added Intel Core i5-6500;

changed results view of gcbd benchmark to relative performance gain;
changed reference CPU and GPU;
target architecture detection fixes

README.md

@@ -1,9 +1,13 @@
 # BLAS libraries benchmarks
 Andrzej Wójtowicz  
 
-[![DOI](https://zenodo.org/badge/22705/andre-wojtowicz/blas-benchmarks.svg)](https://dx.doi.org/10.5281/zenodo.55662)
+Document generation date: 2016-07-14 17:20:41
 
-Document generation date: 2016-06-06 22:40:18
+This document presents timing results for BLAS ([Basic Linear Algebra Subprograms](https://en.wikipedia.org/wiki/Basic_Linear_Algebra_Subprograms)) libraries in [R](https://en.wikipedia.org/wiki/R_(programming_language)) on diverse CPUs and GPUs.
+
+### Changelog
+
+ * 2016-07-14: **results:** added Intel Core i5-6500; changed results view of gcbd benchmark to relative performance gain; changed reference CPU (Intel Pentium Dual-Core E5300) and GPU (NVIDIA GeForce GT 630M); **code:** fixed target architecture detection for Intel Core i5-6500-like CPUs in multi-threaded Atlas library; added info how to force target architecture in GotoBLAS2 and BLIS libraries.
 
 
 
@@ -14,6 +18,7 @@ Document generation date: 2016-06-06 22:40:18
     * [Intel Core i7-4790K + MSI GeForce GTX 980 Ti Lightning](#intel-core-i7-4790k--msi-geforce-gtx-980-ti-lightning)
     * [Intel Core i5-4590 + NVIDIA GeForce GT 430](#intel-core-i5-4590--nvidia-geforce-gt-430)
     * [Intel Core i5-4590 + NVIDIA GeForce GTX 750 Ti](#intel-core-i5-4590--nvidia-geforce-gtx-750-ti)
+    * [Intel Core i5-6500](#intel-core-i5-6500)
     * [Intel Core i5-3570](#intel-core-i5-3570)
     * [Intel Core i3-2120](#intel-core-i3-2120)
     * [Intel Core i3-3120M](#intel-core-i3-3120m)
@@ -33,7 +38,7 @@ Document generation date: 2016-06-06 22:40:18
 
 ## Configuration
 
-**OS**: Debian Jessie, kernel 4.4
+**OS**: [Debian](https://www.debian.org/) Jessie, kernel 4.4
 
