colRowDiffs_subset

matrixStats: Benchmark report

---

colDiffs() and rowDiffs() benchmarks on subsetted computation

This report benchmark the performance of colDiffs() and rowDiffs() on subsetted computation.

Data type "integer"

Data

> rmatrix <- function(nrow, ncol, mode = c("logical", "double", "integer", "index"), range = c(-100, 
+     +100), na_prob = 0) {
+     mode <- match.arg(mode)
+     n <- nrow * ncol
+     if (mode == "logical") {
+         x <- sample(c(FALSE, TRUE), size = n, replace = TRUE)
+     }     else if (mode == "index") {
+         x <- seq_len(n)
+         mode <- "integer"
+     }     else {
+         x <- runif(n, min = range[1], max = range[2])
+     }
+     storage.mode(x) <- mode
+     if (na_prob > 0) 
+         x[sample(n, size = na_prob * n)] <- NA
+     dim(x) <- c(nrow, ncol)
+     x
+ }
> rmatrices <- function(scale = 10, seed = 1, ...) {
+     set.seed(seed)
+     data <- list()
+     data[[1]] <- rmatrix(nrow = scale * 1, ncol = scale * 1, ...)
+     data[[2]] <- rmatrix(nrow = scale * 10, ncol = scale * 10, ...)
+     data[[3]] <- rmatrix(nrow = scale * 100, ncol = scale * 1, ...)
+     data[[4]] <- t(data[[3]])
+     data[[5]] <- rmatrix(nrow = scale * 10, ncol = scale * 100, ...)
+     data[[6]] <- t(data[[5]])
+     names(data) <- sapply(data, FUN = function(x) paste(dim(x), collapse = "x"))
+     data
+ }
> data <- rmatrices(mode = mode)

Results

10x10 integer matrix

> X <- data[["10x10"]]
> rows <- sample.int(nrow(X), size = nrow(X) * 0.7)
> cols <- sample.int(ncol(X), size = ncol(X) * 0.7)
> X_S <- X[rows, cols]
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3113624 166.3    5709258 305.0  5709258 305.0
Vcells 5834546  44.6   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colDiffs_X_S = colDiffs(X_S), `colDiffs(X, rows, cols)` = colDiffs(X, 
+     rows = rows, cols = cols), `colDiffs(X[rows, cols])` = colDiffs(X[rows, cols]), unit = "ms")
> X <- t(X)
> X_S <- t(X_S)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3112797 166.3    5709258 305.0  5709258 305.0
Vcells 5832508  44.5   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowDiffs_X_S = rowDiffs(X_S), `rowDiffs(X, cols, rows)` = rowDiffs(X, 
+     rows = cols, cols = rows), `rowDiffs(X[cols, rows])` = rowDiffs(X[cols, rows]), unit = "ms")

Table: Benchmarking of colDiffs_X_S(), colDiffs(X, rows, cols)() and colDiffs(X[rows, cols])() on integer+10x10 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	0.001091	0.0012340	0.0023884	0.0013290	0.001448	0.101899
2	colDiffs(X, rows, cols)	0.001271	0.0014815	0.0016570	0.0015785	0.001671	0.003063
3	colDiffs(X[rows, cols])	0.001570	0.0019545	0.0021673	0.0020585	0.002199	0.008013

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	1.000000	1.000000	1.0000000	1.000000	1.000000	1.0000000
2	colDiffs(X, rows, cols)	1.164986	1.200567	0.6937606	1.187735	1.154006	0.0300592
3	colDiffs(X[rows, cols])	1.439047	1.583874	0.9074302	1.548909	1.518646	0.0786367

Table: Benchmarking of rowDiffs_X_S(), rowDiffs(X, cols, rows)() and rowDiffs(X[cols, rows])() on integer+10x10 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	rowDiffs_X_S	0.001095	0.0012615	0.0014586	0.0013540	0.0014220	0.006309
2	rowDiffs(X, cols, rows)	0.001318	0.0015310	0.0026410	0.0016080	0.0017385	0.093415
3	rowDiffs(X[cols, rows])	0.001568	0.0019345	0.0021375	0.0020505	0.0021545	0.006355

	expr	min	lq	mean	median	uq	max
1	rowDiffs_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	rowDiffs(X, cols, rows)	1.203653	1.213635	1.810662	1.187592	1.222574	14.806626
3	rowDiffs(X[cols, rows])	1.431963	1.533492	1.465429	1.514402	1.515119	1.007291

