[i];);
+	  REP20(a[i]=a[i]*a[i];);
+	  REP8(a[i]=a[i]*a[i];);
+  }
+}
+
+inline void kernel_8_1_fuseaware(double* a, size_t size)
 {
   #pragma omp parallel for
   for(size_t i=0; i<size; i++)
diff --git a/roofline/src/aikern.h b/roofline/src/aikern.h
index beaed76..2749a43 100644
--- a/roofline/src/aikern.h
+++ b/roofline/src/aikern.h
@@ -2,14 +2,16 @@
 #define AIKERN_H
 
 typedef struct {
-  size_t runs; // also # of start-/endtimes
-  double* starts; // starttimes
-  double* ends; // endtimes
-  int flops; //flops per run
+  size_t	runs;				// also # of start-/endtimes
+  double*	starts;				// starttimes
+  double*	ends;				// endtimes
+  int		flops;				// flops per iteration
+  char*		kern_name;
+  size_t	size;				// size of arrays handeld
 } kern_result;
 
 typedef enum {
-  SIMPLE_1_16, FMA_1_16, SIMPLE_8_1, FMA_8_1
+  SIMPLE_1_16, FMA_1_16, SIMPLE_8_1, FMA_8_1, SIMPLE_8_1_FASTMATH
 } kernel_t;
 
 /**
@@ -152,6 +154,43 @@ void kernel_8_1_simple(double* a, size_t size);
  */
 void kernel_8_1_fuseaware(double* a, size_t size);
 
+/**
+ * @brief A simple 8/1 operational intensity kernel which
+ *		  undermines evil fastmath optimization
+ * @param a			An array with double values of size param size
+ * @param size		Size of the three param arrays
+ * @param result	Pointer to result storage
+ *
+ * === Warning ===
+ * Don't use with anything other than -Ofast / -ffast-math
+ *
+ * === Description ===
+ * Uses a simple floating point operation that more closely resembles
+ * that of 8_1_fuseaware:
+ * a[i] = a[i]*a[i];		# 128x
+ * 
+ * Runs in a parallelized for loop.
+ *
+ * === Analysis ===
+ * -Ofast/-ffast-math does not preserve strict IEEE compliance. It
+ * therefore is allowed to ignore non-associativity of floating
+ * point operations.
+ *
+ * x = x*x*x*x*x*x*x*x; is optimized to x *= x;x *= x;x *= x;
+ * 
+ * This cleary breaks the whole OI calculation of 8_1_simple.
+ * 
+ * This kernel does not introduce more byte write-outs than
+ * 8_1_simple at a high optimization level since a[i] is held
+ * in a register and only written out once at the end of an
+ * iteration.
+ * 
+ * 
+ * === Optimization ===
+ * Nothing special
+ */
+void kernel_8_1_simple_fastmath(double* a, size_t size);
+
 
 /********************************************
  *  Kernels which potentially compile to	*
diff --git a/roofline/src/log/fma16 b/roofline/src/log/fma16
new file mode 100644
index 0000000..00e53e9
--- /dev/null
+++ b/roofline/src/log/fma16
@@ -0,0 +1,5 @@
+run,start,end,delta,GFLOP/s
+1,1466732365.5426,1466732366.1946,0.6520,0.9202
+2,1466732366.1946,1466732366.8410,0.6464,0.9282
+3,1466732366.8410,1466732367.4875,0.6465,0.9281
+4,1466732367.4875,1466732368.1370,0.6495,0.9238
diff --git a/roofline/src/log/fma8 b/roofline/src/log/fma8
new file mode 100644
index 0000000..18e1f22
--- /dev/null
+++ b/roofline/src/log/fma8
@@ -0,0 +1,5 @@
+run,start,end,delta,GFLOP/s
+1,1466732371.5080,1466732373.2547,1.7468,21.9836
+2,1466732373.2547,1466732375.0033,1.7486,21.9603
+3,1466732375.0033,1466732376.7499,1.7465,21.9864
+4,1466732376.7499,1466732378.4990,1.7492,21.9534
diff --git a/roofline/src/log/simple16 b/roofline/src/log/simple16
new file mode 100644
index 0000000..d480224
--- /dev/null
+++ b/roofline/src/log/simple16
@@ -0,0 +1,5 @@
+run,start,end,delta,GFLOP/s
+1,1466732363.5905,1466732363.9157,0.3252,0.9226
+2,1466732363.9157,1466732364.2410,0.3253,0.9223
+3,1466732364.2410,1466732364.5659,0.3249,0.9234
+4,1466732364.5659,1466732364.8925,0.3266,0.9186
diff --git a/roofline/src/log/simple8 b/roofline/src/log/simple8
new file mode 100644
index 0000000..d04184f
--- /dev/null
+++ b/roofline/src/log/simple8
@@ -0,0 +1,5 @@
+run,start,end,delta,GFLOP/s
+1,1466732368.4592,1466732368.7843,0.3251,118.1190
+2,1466732368.7843,1466732369.1090,0.3247,118.2713
+3,1466732369.1090,1466732369.4353,0.3263,117.6684
+4,1466732369.4353,1466732369.7596,0.3243,118.4045
diff --git a/roofline/src/log/simple8fastmath b/roofline/src/log/simple8fastmath
new file mode 100644
index 0000000..6ee022d
--- /dev/null
+++ b/roofline/src/log/simple8fastmath
@@ -0,0 +1,5 @@
+run,start,end,delta,GFLOP/s
+1,1466732382.7768,1466732387.0594,4.2827,8.9664
+2,1466732387.0594,1466732391.3489,4.2895,8.9521
+3,1466732391.3489,1466732395.6322,4.2834,8.9649
+4,1466732395.6322,1466732399.9114,4.2791,8.9738
diff --git a/roofline/src/roofline b/roofline/src/roofline
deleted file mode 100755
index 617c4c551a71c30539a445c0a8485d2c365b78f7..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 31088
zcmeHQ3ve98nQmz<EWfm}F+tb_CJ81s*g`T!C&0kS_Sz#(vK5eQNHD9lw0gj5SL{Q`
z1P9}_I2NIx5RwukRp$dP=cMj}D<PaA7|9qcL8wGPl2ZkeQV}4l$Rx5u@LuAu_x=5t
zomuVrQORAZuC`S(fB(<!{`)aKJ=*SjTs7<S^YWMq`RrOom_A$QByneK>JE-dS|yvs
zbanw-%+3O(0GG}=1Vv22gfy*Sp_Fp~&cIcfF9?<SI*%xLz92INRSpSKda0nxsDO8G
z7N~-ZX`DcMl-H@P90hFT5e1)?`6#4}@u2ZLkx8%cR*nKT@`!>{GGhv=@=@97s+V@_
z^L1c^>jc<3eosM?rBL@?xS8TpbsPoMc|^e(GCdOtG4?s=Q9e&M6idMyRC-zbQY`bY
zpiG`!=4;<pvFb8k^9o<PKhnLTyRu?M#j4U!pmZfKH`yoI4I3MHO3IHKFe=|dT;yuv
z|7yXdtq;alo%80v1yw)mst!F=)3};==ink6yj%qoo6B=U^iSZr@>uWrD_=Uk;QNms
zUjQ^?_;*k<F^fRgGHAS+MgBKg=&>yH#mH`^bmwKEUz&v;$zp#j3w==*JC{IS!<8P6
zf{-bnSAwpk3j-g=l}Ua$3;nDt`Kg7xhATa^f{@Anr?c2eLVi9wkBv7ra#O1Ry(Z~H
zlCJt!LefXl<Qdv?YoMc(T5vgYSJ&6No4rAAYkMf{4c6DL^#%Oide1hWm$}`o9Ra^P
z6!rweZZ~sZ-{rnR%B}TzLLqNRsAp8(u&K7r-RTK>d_J#_Aw74iH|+LDI^5y5px4tJ
zGHC)Q7G<qdpp?5Z<iHyY`U7rXpve<%5BN!-iC7iP+ua^!fzEKdznOJL!XXw42g3oM
zk0BY5Sy#y48Ep55TUZCO5NIM<Pp}nYC~c6n+!<^Sdr3Il-r)_jG<&wQ7O4PB5D5g4
zJf$Zc(BWzKGhV7y$QhNc9B@NT^_sQrm8C07uYkX)(>m}sgWnYt{h#R6*9#<%;uP|r
zg(_z>j+EgV{mjLv!UA>;h(yybB76p0PeKTJ_&WedJPlW5QTr@PVWWlrP00t4{Ctz2
za5oZ}k2Q07Ox#rap=S<_IV!03L-cbM5m2>5qSH8~LWyi|GeMA9;w;TVS7j&0QVU(R
zU80v+=+@_8rG;*N4(Jv-rLBTtp{qVaygCaV4m86}7W!<H%2=a?KF30Dv(T;n?6lD5
zS>(Gdbf<;B(?X}ZRpEXMeZC?B_FL#@Tj&E8Iz4+-Frr5ajoyL}Hfzl2844G~e+Z5d
zeW~z}kVszf4p_-0@8a&fP)CgLHeybUClQwXf^e$RiBXQfK{z$}i6M@^O1OjYeH?$0
zaH`UYJsf|IaH`6QevW^SaH_(IogDuT;Z$`Kog9ChaH_J2Mvgy5I91g|9mgLgoT_L-
z=lBDJQ`JnAaeNoyR3#G{$L}JXs$!y;<3Ymd@i@UaejDLbB@<I0gRo>9;Zzk9<G{;5
zsE)mUs}Xz0h`u{jS6@9mG+wJSWB6bv2g4O}bQb?C^ge8HUbv32{ZFAH8jKZt0gPy(
z$cVkywRk_Qare*PI!KJ<ID~f-k!?NHLZMXQFA%@Mxc7J5z`>6lFkr-{jDzv(jJ(6f
z>mP^DF|Ex})|?lT9iINqYmPQ(=>GE?pcwZfV_Odw{0gMJKOCa^9b~ss#;#+Lg2}H!
z&op+F6q2J`C+>m{#f^K9P*08>m<xgt`vG~&=<N}J8L@(|z^XCqq9+c)DT0%gy{>pP
z5e%OfO@zWTqlxRn=TBaU2DqApW3I&Hxj-%x$kgQQ_*%4#__eqhy?tc5w}+B@j&0N-
z{^Qti{8Cgx3^sSflaVKS3y#!l>}oEv;~#qqeg!P%Ix<<->pEf#7<~(0=9!VX%o{a&
zVdd~(0j%+Y#tV^B)O92e^1{m9FUN<_SdHk81Pjjun1W~G|3Rc)SHeIB92Aea#(_@}
zemU@%>usRpWLq<B=luoQcfAc-ok{!P1UU_o2ggy9!|__kz!)-?qFpIyBUaiHK^sDb
zMp?KBU_@kZh(#_idTWP_-Ud<_r7Y|;lUv5~-GGdZ8hJwo96;&z40X?GasGbjnBlx+
zX!1NG_M-9nZ=>UR#;g|=`Nd<-n}&`#i{~9#!`LzBtp_PSZi)XC@fPROQSe?jGkPY5
zziKh|PcjaJuMX~`+~=d-;zNH<CI`oX#_vOnrxJ7Zi&DA<07u}uJ%B@idyPI<|Fcxx
z&+$5kAGR%k>ZQWt{j$miq__6~lEd}_4h?FXAo(dI5+6YmZ!w%-B9r^!SX6TZoEwqO
z9gdenApSUR$DFsg_8)W3r(1C?-3o@Qh<r0St9JijC5(!K9RCw<2}aYD(bq$xS`Afq
z?+zHQh2e%#8kINK#tv4;5-*6{565R<OvJ!kj3y)$X}CL?BJ#wjP5p>^k5!Sz!ILSH
zvq|ycbVV8<Pp0UQiZ@F|Svs;c38RBll=uTEJRd)zM|2~$W6X#}&`Qx-W7KM0z{wZX
za$Rvis?#Yzs(=LQblhI2EoY;qw!kDr#vw8W7zd;tGzCcYgl_5@>VM^sOjRQ*rdz-G