 **R software**: [Microsoft R Open](https://mran.microsoft.com/open/) (3.2.4)
 
@@ -54,11 +59,12 @@ Document generation date: 2016-06-06 22:40:18
 |1.|[Intel Core i7-4790K](http://ark.intel.com/products/80807/Intel-Core-i7-4790K-Processor-8M-Cache-up-to-4_40-GHz) (OC 4.5 GHz)|[MSI GeForce GTX 980 Ti Lightning](https://us.msi.com/Graphics-card/GTX-980-Ti-LIGHTNING.html#hero-specification)|
 |2.|[Intel Core i5-4590](http://ark.intel.com/products/80815/Intel-Core-i5-4590-Processor-6M-Cache-up-to-3_70-GHz)|[NVIDIA GeForce GT 430](http://www.geforce.com/hardware/desktop-gpus/geforce-gt-430/specifications)|
 |3.|[Intel Core i5-4590](http://ark.intel.com/products/80815/Intel-Core-i5-4590-Processor-6M-Cache-up-to-3_70-GHz)|[NVIDIA GeForce GTX 750 Ti](http://www.geforce.com/hardware/desktop-gpus/geforce-gtx-750-ti/specifications)|
-|4.|[Intel Core i5-3570](http://ark.intel.com/products/65702/Intel-Core-i5-3570-Processor-6M-Cache-up-to-3_80-GHz)| - |
+|4.|[Intel Core i5-6500](http://ark.intel.com/products/88184/Intel-Core-i5-6500-Processor-6M-Cache-up-to-3_60-GHz)| - |
-|5.|[Intel Core i3-2120](http://ark.intel.com/products/53426/Intel-Core-i3-2120-Processor-3M-Cache-3_30-GHz)| - |
+|5.|[Intel Core i5-3570](http://ark.intel.com/products/65702/Intel-Core-i5-3570-Processor-6M-Cache-up-to-3_80-GHz)| - |
-|6.|[Intel Core i3-3120M](http://ark.intel.com/products/71465/Intel-Core-i3-3120M-Processor-3M-Cache-2_50-GHz)| - |
+|6.|[Intel Core i3-2120](http://ark.intel.com/products/53426/Intel-Core-i3-2120-Processor-3M-Cache-3_30-GHz)| - |
-|7.|[Intel Core i5-3317U](http://ark.intel.com/products/65707/Intel-Core-i5-3317U-Processor-3M-Cache-up-to-2_60-GHz)|[NVIDIA GeForce GT 630M](http://www.geforce.com/hardware/notebook-gpus/geforce-gt-630m/specifications)|
+|7.|[Intel Core i3-3120M](http://ark.intel.com/products/71465/Intel-Core-i3-3120M-Processor-3M-Cache-2_50-GHz)| - |
-|8.|[Intel Pentium Dual-Core E5300](http://ark.intel.com/products/35300/Intel-Pentium-Processor-E5300-2M-Cache-2_60-GHz-800-MHz-FSB)| - |
+|8.|[Intel Core i5-3317U](http://ark.intel.com/products/65707/Intel-Core-i5-3317U-Processor-3M-Cache-up-to-2_60-GHz)|[NVIDIA GeForce GT 630M](http://www.geforce.com/hardware/notebook-gpus/geforce-gt-630m/specifications)|
+|9.|[Intel Pentium Dual-Core E5300](http://ark.intel.com/products/35300/Intel-Pentium-Processor-E5300-2M-Cache-2_60-GHz-800-MHz-FSB)| - |
 
 **Benchmarks**: [R-benchmark-25](http://r.research.att.com/benchmarks/R-benchmark-25.R), [Revolution](https://gist.github.com/andrie/24c9672f1ea39af89c66#file-rro-mkl-benchmark-r), [Gcbd](https://cran.r-project.org/web/packages/gcbd/vignettes/gcbd.pdf).
 
@@ -176,33 +182,33 @@ Time in seconds  - 10 runs - lower is better
 
 #### Matrix Multiply 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h1_b3_t1.png)![](gen/img/img_ph_h1_b3_t1b.png)
+![](gen/img/img_ph_h1_b3_t1.png)
 
 
 
 #### QR Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h1_b3_t2.png)![](gen/img/img_ph_h1_b3_t2b.png)
+![](gen/img/img_ph_h1_b3_t2.png)
 
 
 
 #### Singular Value Deomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h1_b3_t3.png)![](gen/img/img_ph_h1_b3_t3b.png)
+![](gen/img/img_ph_h1_b3_t3.png)
 
 
 
 #### Triangular Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h1_b3_t4.png)![](gen/img/img_ph_h1_b3_t4b.png)
+![](gen/img/img_ph_h1_b3_t4.png)
 
 
 
@@ -314,33 +320,33 @@ Time in seconds  - 10 runs - lower is better
 
 #### Matrix Multiply 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h2_b3_t1.png)![](gen/img/img_ph_h2_b3_t1b.png)
+![](gen/img/img_ph_h2_b3_t1.png)
 
 
 
 #### QR Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h2_b3_t2.png)![](gen/img/img_ph_h2_b3_t2b.png)
+![](gen/img/img_ph_h2_b3_t2.png)
 
 
 
 #### Singular Value Deomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h2_b3_t3.png)![](gen/img/img_ph_h2_b3_t3b.png)
+![](gen/img/img_ph_h2_b3_t3.png)
 
 
 
 #### Triangular Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h2_b3_t4.png)![](gen/img/img_ph_h2_b3_t4b.png)
+![](gen/img/img_ph_h2_b3_t4.png)
 
 
 
@@ -452,37 +458,37 @@ Time in seconds  - 10 runs - lower is better
 
 #### Matrix Multiply 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h3_b3_t1.png)![](gen/img/img_ph_h3_b3_t1b.png)
+![](gen/img/img_ph_h3_b3_t1.png)
 