z-!y)li0(MS!p_0_sg8IyKc=$4CVIdt)G^3VTy+QP3g;ZXj@K@Aje{2Bu^p7{_;d6K
zGE;FO733np$B}yq;yz)Waz=$qoJzkSO@Fyn|K1%_sF-`Y7QR4WaKBlGtE7~)-dpgy
zO&V(U;B%j)yixlY9H*8(6V`d<h@w+9_HPyLy7x#Y<c!#X0gSU4<!?d~M(k^i9Q3sT
z)Wk#+j`ei|+dUN4oXcE0oy%(boy!{bIG1hR=UnC;axUu_buNpHJD2U4GNRlXbMAf!
zDGm2jLXRHk!#%4JD?<!5RBXh0bf89G4>fdP2FLnH+UWfn&re^^9%So<uagQI**(XT
z$)2I@GY@kztZ?Xjh9Os<HG0oUZbb2W3*rR&`nYOgoldHv;7>t6z6iHohPhCA+{vRb
zd{%;Ta1bV*7bgA{;Dz~C-4J94$))jcBgUIrY)3-6v8w(?bZLz@_hDBYQq;E4R!fYY
zx176wN@@$BR?{>dfAu4zg^?jP#s{wI;~!GWR&3O`^00H)SICI-siQ{h7!Blo3k!`a
zN1f68i7dKK+~}M8GEkY8UTEZv(6FRmLk)L6k<&$rrO1-(A|+B}wi1cD;(5lE4e{WE
zlSSN*M)Y9bgommr`jYmAs#`=;%IbrggcE4RAU+R;Cnvo8$J2Ad7%wgQ=Tom+o$!;a
zPGCkJs*%dQ%HkSOr{EcQg`RTr^D=k_&dZ=kK0TEte40rUWV@#FqjQ%FzO(vrQMxaw
zHFcvk?TtmA#Mn82v9rG>=DNQoR=X2<N8fSn<4{g3KKhW_FFnA(leBjPZ3wjeH1(f_
z$PgltMdA?$Dos_;#zp7Yh(W#A^<*@W=iL2e_!p7jxzR+vbN6wMIh?y2QH2w);x_D}
z=O)eicw3*mfhW@25H}9K<1qTpFOENfnTUL@I*=aa8m0I-_2M5vmFLAOAF|4GPf;aa
z?J5pFIpGs2^2gS&Ya9u)$t^NP%&cUJ<yH&5<fn%F@gVOPDG&2!xJ)*sbOK|cna!vR
zb0T`oP_q9#@kgYCg6;%BL8&D&v!G*Uvd>v9*bDltR6$en4_M`+t}#k+OcpeXBvL7k
z+ET<+P8QU#TCk_ME0tnOzQigYb&XPrqawx0g{X0~f_u$YK*qt}3-9eSV#NSxCUu2?
zLx6<ifn8U^MWQcte?dM*#NdYqb7UMWdO%MBjswy&eaik!KeQOnr06j;S@8^~=V=0>
z<Edx3{@2eOzI{GY`J_mNGIClo2YR@lRvG^DldwK`^kikwP=?dA*1H1M?l`HnI#?S)
z-i9z9@X2R>B_HK^xxnC)Dlp-E>hP!l40x_Fny_Y=R$Y2M5@aGla?S{pkXH;4qDm+Q
zSAvws?3Kc^Ko6PdKV(VXOcs;h^x53Ulxz&M86;y%mXwg;%*pXt8S$WHE9uFS`f3f5
zy5KaEvQHU}-&?1Q17_clld_Xc8M`W`O&RY!vXeSx?7;nh!wJ?fpDc~g{3OZ^n{%x)
zJW2{_80lGsm}RcTr&Z=$OU%6bNE&l3p0&Q7e(PN8?BA)m7LjSLMbMmU6+~ScO|pmx
z3p96o;U|<IHP>1T{rGo3Fy~rN@F<4=$H3rotpc4)NZm23?o5b?xz;s^(OfJ0<f4$z
zuzr9d$M+!#nq0I2o_dBA?3y;a3f&?5PujHUq>~Dajp#aArW^Ebi0OuU^ar1?KKAb>
z@1MyO>X%f?SC5EN?gBh@DV@6>gA&al)22{-dYN$^d8Vbc%#m(B$wa7--lOc^h+4gI
z)9HId)cF%mc;op=;f*lh>3E|NN*TPdg_HT%`8TZltStk+$mA+HP0exnyHm~^8D3#>
zZ6nt<vORBdZ6nt<a&1E_yX5*pt}o>JLar}NUj)vL47rgZH!}Q<Mh4saymQw9thZU;
z=^v)`x6@nU5byK>{6BO&{+4*B-vf9$@AS)|l;NHJGo1W?`A#p_H+bu^ewjPd6&3mB
zi_cXjUnR(;m+1{T>#9r6pJM$j*FJ4ez+8LDxO$CwQf_?7jW6`oE%!Xam&ROw$o!S-
z6yLwHj;=X>=KM+jYLXjY%y+Te^EUUq&Gm=Wci&up$n}R@f5`O*cp$?+7Ae1EonhwI
zdvfbN8NUDI+DmS|$NtNi-1w5QMdZfM-1zzbVf<A8eeK+J^w;+9lZSzy=J&~B{(t6=
z(|=$agW~%n13dNj$sm4Tl=dsW5H=LZ@?CO-RvN_<%K8OS{kBhaN%~!fUXuF1^6yX1
z;lJXdpXN<IbecsHpIn>q^MrWh1yQ6GfTu2!bJwTnHx22<az;z|PaV#zXc&B{(5ht9
z|NWY~B=srO>K3Vsr9}0i^zXvuwNY%=hI7{m{3^ow9sG6rmBg8R2mcppXXDR{N*Mq=
zb)^Ko(<&si^-SAYT4kgaEcmpwnf03`N^M`!U$v!bb5$%^)lgMm)3?N5tYI^Yz7+&-
ztd4zD9s60$viJC&BnSWEh`WD?g7th5*2>>f`(M<=ep?g!XdNsh7rt#o58;=mlab#d
zrNS?4t-7UZYt^k)Zfu!h?w9e0LuiP!$BEq1g!a|3hR{{PtJubXe+6wC6!2@<3@Fm!
z^@p`3A->JamUf>{3r76fo$cW^aD$%hTB!XlZ%7NYXiM&jXkMR}<b^b2T)xTUYl`?h
zVXxNF?q9BTc)FKso~~A{B@on>G?O_c&NlhIWGUPRS-vAtNNe%5`@GGwOEr&njkct7
zxwh>Zx;Np@SQYJB1k26Z<svsBZRzEgEGb>tqSdY;d5UdVvrO$#vmW_x_qQrpUW(>O
zN5^)p%j1i9L#0}MTYCtlLoq$RP(ZuW6YSvGY4S9+d70tG7E92NbRvIEZS8)q@LsJz
zv^L2P2LiVv@l`D-a#Ju6@-_wh%^{v4&=k6qryur)!rG>4?RIa_@AZkS)?>dVt;6dH
zMS|R)T6;%lAQWoHj#H(Ybw4IeY{In3?-LcFZS#ij^m_eTxfTwPcgc%)dYel{J{xYV
z+TglME4?fj2(-WxUTsB4TTL~1t+pblt){xTR-4V1gswuBn-S$c*;M*<itOB7TGjW#
z5B3~Cyzg4Un8FqSJm>-x>L;sDzGW8o398cz_fU;)3}~T9Q=4=K@*pgosXsjJ<jv{D
zc<fOM&t9kUowwL=HSL{uA(AQEpG?*vUiWM=sR6EkE}85Gybpu?0N@G0{eU;)IW`9P
zC}0BcIA967(WOJlB<)|=dN7&12`~uQ4cPT!GPwuv#Y4&D2+<EGlT(00fW`RzXvHvo
zc?P%}unzDIz)rx1kz}$T@Fl=~fS0|JOpXFR2sj1!=Bv=huXvyR5z+-5el3}71pLO^
z$>jZjS7RH4eSk&S4q^=Oi9h3qX$7da6UpRKz$w5^Kne>5V0m}l!1B6_^A^vYS=gU9
zvzX}kzaX|0zPUuo5Xdag*hS!x{Ma+Xhpec0eNoBv&N+7$?qt_3y84pKFZwj;lgv6?
zhhbMo-dI&p@qPJg=c&D_h~JFseT?N@fF?ix4&+Se?*@7G0qodg<KJZRp9Fn7__~$<
zMb75wzXbZD;NNQH=ihAVzYF?*fxpVi52W;|-AsbN8qmxi$s$nxF9Llg+6LbTOO<cE
z(7y=$b>JUEn<6`={)Pt~(HZ#(Atn4tzvuoj?9@z?bLe?eknBA|_Rt3Rp>3J=Tv3PP
zTlvZy&)4&iy9;e`Bg9O({Aw;k`F<Pxz2Il_`w8$rK8gPMNPqcD+3l|e|MTESZRvl>
zO#c?}4}c%A@ozKvcY{Co<z(`98^6Wme-r#I;Ad+W&x8L6_#03Tv;1TWf&BP3_&-G7
ze4hxa-bB6pf+r&7W}pvlM_(<n+Tn7Jd~3ZdhnyRI_^0^MoZX&d#3>C@y7iEoGlsE(
zD#4bn<B{_%a(6>6irormZ%Vsd{;)NFPeQH~8%KO?dj4Y8{JjRbZ(yT|H>TS=tn94=
z??cGdVOI|R$XioX{80XyqLME=Tt!;nj5S3|A1pA6%I-VMD5{LktS!=mMU_=WWmQE>
z*A!`xgZ!GJLOyP+p21i*?7+>rkn=#!133@mJdpE1&I5l34-{iRP6|`A1>^{BG-aTW
z4!<eLCjEfK=Ss+C={z=1;_uc9{5*-P{nZysoc7+Jut)%w*xJZrGzFrd_<#IZOa}+@
zbt0R4z84k<T)j7p<O^KAuzp?^fX~69SAvVeIRcn>Ol?GHY~)bQP3XOkf_mRkd*+@i
z$k>Th=P^Yul(@=wqm-K^p-LYMTzq;vaF)(FNr|g`;D;cbzfJO``u{pm``D)^R3Yni
zy@Z=3Y>}`_!uurrx`f}7@EHjYOZamM-<R+s3D26X^Q1o^;l&bGNJz{LYu8?-EyYgp
z%d{&>FE1_ER+g1ll&vbO(3akSg&@Nd7W5Sr%P;{BZSM$swgF=2YYy8~4EtVtgPp9@
z9|(I(tJYMn2zy#3*6NRxZi}>IJNx!#R!Uf#C)CDDo45O6Prz_caJsy~5be6n#U$be
zFX;7oNI}LrePLGG?#F)IrD1P3?kzyT3p9Je9#-mYbGHON9bR`^GjtUuG~9fhO=zh2
zZA}n^br05;kaPeBaVzDi2#ec7Ay(QH=%D2|?$c~H2YPd5zCMlZz}jGaM!JKAGty@;
z)jzGq9O$hh8RgGns{h*MXR=J~-ht6T^%uMRET;OMoj%*_pLY5jrp75deJ;~vJG9g1
znde{F=|xP9Z+5zqP04Z3PA@jkZ?Mznv&83gD~ZP)*!f<KmsX~Ooo$Yjb~;9>TN|w;
z9(Q20%hazO*uh-wb8b~~FpO?;9JJF**q)5^^Vm>Ex*F9}R%ke#fsM3M6cMB6M;?u`
zY5Aw;Mn0RzIxYR5G-x@K==;*>Rg#|mJlg_#9?k`b%Y4dVowacKbbt0qJL>rel!r%g
z&&yCR`$5lCu0x<xK7UiJBgWV&erN0zK%y_6uX74tLqd$+!HHfi<pr0qw<J9#>3q!!
zv|md)?d3~>uO$H<!$o$EICaFZZ#KU(_Gdt%cS$;5;{lEIh`vSIK@Z{K97(5rWhwAA
zD8QwlXDY8M*WHVB5@&pk1yJSZLP=Noq3wbxUE1T00$+0hq<#8{PP@`k;A<>^eo3eO
z<S6hp6~G51opy$!z}HX!zb)ysfg1(BW&$`W>9oTd1-?cC_>rX3c4!pd;CC?(%2Zxj
z_szqJ6WyD2DJj6koIbrh6E~Bcm3+dnU|PRk%k50>zco@meSC0p`RV;P0{Q~@lXhdL
zz|{eJIGvMhw<kc?Y!rbEa{Baf;8kvC9&??ib3s0NLG1l3_7hp?P6wUYP{N*+{Zyoi
zvlc+7b3N2JgC!&$E|K&pDPJh*t3Y2M<5p0|>C?yU&73}uZ97-z;(Tosv35!S-a<j=
zYpbAd&tktX3;o}+&<}%7<xL+~-^?OU>;9R_RRD*Oe?}S`k-OBq`8?1w$(MqzaY_5_
z8nTn`WQnxr*Y(_f>bVA#hno?hb`>tsIj;QXmvq|pp8{Vq1pE@GbCT`$u#|sVwx8Q2
z_dB2$qg)ADE~WI8lvn3csc#d{NcxL1KT)ahrlen-me2R3{q**BRLawN1r&7P6g~pf
zc+7sAU69$&7i6KAWudPFeSUrk+nZM2I?ywvdmFcd=b&u2YCPYW#r{AR`gb{f`n>3A
z&@-K<!<vHOP&k4w%S|`{=K8faxNE9!tjGB`sZ(;?;SP5bos$z{>g*c}$?a|qxLbXJ
zZ62SyIUERv+@44`!*sRN=f&sT(uylr(aAO$gxoZ77SmuirnW&GNy8`3{4*|8ESQ_-
z+cr`4*%R50ggmWYwhdpG-GNA$4!LpDcTgU|s{x~<Y)aj9CQpkWXV`SM``vU34x`Vt
z{No}&2**v)pf?o3^j>jHt#T&5Ln1Ztao1h$F0Vi`R7z8-%*JeqguEVnbx!fi6~7Yd
zX}T+|yy>c!E4C$FceBUe>J0`WAu|W7lvpZ-3{qDpsd9IvyDN}Eq5|$;e?wKR%kA2@
zjt)cOEkng^mC?-H>u%myRa?ClQWg&XD9qd&YBsH@s&Q{xzy3y7y}Q0@O^u7p@soy{
zLJ{5ufZGo~(sk=WLnk}5iQbUJiAAbRnM%%&OtKp_4@|O?-Cq9AB~DthvpR!;RySVL
zycv#i(sgx0k|@yhMxwIU+#c%mgqzx!IDN^Ug!TUc?3`(5Go`aFj-ZxKoxpPXR4RY+
z+$g)cI4sIelgC5ZiRou#xtl`)cbmuGOy_4+Z$eEpqj#a<H>>heb(qcBTJh9W&^77b
zKJ00jzO%PYhA{$dO*AW0#_WWtMr=)4bTTB1cW9}~54Os5xEEXoKiKm^$8Omv;v_9Q
z&F#iPVydUm31e!Q1UYrSm^IHSkJt_lgIQ`vbmEpel`AbJ?lp0!mz`sOik&&O7%F+V
zFrDL}GGCXF1OB~P#KfIst>H63Xzfht*XQfN=v+h&raklIV^}b8FUMuo59jNkFsx5-
zz`y5;n7F6wuLBvMJ_9QKfqb0{4oCv+2uuNwK_1lnQlgCJJM^xk^hfe_U<~UW957u%
zTA!q{EBz9|W}5*WxRkzH4=I!hs`X1L&TXhTL3$6QwL_(^)=jp_1<~~UD>(&eo<(bs
z3RmkZ15&?}5<wu8@~QGqGzxd&BA+UKwGLA#7epJuqmb@DRsP-Jk^UMILTH<*OUMCJ
z>#w9wYXmBNrT=Bn$geWHsdb%MAR;Uhz`Rph)(`_?)mQ63n$*{bX$i$LuHYlUt@`?G
zolEF)0s8T^Qb7Wxr>Ku1V%1mcN$=JQ37HKg&V6Mfi0r{d`pW-m-Dyhd^GX9>1?4}5
z??uF>U#D}HI&W01Cn))J|2>nYuhy{=pA#Cp(-hLvAA(pBE^1p8)cdSDmo;5J9e)iv
zPFzY~eLqOZ1%`C_bbJiDRDPvzu0P2ImUQ`a{Fl&GeFIT*olTt&tImN{?NjNicJMxQ
ztm&)uxP+UmrP{l|Qn<1+35HdlH8u)S3HAugyemBgKLp3BuiiuVGzyip>f^GeF+~1^
zi)<<XsrU0I8$m><6u`V|W=uqAA&}B9OQS3O<!N-1rJ&$^#8ujyA#bCNB2uUXB|Q!0
sc2wVyA{>l)1Xd^e@h5@mxRl@2_(T0OQ~FDNU?7aBflr`jz9q{32SwK{e*gdg

diff --git a/roofline/src/roofline.c b/roofline/src/roofline.c
index aa91fdc..5f8bcd4 100644
--- a/roofline/src/roofline.c
+++ b/roofline/src/roofline.c
@@ -3,6 +3,7 @@
 # include <unistd.h>
 # include <ctype.h>
 # include <sys/time.h>
+# include <sys/stat.h>
 # include <errno.h>
 # include <string.h>
 # include <stdint.h>
@@ -67,7 +68,7 @@ static void testkern(double* a, double* b, double* c, size_t size);
 /**
  * @brief pretty prints a kern_result
  */
-static void print_kernresult(kern_result* result);
+static void print_kernresult(kern_result* result, const char* logname);
 
 int main(int argc, char* argv[]) {
   prog_name = argv[0];
@@ -108,6 +109,7 @@ int main(int argc, char* argv[]) {
   size_t size = get_size(size_arg);
   int runs = get_int(runs_arg);
 
+  // Allocating arrays
   printf("Will run with array sizes of %zu elements\n", size);
   printf("Will calculate min, max, avg for %d runs\n", runs);
   double* a = malloc(sizeof(double)*(size));
@@ -119,7 +121,8 @@ int main(int argc, char* argv[]) {
   
   printf("Allocated 3 arrays (3*%.2f MB = %.2f GB)\n", (sizeof(double)*(size)/1024.0/1024.0), (sizeof(double)*(size)*3/1024.0/1024.0/1024));
   printf("Filling arrays with dummy values. This will also warm the cache\n");
-  
+
+  // Filling arrays with arbitrary numbers
   #pragma omp parallel for
   for (size_t j=0; j<size; j++)
 	{
@@ -134,13 +137,26 @@ int main(int argc, char* argv[]) {
   testkern(a,b,c, size);
   t = pin_time() - t;
   printf("Machine heating took %.4f microseconds = %.4f seconds (with test OI kernel)\n", (t*1.0E6), t);
+  printf("Starting tests...\n\n\n");
 
+  // Executing kernels
   kern_result simple16 = kernel_dispatch(SIMPLE_1_16, a, b, c, size, runs);
   kern_result fma16 = kernel_dispatch(FMA_1_16, a, b, c, size, runs);
   kern_result simple8 = kernel_dispatch(SIMPLE_8_1, a, b, c, size, runs);
   kern_result fma8 = kernel_dispatch(FMA_8_1, a, b, c, size, runs);
+  kern_result simple8fm = kernel_dispatch(SIMPLE_8_1_FASTMATH, a, b, c, size, runs);
 
-  print_kernresult(&simple16);
+  // Freeing arrays
+  free(a);
+  free(b);
+  free(c);
+
+  // Printing results
+  print_kernresult(&simple16, "simple16");
+  print_kernresult(&fma16, "fma16");
+  print_kernresult(&simple8, "simple8");
+  print_kernresult(&fma8, "fma8");
+  print_kernresult(&simple8fm, "simple8fastmath");
 
   exit(EXIT_SUCCESS);
 }
@@ -233,6 +249,72 @@ static void bail_out(char* fmt, ...)
   exit(EXIT_FAILURE);
 }
 