 
 
 #### QR Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h3_b3_t2.png)![](gen/img/img_ph_h3_b3_t2b.png)
+![](gen/img/img_ph_h3_b3_t2.png)
 
 
 
 #### Singular Value Deomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h3_b3_t3.png)![](gen/img/img_ph_h3_b3_t3b.png)
+![](gen/img/img_ph_h3_b3_t3.png)
 
 
 
 #### Triangular Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h3_b3_t4.png)![](gen/img/img_ph_h3_b3_t4b.png)
+![](gen/img/img_ph_h3_b3_t4.png)
 
 
 
-## Intel Core i5-3570
+## Intel Core i5-6500
 
 
 
@@ -590,37 +596,37 @@ Time in seconds  - 10 runs - lower is better
 
 #### Matrix Multiply 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h4_b3_t1.png)![](gen/img/img_ph_h4_b3_t1b.png)
+![](gen/img/img_ph_h4_b3_t1.png)
 
 
 
 #### QR Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h4_b3_t2.png)![](gen/img/img_ph_h4_b3_t2b.png)
+![](gen/img/img_ph_h4_b3_t2.png)
 
 
 
 #### Singular Value Deomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h4_b3_t3.png)![](gen/img/img_ph_h4_b3_t3b.png)
+![](gen/img/img_ph_h4_b3_t3.png)
 
 
 
 #### Triangular Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h4_b3_t4.png)![](gen/img/img_ph_h4_b3_t4b.png)
+![](gen/img/img_ph_h4_b3_t4.png)
 
 
 
-## Intel Core i3-2120
+## Intel Core i5-3570
 
 
 
@@ -728,37 +734,37 @@ Time in seconds  - 10 runs - lower is better
 
 #### Matrix Multiply 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h5_b3_t1.png)![](gen/img/img_ph_h5_b3_t1b.png)
+![](gen/img/img_ph_h5_b3_t1.png)
 
 
 
 #### QR Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h5_b3_t2.png)![](gen/img/img_ph_h5_b3_t2b.png)
+![](gen/img/img_ph_h5_b3_t2.png)
 
 
 
 #### Singular Value Deomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h5_b3_t3.png)![](gen/img/img_ph_h5_b3_t3b.png)
+![](gen/img/img_ph_h5_b3_t3.png)
 
 
 
 #### Triangular Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h5_b3_t4.png)![](gen/img/img_ph_h5_b3_t4b.png)
+![](gen/img/img_ph_h5_b3_t4.png)
 
 
 
-## Intel Core i3-3120M
+## Intel Core i3-2120
 
 
 
@@ -866,37 +872,37 @@ Time in seconds  - 10 runs - lower is better
 
 #### Matrix Multiply 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h6_b3_t1.png)![](gen/img/img_ph_h6_b3_t1b.png)
+![](gen/img/img_ph_h6_b3_t1.png)
 
 
 
 #### QR Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h6_b3_t2.png)![](gen/img/img_ph_h6_b3_t2b.png)
+![](gen/img/img_ph_h6_b3_t2.png)
 
 
 
 #### Singular Value Deomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h6_b3_t3.png)![](gen/img/img_ph_h6_b3_t3b.png)
+![](gen/img/img_ph_h6_b3_t3.png)
 
 
 
 #### Triangular Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h6_b3_t4.png)![](gen/img/img_ph_h6_b3_t4b.png)
+![](gen/img/img_ph_h6_b3_t4.png)
 
 
 
-## Intel Core i5-3317U + NVIDIA GeForce GT 630M
+## Intel Core i3-3120M
 
 
 
@@ -1004,37 +1010,37 @@ Time in seconds  - 10 runs - lower is better
 
 #### Matrix Multiply 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h7_b3_t1.png)![](gen/img/img_ph_h7_b3_t1b.png)
+![](gen/img/img_ph_h7_b3_t1.png)
 
 
 
 #### QR Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h7_b3_t2.png)![](gen/img/img_ph_h7_b3_t2b.png)
+![](gen/img/img_ph_h7_b3_t2.png)
 