-static void print_kernresult(kern_result* result){
-  return;
+static void print_kernresult(kern_result* result, const char* logname)
+{
+  struct stat st = {0};
+
+  if (stat("log", &st) == -1)
+	{
+	  if(mkdir("log", 0700))
+		{
+		  bail_out("Couldn't create log directory for %s", result->kern_name);
+		}
+	}
+
+  char logpath[20];
+  snprintf(logpath, sizeof(logpath), "%s/%s", "log", logname);
+  FILE* log = fopen(logpath, "w");
+  if(log == NULL)
+	bail_out("Couldn't open log file for %s", result->kern_name);
+  
+  if(fputs("run,start,end,delta,GFLOP/s\n", log) == EOF)
+	{
+	  fclose(log);
+	  bail_out("Couldn't write header to log file");
+	}
+
+  printf("=== %s ===\n", result->kern_name);
+  
+  double min;
+  double max;
+  double sum = 0.0;
+  double deltas[result->runs];
+
+  deltas[0] = result->ends[0] - result->starts[0];
+  min=deltas[0];
+  max=deltas[0];
+  sum+=deltas[0];
+
+  for(size_t i=1; i<result->runs; i++)
+	{
+	  deltas[i] = result->ends[i] - result->starts[i];
+	  sum+=deltas[i];
+
+	  if(deltas[i] < min) min=deltas[i];
+	  if(deltas[i] > max) max=deltas[i];
+
+	  double gflops = ((result->flops * result->size) / deltas[i]) / 1.0E9;
+
+	  if(fprintf(log, "%zu,%.4f,%.4f,%.4f,%.4f\n",
+				 i, result->starts[i],
+				 result->ends[i], deltas[i],
+				 gflops) == EOF)
+		{
+		  fclose(log);
+		  bail_out("Couldn't write to log file");
+		}
+  	}
+
+
+  printf("%d flop(s) per run\t %zu run(s)\n\n", result->flops, result->runs);
+  printf("Min: %.4f \t Max: %.4f \t Avg: %.4f\n", min, max, (sum/result->runs));
+  
+  printf("\n\n\n");
+    
+
+
+  if(fclose(log))
+	{
+	  bail_out("Couldn't close log file for %s", result->kern_name);
+	}
 }
diff --git a/roofline/src/roofline_fma b/roofline/src/roofline_fma
deleted file mode 100755
index 3d5832b7ec068b8c2c3da3fde7ba2a999d7e0cf1..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 18800
zcmeHPeRNyJl^@A=B0^%x5Wo;h9z>yb8Vfsw9EbAplvv3#*dZkj2`xcZY}r<kC9m`p
z%twmBHjC=GxP^8DhjTWxbeHXROU@S9lol-K%ZB9?2fA%eOF1sf$1@Jp2^38E(Ejec
z8Tm<;Y1(u4pIydB@6P?ry?18re7t${-uMn*-8#F?#$>RwD;RO=Je5P{m9a@LX9ZTx
zPGKrLk1b+zfE3}U@)V+GOwuV~T2hCQD*!FTui9<`)pnIfBt6%JnWVCWL@B@Npo*wT
zS1vPINf}c(KzdZx$qLSbHu8w1kBf3Rs9-!O{7NwCIXs*NZR8P2Cq=;|mGz^#(XT<+
zZLq80h*ycQ8h)LoBuioA>v}WA?{RY$ROJy#3q^hzYKpP1K#$6KvcV+;uNC>t<`<VJ
zzoa62c2PLAebvf~!p#-oP$bb=(OJE!V%17dJnC7&>rM7acKwFUJSXW#1sv6HA%5g)
zO0RhRsc7qOmOXay6Q5pv^Su5e_jW!UrL;5fBOAP4MHHLQOGEgX_^m$CckYUVADs5n
z?|*O_*u3G#K=UOQ16T5Byflmao3r4_S@1<DZoYgM%z}Su7CbSF{iCzs=geY<hJ}J(
zZa5AgUpZF;S8|2HkKvb3es~sq&Mf7rhrEJcZfFIN&;A9o*hxdagq_XCDjT^eIsTp(
zcuL@Md`$^_BuAd1KetBPJE#X&F@J4Cy}vma3$}*hdN9^dzcw6=1RDa|!$Icvx3)(k
z{<t29>3%=+U$w)3t&m$A4#eZZxT&63dHpr@8~q)DSRfn@h8gnnw+3~8B+>5I+hV~$
zb3B8YJS9=lI!u;ISBVk?W3fooAC5K!^iVWH0!@^(iUm7EI*WGbp-409Na%4E*JFA#
z9A?M{FxwH2bi_gty@j=-2+<~z4a8a@hT6th%gwQn9wcEs)E<nsGzV^BEkXg7AQFus
zd&*BZpgj<ZFkY*bC>hnR3Uqy4?V7dz6`mEIOW<!Bv<3Jp<X1_}@lQC7^&*ibr(YYi
z(BzEfkxKkVmt2S@EMk`fBusS^U&z*x5P}VVPlHmLf?wjC&?%I|221)wA{|7e+cW7?
z7c~;h=bHIEW?p3fp?eO^IWoxpL-<(|1eN`eaGIxNC>8x}E&x#^o}^fCS$9e(v*5Dt
z5?*P+t@pud3vRs+s1}^^mO-=Na?DWLMhgxHX2MnrJ}-kZ)@Z>`wcu?Q-0IH`3%<Z2
z-)X^}7QD-XQ`^dLj|DH0An1SvKiz`ww%~N{kwNP|=Fs|z-o8*_T5n1(GJcK}t^1(k
zfGLr_<W(f4m%fIpbFqpT@okhiK9)vU`di|uNykSy|040!<;PQ;e~x%++VQ=d-%mU>
z>G&SbKSew><@f;S|DAYh!tpN7|A_dx#CLH1`@~a|jW=@sLE@>Y#y4{QJH%5HjjNph
zCh^oX<CUDhi+F01afS1@5>HJr?&5rmc)C4~GtPgVcxsaI$xi_+-A+6;#rPQbs<&&C
zM{d-TuWH?|O>S(c9Zq$euQF|Th;cH!>Qt2(UxeO!EzZR?jO}|24Y66PcoanIo+{Ro
zukTp257xN*OSTPCLV66sJp{6C2U;k2B>xPhZ`SU7hZ`9Bv;YRQ<fJxaT&dX(YezoS
z&&XIiRa$c{CObU;uFH=vRWSTZHbc>fpkUh$7rg?=_U-{{-$8~2p4NROOB7Aq2fd83
z<D`%t-8OzJbSSRfd5lJK^6B{iwB!@yEv>KDBuq;deH&J_VISRbh)xonsP6L_-BU6B
z?Cz<!KDT@7O8wl4#pr;Sk#N#CHE|Z0i%e#6VxF-UJ;S&H7p<?KO!xItc25nEcay(N
z4jW%WBP3ySr;$$lps(ndTVY@3GCQyCD|!W7(syj4vd?!++pYC4Jj4qlN@*L_`e5bo
z;CEq-SJZGIr*7Xd8{|zZw;wW6g~+{o=M>ZDf=t3Q#=jBN=bO?{fC7pqePiG!iN6?p
z()TjhF|xfkW83x#itl?F*jNU8`(tt%B!|Y(lEX$lWMB*ho5og~*oYNdYGNrAXq4&2
zAR}h+QY>+v)>og>`ZkloC>6nVfvjFRFZX5?Y*e$QG&q3r?M-!_+T#3U>V)Q8mYO(Q
zOYYZ>ywyEs(@xnh;r%C^TT>^Tt_8Q_CUC-e;}FF=Eb(T<TbyO1NIQ}#=wD-aiN)Ai
zA`XMk4XRXnnoo?>N9pun2Uz27#CR@A-+)<5-)_(mxNZ+<3iMH}-#73xsIjN$LVnnO
z8dSezdVD}M*>2(OJ)q>UM?q7AdoO_G`N+f=K^Je)oZldm``}nK^JX|VA)GsGR6)S_
zJ}xJmH~99QaF)=;RZo|qEw}@oa9&T&s^2#_0;6U{KKL{530l*n*56CBS{*fa-%c2=
zhvChmG%H_MpB$=9PCa9me%M%nISCW93th-mr0MRn6j3Hrd)W98lUtS|&4Zt%NX{n3
zRIVaTke{VkAQb;zD2m#VtqB+%q^cO-MCJMX(S1zSk~@!T$pm^SdTWw;tq(l;f_kpc
z0Hrpa1f>R;LYt0dw`t4iXsInQ36U{~90fH%X#`DzQafRo`cebW9uT=IWF_O)+dicE
z<Kym~A2Z!KSh_?Gl~1WIu!#}yEDa1YWXRz_L*a}&Pv@;mLu1HdJh_vKZ9GM{piC}4
z<bqO6@p<HqA|q^Cr;<_Srv8$CQI7s{tNxukC($r>>{$2=kwGm}hf9T&u-;en&SC{E
zJM`2Ssch6g20I{Z%!PH{IOZfs%{Z{lbl087Iv}SdpWcmm7PI_TWT7R$+sH|O8%SN!
z?83?ZPH;UbU2&HCx}4?p1J3fzdz|Il_BzXhDQ9{6sIxpV<}BYisdaN}%-M4<avJWf
zh92F}hkGwWtP(M_kV{MUs$jMLUh3%JG|u&twAS}sUY`EmJt)>Q-y;=tvO7LVr+ZVk
z%stFuSmD(9BtxmbsP&zZJ`dIJD>8`m_jA?Vs6SNAia!Q@;~ZT27}i2nhLcBOc+M1+
zU<f9DVVd|M$TJUHb>om7gsYhGFNpE3mfSfd+~{rCgdwf)?mp}@AVqx(eYI5U{hhPt
zSEP0t)ashXjOX4*UYHq@NBP86`@wsZvy~gwRv&iW^=&fZeC)WEJV6tA|3Zhhdeqr{
z55eZpF|_{q;)X|gsSeFHLetVHRxg;Ss0g_tE+O*7>>{N?<UuLY?K5oJ>di)M*F-V*
zqt-oS8xK%3bsto|=Doq}O0)W4tLX&VF)$XO^5ld=e>yoQ9Obpe_<Zb$)d?@m>IA0Q
z(2P{?l@`|korGuL6}ron*z$M=&dZ}nKK)CY@M%6xz}cG8kIuV%@SW9{=U9C?y%wbP
z8dhPQa9mGHOFr%%;nlw1J<91WH+iQn+2bDLbeEeto~|95D(`VmB8tc^H?83ix!>(F
zBM-S1GxCVL(v1Art(uX?-5brwFWilYXuWB@PV4V+cM+5Pv({hdc4+;#yA?hlhilvp
z2XfJpHEtJ)rN983Un#&+<Arx=ymoHuf(}cZVf-WbzUtm|qTC!w{`b%e`Yx|tq+q<(
zxs|8JmZ<ERH*u*-8FnO2Ja61&$}A)q%xqj@(v%p3xrrwZ*TDG1L(aur$QL)lW2IH^
zYsoi_yLje_C5JxTk~(yJ%d-dn@X<09`QRVkT{ho+@YUlF6m2!{xTOaMU2dM`y_jtH
za(ucCybYT-X?=bN#<70Ami)xH8fK{n7l2m1s5RmKb(F^N>Lc+vT6f9@%=y@eW_ynM
z>+1b+H}3hU-NGYUf8kKez@d*^cCCBA<IqRfrw)BovStc{I`P)mws~*B^1OyIksl2i
zJwW;W?u41&eV`|k-*TQ`-%bY>F)p58;Sppy|JPLI!rvgz!XqdEM)YXnRi1Cw^M|Fr
z&YmD{>Qsg;q72THD}%FpIVoD}oq4`NQReabzkB@<eDL}2|E&+qzVGaP3y(T8`ag{~
zXU`A$#Pf;qn?IUk;9k&^7z57kFe%Q~|M>Vj`09rbq&~m=%|4KGm$Uarm^5d!pNr<N
zuhHDG=pD1|UV8ha+Ae+}<}GSFhq<4D`J(XGm?$!n7BcpgCJd3IoG<zO9HadH75UKo
z+;o!pT~7UEdcLQrQqGTuK68#L@wpO5crmQ|H<tNR|0=J)D9=mBPOkn-XOzcMb$-u4
zzT=|$D)VmfUgu4I;N9$Ps7?OY=1uj<U*Z<1X?+j5>3*msZ+Dk!N!+;W%-i(x-vb5w
zM`J$<1X`OsX#5AF^}VlY$wSVbw`u6~rkp*mA>!>$+nhbL`{%st1oXh$ojtS!MXbQt
zdj$#i;|A=jN~6WKWT9L4^81((rzLgYN_W+>RWA<iJzHf5@4T&~@Z-8<*sathb$6L@
zCvr9FAa?NK+e&IaVYS`Ewz}0HB@XWEq9XRyxl5B?6doHEBX!AuLraESSaE7pckedu
zjb8tOmVw;daTg_RX1t5#b|(MIv-0F`qx_$TSZ@B~w_!ad*AqFlTPy9~XSJ`MnaBRa
z6R;o7V}BpK#>N*Rla_&+<fnsEOZe)e1Z(3N_1PlxTf+0h0wz<by)!edro~v9)<-Kc
z#43$1P#Fvu&Z*;GzELs$5B#C;t-9nJgQ~d%+J%L=wLkOq6q-;=Ugp-&Y>q*ytCoDj
z*#8zB+<mZ=Z(wHb?~pH>mZc(U$p$J7S`%8-8;pBQ%`&u=xn9e%Pi;l^x6fw3l#E#N
zK;I(y6|>1#S3QeST@>T{csh#FuTz0;r%Qud8SGf3GR!J)ZPkkq1m#?rF*cq~bD+#j
zMs2Ij?b_-20`-M`u2|%nnZw{%#<<0_V)RPWip4ao_i43CHp$nPlJZg1ajq<AmaWzK
zZS~MF5c5Nt=7+lerLW9au({k{L3C4X@_kGhb>*)kKZiE-NrCagn`qMBw{@lJcXB_v
zE{WdpeoZp%O)q>|>pp;W<3!>uc*0>FhSUCd_W+|db&v$2jNWruL;TX%rEEhqQb8~A
zq7elz=MwF~h^{P+^Vcq0LgBCyOGK2LLwXxhV}V<gc<9z(T#2?QOK(jm!Elh|O=-rs
zd{ZFYln4j(pwb?SELYkCoy(QLj#i~58dH`wlQ}8Qu89Q6lHLYc{)RBFv;;!oVDmhW
z5>PHzmUb*xwqH)yCS2h^dW#6l&C11QY2r%R#mkm@R<tPfYe=4A>(`XacSh?_{!pY<
z%JNz?C)(R@QFa8viD2BLG_-}{s2!>q2*;z!&4E}uFHTdSsV&H~AYOSwKhc5mHMNBz
zLDPHnCZn{8bUhlq37M~KL6w_g(Ri>a8flL60s)h38P8u2#&zYITIHr-ED{WxMQy<Q
zPo+H=h$mv)pGv5`BN~r~@CMkUSl^E-=1b9QB4M*3l<h(N=3p?QR4ICtyh~oZIoRwm
z%ei@zcfIdY#dA?C8f}3mf=WePxr|!y3Z){ZTt;nig))yVjbDl;&qSpAL{}M5Dbm&B
z@%F#{#GVfh@4doIn7q=YOf{)P4TPvxGtF&?D)B^9o2WCg5{iF)y=U(){1rL(Il4;_
z?&m$m%t)5|!3QY(>`L3Bf-loIEsK%7`?YkMp0xM9o=z*EKY1garic7@FoW#|ZN{Wc
z_lP0Tqo8!7ngaEMmSQ;F2U-pK0+tqAL4OO{3Hl0_T6;jNv0XR<+IBph9s}J6N-v?y
z@c32<`Y324@tEreK<WOF0^N#R%p~!k3g)t{xHoSE-G}YhJ)j2YKG0HZoR5NPpay6s
zD1Eea6tony3|sJ(pq;c84{Cr8fKpg!5@x&gT4w8X*%r;8>lm=jbrDYALJhz-%Y+P(
z{PK*skw)_Bf15sJ#jbV5rB^vmz1h*lt~}?<%Pw|bK>8$8gP#Y!R#8^wEq2{)U%NoQ
zL#Fg*{5D_`-2s|OxBneVX6p9<egx^;veLI^(jNi-2c)ak^xJqc&;KBB8`}0pYr6gV
zjQ(rDuSEJvYkG8ApZd)vq+bS_DIduqQ2E`!qv$LA{j{v#I#b_`^ctjp73pLrqrZMv
zL3g2j%9JwwNl#6y(6{Pl$Q7tIAxQ0UKhoRK=gLWLeQ_7M3@^uz!FQn#dcm`rVx?nV
zvBN7~n$w2%^GM%|^qVMsO|fg2-CJCGcY(K9K}nj5OHuSS#k8e_K=L0$eiHJ0?B>ef
zG*g~-9aKCb-Uga$zaVbUQXu;`KyCnI<LAg{hWt-$Gvs$cUct6cJH&J4?ZGT@UjAP|
zz7g_0`Q&e(DSr&|DaeQO$v0=q*Fol7ay+)j_&0d)&%O3F#iidW@D(flg=>n-b`{kW
zSKd9xTU^~e7p-KozdWzF+FM-dEiPMAtUweh@b_BYLe#Ed*mGenSVH-u{cc2K5Z(j4
z40wf!SGEcKA>e8r`Nx58&4WJ=yaV`GOgpN*uGm5LUjx1mIMoUHs`m9J{xR@L;2Pmt
zv3dvctwD5QA<Ba}DoA*~dTUf~jp|)rtj6+Hn%C<F*x3j>aQo-sa}Rv(f&T*@aN&C#
z3X}6p${Ac}il>lEA2Q(#{<Prd3(8JWd2E5;_47^sY{AR#s}~7g+`ia3f}g5r<T1KW
zQ;_L@`ZOISez#pElKJM_QDpLReKKM<dAUITs;Gdtt+G<+Qm}rDM?9uZGvIR~3UZF4
z^(6(GcPY#@N#+~Bwz(=tB<>KrEO(=jKSfZPKQ<D~`pi*z3N|-5FUx@?Dd)F~bfNyg
zP8^=bym_n=?Yd6T>jZ5PbcdjK3;I1l|4Gm%1wAb2uLXTW(DwzMGf(CDpDE~tg02#j
z64$R?d#O@}FU8B1OFS2Qs+1L#RjVpjR<2UYuEm3=7SK(+VpTa7vhiEm^}u#eeEG_0
zn~dSR*I=xJc_L9g=<%+pt<VFlf@_T=Jlhi?e9j(fW**|&0`WHHX}%=_dnVOmX3CCW
zEKXnDaxua9kroSv1Ee5g9buh$LJ@qA>(PUqxVC^nTC_Q!2bd?==5L7w+JpYKX6Q=J
z)bR7?dQ(Hjzup8fSP$TN7FkDO5El>6#k9CR9%r7WXgfXrbDz$3DZps1wyQIe3or^A
z^5O+d&x;pk#;4U-0Y>XcUimppj^Awgxh!A5FTiZDJI{D1!08h??y~L7!<Uk(72@#%
zb}Ezelr^z{&1Z_}huQdo%=wpWyqL-PEgN^TE-~+A<F3s4jcmMx4J=iy5RVt&%Xc|n
zS`!P{=}gX(**IpYJma?jvt7P%U4So`<#*26cJKwWm<O}*Qnn{Ael|<x#pSF%ZH1=O
zLUueojhHdIf7oc2%_%?KH|%Ty>#&S}(xB}V!uRIjUV-P{&$a-!;aq?r$|<IG*23|b
z{_Gca<oyvW4-epK%hN9Vfaj~%5O6AIYN3i4WBeHq^f~+ppFBfF43C-o%GgVwgsW$&
z9Jxv0^nC)aufXF1r>_(Ur(IwQ{8<ikxxnd70tNoe270x?X%m-1JHIj(6FBX^Qt07V
z+;62k_EIT?`4t}#0?$|9Lqh&dm&zqq3i%;WD(9jS6X(z7!2VOn*9v(vm9bw5JSlL#
z7XbN}1y0{CQsB?%pwgemoGQVL?>&G{2>BfX=X(~w-xK&2VFxn;4@J1~Q#t87Itq1=
zqd+8IeQAHmh7%{1%TyA}gbO)7vp<tSK07P;g5$Iq<9aQ(Gjsgb3HjXl!O!Jqj^70E
z(@-z^8k+)F2kqrJhq5j|0Ip;qCNs$Kne)JN+|B~lsHj|!Kbs=<2FFcZ>vd`t+=&H{
z3#+t8#5k2YX9A~lJ#wBY6bZ`&J}KlK0$&OIG^uQ%8#z96-oB3G7<Y?RF1<wPgake&
z`oBECcMHcYhDEf07W|)Q!4Ctc`sU87FU=xP`;GbPRRo8Se@1fra5nIK@*d#I%;K}l
z$&TI0?m1u0O1+BPpT4hw<>7ins9%ld_$?xE`s|+q-+KW42FE#+b@`5v-z)l$CQ^R{
z+=Y5g<@ATgguFbLD)&I26gU%h_}&R@y(I8GIpus^*w5{6$AvtdSD>JRr|>?g!eiN&
zc}4mCeBLZL?f2$umm1(D_EJX28LVL=aK#FU_}94|jB8P^3W0acVt@B6_)j=Kb6xZ}
z@O<a#SW`@o>j|91Xu<)otJYrYudChEfb(zDr{w&4yT6If$;FvG`(}as{^qE^H5}a@
z2>YA$Xe{mzBsv+Es~zDWULSc@tzJnd+wutcY2j=xgZ)_A#&9H!FP!;nBdAz7Kdra3
zM9udfDD-x`#qviJI_n5U{B*jFnXh~NRsO0~{y1Lm;LVPoPUN*jaDvU6*pi3`19%TL
zon9r=tD&BwyTY0_Q}tq*Y{}Q(9Eh|AW6?xBQ-YOJY&wTLQkO`nDu1<qM>LPbDkfhv
z!3T+Wpp~EQLz!<iBX}mlQNU0nlW`@Z*J1pn5kL6nr)Vq~PvCJzreuV1QqsTfT5rA2
z@7qvAhaq{-knya>XlDMJ>o<7oYu7@`lEPmSG5`9yYu0${{MW2ox5?MwZ}6_E^N~4z
z(y%F>;C%pm_Q6M0wH`G5tV5ghgk+vrlpf3HSAJwN+i2#%WH#&%@@Fpdq-AzeM=aXv
z$75PB&rwcQl_w<43R!!Z*^OlJnnUr9fZo)`%+r_ISy<mcW~a<Jo0*$zaYUZer_<#V
z=SH*D&BLPEm^dDq4d$Ma^*6_({<c7*na<DFUW1ls#wf(FYL@k-=Ew{dYr|7lK@X+Z
z%-MNle3#wx8RiJ|HM3h~WHLj1dWcwa7QXlU@eD0g`N7tFhkMPFwAp2$W4GCed6G68
z^ZRj-SdJ7rVJxS~tn<azGEaLX>)<fVQa_>-xAIhOPEOow=Aqu~lsvbTzY3y#JR5#k
z&ha4k5vm0s_whtBmz8m%OYoz;GpXNTSHVsHp9F<-(-(0Y&Hz=2S@m_h%F}dV(9OB&
zORjzmlJLqTD)ZlMSGn+R0aS9%vfn6pNj4xt&kj<5#IAy4*ze$kse;n}B>73MPfN{Y
zb{!~|bQyiQAL77$mxA2CG&$3Tj1#5jBHBBY`OAHi?P5bTxBOC0QhL6my-3N+{gvH9
zzk@PCAePE0%TE}ETk#{GN`1Kx;}9F7jYy*)s*}-|_3uF%>8~Ln1i4?+CiLb0E9ui7
zfy`g(e+wA-RTOu6Uq@`X=I%gHUe=HVW7U`YKZ?*-DA5vJA};Cu;H~=VJe5nRVgq^w
z9>uId>PhTDM6CL9KS@8|ln}*`;@npm29Z7Zk-qf5+;{2{`qTfv1VRYXe<p`-8xXPT
z%l|7P|KEswzLxU2{(CY<U+!ZKEHyPA$x+D7KLxR3{HSkHkk7O7|BvL#=km`($BCcR
zm-}K<Vj-L>pUWSGF4bS^XZD|Hg9#y5K9~PJWaQZ3@}5Q_ng0(nNyZUm|CIW&AH0Dy
zYyNUSZt5bgHf#RU&P0wrlg|y6rv4t2WUf+A()W;J&0ju;?h*Pq&BtXk#t`{4eq>Af
zPd=YNBJ``R!iXz5JZ%J0{z5#HD)pCV5Hmuu6eKM{T;|Oa#AQ?%bVxyg=dj$496MJ1
oqjY0I*eJ&F7Z6wRlYW!)4~@@!`IlXW1cVVe@tJI<+=6BQ1CN{%!vFvP