 
 
 #### Singular Value Deomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h7_b3_t3.png)![](gen/img/img_ph_h7_b3_t3b.png)
+![](gen/img/img_ph_h7_b3_t3.png)
 
 
 
 #### Triangular Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h7_b3_t4.png)![](gen/img/img_ph_h7_b3_t4b.png)
+![](gen/img/img_ph_h7_b3_t4.png)
 
 
 
-## Intel Pentium Dual-Core E5300
+## Intel Core i5-3317U + NVIDIA GeForce GT 630M
 
 
 
@@ -1058,8 +1064,6 @@ Time in seconds  - 10 runs - lower is better
 
 #### Eigenvalues of a 600x600 random matrix 
 
-BLIS hangs in this test
-
 Time in seconds - 10 runs - lower is better
 
 ![](gen/img/img_ph_h8_b1_t3.png)
@@ -1144,33 +1148,173 @@ Time in seconds  - 10 runs - lower is better
 
 #### Matrix Multiply 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h8_b3_t1.png)![](gen/img/img_ph_h8_b3_t1b.png)
+![](gen/img/img_ph_h8_b3_t1.png)
 
 
 
 #### QR Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h8_b3_t2.png)![](gen/img/img_ph_h8_b3_t2b.png)
+![](gen/img/img_ph_h8_b3_t2.png)
 
 
 
 #### Singular Value Deomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h8_b3_t3.png)![](gen/img/img_ph_h8_b3_t3b.png)
+![](gen/img/img_ph_h8_b3_t3.png)
 
 
 
 #### Triangular Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
 
-![](gen/img/img_ph_h8_b3_t4.png)![](gen/img/img_ph_h8_b3_t4b.png)
+![](gen/img/img_ph_h8_b3_t4.png)
+
+
+
+## Intel Pentium Dual-Core E5300
+
+
+
+### R-benchmark-25
+
+#### 2800x2800 cross-product matrix 
+
+Time in seconds - 10 runs - lower is better
+
+![](gen/img/img_ph_h9_b1_t1.png)
+
+
+
+#### Linear regr. over a 2000x2000 matrix 
+
+Time in seconds - 10 runs - lower is better
+
+![](gen/img/img_ph_h9_b1_t2.png)
+
+
+
+#### Eigenvalues of a 600x600 random matrix 
+
+BLIS hangs in this test
+
+Time in seconds - 10 runs - lower is better
+
+![](gen/img/img_ph_h9_b1_t3.png)
+
+
+
+#### Determinant of a 2500x2500 random matrix 
+
+Time in seconds - 10 runs - lower is better
+
+![](gen/img/img_ph_h9_b1_t4.png)
+
+
+
+#### Cholesky decomposition of a 3000x3000 matrix 
+
+Time in seconds - 10 runs - lower is better
+
+![](gen/img/img_ph_h9_b1_t5.png)
+
+
+
+#### Inverse of a 1600x1600 random matrix 
+
+Time in seconds - 10 runs - lower is better
+
+![](gen/img/img_ph_h9_b1_t6.png)
+
+
+
+#### Escoufier's method on a 45x45 matrix 
+
+Time in seconds - 10 runs - lower is better
+
+![](gen/img/img_ph_h9_b1_t7.png)
+
+
+
+### Revolution benchmark
+
+#### Matrix Multiply 
+
+Time in seconds - 10 runs - lower is better
+
+![](gen/img/img_ph_h9_b2_t1.png)
+
+
+
+#### Cholesky Factorization 
+
+Time in seconds - 10 runs - lower is better
+
+![](gen/img/img_ph_h9_b2_t2.png)
+
+
+
+#### Singular Value Deomposition 
+
+Time in seconds - 10 runs - lower is better
+
+![](gen/img/img_ph_h9_b2_t3.png)
+
+
+
+#### Principal Components Analysis 
+
+Time in seconds - 10 runs - lower is better
+
+![](gen/img/img_ph_h9_b2_t4.png)
+
+
+
+#### Linear Discriminant Analysis 
+
+Time in seconds - 10 runs - lower is better
+
+![](gen/img/img_ph_h9_b2_t5.png)
+
+
+
+### Gcbd benchmark
+
+#### Matrix Multiply 
+
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
+
+![](gen/img/img_ph_h9_b3_t1.png)
+
+
+
+#### QR Decomposition 
+
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
+
+![](gen/img/img_ph_h9_b3_t2.png)
+
+
+
+#### Singular Value Deomposition 
+
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
+
+![](gen/img/img_ph_h9_b3_t3.png)
+
+
+
+#### Triangular Decomposition 
+
+Performance gain regarding matrix size - reference: Netlib - from 50 to 5 runs - higher is better
+
+![](gen/img/img_ph_h9_b3_t4.png)
 
 
 
@@ -1285,33 +1429,33 @@ Time in seconds  - 10 runs - lower is better
 
 #### Matrix Multiply 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l1_b3_t1.png)![](gen/img/img_pl_l1_b3_t1b.png)
+![](gen/img/img_pl_l1_b3_t1.png)
 
 
 
 #### QR Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l1_b3_t2.png)![](gen/img/img_pl_l1_b3_t2b.png)
+![](gen/img/img_pl_l1_b3_t2.png)
 
 
 
 #### Singular Value Deomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l1_b3_t3.png)![](gen/img/img_pl_l1_b3_t3b.png)
+![](gen/img/img_pl_l1_b3_t3.png)
 
 
 
 #### Triangular Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l1_b3_t4.png)![](gen/img/img_pl_l1_b3_t4b.png)
+![](gen/img/img_pl_l1_b3_t4.png)
 
 
 
@@ -1423,33 +1567,33 @@ Time in seconds  - 10 runs - lower is better
 
 #### Matrix Multiply 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l2_b3_t1.png)![](gen/img/img_pl_l2_b3_t1b.png)
+![](gen/img/img_pl_l2_b3_t1.png)
 
 
 
 #### QR Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l2_b3_t2.png)![](gen/img/img_pl_l2_b3_t2b.png)
+![](gen/img/img_pl_l2_b3_t2.png)
 
 
 
 #### Singular Value Deomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l2_b3_t3.png)![](gen/img/img_pl_l2_b3_t3b.png)
+![](gen/img/img_pl_l2_b3_t3.png)
 
 
 
 #### Triangular Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l2_b3_t4.png)![](gen/img/img_pl_l2_b3_t4b.png)
+![](gen/img/img_pl_l2_b3_t4.png)
 
 
 
@@ -1561,33 +1705,33 @@ Time in seconds  - 10 runs - lower is better
 
 #### Matrix Multiply 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l3_b3_t1.png)![](gen/img/img_pl_l3_b3_t1b.png)
+![](gen/img/img_pl_l3_b3_t1.png)
 
 
 
 #### QR Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l3_b3_t2.png)![](gen/img/img_pl_l3_b3_t2b.png)
+![](gen/img/img_pl_l3_b3_t2.png)
 
 
 
 #### Singular Value Deomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l3_b3_t3.png)![](gen/img/img_pl_l3_b3_t3b.png)
+![](gen/img/img_pl_l3_b3_t3.png)
 
 
 
 #### Triangular Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l3_b3_t4.png)![](gen/img/img_pl_l3_b3_t4b.png)
+![](gen/img/img_pl_l3_b3_t4.png)
 
 
 
@@ -1699,33 +1843,33 @@ Time in seconds  - 10 runs - lower is better
 
 #### Matrix Multiply 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l4_b3_t1.png)![](gen/img/img_pl_l4_b3_t1b.png)
+![](gen/img/img_pl_l4_b3_t1.png)
 
 
 
 #### QR Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l4_b3_t2.png)![](gen/img/img_pl_l4_b3_t2b.png)
+![](gen/img/img_pl_l4_b3_t2.png)
 
 
 
 #### Singular Value Deomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l4_b3_t3.png)![](gen/img/img_pl_l4_b3_t3b.png)
+![](gen/img/img_pl_l4_b3_t3.png)
 
 
 