diff --git a/roofline/src/roofline_fma_fast_fastmath_o3 b/roofline/src/roofline_fma_fast_fastmath_o3
deleted file mode 100755
index 1bd004ae0e874182426b34a514f0e54f85afc5c2..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 22872
zcmeHvdw5&bmG72hCn66^Ndp*2;S4BXhs4;yWNaG3QDP;>R3>R~2o!>>*mCS*TOK`}
z@W_y0o2lxkO<Sg))8T$MW#~+%!`yIZpkIb5SPo9?q+BqiLo-uKos@@Mht>&?7)av#
zTl*m&$ugw<zI*@5)_2a@YwfkxUVH7e_c>>uqy31-yUJ>@FqN!qIU`J;p>vWrGj{ZA
z9F?>(b`jIrmFx;O4U}9QIxit8ate+~)e71rpA9evN10U+%B(uiC^%n`nS#nff>hr`
z(q&e_$P$4n$e6|nBuBQ6-ojD98lF+`S!suz4C6`TXCjlFeK|(~Yj{S%QE8Zhs(sWp
zI;y4KYO4;6aGe0F;O7ZQyp-0y6RD&8-bEY*)Okk1GlI+%RB~T~9NBrk$tfAv$@-@9
zi&NUKplmL?p{a4x($X86>J~ROwnVos-deVF@zT<waIk0zZ#U^D-qp9S=QSxmYQV_m
zc{s?`RKBlu@t4m%@gLXjk3UrS)b*F|c>0&qze{Bo<52y^aw#{9TSN3qaFn0wn!lv)
z^xS`Y^7LGwS<};?W-H7CUCScz`V{={O+k-MLB9g#W~+Dh6!hz-phu_Be{>4^<x}Wc
z2!0JmW;zK%Hap8f*D{%b58=qh-#-O?+7$LwfnURsnKpxvP5;$X=ott99CjHS(rdUX
zHU3_f^n|3VaWW?90~!1b@w_?M+)5E#!hDt0Rld4FD6qLP90`P~t5!AzTLRVoO-%vj
z^KEVpw)nyke<<SfG2d-%zIBpsWs^S~4upkxR^ioas@D2i{ULu-Q=o~VKHug*#McsS
z_C*>(0e@XMMH4s`k!h_0CF_b|K_C=r3Hq9Xwf;zBu!R_Esc0z+Y;BCNU~8nYrH-{m
zBViVfgd)MFCWdN2W^Lh?)=*<hq@FdygkUZ4`a_$+hSr8y{gzN;BtXoO#^yk<zRv#u
ztCs@M1eRb3)l+@a0nPr#7RFmu3d^W<C4j5Fl`B^ImJ}^1x(WWKL2JV=ho2P`<Dcj>
z)^lYU<tbSpg)V0_j}+q=yyjYTVJ^D`M55^~!gJUvVnVXu_W~f5X*i;nH(o?F+-@p=
zLY4!_a%-x5?A970^SOQ&&xw<YAG+qyoTHM8AEIBXh=3{%iB9vBO8GM0rh_0&;zgQ?
zuG&ro1tz+RU7{D8=;rHSnTc+`4(KL2)vc0YqN_1OWou1zI53s&GSO$GsEpN^=rc|9
z1{2-v&sGzCwuyhMiS97bx0~qHw<>+pM4zLGfV)ieFPZ4wCOTbvR5IF6*p05-_m^tS
z=uAX%lmCGdqrK1mlHeG>={SnU7rc$L<0>6F!W*b?cxW7H!G9A@T{=9-@m~;5AwQhp
z_z}Wwgg?Xa1B6qT4)5Xk^Mq4Z4)5alzY<PeIJ}+XKOvmDZn%}>PZCaDHeAE;9}-So
zHN2MN-y@v5Xjtd?w+N@M87}7dql8nJ3~L;JkZ|gXVJF8!gwy45m~s3Ygj1IckDdcz
z!6w3~D~5-Fm%LvYd+k0WcHC%xdvtAeWq)E<q0Wr{UX6qPr89Mwybf|7);q4MU~K<B
zbi{gN@m>I<eJsz2z0-EZerV(FpR=)t3dV=P+(AUP@uhl7MGAkB%GVn^{=gOVp0h!L
z5gRpnleZd{L&j_8A{VE$%~aYPSCJlG|Mpu>-lSpp&sh(_WD5-2cqsQRkSrg(ME%>t
zaD!*G-^!x7Bj1NyO4&&g7$4j?{2*i~Z|pchBRTfMED(&?bL1_ft5X1G#B#q2twz6x
zt~dlo367L?d6Mm8p~z+JW8uj3_OV+d^GB{i04^ium}hL{QXn@7WOQUkawQ@oxf~~>
zD^9ArI;pznpCRwYUWxT5uSX}upmSStJo@)txhG0B_7%>v?W<k6Zvl&WPK*?Hc}^JJ
zMtt5uZj7|jGH7%`%b}j<pp7>)X-7@%o)Z@E3oQ>HOePSlM*Frg7MTt(3eP0}g-BhV
zF#`tJC?E3-0Uss&M&L2en?Q$1_h3r5<t)tiya^gxsOo$F3^@&)y+i29{$v$+pbUmh
z(8>jEz)V{pXbBiJ$Rc?F1H!xni(YATRV9qB^&~JzCKRWtE97>shhc+;C1Jn;RBvZu
z>&$w`9}}kx$HK(OWk&3P@!Id(hb+cL2Ne0hDaT!jQx50sC$41dl;gf$%I`AeJCLt;
z6bz#5wUnWMisF?fWq&F2P<W)LhODRgB$@bZe7t8D(Bxyt@mgY@U80qqZomP!ZVzAr
zaIX>f?D}`=?&o=*!w;M0LiBp!@m;dZx}~@G0FuM@0w#I}sgl1yCCLGVc)j8HYf`x%
zjzu@GhjXLSx&6r!FeIPE>6GJM&;C=6IdpPX(J6P!mx+8gIjd@aPcqez)1UH4FltAQ
zcqh$jUh3|yZBSeV#p?%YR=%?;)>|1Hdr?^5pL}v2g<+o)AtVH8y8Apq*u*HN8Z!lH
z9{fB(ayAJTXA05;`FVmiDR>%J9x_$7j&zMcX%96e`7Jb_&mZk4bR)Lys1b`IN)fFw
zidqkF@&!e%CkaS>ItoY~For%IO7GM9FQKP4KqXj)z;YBY2}mPo6p;D}!_<@5b@(M&
zt43N<Zhhzznm^98Z#%;xj-ImX)yO<YZGldVfWtH}NKsM^2O0_&|L6+dyEHU<P0C~2
z$n50vbO}n;;z2F2VvNrtkK`tsgm$uw8aMVu^0^uEi_G#nwvD1=9%-BRB7vTUR2y!V
zTvB^i?jLT@(6hbIUq`l4eDv%Bb8<Sg^Ue{IAa&!ejlx|!PPBs0h`rE_c^0$$U8ur{
z{cQ~g@df~IOoVVOz7<$UBBD78J=-0HRl6L8>-RVcH$LMi3?v+d&4Z4@=#Zmu+o;jb
zwJ}G>cTiJ*XBp(|Jhu$FV&u?6P9xT-12y8E6zISVj>U=F==xi3PrP#v%zE($B!VD&
z<n;J>XX1hBhd3EpICT7!!K&+wu8YTu(EP64B!PIGi|$AJAu1Za5Aw;&aq42Y7b;0Q
zcovGMjgbYtQ1LUN;>Q3l_L^nG;O!xoCjSXJ9%`{|W73W8>N_x`H6HH$o+P*^wh*iN
zM(3{`9ls>8xe)W#4keF#jJhy0#E$ZbtMc@RRI?cyG|CS-9{nyUaqK&3#7@yf9-n77
z$_E|o-z2gaI!PlwOJ4A(F5PZe254Fu&8NVfLndUhI3>$}PR)`pS)Nxc?VhB?C|{op
z?HtMDel*&9EyI55ruIJV9{0T>l&18-UBU^pVvwAT#*-5cesq3LILcd#@wx9cvlCvO
z(h1D4pc|>(r6$*aIttIgD|D5cW69zfI4_GJ`Sgn<;nQrAAg4<zKRO=uz;|X}UT*f~
z#JwQhui-9iD;(F6Fk;WT26(d{cMWp5(?#C##yVU>9PV^c;6;psV}%_qx*J1gr;G04
zka^tY6qzSnn#ermDi)ciUAo9T>sl)^KXcU}V|0#3yhgmkwVjaIr$*fCvK#S-T^b*d
z{S_{|9km#-3Ky;f9j*ir5NGiL4q|@t)emUAZr!*YGAw#=@_z#FD(f7N7K)MN`w&sk
zbxYax8pf;Fr9IEE8u;RLoJtZ&D+)(mPTntg<`EBOHqPM`979ky^4y^cC?9#kan*Lz
z&%E(!7}@^zkb68dMx)fd-d*ipN5j);bouOjJl>B6RmRRFAAtgIZ8G_nfF-{$YH=+(
zT6y{-qx`k-G^0IX0nM>*z_1+g#%DRb<wqRtbT?4hwakU^`jrto*?#OQqj%V5oc^!&
zV>Y9_*Kn*Jz+Hci;aD|bbRKqeTnzq$pE;rJ6U)KR?oJ&1Y|b<z9<oFXXgCe3l}b$|
z)<}^o$;l|OMdwjT1~tbhA8>U15G}8a9WdIHmde5-M&UqJe8ri{-lVk>t$;z++vD>)
z$vtf)XiZ516UVPgesS<@K@Lg}SPq_rWoHWsNobYjgInJ5THdNE|L=(Z;Mf);o<o(Q
z8QgsI)GAA5d2d@G3i>k&tW<zt$SA<A!aeS<yYF>x+*ldkG=>Ke^yM4Fw;}9N!N_k-
z&p(s<Z-e_z)rVa^F_nD4`|oVB=cJfTwg67_--qV@i@ky_97TJVpbzcd^23hy=Xf7(
zV^v-2VQTe~9B<t9f5<siT^_s9d)!tTFP#gAUFVILUi#Je{Hxc*7r`qZxE<f`_m+R4
zJHCI&YdKJH(Yea<*NIkD{*mMRz(3VH-f(nKY{BEx9gq2u^Tyt)jJ?wSF2dvw2$Q$k
z-$j@lLYRyiogYW;?dyHWX*1f-S-kPTLVP%{=sR$1w}*i|#}-+BTYTQ>z8w!aU8lfe
zL$S5*(f`_=GYV#e%^YiVzT#MPN|e4PO2IqFY7~wdmXn^6ac}&tGe&&=tCjItAC`Pv
z**jvbiZ8NP#a%Nh+h4L6mQSGNe^C4TdLMA+lvcL)Tl!wwmhUZfT8L6{6C<?ZCZNYZ
z>dv{5BBjJ=KcrQa|2Fh<T$G^r5)p&w`B$k;IRkWFcBJok?`{MRE=6%S51OsGj`dSn
zDT1|^f*&DsxbMK>-8nQitZf^Viqaz}WV~d6G$2r947u-f`%-ZPCQ;?QBYg*kcjpjq
z&Z~5Vg5Kz{k)O+Xf#R$OaW^@>vdj;6;AWQQ2j3sX{BSAYM0|bT{BRjb{t5HLS;PpM
zfCoIx4<Fp!D+U=2v`Gib#N6TNpoKou{E>H_lGaGQX>K`RDO<Ey&O7HTA?Bn!8fND;
zCw(n_PMRg=q>tbYJ}=SSH0eN@X}&rv<|{|X3-HXuoMk@`$E4;nznr$tSBTGbi~b~`
zDdxd9aF46je<)6W^||w5mUtb+V?V{~z7rx|Hv^tuy#4@^Rb4qQ4B^2qyqImL)rENJ
z3?ud{KFG)Vj(@T{apCmm`=IxsIXQS{?8p7}IY;L|(jypM&gqRo+R&+tKX}II{iBt$
z;&Sed&!<Rdj>jIR%MD%9sRvD$^uFU~yU{flPJgBkQqzl3{;A{9AERgZ6!W^H<6AUk
zk8P3H6+RF~FN}dde+=+Bga!iE4G!U|ebVtad>~{_mSTOI?i-I`iO)P={?asGx~tdU
zf%|+X?(-w}7_qa-zk5&IBi;cxF(;<oBZ_-`dcG__5WW_Re|U1PvWxoY{^AW>jdNb3
z8)e+=;419Nk0$Tg=)RZkV`Y8xToKts^<6LP``x?eSKngP#U`HlsJ@A+%r^mZJkIDb
zUak3kfw}&#Q~m#pdg%$f_B`uf@J01gy~DD(I&<T8yejqz-H}(u{u935ae~$ajIJm6
zBc~C2*p(lPxaJwLSuQUg0`0iD{tZo4-q?G|HDp`tJtKB7S%Iw4iI*#Vj*f0jcUYwE
zf|#SD3kYtA9Ub4pt(4=@moeJf&siNEHlDFLI@f^#ceVUMj@I8~T*QCo+6AoS<0M{V
z;HlHmc@r29ezGB9#Ge&KhYz0IaJb|bJ&E}`>pS+@!VXtAH|$vs`;L9Ea3^Q%JAQI^
z?mc+6?fcO~b1Kfl^rtyJ`6@IUsA#*WNRt)43`?mZQZMv=6AVHhv2uMgN#72H`5C&w
z_LhI<;0s>+iNWad!fWn)BZkF!JcKsdy)nPj*wII>CD*$*(oHYLn`n7tK*c{jkGG=H
zkH0M%txH=bHQIT(o@$5}YYEhJGPc6Cl`QqTTDh5ZcoQd^og9Kr(q9AprBeS8&3>k~
zR=663cnud{K;r4G{5qcUhA+m6zn@Y)_Y+jl*EgWQ82W$tJ7GVqHDv9y(Tk_Dzf2fV
z%nf)7Zx)sPnzUbo{)p7CO|rj0h-+LtK;o0_-w!D^ydGBpWj`&aD*M+$|2(Nb0lI-3
zD&?bzen8+obE%F8UlDdj7Y2vufV+eRYq<r#%S|;v#T|9mUeXKEJ%Gv8sEj&BN5y}4
zoE#VZs;W?^G=xf>tK3B@(;G!AQ2eFK@OHduLR|Iai#3&<?_-&>LFBh1@5P(X*dKa|
z1*HX)^!JrmSIIBIpfX}u6v4o^_fSU}v3HYC!Z0il=JOYvsrQ@UFPPvJ4ynfaQ4Avh
zQlu)`AS4U;I>99U6v(81#Z>w;?4i+4{Bx(`FDp5WSvNOCceiLkz7da*K@ZcZ+NJf3
z{)JA@u;?z_Uw!5T<XAy6aSZDhL?{r&Xl<EzojWm~qrK421#_K~Yv`%DS}8ZF){+6G
zS~IDpc;jOOIx{S#Dtd;p>$)tvt`0;J<MX*YHcs=oH@@J-c^Y0m#}^a4qcZj}rg(4R
zJ7@#eMb6ri`+kcq?R-C?mHb-GKi=5yy|IrgpkaL8n+B~Xl#fJz4^P-l^OouQ{J~4`
zGU)>cvW&iiF{kj&p_|$5!Is7Jkx#Hi!^b<(=0HnCTM(ufE^I?%Q<D~owrE=#BMm4G
z`5(~2jSmLGTCiSQ@L*I6GzExXa7+GLe^YI=$sY-5&5bRKv}XU-MVh~DvsNDrX$$H|
zonmKeS^}gg(g0rm1yNY5_ct~L>Sh#ae(e@*LF*!I(=Bwa#ToviuZ*C%PP<W96V?iD
zT)3cUNxfFJg7_)7dPN~O9`>(-{f#Y~6)$f^U9`FR0j<s76b*!nwCaY&Fj|LZ`kTT*
zZHqtD%+0Cw*ER&05x^%=kdL;)zS@SymVoeHl|ZxxSsn=n??>gO^=NW!C>RdZ23zXF
z+#t{tTEOd%1i}$*O{I2!Ak-3Q5~fz;`zWnB;15Sb+@D%wb89dhZp0U2MVk5hDNTGd
zwWg&>bcD7k5ZMw4v}h$-BuL&RFK!9c6$v}n-{D^Exmhc^AruVO!xI5*aadbMJ-A$3
z9MYCiUo6*Vum$0p(dDU(a-UFwuV5&%y`#t-fB(5Xrw=`|TojDnDj?GZwCnyxYSm<Q
zw+FRww6;OE8C6Na&+C8wNgVefx<3GS@(2)uf2zNf-9Ow->ECa)Tw(jl8K}Av$JRG7
z%y7l&{>}J!KH#_C8Xu>(bgy6r+XYyG>HitP9{~;m9s?W&ycO3w`i?aWSPZxi_a|!s
zp9gFOd=d9t-GKSY@$m$pF+4tg6wp3`dTfj}V0Nzo%*X4-J%9#Y(hLFa2c)^H3@>kV
zz`cO2fcx=UmOh?H0v-j-#|!mQKm(9IwAc!$0UiY`11!LO`&yy{(icHVz-Is{;Z-zG
zmIv1{%T}l5idoa`yDZb4M4yj?e6mom@N;&4#$1xW;Mb@dK4E#zReAZhIc9FLZ)dk&
z{*{F{x~?V#q_YCYCGat|iMjKfk6BmFR^OIUc^!_Wm^|A6Q{~pbf@MO!1LU13-<Vc@
zSE~Fe&<~?rH<v%ei+TNhpbw+`K6APC?v(u7pkIeRFEy73C**U`zOSHs8MQNu{Vvcq
zAU0;2<yQ%L7ibkIUx9MclagP()7GA29TQx_pG_#g5;5hS#Ankjl9Br3ag?t`tQ8X9
z>b&g;9kvX|)4&@Ld$jP*JhNu7O6N3d8>@8K+8U-gXItC0I%iqiT0KsiwXMeXh-C(`
zKyL*MuYkx2$R+VuwGpvulCOgNUDKQnlFx%YX=)%%HEz<h)@HMkrYf$9>?kGq-;a;~
zI9<;ac5LR-$_|o`K&}-_YYQl5zpx!|=csn9Mmy+*K7E5<gI5*J$g>rB&Yf0wUjAb?
zcb*212lDdKg)8z9Ta<|ZZSd~_e}o8`{7sYj=fUBl;BU&tUzg7B1y2R|i%*S@TXTf|
z?^svl<$v4e$<yLFEAk3<=2qktKQ_&sSJpm#b>67Oc8_I7UYR?u*qv9fB2NP^Y=Xhm
zO?>=5PW5B1xQ6Pl$kXmaV<;}32HlQ1W3iwYH%NLP=o;vm{BMFzb4@1wL(tcP{xzXT
zw|evJq<<Fd?*^UPgnD)BYC(5_J_x!&bR$oH1ocupEd$+-xA6g@qYabVZBV-nYIjwh
z9?I5c-mVVlDTW>jrwi$V2QGNvf(I^m;DQG(c;JEuE_mRA2QGNv|Gx*EGj(3+=nMgs
z1@!1iDHA>+$SL{@5}zd@yGZ9b`U4TA#|w21FXLzWzjY1I;ahQ@<l`H=T;gN0e11mx
zN9V?aMC!Kc1n|#4k?grT&!}~X0jt2(0@&AO1L&EHQoaCb|HUH`9+gDWYdBQTDBV(?
zo|`D)eGN~SO4u-6=Q%~UOI)?9M)F-Gp{gIRMnrq2>AVE*7dWo$z>*TjH_37-{{I}P
z|FLIQv{d%%DhcnDuwKG82_KX22NM2+gg=$=kc4kY_^yN>OE^u&;UyAYE8$WJ$*$Ea
zSKh1@;Gg1!+D%0_7L{mAic6LjmliM83f5s^+3-gMeeu#l#)`rZG)MfK0P)XP4jWVs
z|9cIDT3Jy`FcK(ouc%xc@o$#c=9XyDrf4Jn&fZwZiU@1)hZ|T?-2*MqCtxHbO4<UU
zF#YS6vq{8<vQVJOPXaR6+7w|$jV<^;Zc!w#73X>&C=1s4BYsvCXz<mC{LKMhLmgxl
zCM0}(-Cjtj{5NXB2JL>VSEK466yj9GYY`eZg~P0<HrPzd`<dODhTAZj%dC1DjSw5&
zHB?)5%BCgyhi=5Gr_l(ZfAXsFX)d&3v<_HxGl}PI_;;-uzve<4Mp?GFw=p~G&N3cs
z*nL8cyL3G>@J~tIOyYSPo5|EXWiGU_Sxl30m`<Oa+W#V*o|l^6(&-Mir%*SOc;3dG
zsr?(wg*N<SeedFI{Bu+DWjg<tn3^Zk>6oRK*O*B>Z^LYtZCu;%4`%g0XS0xv{=qEg
z!E|~)+mn@k8B1iPt66<Q3r(jv?Bw_aMdaxEVWCx`4EyQ2VP&&ft7-g`1ic<1`ZF1H
zx1?uY&o+Q=!M*@VX{VglSv{vu_Get`QP)SHJnhEWf*lFebxe}^-zBrPs}~GpXJVet
zS@~Koaz`Yc-u_bJYrTN);o$AUt8Yqt4Hs}a3?@3g&85WGWC5>{bb4z`iLbQ+u8?$k
z8B2+;p#nBadd<~3Fs#w?Gh^S8^r5SDPHB?#{|$P!cKuB9zvmSErINoFknFr-j-d0k
zcF+z;{z}O&N*VjPq{k$kukC~XbxEiHC{p5U?tsT7{e(j&GTX?{_#>^Pw@EsG2Z6E=
zC4Ga`gOSfuE-vt7C;eZJ5?`wa1bf!@=7ZmYaoV~>2ggLZmeVK4EpcS?+Y)|bF?Z7V
zUCH%K9;aT(pBZmH&OdpaMnRv8cG17CDRFVYPEO~fwA0^%uBA}~(!=SK$NLekXExhj
zpmRpPhK$_1Q|KR?g6^=<-VFKdsf>B!63}V?4mF=($(W~wl0GW=?UG&!`dpbegSDJK
zd49f=(`U1q>vYb3jTCB>^tIOrI$z^Q*#n#|nM@!)1^pkVpdSLA+M79lzCH#2=oIu^
zIE4H&px1C~lpihwJ)0kjK-WxmtL!bL$Le5vGp<v&as3n5FQ7c#jSR)rXolZfB%S`=
zPbt99jQusIOT+}eC;10u{1~$AC!jmgt})pzC9+TQtNo`I%7ULtx+eATcN{2tUD6X7
zcD^I^XU5w}$xr(dP||@@`WR5-x%AVF-0Xh7ateC!6!Z$v=UDUE-i-Fr``&EzeuL}5
zxV}#31}&CmZRd1}m_YXw^#8@_m<2D=P270b270!Ab69OC5{^XiMM^C;dbw@oI-j@l
zj%w_4GqD?vFVgI*rTuWiOzm-FBKdrELEq-4;3j{QuPzb{g?;|$R)*Ws)}{bH11VZs
zUP?RIWMT5r4YIi1_2Cvagso@z4Kn}y10p8ONB7lftl~Qc7`+K!qxgc+2oqmx_)2^w
zOMPK{oq-QCe6-<CeG4|IF&EZH!vQ}&Bbg{KQRQV2&yZbWE}JZRqbfGl>#Os(Yz~Bi
z(QwLwQpGk=Ll&-^6jzC_%-0so!m*U8??lk_XxP7*Z{7o&9~2op@?fW*#+FpY#jLfl
z#YemQF#4>8fBeBW`0-IT6bMK0*rG~OQrHp6w`!fc%H#9gUO}54@yJm5w9cqwzKXkV
zcUM)e1ed9Ve;mSmtG#PhxV^qLt5)6NsrFU7S9m?7j_)W`8;<fA0G__Vk*=FJ6#Bf)
zndnJK>`bIwm(9U^Yov6gsg04+$-V%8wh}ukr5Ck^f}4GK91CRGvPsv~PDr9b=5Q17
zq|B>p47d6twGB+{vXowh`8%TYl1Y0oWfq$pk!6>u3H#*!Q0d}gQ>b*B+y*M0n7J30
zuPz+)HTYZVny^n+<r?%v9ik26t4_6-x+67e%pFgmf>5Q;#nS6Y`7V73Fw7B%H4#=R
zp43>M7%S$QrSE+{JS$64zJXP?O}xa8S?RXW_FCx_v13*`&F900VQQq%PGV}BOxs7y
zZ1aRi(l!W#T8bmuSxfD{l~EJ-n%KN6y(De2|1%I;ue0F5?Ho^Py`W4I)OwvP=DaFT
z@ERPnMyBMetvaxY{SY;dO`POqI7KKy&MY6X>b#7;F`?uVAa%}^uRsYtZ3wFR)jCDD
zBos?YTn2ek_f-;Qw<CkAX3G8ns}77|U4sLrOGxXG<R>McFN)cnfHoYeezoplmldjY
zO36OSfAm~LYlKQ(t&ePy3!j<xD?SD9hdix0DqO9jbW8bGSuvqxr?Ov?*@MWCPnEn{
zZ?VgT&l<^?=|9!}4k^Dv@~L&11}U%BSxKJO0#yA<{@b9DU!}Pd_b+l`HFKpQv;Sim
z@@kz&lkyrVHzlXcEBH8YbN%`Zopb0j1mn7^1XO1wr|=&lW0qIz;*S>!4rzvB=e|xM
z$n3#E^2+~eeQA%BpV+4qOi0Rq0>d|m$e88T{-J7L(KCus!c71DG*e!#SM8NIqAzC%
zWY(X6d>#&pElTQnR_!yI$)Ab83>gOwC9l@M#%>{PliQmqe-u2_ekGq;XIf6uUo8I{
z@TjrD`Q<v5+6R`tC#9s~Q`N8H;9ZoN>sRY;WAY1J#m|-Cq;LgCGUS<FBUp6-_6U$V
zD>(%}M2We6^&GlK7a|$m$9YrAkogn`=~Dhv&*x9+AR?6skUDFroXF6EAJt!+L09sN
zGU(EsDLdw%T-9x69n=M8R|2vy1Lb<u*fGl=bqlOaj^i%_r7lr^Q}Yjv&usNO<pRin
Nn)rmel-&|#{}<g(EinK9

diff --git a/roofline/src/roofline_fma_fast_o3 b/roofline/src/roofline_fma_fast_o3
deleted file mode 100755
index 1bd004ae0e874182426b34a514f0e54f85afc5c2..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 22872
zcmeHvdw5&bmG72hCn66^Ndp*2;S4BXhs4;yWNaG3QDP;>R3>R~2o!>>*mCS*TOK`}
z@W_y0o2lxkO<Sg))8T$MW#~+%!`yIZpkIb5SPo9?q+BqiLo-uKos@@Mht>&?7)av#
zTl*m&$ugw<zI*@5)_2a@YwfkxUVH7e_c>>uqy31-yUJ>@FqN!qIU`J;p>vWrGj{ZA
z9F?>(b`jIrmFx;O4U}9QIxit8ate+~)e71rpA9evN10U+%B(uiC^%n`nS#nff>hr`
z(q&e_$P$4n$e6|nBuBQ6-ojD98lF+`S!suz4C6`TXCjlFeK|(~Yj{S%QE8Zhs(sWp
zI;y4KYO4;6aGe0F;O7ZQyp-0y6RD&8-bEY*)Okk1GlI+%RB~T~9NBrk$tfAv$@-@9
zi&NUKplmL?p{a4x($X86>J~ROwnVos-deVF@zT<waIk0zZ#U^D-qp9S=QSxmYQV_m
zc{s?`RKBlu@t4m%@gLXjk3UrS)b*F|c>0&qze{Bo<52y^aw#{9TSN3qaFn0wn!lv)
z^xS`Y^7LGwS<};?W-H7CUCScz`V{={O+k-MLB9g#W~+Dh6!hz-phu_Be{>4^<x}Wc
z2!0JmW;zK%Hap8f*D{%b58=qh-#-O?+7$LwfnURsnKpxvP5;$X=ott99CjHS(rdUX
zHU3_f^n|3VaWW?90~!1b@w_?M+)5E#!hDt0Rld4FD6qLP90`P~t5!AzTLRVoO-%vj
z^KEVpw)nyke<<SfG2d-%zIBpsWs^S~4upkxR^ioas@D2i{ULu-Q=o~VKHug*#McsS
z_C*>(0e@XMMH4s`k!h_0CF_b|K_C=r3Hq9Xwf;zBu!R_Esc0z+Y;BCNU~8nYrH-{m
zBViVfgd)MFCWdN2W^Lh?)=*<hq@FdygkUZ4`a_$+hSr8y{gzN;BtXoO#^yk<zRv#u
ztCs@M1eRb3)l+@a0nPr#7RFmu3d^W<C4j5Fl`B^ImJ}^1x(WWKL2JV=ho2P`<Dcj>
z)^lYU<tbSpg)V0_j}+q=yyjYTVJ^D`M55^~!gJUvVnVXu_W~f5X*i;nH(o?F+-@p=
zLY4!_a%-x5?A970^SOQ&&xw<YAG+qyoTHM8AEIBXh=3{%iB9vBO8GM0rh_0&;zgQ?
zuG&ro1tz+RU7{D8=;rHSnTc+`4(KL2)vc0YqN_1OWou1zI53s&GSO$GsEpN^=rc|9
z1{2-v&sGzCwuyhMiS97bx0~qHw<>+pM4zLGfV)ieFPZ4wCOTbvR5IF6*p05-_m^tS
z=uAX%lmCGdqrK1mlHeG>={SnU7rc$L<0>6F!W*b?cxW7H!G9A@T{=9-@m~;5AwQhp
z_z}Wwgg?Xa1B6qT4)5Xk^Mq4Z4)5alzY<PeIJ}+XKOvmDZn%}>PZCaDHeAE;9}-So
zHN2MN-y@v5Xjtd?w+N@M87}7dql8nJ3~L;JkZ|gXVJF8!gwy45m~s3Ygj1IckDdcz
z!6w3~D~5-Fm%LvYd+k0WcHC%xdvtAeWq)E<q0Wr{UX6qPr89Mwybf|7);q4MU~K<B
zbi{gN@m>I<eJsz2z0-EZerV(FpR=)t3dV=P+(AUP@uhl7MGAkB%GVn^{=gOVp0h!L
z5gRpnleZd{L&j_8A{VE$%~aYPSCJlG|Mpu>-lSpp&sh(_WD5-2cqsQRkSrg(ME%>t
zaD!*G-^!x7Bj1NyO4&&g7$4j?{2*i~Z|pchBRTfMED(&?bL1_ft5X1G#B#q2twz6x
zt~dlo367L?d6Mm8p~z+JW8uj3_OV+d^GB{i04^ium}hL{QXn@7WOQUkawQ@oxf~~>
zD^9ArI;pznpCRwYUWxT5uSX}upmSStJo@)txhG0B_7%>v?W<k6Zvl&WPK*?Hc}^JJ
zMtt5uZj7|jGH7%`%b}j<pp7>)X-7@%o)Z@E3oQ>HOePSlM*Frg7MTt(3eP0}g-BhV
zF#`tJC?E3-0Uss&M&L2en?Q$1_h3r5<t)tiya^gxsOo$F3^@&)y+i29{$v$+pbUmh
z(8>jEz)V{pXbBiJ$Rc?F1H!xni(YATRV9qB^&~JzCKRWtE97>shhc+;C1Jn;RBvZu
z>&$w`9}}kx$HK(OWk&3P@!Id(hb+cL2Ne0hDaT!jQx50sC$41dl;gf$%I`AeJCLt;
z6bz#5wUnWMisF?fWq&F2P<W)LhODRgB$@bZe7t8D(Bxyt@mgY@U80qqZomP!ZVzAr
zaIX>f?D}`=?&o=*!w;M0LiBp!@m;dZx}~@G0FuM@0w#I}sgl1yCCLGVc)j8HYf`x%
zjzu@GhjXLSx&6r!FeIPE>6GJM&;C=6IdpPX(J6P!mx+8gIjd@aPcqez)1UH4FltAQ
zcqh$jUh3|yZBSeV#p?%YR=%?;)>|1Hdr?^5pL}v2g<+o)AtVH8y8Apq*u*HN8Z!lH
z9{fB(ayAJTXA05;`FVmiDR>%J9x_$7j&zMcX%96e`7Jb_&mZk4bR)Lys1b`IN)fFw
zidqkF@&!e%CkaS>ItoY~For%IO7GM9FQKP4KqXj)z;YBY2}mPo6p;D}!_<@5b@(M&
zt43N<Zhhzznm^98Z#%;xj-ImX)yO<YZGldVfWtH}NKsM^2O0_&|L6+dyEHU<P0C~2
z$n50vbO}n;;z2F2VvNrtkK`tsgm$uw8aMVu^0^uEi_G#nwvD1=9%-BRB7vTUR2y!V
zTvB^i?jLT@(6hbIUq`l4eDv%Bb8<Sg^Ue{IAa&!ejlx|!PPBs0h`rE_c^0$$U8ur{
z{cQ~g@df~IOoVVOz7<$UBBD78J=-0HRl6L8>-RVcH$LMi3?v+d&4Z4@=#Zmu+o;jb
zwJ}G>cTiJ*XBp(|Jhu$FV&u?6P9xT-12y8E6zISVj>U=F==xi3PrP#v%zE($B!VD&
z<n;J>XX1hBhd3EpICT7!!K&+wu8YTu(EP64B!PIGi|$AJAu1Za5Aw;&aq42Y7b;0Q
zcovGMjgbYtQ1LUN;>Q3l_L^nG;O!xoCjSXJ9%`{|W73W8>N_x`H6HH$o+P*^wh*iN
zM(3{`9ls>8xe)W#4keF#jJhy0#E$ZbtMc@RRI?cyG|CS-9{nyUaqK&3#7@yf9-n77
z$_E|o-z2gaI!PlwOJ4A(F5PZe254Fu&8NVfLndUhI3>$}PR)`pS)Nxc?VhB?C|{op
z?HtMDel*&9EyI55ruIJV9{0T>l&18-UBU^pVvwAT#*-5cesq3LILcd#@wx9cvlCvO
z(h1D4pc|>(r6$*aIttIgD|D5cW69zfI4_GJ`Sgn<;nQrAAg4<zKRO=uz;|X}UT*f~
z#JwQhui-9iD;(F6Fk;WT26(d{cMWp5(?#C##yVU>9PV^c;6;psV}%_qx*J1gr;G04
zka^tY6qzSnn#ermDi)ciUAo9T>sl)^KXcU}V|0#3yhgmkwVjaIr$*fCvK#S-T^b*d
z{S_{|9km#-3Ky;f9j*ir5NGiL4q|@t)emUAZr!*YGAw#=@_z#FD(f7N7K)MN`w&sk
zbxYax8pf;Fr9IEE8u;RLoJtZ&D+)(mPTntg<`EBOHqPM`979ky^4y^cC?9#kan*Lz
z&%E(!7}@^zkb68dMx)fd-d*ipN5j);bouOjJl>B6RmRRFAAtgIZ8G_nfF-{$YH=+(
zT6y{-qx`k-G^0IX0nM>*z_1+g#%DRb<wqRtbT?4hwakU^`jrto*?#OQqj%V5oc^!&
zV>Y9_*Kn*Jz+Hci;aD|bbRKqeTnzq$pE;rJ6U)KR?oJ&1Y|b<z9<oFXXgCe3l}b$|
z)<}^o$;l|OMdwjT1~tbhA8>U15G}8a9WdIHmde5-M&UqJe8ri{-lVk>t$;z++vD>)
z$vtf)XiZ516UVPgesS<@K@Lg}SPq_rWoHWsNobYjgInJ5THdNE|L=(Z;Mf);o<o(Q
z8QgsI)GAA5d2d@G3i>k&tW<zt$SA<A!aeS<yYF>x+*ldkG=>Ke^yM4Fw;}9N!N_k-
z&p(s<Z-e_z)rVa^F_nD4`|oVB=cJfTwg67_--qV@i@ky_97TJVpbzcd^23hy=Xf7(
zV^v-2VQTe~9B<t9f5<siT^_s9d)!tTFP#gAUFVILUi#Je{Hxc*7r`qZxE<f`_m+R4
zJHCI&YdKJH(Yea<*NIkD{*mMRz(3VH-f(nKY{BEx9gq2u^Tyt)jJ?wSF2dvw2$Q$k
z-$j@lLYRyiogYW;?dyHWX*1f-S-kPTLVP%{=sR$1w}*i|#}-+BTYTQ>z8w!aU8lfe
zL$S5*(f`_=GYV#e%^YiVzT#MPN|e4PO2IqFY7~wdmXn^6ac}&tGe&&=tCjItAC`Pv
z**jvbiZ8NP#a%Nh+h4L6mQSGNe^C4TdLMA+lvcL)Tl!wwmhUZfT8L6{6C<?ZCZNYZ
z>dv{5BBjJ=KcrQa|2Fh<T$G^r5)p&w`B$k;IRkWFcBJok?`{MRE=6%S51OsGj`dSn
zDT1|^f*&DsxbMK>-8nQitZf^Viqaz}WV~d6G$2r947u-f`%-ZPCQ;?QBYg*kcjpjq
z&Z~5Vg5Kz{k)O+Xf#R$OaW^@>vdj;6;AWQQ2j3sX{BSAYM0|bT{BRjb{t5HLS;PpM
zfCoIx4<Fp!D+U=2v`Gib#N6TNpoKou{E>H_lGaGQX>K`RDO<Ey&O7HTA?Bn!8fND;
zCw(n_PMRg=q>tbYJ}=SSH0eN@X}&rv<|{|X3-HXuoMk@`$E4;nznr$tSBTGbi~b~`
zDdxd9aF46je<)6W^||w5mUtb+V?V{~z7rx|Hv^tuy#4@^Rb4qQ4B^2qyqImL)rENJ
z3?ud{KFG)Vj(@T{apCmm`=IxsIXQS{?8p7}IY;L|(jypM&gqRo+R&+tKX}II{iBt$
z;&Sed&!<Rdj>jIR%MD%9sRvD$^uFU~yU{flPJgBkQqzl3{;A{9AERgZ6!W^H<6AUk
zk8P3H6+RF~FN}dde+=+Bga!iE4G!U|ebVtad>~{_mSTOI?i-I`iO)P={?asGx~tdU
zf%|+X?(-w}7_qa-zk5&IBi;cxF(;<oBZ_-`dcG__5WW_Re|U1PvWxoY{^AW>jdNb3
z8)e+=;419Nk0$Tg=)RZkV`Y8xToKts^<6LP``x?eSKngP#U`HlsJ@A+%r^mZJkIDb
zUak3kfw}&#Q~m#pdg%$f_B`uf@J01gy~DD(I&<T8yejqz-H}(u{u935ae~$ajIJm6
zBc~C2*p(lPxaJwLSuQUg0`0iD{tZo4-q?G|HDp`tJtKB7S%Iw4iI*#Vj*f0jcUYwE
zf|#SD3kYtA9Ub4pt(4=@moeJf&siNEHlDFLI@f^#ceVUMj@I8~T*QCo+6AoS<0M{V
z;HlHmc@r29ezGB9#Ge&KhYz0IaJb|bJ&E}`>pS+@!VXtAH|$vs`;L9Ea3^Q%JAQI^
z?mc+6?fcO~b1Kfl^rtyJ`6@IUsA#*WNRt)43`?mZQZMv=6AVHhv2uMgN#72H`5C&w
z_LhI<;0s>+iNWad!fWn)BZkF!JcKsdy)nPj*wII>CD*$*(oHYLn`n7tK*c{jkGG=H
zkH0M%txH=bHQIT(o@$5}YYEhJGPc6Cl`QqTTDh5ZcoQd^og9Kr(q9AprBeS8&3>k~
zR=663cnud{K;r4G{5qcUhA+m6zn@Y)_Y+jl*EgWQ82W$tJ7GVqHDv9y(Tk_Dzf2fV
z%nf)7Zx)sPnzUbo{)p7CO|rj0h-+LtK;o0_-w!D^ydGBpWj`&aD*M+$|2(Nb0lI-3
zD&?bzen8+obE%F8UlDdj7Y2vufV+eRYq<r#%S|;v#T|9mUeXKEJ%Gv8sEj&BN5y}4
zoE#VZs;W?^G=xf>tK3B@(;G!AQ2eFK@OHduLR|Iai#3&<?_-&>LFBh1@5P(X*dKa|
z1*HX)^!JrmSIIBIpfX}u6v4o^_fSU}v3HYC!Z0il=JOYvsrQ@UFPPvJ4ynfaQ4Avh
zQlu)`AS4U;I>99U6v(81#Z>w;?4i+4{Bx(`FDp5WSvNOCceiLkz7da*K@ZcZ+NJf3
z{)JA@u;?z_Uw!5T<XAy6aSZDhL?{r&Xl<EzojWm~qrK421#_K~Yv`%DS}8ZF){+6G
zS~IDpc;jOOIx{S#Dtd;p>$)tvt`0;J<MX*YHcs=oH@@J-c^Y0m#}^a4qcZj}rg(4R
zJ7@#eMb6ri`+kcq?R-C?mHb-GKi=5yy|IrgpkaL8n+B~Xl#fJz4^P-l^OouQ{J~4`
zGU)>cvW&iiF{kj&p_|$5!Is7Jkx#Hi!^b<(=0HnCTM(ufE^I?%Q<D~owrE=#BMm4G
z`5(~2jSmLGTCiSQ@L*I6GzExXa7+GLe^YI=$sY-5&5bRKv}XU-MVh~DvsNDrX$$H|
zonmKeS^}gg(g0rm1yNY5_ct~L>Sh#ae(e@*LF*!I(=Bwa#ToviuZ*C%PP<W96V?iD
zT)3cUNxfFJg7_)7dPN~O9`>(-{f#Y~6)$f^U9`FR0j<s76b*!nwCaY&Fj|LZ`kTT*
zZHqtD%+0Cw*ER&05x^%=kdL;)zS@SymVoeHl|ZxxSsn=n??>gO^=NW!C>RdZ23zXF
z+#t{tTEOd%1i}$*O{I2!Ak-3Q5~fz;`zWnB;15Sb+@D%wb89dhZp0U2MVk5hDNTGd
zwWg&>bcD7k5ZMw4v}h$-BuL&RFK!9c6$v}n-{D^Exmhc^AruVO!xI5*aadbMJ-A$3
z9MYCiUo6*Vum$0p(dDU(a-UFwuV5&%y`#t-fB(5Xrw=`|TojDnDj?GZwCnyxYSm<Q
zw+FRww6;OE8C6Na&+C8wNgVefx<3GS@(2)uf2zNf-9Ow->ECa)Tw(jl8K}Av$JRG7
z%y7l&{>}J!KH#_C8Xu>(bgy6r+XYyG>HitP9{~;m9s?W&ycO3w`i?aWSPZxi_a|!s
zp9gFOd=d9t-GKSY@$m$pF+4tg6wp3`dTfj}V0Nzo%*X4-J%9#Y(hLFa2c)^H3@>kV
zz`cO2fcx=UmOh?H0v-j-#|!mQKm(9IwAc!$0UiY`11!LO`&yy{(icHVz-Is{;Z-zG
zmIv1{%T}l5idoa`yDZb4M4yj?e6mom@N;&4#$1xW;Mb@dK4E#zReAZhIc9FLZ)dk&
z{*{F{x~?V#q_YCYCGat|iMjKfk6BmFR^OIUc^!_Wm^|A6Q{~pbf@MO!1LU13-<Vc@
zSE~Fe&<~?rH<v%ei+TNhpbw+`K6APC?v(u7pkIeRFEy73C**U`zOSHs8MQNu{Vvcq
zAU0;2<yQ%L7ibkIUx9MclagP()7GA29TQx_pG_#g5;5hS#Ankjl9Br3ag?t`tQ8X9
z>b&g;9kvX|)4&@Ld$jP*JhNu7O6N3d8>@8K+8U-gXItC0I%iqiT0KsiwXMeXh-C(`
zKyL*MuYkx2$R+VuwGpvulCOgNUDKQnlFx%YX=)%%HEz<h)@HMkrYf$9>?kGq-;a;~
zI9<;ac5LR-$_|o`K&}-_YYQl5zpx!|=csn9Mmy+*K7E5<gI5*J$g>rB&Yf0wUjAb?
zcb*212lDdKg)8z9Ta<|ZZSd~_e}o8`{7sYj=fUBl;BU&tUzg7B1y2R|i%*S@TXTf|
z?^svl<$v4e$<yLFEAk3<=2qktKQ_&sSJpm#b>67Oc8_I7UYR?u*qv9fB2NP^Y=Xhm
zO?>=5PW5B1xQ6Pl$kXmaV<;}32HlQ1W3iwYH%NLP=o;vm{BMFzb4@1wL(tcP{xzXT
zw|evJq<<Fd?*^UPgnD)BYC(5_J_x!&bR$oH1ocupEd$+-xA6g@qYabVZBV-nYIjwh
z9?I5c-mVVlDTW>jrwi$V2QGNvf(I^m;DQG(c;JEuE_mRA2QGNv|Gx*EGj(3+=nMgs
z1@!1iDHA>+$SL{@5}zd@yGZ9b`U4TA#|w21FXLzWzjY1I;ahQ@<l`H=T;gN0e11mx
zN9V?aMC!Kc1n|#4k?grT&!}~X0jt2(0@&AO1L&EHQoaCb|HUH`9+gDWYdBQTDBV(?
zo|`D)eGN~SO4u-6=Q%~UOI)?9M)F-Gp{gIRMnrq2>AVE*7dWo$z>*TjH_37-{{I}P
z|FLIQv{d%%DhcnDuwKG82_KX22NM2+gg=$=kc4kY_^yN>OE^u&;UyAYE8$WJ$*$Ea
zSKh1@;Gg1!+D%0_7L{mAic6LjmliM83f5s^+3-gMeeu#l#)`rZG)MfK0P)XP4jWVs
z|9cIDT3Jy`FcK(ouc%xc@o$#c=9XyDrf4Jn&fZwZiU@1)hZ|T?-2*MqCtxHbO4<UU
zF#YS6vq{8<vQVJOPXaR6+7w|$jV<^;Zc!w#73X>&C=1s4BYsvCXz<mC{LKMhLmgxl
zCM0}(-Cjtj{5NXB2JL>VSEK466yj9GYY`eZg~P0<HrPzd`<dODhTAZj%dC1DjSw5&
zHB?)5%BCgyhi=5Gr_l(ZfAXsFX)d&3v<_HxGl}PI_;;-uzve<4Mp?GFw=p~G&N3cs
z*nL8cyL3G>@J~tIOyYSPo5|EXWiGU_Sxl30m`<Oa+W#V*o|l^6(&-Mir%*SOc;3dG
zsr?(wg*N<SeedFI{Bu+DWjg<tn3^Zk>6oRK*O*B>Z^LYtZCu;%4`%g0XS0xv{=qEg
z!E|~)+mn@k8B1iPt66<Q3r(jv?Bw_aMdaxEVWCx`4EyQ2VP&&ft7-g`1ic<1`ZF1H
zx1?uY&o+Q=!M*@VX{VglSv{vu_Get`QP)SHJnhEWf*lFebxe}^-zBrPs}~GpXJVet
zS@~Koaz`Yc-u_bJYrTN);o$AUt8Yqt4Hs}a3?@3g&85WGWC5>{bb4z`iLbQ+u8?$k
z8B2+;p#nBadd<~3Fs#w?Gh^S8^r5SDPHB?#{|$P!cKuB9zvmSErINoFknFr-j-d0k
zcF+z;{z}O&N*VjPq{k$kukC~XbxEiHC{p5U?tsT7{e(j&GTX?{_#>^Pw@EsG2Z6E=
zC4Ga`gOSfuE-vt7C;eZJ5?`wa1bf!@=7ZmYaoV~>2ggLZmeVK4EpcS?+Y)|bF?Z7V
zUCH%K9;aT(pBZmH&OdpaMnRv8cG17CDRFVYPEO~fwA0^%uBA}~(!=SK$NLekXExhj
zpmRpPhK$_1Q|KR?g6^=<-VFKdsf>B!63}V?4mF=($(W~wl0GW=?UG&!`dpbegSDJK
zd49f=(`U1q>vYb3jTCB>^tIOrI$z^Q*#n#|nM@!)1^pkVpdSLA+M79lzCH#2=oIu^
zIE4H&px1C~lpihwJ)0kjK-WxmtL!bL$Le5vGp<v&as3n5FQ7c#jSR)rXolZfB%S`=
zPbt99jQusIOT+}eC;10u{1~$AC!jmgt})pzC9+TQtNo`I%7ULtx+eATcN{2tUD6X7
zcD^I^XU5w}$xr(dP||@@`WR5-x%AVF-0Xh7ateC!6!Z$v=UDUE-i-Fr``&EzeuL}5
zxV}#31}&CmZRd1}m_YXw^#8@_m<2D=P270b270!Ab69OC5{^XiMM^C;dbw@oI-j@l
zj%w_4GqD?vFVgI*rTuWiOzm-FBKdrELEq-4;3j{QuPzb{g?;|$R)*Ws)}{bH11VZs
zUP?RIWMT5r4YIi1_2Cvagso@z4Kn}y10p8ONB7lftl~Qc7`+K!qxgc+2oqmx_)2^w
zOMPK{oq-QCe6-<CeG4|IF&EZH!vQ}&Bbg{KQRQV2&yZbWE}JZRqbfGl>#Os(Yz~Bi
z(QwLwQpGk=Ll&-^6jzC_%-0so!m*U8??lk_XxP7*Z{7o&9~2op@?fW*#+FpY#jLfl
z#YemQF#4>8fBeBW`0-IT6bMK0*rG~OQrHp6w`!fc%H#9gUO}54@yJm5w9cqwzKXkV
zcUM)e1ed9Ve;mSmtG#PhxV^qLt5)6NsrFU7S9m?7j_)W`8;<fA0G__Vk*=FJ6#Bf)
zndnJK>`bIwm(9U^Yov6gsg04+$-V%8wh}ukr5Ck^f}4GK91CRGvPsv~PDr9b=5Q17
zq|B>p47d6twGB+{vXowh`8%TYl1Y0oWfq$pk!6>u3H#*!Q0d}gQ>b*B+y*M0n7J30
zuPz+)HTYZVny^n+<r?%v9ik26t4_6-x+67e%pFgmf>5Q;#nS6Y`7V73Fw7B%H4#=R
zp43>M7%S$QrSE+{JS$64zJXP?O}xa8S?RXW_FCx_v13*`&F900VQQq%PGV}BOxs7y
zZ1aRi(l!W#T8bmuSxfD{l~EJ-n%KN6y(De2|1%I;ue0F5?Ho^Py`W4I)OwvP=DaFT
z@ERPnMyBMetvaxY{SY;dO`POqI7KKy&MY6X>b#7;F`?uVAa%}^uRsYtZ3wFR)jCDD
zBos?YTn2ek_f-;Qw<CkAX3G8ns}77|U4sLrOGxXG<R>McFN)cnfHoYeezoplmldjY
zO36OSfAm~LYlKQ(t&ePy3!j<xD?SD9hdix0DqO9jbW8bGSuvqxr?Ov?*@MWCPnEn{
zZ?VgT&l<^?=|9!}4k^Dv@~L&11}U%BSxKJO0#yA<{@b9DU!}Pd_b+l`HFKpQv;Sim
z@@kz&lkyrVHzlXcEBH8YbN%`Zopb0j1mn7^1XO1wr|=&lW0qIz;*S>!4rzvB=e|xM
z$n3#E^2+~eeQA%BpV+4qOi0Rq0>d|m$e88T{-J7L(KCus!c71DG*e!#SM8NIqAzC%
zWY(X6d>#&pElTQnR_!yI$)Ab83>gOwC9l@M#%>{PliQmqe-u2_ekGq;XIf6uUo8I{
z@TjrD`Q<v5+6R`tC#9s~Q`N8H;9ZoN>sRY;WAY1J#m|-Cq;LgCGUS<FBUp6-_6U$V
zD>(%}M2We6^&GlK7a|$m$9YrAkogn`=~Dhv&*x9+AR?6skUDFroXF6EAJt!+L09sN
zGU(EsDLdw%T-9x69n=M8R|2vy1Lb<u*fGl=bqlOaj^i%_r7lr^Q}Yjv&usNO<pRin
Nn)rmel-&|#{}<g(EinK9