 #### Triangular Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l4_b3_t4.png)![](gen/img/img_pl_l4_b3_t4b.png)
+![](gen/img/img_pl_l4_b3_t4.png)
 
 
 
@@ -1837,33 +1981,33 @@ Time in seconds  - 10 runs - lower is better
 
 #### Matrix Multiply 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l5_b3_t1.png)![](gen/img/img_pl_l5_b3_t1b.png)
+![](gen/img/img_pl_l5_b3_t1.png)
 
 
 
 #### QR Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l5_b3_t2.png)![](gen/img/img_pl_l5_b3_t2b.png)
+![](gen/img/img_pl_l5_b3_t2.png)
 
 
 
 #### Singular Value Deomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l5_b3_t3.png)![](gen/img/img_pl_l5_b3_t3b.png)
+![](gen/img/img_pl_l5_b3_t3.png)
 
 
 
 #### Triangular Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l5_b3_t4.png)![](gen/img/img_pl_l5_b3_t4b.png)
+![](gen/img/img_pl_l5_b3_t4.png)
 
 
 
@@ -1975,33 +2119,33 @@ Time in seconds  - 10 runs - lower is better
 
 #### Matrix Multiply 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l6_b3_t1.png)![](gen/img/img_pl_l6_b3_t1b.png)
+![](gen/img/img_pl_l6_b3_t1.png)
 
 
 
 #### QR Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l6_b3_t2.png)![](gen/img/img_pl_l6_b3_t2b.png)
+![](gen/img/img_pl_l6_b3_t2.png)
 
 
 
 #### Singular Value Deomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l6_b3_t3.png)![](gen/img/img_pl_l6_b3_t3b.png)
+![](gen/img/img_pl_l6_b3_t3.png)
 
 
 
 #### Triangular Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l6_b3_t4.png)![](gen/img/img_pl_l6_b3_t4b.png)
+![](gen/img/img_pl_l6_b3_t4.png)
 
 
 
@@ -2115,33 +2259,33 @@ Time in seconds  - 10 runs - lower is better
 
 #### Matrix Multiply 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l7_b3_t1.png)![](gen/img/img_pl_l7_b3_t1b.png)
+![](gen/img/img_pl_l7_b3_t1.png)
 
 
 
 #### QR Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l7_b3_t2.png)![](gen/img/img_pl_l7_b3_t2b.png)
+![](gen/img/img_pl_l7_b3_t2.png)
 
 
 
 #### Singular Value Deomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l7_b3_t3.png)![](gen/img/img_pl_l7_b3_t3b.png)
+![](gen/img/img_pl_l7_b3_t3.png)
 
 
 
 #### Triangular Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: Intel Pentium Dual-Core E5300 - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l7_b3_t4.png)![](gen/img/img_pl_l7_b3_t4b.png)
+![](gen/img/img_pl_l7_b3_t4.png)
 
 
 
@@ -2253,31 +2397,31 @@ Time in seconds  - 10 runs - lower is better
 
 #### Matrix Multiply 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: NVIDIA GeForce GT 630M - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l8_b3_t1.png)![](gen/img/img_pl_l8_b3_t1b.png)
+![](gen/img/img_pl_l8_b3_t1.png)
 
 
 
 #### QR Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: NVIDIA GeForce GT 630M - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l8_b3_t2.png)![](gen/img/img_pl_l8_b3_t2b.png)
+![](gen/img/img_pl_l8_b3_t2.png)
 
 
 
 #### Singular Value Deomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: NVIDIA GeForce GT 630M - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l8_b3_t3.png)![](gen/img/img_pl_l8_b3_t3b.png)
+![](gen/img/img_pl_l8_b3_t3.png)
 
 
 
 #### Triangular Decomposition 
 
-Time in seconds regarding matrix size - right panel on log scale -  from  50  to  5 runs - lower is better
+Performance gain regarding matrix size - reference: NVIDIA GeForce GT 630M - from 50 to 5 runs - higher is better
 
-![](gen/img/img_pl_l8_b3_t4.png)![](gen/img/img_pl_l8_b3_t4b.png)
+![](gen/img/img_pl_l8_b3_t4.png)