diff --git a/roofline/src/roofline_fma_o3 b/roofline/src/roofline_fma_o3
deleted file mode 100755
index bc3903e4cd4b5edfe94fa86a862c51cff71a2187..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 22872
zcmeHPe|%KcmA^BSK*W%lSgcXm`c?-e7Gi>tCKm7wWZ;dRAk`pP#mSINNNSQv=SPS?
zSkWYIr(w3V+I~Vm?k?NfE!&T~vRm7wu9^u35~y2)?ka7qM%Evbf;Qj}q5|3Pxj!av
zCKGG>`RpJ2DR(~aoqNtb=iGD8J@>x%?t63Zb$eG@Ef%Jdl`Uh0>9cfB5@*H^f0?6_
zR>96;I{O^En9Tqs7l+PE2#TD7<5IPPd6Lftn1iFjDhL%;oo5uBE67YiWg$VTZ!+mJ
zD`3?Z1gaoo8Yhq(**dnAqky$Mqu}Gxjyy7qCyk$pOmcb4I0{(HGYXDL!xU8Qqqfme
zBlXr;bzp?+1XvY6PeS6QwC0USJ>@&fI0~rqjDn{GnJK8`z6?3C^V24WWV}(<H=SP`
z(tZVHbJ<nRO&b@NU)5Z{u(_!<+Oe>sV)4So<t5=@$s*ov(oejrR<GqXDL-n!$mTgX
z$kkN-@=t8u=kEFUpRE7U9bf%-$5UlLxa=RcQQ7%8RKKxY%H?xwi2hj|l_z@UE*dy_
z-cP=N@;snf)1#nfD=YwA%Odf@H2iN(Lyt{EzZm9bt9SM^^v_R2k4~fi@HF&`rqOdH
z_%$4v={N}4?5qS`%VY*Vf+HLM;576Z)7VoDeho)v+5|#2{g+OoX9E0%>_Rr8*K$>A
z{5>n_2}xJuWL(mRGWZ$dc~h{ZjUu>=`8+k%zWP8Yu&F5=3505@S2PD(12z7Q%>m}~
zZE6X&`oa-^DB|-m-}UXj8ztY0W`8&w2n+G7!mDnmUgK-?hy2aWfo6vKe47FhUu(3*
z7ikOy{Pp1!P2f~SrnL!_tSf~Dfl#P5=xYwv`6EriR${25qQxxG(G+39wn$TJJ!^|b
z!YmvKMS{)E4Ap?l+QY4Fp{CYI18ac^!8+pghc<x?tqrk;&7r19fS4mqErDP|z5h<u
zAO)ZaEWr?}r~0G=TKr9|jJK*BmQm}<09ScE%UAdol`JZ`2L7f&Yr`*xpA{72pXfB!
zb7dLjDOn(eE@w24l;Rk^>~eHrE?WvB(R3%_Icy~{AzAQy3XsY)9MOxK&Y>Dso5~-S
z<p8qWnkpZ^u9nDruFvN=aZ>R^*BqL2R8sLn^a~UbP{kq9X`WK4NXFYt5Tr@GNHft@
z+o@o_iLPRo=%psQ`8rr(qMNS+x`|G8t7MqyYRpjC8WSB3Or`ZE`m7X{v04-TTob*~
zL^u1h%|xGV;_oog?I!wG6P@~2rLUXlg^CEc-9-P4iQa3X)3rw>qw`pv(Ubf3VvQNy
ziAZkpKTu+H4&*%}I3}(+f})9euj6dLL`RPBMk*W~nLwKNYr?5ZM~6B7JmD1bqX~{5
zB-}>$ZjSFIoVs*$C&!;4oVs#!JIDWxaO%R*tsMU`;na1bZ5;nT;nZcLwH*H*;nY>5
zYdHRG!l{c!b&h|7aO#@TQjXtGICaUW#__udr>+=va6CjfT^>gn$G=KAb;;=184%`e
zB%HcpbOd<W+n(5=+l|-}qx1E#H8q~WL~V)AjKTh~g#-o{pR2Rv6_9(c!G1{<V|yM$
zN31m#?gB76#|w<u8|@eGfi~{`!VP^?Ffjt=E+VoG&oodfQTWqTzSg+!9j>7Nj13Bm
z*qG6uyw0%fHx8YNoS)KmuF__|g!J(Gw=O+?jfUZ0xE6xRRv5Nnf9@+FS>Any`nQka
z2G8icjzx2iJ_xy#vg0H$F}z{)F33>cxbGN^<k(aBAQ-WqlDCYWZULAP%l#I#8iQ`S
z;t(7oc(kI&o$MSBMK0_d4@YKpj$ap<d-M_n;1XhvxyO%Q0OTrxj2)eoT!F|)F2l*_
ziIeJ{ZmRByt>om`FJptr&!ZD!(77c!5&c0=?y+)>eUbBQ`BG2rE5KszV@FGS+{cVw
zBR*#zH%3}%88&*LWq)5cwDE=}^H5W#`<Mm%Ld!k-k_iN>(Ya-uMP>qw!86H!CQ^@k
z+<*Z#%E#Ozz{d!`8hFh8GSCsyy)&iTavJ8lUj}VBMSJ@cISrisBk0M&WHor842Dh8
zDg|xGOq(ZY2^ciYA_V|L!n_2Fe$MErP8dCFNnn^vU{|T;6?41S!mwe(k}%)^s<%7Q
zac+bC_lXmR{mR7A3ys)b<IwLqM=ZuUdlh-_3H$oQ3A<x<^JR>ku;1QK`C3!H2Kfg2
z{9%+GN*VgcDE^E|*-V*-!h?OKWIfF%$;8JK6MeNnlMf)rYl*qHi&naO0f*qaoq!3z
zT}Ir!{pZx(Pw+m6A2yx`(a#HyZ<k%xE4{rFkQ}xPFwwV@D)~!Pk{m*aHyHN6B9(jK
zSakDRI5#StJD4m3L-PALov`2L-gClUNGC@%opM)QO5|I}S=D>`_M{qe@*^GzM%|bZ
z@1|MJOWobG1&XVoc<nIF$~RZX`aQAnr-k){$<T!qh64_SkPxKl?vn&z6Qh`_%oL<~
z@RJ0|*(8YD%Jdp(g8U>wn-m-@5`waIr0Xb@_EA%k-$3K}{Ly(#H)2~38?h*&6ww-^
zsC5G;Ur^+_lYrExV}R5F<LJ|o^geC)40>uERDxv$EQbM;fHZ=}0I8oaOx=m?2cD6&
zYNRFQ*1JEX`QudQmQyTZ@5?JyL*)#$1v)VT4$#0LMM*UrXegZjHI4T!4UK-2^4JzK
zJNX1%f>O1(Q46dX=kv(Dxyfdsoh+lqjsGF}+zj~zX8HTJjG<%hZJ+Zrfxe1V8?KdH
zQhQJCJ6CDw+5RW4AloQD`f9<PoC)o`bHpS_-MD>&aMyju+Q4VTp6bOsi&=g>sxV^T
zspTNv2;hy05RSz=fORDzn!VV))m~h^-Cn$Qr@eT?ZhLVcVJ~hOwiic7?8RHgj83kN
z*}J}png+WoAV(MU!R{r<l_G~8au~609jFoSra%W~a4b&TM$dP+J@M|HFze~>k_dwA
z-jfp(-HAJA?&oA^;n4mw2CJ?xdd{CHLGycZlLX>%F8Xz}AEKh+k3l|p5l%e}_d;b!
zJI_M#jB&D{A1a;{Dt-v?^ta5iVes~mOOyYE91peFmT~DuSItcr(i#u<L3a{d6kCYZ
zBBT2^_OAaTvGXA2ts6-m`~Y=fW{4f;6PM@YdsMR-8#XHU+wcDtDX~9x+=!i^i99|h
z&!`->cYdA7V(27|c)q;gQC)hTVHu)nDe)H+xP@dwCW}L|{BU}fBFXX%#nS0cT8zrI
z$<Vf=1>BEDXTN3CPu<ixpxx@aO@z|4K3FfDKr05x*=RgDVc+|o&IyNkYcW0_J7jjk
zbJIG385VRSwY%Kp8c@gJ8F+=Ra)p*Go`LhS2$E0#kR*JXO%mjEN##fT{ciZq?8}SH
zzMQ-lr2948g>}GjT?r%hxO0d%`yuBrhufUw9dE45Il|#KCk0-_*f(C><s3s6nQczG
zhePHer$b~Oc4{K?h_h5=9(C#>^SE=3$UNz+MaJl!h<J^7mvbv2v5$<n*O_O;?{R8;
zKn_+p^YT!O5vy`Kh%EsG#C$%$K`cr>_b!dsjtyHO!=mRW{~mZxMfXIsSd1j!dx(Ob
zr4^soFkZb*?TKNGxTjyjsVtGSqVVXm$vXtk9OA*u#yLg=#|YFN{ptQHC_nnJ{gSPy
zpLydoFtYRY5!Xa$oJOf@t*gd$BMnc7(c{bG<M9qO$P+u2ycY_*b;;xx0Lz{?>ToSO
z>^b?qQF$ml!{|&{K(jwKWLOS*<M|G6<w1KV-3@qpmN*e!FB!4pov&SD^pDz%lfUYG
z&1O{g8}?O0xa-d`>??<i?gRF&^TEIGV+XW-XxaDiEs1>}7tS!^Axp%7hLfOLsnk?r
zjg-ieoQx7%bPkndP;-pRA$!;N&~i^~uhE&bc#01i#Y5Hc<)=LTNvj8~fI-&P@wwgP
zp7t`drYwPp<E66a_nn@fgVMd0eWzjB>0&|>nx}Gj^BZ2vE7g_1j`;VDZ#LpNR4JOl
z%|}nIw0J7}+lx^!m{DM*0t7=w0q$yUb$!Kkn`^@cPkiGz9z@WWFOFW1utx<)Uo}1d
zOzppQuA5aK_V~n9@-FYc)5*T$Vm8?fIN5*infovHOLQS_!}1rQ5A(d02kf0c<$bt?
zRrjogsWppoym9A$l5?ti+<8X-5t}DoejXflg*RS)!I$E5FTEkY0A6|5Wq)wcTlubT
ze{jFovbXA-GoH#9h*n+szWqVqAL(5$+Pf&W;PILE2mHu+W3PB(zwCSyVe$^b<dx1h
z5hnW)CSykT2a($b`tNqwjLtI_Z~SYB568s=dtck(W*`^ZBFnCi&pA18-`x)939#5u
zY#q4&S37dXz>Kgdv_|JDjs+(~=^;@H-a@NUJZ4yqyUQlL@%5*S_}u3_@%;D7KJfG(
zwN}R$<W<L=vpk*8SPaXD(DJv`{(=5G9XaKm&Oyt-Gh2$h<qiu`s;*&#R$T-1$osuH
zS5u^v8J+vJ>dIF`zraNaia#r25Iz4KwJB$a&Pxss9O>VIz`>;`?&3kyf$P{Hm6an{
z`ziPlG6x3s9@vpXW5e3MPN^tAh(g9ohDZYfMaGcpc9$;|M_>|F&N(=+cXUS%@#Z{7
zS19O>zIOB%a$cY~>qFd4jjt^8g9FPEG(Y%$FXo2}04L+?ljesPv9M9yljDR-hX3H?
zkN`7=^eiLx61gOQeBZmb^kbNvJ^oC6fG8-;!861l?zc|byWgTk0Gi&n;;}w$diTWd
zI%V|#-pW}qy?f(xpP~_Le_#oo!}t_W5y+?bd@;qpkFYv@_W1MlfxNsMl^@yfe*=-l
zFYzzfyRJvXV0M)AB%g@J&Q3&s-jgt&i0~Z0A6L=i_P^#6QRYpYSZ}8LI6oHB)cQ4D
zXaDOH@8_~yZ-@E(b}IS%x5V}KM}VL1diz~Se(LKjO{1pkZO-_>kq>ty&Yu2sAEaNe
zend-1biJZCo}co1br0Px(<4dhdL<r729BKWMc14?{fRzEJ<jp@{)gxpHQ#>&9#5a|
zbH>h&fj@l=@HvDA!sL9(2g3C8<!g9FG2MLWs#$vz)+roVr-<BY#7-xFbA;x^k>rDb
zlXGI)Gp<;pp!KoJz2VF8@&YS=)p=rmrsty<@l26(h#vIuu!iR&cTqHX>ju|t^c*hh
zqjix;i0ZpT)_2*btZymmVv}oDs4qWN)h9mA6uKSJL);AW`%kkyU*YY4O|~EK)3kry
z6njif=PMYfcT_fCXD%!+RL6cvD-534f5Z3p9iuf)qvv71gl5F<aTdiQ&N)Ub-|5Ao
zRvs1$eoMF7-q>5od&#!gTSjbOGJ>qpjhDp(_O4z`e|YKE12KD74-h=y+q=Gv`z-tY
z&tmj-p0V1yY&>JLci#vGJf-u6DSDqEudDGVo!fzxeUSY52(6;oyRQM`z7N+WjQHcC
z=)k_?>kgDX-?#NIbvE$Y$5(bad%0ncb2#wYyH{@Gi~~oG@5sFsYjp$vaCc$VX_)>f
zrza<%*+4~GMMZ05MbFY5Au1yELhskXAoLL{*LN=IyANT1iWUsKl^@&r0{$LiFnYZ3
znybi&;bkKh)td6WF~7sOZ-87&u6J#qhj@xV(elWU8V|I-*oH<Ayeb;4OIxNi+HsMd
zYKRwWqttXVw#wN-mU^9S+{}7B%FAXa9lW{sIBTK5Ug{sA+0fM1Drch*ujS(NNIboj
zU%{%?==nJD_wB0ZevIn*dM)fPh5j6=pVm&ZcG}qa)7f7k3@GIWJc2a>W&axJ<m<1{
z|Hv!CKiU-g=L>O-iw8)2s{MN)#YWfSs-Wzr*F(zwHPBxr^(R0#u%MxQG}#XbyxkY*
zSjZD$XY^ojhz?jUELg)W_@CTV1615mckLp*5Zwuw+>FYoVLB@Q#c!!$(J!eAg-S!H
z)Va#-q%yrxw4%;mB9CsxqYdJ!FDcfDpG@MlO$9zO#ftK7<h}TkBlb=o6O_LLCH?yg
ztP>ZNVNe+{Eb?OD=k-xX8L>B$&%-dh@F?ORY^1)|0RQ|+Ug3}$tcS-i0w6`IlD$H5
zK415oqMrhp^oOR?pJ5M;cH&<?9e+jH0nEa=A-dZ|3yO?*gbccePBl)gZ};Uot;wOg
zus-my1CV3$lVdMqJ(dXbMKM}iAwEx;oZr!2=;wmDj;S^DX;&!aCe>Q<J4&@?Qcdy3
z#|Cs}SV~p&?ar>tlU-LAB8l<&+!dRk`P>_y_l{k|$29SU1aI=hKEM?3Eq()Sz&h_~
zTk_Rkqf5Krj%a1SQS*;C_B(IvgDPm4nDeqh>!X!NqrZbE@=WuV>G|f}XW(Vh2M%Nz
zeb;AB;cG+Jvem)Xh4j&4uvNpyh|!inYebtDrWeR;T~l+j7K*lNo0}qyC=L1V)WS`7
z1;SdeL7R71R0}i*h+l9^{yKkiU9{OB31}@%tqZgkf5!sN-@Zv}2!^zI^`uU*vm06i
zq$$z}Uj9X7SZnY%H3#Zvm1ur#sWz`|fwpleo$GLh|LChpXs*|;7S@Ed`Bz^#uVhh!
zR=u3~DYt5QF*hFeuY~<gt(z1tZ$*8yrR7eo-QOGygiExV#-=b@hi3Yl!$EDcKh(m_
zsq@!02AC1RC!Uaxw!yx-#-`SQ@Lsh*v_@GT2?p;#<>d`%a$P7G4%7u(>%-h2&=fkK
z*B=RlBiaof?T$dGHP9?ft-<%6T1&tmj)u5DwWgM~U^v`_FJ4PD^Y@^d_z3id)@IQW
z+QvX+b0E;Fm1&V6d6&GnIZ$6B>|A@3YnA(2t>mguFxUW31hj==Z3*?@GHqc<TS9%Y
zOq<2#g|9`Ir!vZYLJ7X=qs-Q>5?B1~pYA-lfA=y`Fm|1QOcyXu_cu|irm9;V)WXra
zM%iXmB?UjJ|M@4r+=uA?0NlwVKnVV^{!(^-ZwIBHUuU`4_C@-J|8qEYy@X+gE6&j2
ziHRb>r++muL0=u5!3?$?FoNlSH{c<_VZcIM=f(hofb?Dd4**L6U&Zx%4dCwp+W_Ce
zJy$QF@jnw23BZomCngR9R=kONY>f3{cCQ7@!{-w_0W0v*c?575AkAGBcp0q&?gDHB
z+=I_d5`am-!+=Ft*BApd0O`ZA4nPg?Fkl7XeB8&cAvz#^;g|&64M=H@0L*gNjm*;F
zuw0x!GjF?Pri18naga~0lsp8o^E2i|8S$@w5p}~StiZ9dpy+!0xtsI0vg<DT;+0oB
zF9ntJQ~|DpkEu<}Rp5BQx?(o2|2&b7dK?XyJllz)%B^36WkS9S<R?+SA+3CUs{9eq
zKSa51F29=>^ZExszYu+TySdzYOG^HA(AS~7+*}@<l+Qu??m+nxYNxV~c#+6{C+L5T
z7~$XHs`8aW-ih)ml!s7GdQ$SMw%Iy!tmA@9__G=1s}WP)DSS5FA{nVa9zyw6#9A@&
ztt!}x&|yn(JPN!Qu}3cwnLB?L^EhT$+nL91Yj2$4m~CzEaO7Lt+uRPDwY}DMuVogo
zKyMWcuY$-i$Q2=GHy~C`^3{-EKf_@s`2xt3rbg0K>mp5SY&I)ts^*%=j&jJCzA-To
zz)h-2&op*8XDT~Z!48s-K)&`EK6#~>{e$gzJ;y9hA1Kj(iq_!OYYXyhd4Xe_)m2dR
zfX!8)!Q+8~B6QL60>l<2;(s0d+ri&OgiQW>rt;5$!^gnioQ=Oeo!<+dD)5({n3%BU
z2>su*E-xth8=Jd8i{~san7=Kzs-X0N8Long&Y7zU#w@m5Ewc(LTm_}Bg89n}H1NVE
z7);&7$Kyj(Kjw<dsQ#(~?RGSV;^I-z^Dt*D6!g+YNgn`R13i=fWzcD^$)vvr`Wn!`
zEcEDBZ$TdE&xifJpi`SruWnr>=uXgwK{tqQ6zKP&-YR65fS!l1ivmPP8>Y0|pmrP7
z?&<<Pl&#IYU0u*q3O(qiv+1k{&U)ai2hMuntOx#IdBB1H<4_u#B_Ky|qFXgenebsj
zPSKx|c)o<}9G&N8OT4~B;P`Z&C-wj8izO~^ci2S|A75I_b8|Dw-#;@EBvP+cCy@Hj
zcV4c*)qVSrRp9D@@hh?cbdN%*NPx8e;t>hA79#1j9I9)2uau|z3`%tCNa+Fr@J}qA
z=M+6p;;LP>lJ6V|RsClsWP4`lyktV+$__k&@%C<%<x>3r9H{@XXI8XW_UlRsZ<eq@
z!gdKCknp<_{-cCHlW@O;FG~2Pgda#aL&o7}CA?h1#S)TTt5&SIR-2E1iWh6wlw4g>
zrY$NhTU=URx>%ckBNk!|e?-t1E-q%QBz$K}#J>>`|9s`JQRVQz*FdO^m9z#UffCnp
z&%%g*lf*W)MoTtEoA7t`rg~OFSffAO$V%$(Y=u4nBOy`J9tef$U$>l1B0iLb0?mFB
zkh!+z2rFr7#s6_jB7qK^8-SoJSnrScSxKPL*AViz1bmJ4kX4wF@bPsnA))eLtpgjh
z`?0=*s)JC7Qwgs{XxtbMvy!@C3oVCbc4r!H!)UIs>S;7WY*?GCvFemfOY{%jh*eLc
z5kmjuRpZlKXv1h7vg&3M&)e|tS~Y&ng*J?`Y;kX6d8{|fc(7sj2{rE0^~_3*&vg2^
zOwCj2^n9ktI83L{PVIk@PA_0;eoLp@S(}{q(&>)W{tfB$Le{ZJH<NhYhJU`R`O;iy
zW1nGao=m4>mReS8Ch@!tvt71vZNop9)&HE$LN=P+<UE*8FJe2h(l2C*taLT2Pimp*
zG>09Zn52jtT|X@JRFz>rT{o<3HfuADf0Cf(M56D`pt~eJ^Ln-pbPN8#m6UeMX`MB2
z`c!|$r5<&C1j^G6oGn@UWe@1t+SLy_*_oK5BgYtD8v#6sgXr|Ogc4sH0elMwZx>z_
zQQ~VJfHPq*(dq36CBB9Mc$uWrTMkNmZ31w)q|?g_N_@=$uu0PCy#u8#erD{OieJ*3
zCH-GP&(^LdCI4FvopY8;{(eBR^Ws85=WA1-?Uno<$uCM7`-P;(B%QB?f&T?br~fEY
z;%iiZM<o52T_-Zzz|Z(2t)#b0I$yIx*?W?{PU^wP=P4H#c(RlJFGq>5Z2^KkYkP~p
zZ^6zI`WJL?Os2~@eQMkiM>fAL;x`uOO&Px{xSpxw)GPTj=K&w*pE^#Xpr40!(Z8-K
zadE(IPUobw(+@z`(kKGy<MgTH{UFyfo2^->b4IU}eRCT9<I~XXHrkt^h&_@qZ+sSX
z+P_21Cpl98m6ARt`ST>b9Q5;K-VD}o`qcUPW=@~YN*Cyy{W2-kB<b3GLFa2?D7%x>
zC6funr=kC78v1_FslA!==L^&Dk4;0*g+s_cLwYT@M)~1F(6jlW1a!@0x5_RhJytvG
zka42u*K_@o*Ds(v-GU6oRU*T0t&&cE@23>tXU6`D(<Nd8-<JG4CBGrdehj(;?HZTu
zQX-E@ezpIUdI#_`NgtE;@HIgQy&&n^Gwggr>d%a~<C35DC7`4Or}P1!#&hYXS-IK$
z{JCl9rPI)>Krgfwv0WMMrS~J*>isI$gK;g}wNRS1bsGJ>)6oBg)2H75JPvxceREh{
zC=!lD@dmdJ8@*h=;zpm>b5jlWxtZJz#}{ew)zN-9VW#%DF_C<}`k-%9b8w@-*;gM4
zhQdC7w1eTcw5>US*QO<lE6ZsIn=DK|x<MAVyFT2)hOqSvzd`1&OCe&yd~{!(#wy<X
z!RU>6d+Q5EBTT&A^p*L_7W=|@n~4{hKH6}np%oj{m<t=C;ea2nT_?-SRCxu&Gh`Q;
z%chE6t%^<c`s)3yn*yO=G@P=aT(M2okcI0S#Z~63@U;iCa4cr(y(*d>4f{9o&3j<;
zT_S@=9_;kf)S9Zel(jXr`e=6_Mz4YS%Tm6<kB_pUKsbuW7FCjx!j4G3l{dPo-9Go~
zD%$jjM~2F$bw)k&Ro$}MRqa^;E>j7AnaX^tyf-X&d3`smTzQkb##iH7?sbzozN1iG
zILc!Hc=`rMx^CW3=#w^Qq9-A-Gm&y#HV5;qk<yi>HbzP(`vUygO6;hVUep!}Zt~%A
zERbc(CS6xMA&CZ=!%f7KGOxZV+~$weH8QcwQhF8U_wMN>Q}$rWEH*hJ%Pv!s_R0OB
z(#6H5Q0X+e4OBWYb1y7keK_cA^taa2K3Sd{&=d8DHjJ-&)n4k3)Tl9cJcSBEl|HXX
zuOsEV^dZ17M<CWjSfzMUV|{Y0m}{24_xbRwEJgVSR@pZ35<6z4+d|uGrBlR?S?M&N
z4;zN5kwQC(scAB8A2GAdlO9RiAPj0Lj%a5swfj~^P26i@^RD!g^sCq(kI{Oa1qW{D
zcv9;H6_TLV>tr$KRe6G!;h;4#C0}FJflcm*sBvuaBrn4$LK$*q`G{5LW%N##l2d@x
zIa9t0C3qDWRQ0QMie5=5m6Es&@}%yoB+6DJL(dLMe#oi=!#5g|pi4;WkmM&NUqnnu
zH{-D3Q1WWsBTrVS)+vdX66r7}dM=_hLM5-(M>fiZ&rJIjpMrNlp4J={uGUd{rF@&L
zj8L*u*)PfLE@a53N?xtE<jIB4TFIE{Kh^#&DZgCuHO|y|u2IUXbykwk$D!(1@_z#w
z`9zvKxqc%TRx?)`GW$Q4A+OeXG%2rTINSk#N(w%Nyt#gTmQKZ37vg|j>J(6&m7K!A
zhm2WXtsANDqNEv$o%=e4AhQz($t(Y>^`*8F5RoSLDV4YYsq-%6%<^jgP_?h<sT8Nk
zWcu%CnevNt&e*X?@IRX&p!!dgGL}H500+euCG|Y3_8HCO&%~dFj2(xPSL<KnOG(>g
zya;S6ei#MRekGq;XIdsCGV9C4e+xD>Hb_FPr>T8l>H8~6Dn6CGii0;%X0BhYyN$~)
z))YTif|J4(JencT^jg8H3$Rmw)LF?X_#R5k^{eO5ow^Xo=swPyQijY&I7pZBpL#xj
zL<bS6LV(m+OXWm{7W}CG(hRziUywnU?o8QHh;mi8nRQqfSe_D)g&8Q<qsER|{;*46
i6>=Q^1yJe|<u^6|(D=+&zr!o!ht$L;(3IU0W&aBk*|X*V

diff --git a/roofline/src/roofline_o3 b/roofline/src/roofline_o3
deleted file mode 100755
index 43a7d010b1e4007022b4abee11f06bca80b94ed7..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 26968
zcmeHQeRNyJl^^*dQNH}-G%1F#@LCjV)5O>zxONES*|CymkdqV~NLqrT*m7(tOJ3<I
zI4mgv*&I|yt=rPp?ctnFDZ5R3*bTddo;0Oc4m2^e1qWK17TS|?_>k+sIw=H~LbSg-
zZ$^HSB})I;Klb49dGF4>bMKuycRt>{nWyocF88_un~lj(z^-A0X>&DBGVhF?@N!hp
zYS|p7v5VLeHVc$uTpCXyXvQR*6sje(3%)|YB3!ivCZV=K;}HojG|5atX(2(%Zz^aa
zYQoiPOjJU~R8Al{vUQ@Fqk!9aM8Zde9d<H|2bJH6Omg-O90lCYBN9#s!z7gDqq5P}
zBJ{QtXut^92(U(epMu0oVbiNd8^sUQaTHMF5eeTj$xK2i_gTo1ogX$h1mpE0zkGgi
z2>T@z#buZKJ9pHqyxiZm+}|09^epeGtyx~PvN9a3T*1pt`iXb_hRr-D=|>e9*}MoB
zxth|W8{XLcgBLD7{C4B-XD>eB-}cbAmdv8G#kgd>v0{qN=hhJY<GAY1^k2B*$#)k1
z=z(_@0?ip70yS4+3FvALiRbh1zn+I4%R^rRb93dpAP@c1dFYWm`cLGcUyw)7W#CtF
zWrx!s<g$~RpPJ1K{3Nbi{KxXpXXUYnx}l0IJM07@m;O)Y(USsyDLbE?+_asmlKt<O
zf}Rj`*-s_~eI$#Yp*`;mc6Cz=u4bO5mS#_zFXY?V88&>OmgcqoV8GYn-Qo8!k7s9B
zFyILr-jLz(FweESJl6}pwSI3n><gRXIfd76Y~JMQ_J+KEzt7K*pJ%7f@B|`V9-|}V
z^R|W4G!v&JGOgQ0$+{|7;0uKUL61M!>NPrp0b*#Sq#EYy=`>id+vp6mvF?ZwW?>^_
z1pR)7Y(Qqa!h!BkXTWG@T`(coO1$3CPOzc0A=Z9dsMGKfv(ee*3%0j;ce8dO08L;C
zhLAnwCmhh_?F=wps+F*eN>>fI-rdx&*0Z8=Mdg+7H+9-V{1x%Lgl7LII`#EpktT;<
z8>CR>jK+~FT%(s<iYhE-SA$42%}IC>TSrU?HvIhzkkV9KkqbKKP!1a`>Gz9t0Fhph
zPM_SkoydHwpU-3FO|~C;=FpfUgKR%UKTi??WjiD~jZ-p|iS{-d1Yr_SQY~~@c1kF>
z&}G{tdX<H4eGb-I=+@_eW}#EwGUyh%>@$?M$wG$%(_xE+J~vHeY`cX%&qD98(5?RL
zw$K+?_<JmL#X{d}p;O(;aIb}4Dv5vt7W&65^g#=qo;@<?(NlK4zxa)HD%1NCMsf0|
zNYSHD+Mh5vQdho=q}0+^a91wY5F@;U630%a5SIR)aH`U=QI0=HI5qjP1jnByoT_&0
zVU8amoT_weKgSOfPE|QJ!14bkoT_kaFUNmK_-w+vIsO3QRApn^IsP5Osj9{{ar|q9
zQx%PA9RCvGR5fE&9KV}zs**94<F^w|RWat^c!+R%JdQDr-%2=D$=Jj>5SH#BoT_5%
zB=G7tnqtRq(PJ;`(N`umwKN?|4Af{$KQ^p#aI9vY#*&|c+*|F+#f^*|I*5wctS>(R
zphqW5^w_JrmK=gM?*7tkLzIv@3FclRvTaYaQ>c{q<CMNxzw1q|VE9}i6zH)DeK@&B
zw;k1wpEDMxwat^-l#59Z&wuaLr(G(#f9YljCIc{R+tK1bf@J&46I8!L3==#(x`suH
z$G;A_w6fDAkQ&`Kb~|J!uHSWvdUEV%^Fh#KkCM0a{yq~hJy!fxXw{Fo=!rvcg5Y><
zzbhG?3>oJ~C&R|<=;RvX!tsmI09O-p%r!ZF9+1mTWMX`7axGd$@*3Rq{y3@b@1yJv
zKTO_@Jrg^Y{4^>d2AzA7smS;Gi%&JG>?+Q)=eqvlKLU%nPK{UfyH4qYdVJB(xiP{@
z+o;|TEk}nQg*IN$q#Zd$U8ii|H?`dH^JJn3xkvX*GGjKt1U!@c4<hxuCUqE4Nb#8K
zB=8BsuK*r%y#(|m=^jn%w!IJYT`z&gJ|pvc<2`a3IEPQ7CXXeX!2@M5Y>HNA(nhSb
zr6w%_gGQNA0x)8lmtc{L^#10A-oKdyM#+S#baoZo?#(c4RJSE`IDqo)OZ3cZSN<n)
zMprINjGwQ^j_Ak#96f2%=NysbBWIK?i8G30!Tq0R?2K~DFvSNf@m|E+mGV)f9Zwtj
zk0@SiQT7=Thr*|awv+WVo+J~0OQnVefF|!jjOP+_4Va~L4FZn9b^8GmfCu!rYv3nT
z-G_Od!w)+aLiE$7#|K1}4GM4X2PB6b089*xQYN23Cdm;r@pfJLBB?wC$D*1y!?_XR
z++)dVFeD$q?Tm7>>(Cjclx~h@x)tAsS?3vLD><wA&`>g6kaynYEkSRc(BpkHs=2AU
z`}aU`GZb$grBV6D=GbskZ1Qo_`eVtjeS(_dlMXZ?Q;>$c4-$k;jM~)JY(W|aKS+?A
zO@eB+APtZoBuJw?3BInHri#)LX&g$2kj#=_LgD%N5j~~pu{|gBSOl#Utu;oi)&-n=
zK`qyn1f)8h089W*qE1g{)@l34P*XQSC0I^^<pf|7kb2MrAk`DPsVgz?)Dt3Cm9(VY
z`Y(T_@#DSdp7)rc4As`ko_UVS0-fjqPf^DpMM>Ems4Fb~ZaJ@A>Kelq<*_|vcJeSi
zg3`ITkPEDs<m1Sl#Yw-Zoh+lmP5wRk;w<@PR{6X3OrT=!+_mU&0z)0?GOQ9@LVJJl
zn`=~5(D30;k!{pIh6cc#oDJ=~a?C-Hs&QbO>8`s@b%Rfj{cI59EJpb)$U={OV><`&
z4ghz|Y{Id453t^Zp(+)wy-G#%fKst}zf!U7VWq;CP%64cm5RtorDD&79_8AY()$(U
zbgZuya`Zqy*0&n5D#TDj4n5YV0oCJu)X;(H9E%gT-v145PrPqG%zFHrB!WhE=R2uX
zUt;&{qnr#a94bF%u<BEK|Kij&D1Lu&l0ZDpMGv9;5H$;a5c0_jaO-E73sol-9);pr
zlVrg#R6J&?_&0#Zhpn<<@D7nnlRrR=H?`QFN#Vx&mK)HeRo>i>xsu?bwuQD@ruV(5
z^!}Q}7DCM3dNTR+S>%P0A$H<0@io2k7UgWkM)kU*%H3ZjCCb6mdh84h<ncvzy>3*A
z-b-Y&>m>E~eDT1eyfnLR8=+xoVktG;QZgZ%#UWUJlb@wbupE{wQCHHY*KJOQ_KlZt
zKkCt8+nATCDf*=P`TCp9rj*wQTTCa=ia~M#3Qtb>`P(1P2`6}I(LWC!w>sh3yiQ=c
z4b@2HUTJX+s1xuEyh2a8Qd<tsz<D_Y$)|r$5<bl(338^S^rLdO3%;}Z@&c<br{;n*
zU&Ab{2afAa=&?tfBfQx6IY&9%=OpjAW4+Fk9PV>c!!tCcqSr~2F{NUklV)&A#eGf(
zhxa>G4j*(@arlr^<M0vZCJrBSZU@x+QifZP_d53y5_?yVyPbADeuq=#9r0MB({6`)
zJ=W-S5L*HWi1T^J2C*#p+rLmh?b)^$GAyz<`F-I1wSB2bh1r94zJ=QFzgoLWML%^r
z)x!*{fRE3{tvZpMi^TC?Cc`Grg~WpqjdM6mjwIBLKYFwg%E#~L)8{Fl)wgWELGSn2
zdA}RuX1<^N76ho(Z3L`-PH)9C=LGe;y5r$ldNg4JO*uHC+nz>;a_DtWE760HboV#2
zrufR^dhE15{HN9W@You?Zdg~=k6>b7peyS}^uDK*-Y<a_Z3~M2THua1*f1fWHa07=
z=(_mIVNePvRZ{i35o5DnF`~!2&*{hC)T2jj&}MVT7u6WcVL|m%P4U&otDnQvsOUWE
zXzfs^R8af~Onc_G*P3mk?z&OKab3KKCLYL}=f^(3t^VdG2AX1{<NGb|6ZJRNZx%i=
zyWvJ&sqZJFXH1`51(?Stv1i~GOkw7)aK|sSyX$_ZL`zA!e-C_v0X*Kaq5w`p+gSE1
zIDzTIFBigL4ix5ccYN7-P4NrMo8!)krn)or%GZy&>&7(#wx=8CoNcOmj%dwwr<AWB
zX`FLL>wQ7#9U>*sNxO2-qlmd<FEqt|p%1?X*S`tZABF2D^uDu38Pqx8>#a6KwiY?j
zs|yh;Ezm0_-0_yT(c4e!@h#`vap$wuXE(-|m0lNL=BR#3KR(`MdsmM>VZ&&;q#a5c
z7a+f*Hbl1;&2Lv6?gg0SZW*qJ!{@^JaQGiXzs0cI6rcYwqfCz%Jxf^^Jq4dL#I_V+
zye;LPuisXG3tq6OC2*##V3G_)&r(G;)%_y!r}5`MsJ*_Aex=9WPri=lJ+;?=1<X@7
zS?%=+NPfun`T$sACEh8__KKE<c0`ldnd_tLgVI?_-y+Q6En`;6Ov{+*5%eU(GfV09
z($YfC=a$lYB^plVvunP1c9jp~`89KWgnckIcR8)x{W6xdFn?*PJEHW`j;ol%*k?wX
znUO}$q%ezW#A|VEbS#bqx{Ue!8(7w-`O&f0&H2%tfFEvt^dm@q$obLNz)CZA%lrtl
zysfylU?3}>iO=}qBh?0I2B7pguxMzR8_t^P2{R+YN4X88=fzvGm7Q;1T#ZhXIWLxL
zI^`pn7vnM?DKqgIKYWzhK*l@4YuK=r?*#MtQ^s7^{3wquGi9H7x=w!|`UbXf#ru#0
z`y0mR>0Rijzc=57!hj$AE`&|Qx+CFB@o@u|@SE|@P48K>Ui|`IZ;Fm%g8|!o_`svo
zRTfF|-2|*>Ti(BnU6kK-BEP~9nO{BfVpIDw$Zx*P3a$L_tU8yv9FboMQ+jEOA;X@}
z@$&yplppZJlz-_odn`rgJ1NL>OysGt`kU(Up~f?`iP9AN4fdAqIz@Y+djI`=k4=x=
z;Vg?8&P94`zSE7(FFQ6J{y>JhW3MMYWLxZYJ@)ft3!-`-J|=ro>3x`{1$~Krh$+3_
z0}?%FQ+gl3dWdrOud%@rJy)Rg&f}3nrSDcSxT+zHjpQex*sM1_{+M$BSoPWDw@=V+
zo6^?|M*4t3k3V82Jw@L(R6jSA_=Lvj6Ng^sAS6)P#b2cD0&FO994Y&_fIIdOhxDd<
zra>ufe4iy>f^y1hubJ0!k=LJKEagQxn5yp;sxP7K1+J1rxT>{EY0-O4(PK7u-QSeH
z25fd7qBOnV4d2z5=`nngge|#FyF2D}=yyFy4yFpAJ(YToxgKD8z&J|#kliTsgTFNk
zy+s%@t<d|+G!)S+khQ3d&K~l>9kWd{-Og@nao>Pi%3?dRs25qhjhe8Qwb9u@8djw>
zY|l~37L;;qF>cDmRMvD2K8!fu*F`2($mFFL%yJBfOs17%VsXB5)RGmCVERe!b~~%A
z<yb1pp(2ajB8%nI%27@lLTL?Zo^l+5ARBAOml*8>vL({yx-7>gWU@$PG6Fg|OYbjo
z%Cb-+siua=&YLO=8QtjILKaLEqcLC`Tee2GbCcdrT?sAp0Q6DXen4vh8<8JXl!m+q
ze?vZuY|5-8om77Vs!jK3ISRLinj#tyK0U?4Bp(ll%FI3H-jn#;qr;5%AnwK&ps_cH
zs!U2dD54)mPnA*G^w?^rj)A~FL=~pTUQ0dztMGwD8UMgD{e34s6DgnKC5Py-7VM+P
z&=(*@s*)Y1WI638f<H??HEhz~kWYUWKlOg%Uzm@-wi^4Hj^a=e1D^pi;szOX2i;nn
z>d<JV2G^IOLaLtwBOu4hlZjuaQk;NxM9C;^t@%0Y)OZj3p`QyDJErF_q+KqRTU4t_
zpHz)DnO05hjkhi6%(9d>uhHDP&d;q&JWnvb*T+&czPsZ~A3smU$G7q21aD}HoyCyv
zu6Pw?u<OI`7bXurhbrxR!%(YVl;7*RV}EwX&Nf0rYSBx4Yq@Sb@@II$ZW+NW`|*Ey
z0$wJ4;6Rko?*ObJyehPcZ3qUI(~lg20Tn-9h;;b^hPpINADpq9I{kh%6bY!ebs8N=
z4S9E~;m+HAVKvyUF1<aX`uslPH@O9WtJmKe@p}!Q+SM6YrgnLImZ{!dJJt4JNL|`S
z>Lfec81RuMqXWGB7ZhQ&-P`H+wau+mz3SEK((YyIj;rb3iaVai^eYQ!Zd0!?tqH5;
zS6sHVaz(q^+(7&kTi;N@jfef~V1H*|r{v|OXp3}p?N)br{SjZdQf=wz45M@?rq>@1
zs<(MVUEG{jZ)=B-=|21<1oDw?*w@<88St6jYc>(JL!=wQ;1`hj%61gFH53f{T7!YM
zFgFM^iI(&H4PV$$H#VtX@Pz_CziDa<ejlQC`Mlvsi2GCR?CK7N!=3nrOr>i5{fKJ*
zXk=r+Z&rl5!)M&)^99su)d-Sz$&0u7+A2*uH{Vdd-nB}tygU>Pw!;%Xb$M7_O*MFp
zx;&(=rn<OBoy(SnSE0(&5$Qft2^-KTvbVRgKK{m|``<bG@HJ+_#2OPa&4hN%+exLG
zu5Lq64M$o#M46G55d5I}=RXPJK1B8V;7;BGOu>KDU()XHe4D~g*4UO5UPZt8y9ig$
zv*>1c<_wOeQe}W&KAuW-13rThjGmJ;_CE~xUBFSmKLJhvuE9f(c9p__Re%RES=j`5
z7_b}gam>L60n1)Yr4oSpOR3Zez%75k_lbp!P5u%20+uCHsUE=cw^OMDfP;W12tNxx
zjA4U-wSb4-O{MMyOakr)EW>`qAwV5)6tD*{33vjKe$-ixH{CM89>6NVBw!C9g+(S{
zw%e~~wjPIV$^6;&0o!Z`(J#bBKDkWr5XjBXm=kHlU;a!grNSqy#IdfV>{?~sZT7uv
z%>`Foc7^kkpmLr@z>mYn8f;|sC60Rv)-J$gfCtjihN}jH=Pshi^n!neWu|;D$or7K
zEhBwPI{iV=pF+B3P5&33%=3Q|^f9F0VofjDnwEbB^iQGAS6b79Q}RV9-&IIoP34sK
z5ibJS?*#oOw2gU0km>78c_&DXNN+$o=}D)r-&YtdDws66On>^3ei7P~dm3M%W)qB5
zANL`B6WZD*h;Mz#UNjlD4A(=zJJ9y{Pdpn+9Qz9DOUmvktS?cK(>rrZ%3xbViM@gB
zT#f4`@auTJ`YP!G|Cb9IO3LDe4JGQnqWY5Zdy4Bzs-m+Bx7p^FltT#;!2|`3P(b-z
zh&FxTMb!TsQ{I|i;k-HN4A%*MH~0r%LEl-Bga2>yr}5tb{+`pRR2SIEKCAu$pDYd~
zit>a15cuyULN@>2>HIH&zYL$H`E&8NW%9ehL#swxz<)uJssAhFuP+z6O4N8!LrM9*
z;>MDyduG*_)JA8oFPX3vdf~WQ_z=E?Pt)!rckq67v*_!WQ2vc2>MbR9PB%c;(D#>{
z^r{X)-$(g_p3VOt=-oN!kAZ$K=$|$9Xa(*PJGGx*gFXs6l?nN31?x@vDbVfjU~D3~
zUZUNJd>auhLU}dNeME;}r<Gf$a_dy?<`ONGtIWJyZs_TN9=Lrb%y?kN12Z0&@xY7+
zW;`(Cff*0Xcwoi@Gai`nz>Ej}UwFVVPvdq@%r&7jnwC;1WW)PSa+>}#fzKC^&Cz&_
zmPRPtSEF%wKEKm>YukBjiNLWm#=`{yPKtC|kD?&c-#(WL5<Xa<5lEl;W-m5zxehl{
zVB&HC_;aEFv{pf(%mf){@em1124=u<A{6Ai{h*Mi^#%&q#N<Jq@zXI|<G7^T1upI0
z&QlrA$1{V>ADe?_d1h%m0Xu6Pmv-QnY#iSq(uMf{IgsbrWoJ|)>UEufHwxG;;4T60
z5%8M={=0xb7VxNmF9`UWfM*4qHCN+Ve_X&z1*{Q}64$R?yGkv`NpThGm6caiR;w$j
zs%xrNR@JEG*JEQv_ZlXBc})dlmEql8hIa=bPQK!>L&k9Kl`quIDg!~oS6ScCwA}FS
z6xhx{q;f~36UW(gwy{dWI=tZyR@t^Y0DUGjLT1V?UnoqcZgDn&c#szI`Mo3{V%>g&
zRdxn&9#^H|>%qMp2-1RWUc<{OeI1_mkhja{>1czj#7qef-?uU)Wc=1vutB>Q`!&cq
z2!*&+@?1=fJHlaB*&6Jk?Je%pe78dM=Gp>nT5=(JK}$}0Av1E)i_-nms;m&bbtEVM
zEGGMJCjV@ftKAo3G#Jd$9}3aCWWUSQGZ!Z%X;u=C7qWTikh$pdnVOTnAbtL&Raqe`
zNsn)tbcOA&$)(?sKEE-Ozmy$Vm5YC2dc4f!{}_|wWF~zPqvMLQLm@`HT>ZKbCz#1|
z&a5(pbb^@}2Q%qqY=2Jr`7Du>E=Tn#Ei{}KvD2w3iW#HlhmBTOv+SqmMgd#Ex-I>m
zBxrk(=nrSn>jgdgd3F=%Hk=EX6n2VXowalNbbrQ$9{GF(%EPyDx8X!W`5Y5ueq2(n
zat(ul>`W}uI4j@lK<sHjr>`F<@VySe*KrX)ed$1f?`Z(ehQUOquNf%ty$Zlf1f9NG
zpuqPe02>6IK1QI>#qW%D3Oap9K;d3~XY4DIU(o%6{$HTyD%WFz|8<APIadn)VL-BT
zNvTQadrhDn5&TVp-%Mrf7lIxWbiTI*{^tdq&M2b5_nZJ<7W7k!Mr6kK(g4Q=eV3r~
zJvh+c67-vd9`t-3it&IaJL$X}3Vg2$5bQb2TLyj`PL`})qk&^8T*~Rw+bwbA^4khN
zu~;~*|E}eFruS3YYoZe{v)h}8^H1-m5zrT+Ty*Lh1uhQQ$LXAuar++VY6itbhB$qC
ze}9_mS-|#Qsc}ZPkbNzW{>eOa1rr?yW@8U#jT;{aozCx+;|aD}c(_c^Cj`G;&{u-K
zP{gfZ6Q@rfpKs*!1#J6Djk8}OggOOXt1{_)ZwzU>IbASWKs*oq`+4X`L8tO&kDt%y
z;h)GuFNQ<NKO<RwI3M&}ey9Xpwb(7ASCgIsg&h#>MAEP2`lp^>KzZ1T2(_z;EWZT=
zosRdT!1vYwzsTvFlyUo-;2#z3M;EC-1l@shO=h)+gMwe4KP~t4e=O*#(8Kozq3d};
zPh{Eos?eX^-cAdCI+u`w2AslKK$XWbZ*z-t`}v|gboy=~SG_cXURqGb=-5JQ*aW(2
zC5ZT~To3xSDA#g9-<wDOU>^E^a{Bc7&m*AcIyaZKhK#Tg!53MrIOz4-wby&xO*gdQ
zoZG3>a6Lwsr<Kmb4KsPht%c<Av;{po{lOhxzo*RzhQc0iq=#Wz+U@t@GpNd%x|MW-
zZ4M?6O_0s$t_M@t5RRVZ6J-AR6GSYShvwB8tmbz-FnR~R#_|Ls1~b1_@>F}OYdm3m
zorDjQJapi0djJR2S`*tNVV@VDQB9>+%k)}^XUVRxrcD>ULMB`C^|W~dJAI*GB%HQj
zrDU7RAqUr$lB?QN>)92|!BNBHcTy;NB<$VE5ATJ|x0?~X^5CT3&OkcjD%RZ@@X+bL
zj6Un)A3yN}emxWo`N9#rw#byU6i!6;th>Ix+2wI<Xrx1rdCQRTjLK+Zp2n>k>YJO^
zg3FS^KaOIa_3n)g^={9`b?a_$wRl?U8{95Z$4?Y)4M%tz0G@fkv8GuM6#k&Yndwc)
zJegR!E|-J((a4!f(+4AGl082DZe^aRoSD=e3hwmab<CIJ$YxEGCn1{!vNkufJxTN0
zI>X&wqqT#Xrz~e?Vf~ISGiBNt%-P8nM^K}OChIBFQ})rjG0un16gLlr&ZLQBpfict
zXJLEV!a+}mH_+zCIoVAcQ4?)wZRlTZvb<Cs>0V>4cxozWs`R;JW*%wZWp)9EF#>JP
zY*uNWbYGw9E7qKa?>!#8D+^J6fOW1zyx}tVA=4H*UOSUwo|v6U^LTJzxa=u(lDHfu
zGtLpW+C1fvj03`<mf8`Wtc`l&A5v!J#Jy%7-kq6}<B{?Yk!Zcnh6~d<9^`sKtsuzt
zI+4tIWt`w8xM+<`%C{6~z^48mg37U}n|KVT3Dt;M<&6T3rx`+_lVel2Z23kc;ZvcY
z%zv;z<IIDCP{lFJx}v}(*nkMVJ4pGF0u30$x&{YK6Oh&+$xm{AT4pA*8v!w;OUui3
z4?CW_6y!RkiJ3ZNoFKgy(Hf!5U#^er5DTB#_DenqzW{k!bCkGTM;R3I-INIep=77D
zpJ)_r$3;Gs@^Zb!E*3twBaMP6PFh};zZYpF-#|<Va$TlF$jfzBlBcx*nZK0(GHB#i
zVeZuWjaXRCUTL7btRV)(DlgZ0<a@MAiI(6HaS87OZk5;OYMet83(eOr69NKAIZ6Ev
zB3601E`DE)$svj%*}1RM2qOD&k-YT3TwmHR<mr1{Ymokvv;&A(<>mhmk^fiZJ;^9w
zw*P*dEx$_RoCj8!0{3MJWapm%TL~^{rhK23|7RqdKO6rgWEG3NT>qME1`#2fKN~**
z9xA`gKfTUG9UCE=KO6r8cx2z;{2R6tNdF%g`ktDCY@bqIwu9G@X3bx&yG?H7VtMnI
zdd9QlnS5{9WXkV1LHaJ`Bzy}g*8Jsr=zbxeRehW{tqhTOagi?RKly(CppdV%G9#{L
z;k4jK`B!DprF>TRN#{>86iRW)yg5PKMn(R1$tdVqDAy<Zj#d7IcyX*1{rI1dNIwdY
aev{)5_0L@SJ8m|$jmUw|MAP;Ql>IMWNmJhd