Figure: Benchmarking of colDiffs_X_S(), colDiffs(X, rows, cols)() and colDiffs(X[rows, cols])() on integer+10x10 data as well as rowDiffs_X_S(), rowDiffs(X, cols, rows)() and rowDiffs(X[cols, rows])() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colDiffs_X_S() and rowDiffs_X_S() on integer+10x10 data (original and transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	1.091	1.2340	2.38836	1.329	1.448	101.899
2	rowDiffs_X_S	1.095	1.2615	1.45861	1.354	1.422	6.309

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	1.000000	1.000000	1.0000000	1.000000	1.0000000	1.0000000
2	rowDiffs_X_S	1.003666	1.022285	0.6107161	1.018811	0.9820442	0.0619142

Figure: Benchmarking of colDiffs_X_S() and rowDiffs_X_S() on integer+10x10 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

100x100 integer matrix

> X <- data[["100x100"]]
> rows <- sample.int(nrow(X), size = nrow(X) * 0.7)
> cols <- sample.int(ncol(X), size = ncol(X) * 0.7)
> X_S <- X[rows, cols]
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3111391 166.2    5709258 305.0  5709258 305.0
Vcells 5501366  42.0   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colDiffs_X_S = colDiffs(X_S), `colDiffs(X, rows, cols)` = colDiffs(X, 
+     rows = rows, cols = cols), `colDiffs(X[rows, cols])` = colDiffs(X[rows, cols]), unit = "ms")
> X <- t(X)
> X_S <- t(X_S)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3111385 166.2    5709258 305.0  5709258 305.0
Vcells 5506449  42.1   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowDiffs_X_S = rowDiffs(X_S), `rowDiffs(X, cols, rows)` = rowDiffs(X, 
+     rows = cols, cols = rows), `rowDiffs(X[cols, rows])` = rowDiffs(X[cols, rows]), unit = "ms")

Table: Benchmarking of colDiffs_X_S(), colDiffs(X, rows, cols)() and colDiffs(X[rows, cols])() on integer+100x100 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	0.005324	0.0055875	0.0081225	0.0057265	0.0089555	0.039719
2	colDiffs(X, rows, cols)	0.008646	0.0089740	0.0107692	0.0091480	0.0094515	0.046548
3	colDiffs(X[rows, cols])	0.012920	0.0132015	0.0212689	0.0135625	0.0214185	0.178538

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	colDiffs(X, rows, cols)	1.623967	1.606085	1.325854	1.597485	1.055385	1.171933
3	colDiffs(X[rows, cols])	2.426747	2.362685	2.618527	2.368375	2.391659	4.495028

Table: Benchmarking of rowDiffs_X_S(), rowDiffs(X, cols, rows)() and rowDiffs(X[cols, rows])() on integer+100x100 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	rowDiffs_X_S	0.005338	0.0055090	0.0057910	0.0056115	0.0057570	0.010078
2	rowDiffs(X, cols, rows)	0.008235	0.0084540	0.0086759	0.0085615	0.0087395	0.012207
3	rowDiffs(X[cols, rows])	0.012938	0.0131685	0.0138666	0.0132730	0.0134630	0.056660

	expr	min	lq	mean	median	uq	max
1	rowDiffs_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	rowDiffs(X, cols, rows)	1.542713	1.534580	1.498175	1.525706	1.518065	1.211252
3	rowDiffs(X[cols, rows])	2.423754	2.390361	2.394532	2.365321	2.338544	5.622147

Figure: Benchmarking of colDiffs_X_S(), colDiffs(X, rows, cols)() and colDiffs(X[rows, cols])() on integer+100x100 data as well as rowDiffs_X_S(), rowDiffs(X, cols, rows)() and rowDiffs(X[cols, rows])() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colDiffs_X_S() and rowDiffs_X_S() on integer+100x100 data (original and transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	rowDiffs_X_S	5.338	5.5090	5.79096	5.6115	5.7570	10.078
1	colDiffs_X_S	5.324	5.5875	8.12247	5.7265	8.9555	39.719

	expr	min	lq	mean	median	uq	max
2	rowDiffs_X_S	1.0000000	1.000000	1.000000	1.000000	1.000000	1.000000
1	colDiffs_X_S	0.9973773	1.014249	1.402612	1.020494	1.555584	3.941159

Figure: Benchmarking of colDiffs_X_S() and rowDiffs_X_S() on integer+100x100 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

1000x10 integer matrix

> X <- data[["1000x10"]]
> rows <- sample.int(nrow(X), size = nrow(X) * 0.7)
> cols <- sample.int(ncol(X), size = ncol(X) * 0.7)
> X_S <- X[rows, cols]
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3112146 166.3    5709258 305.0  5709258 305.0
Vcells 5505452  42.1   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colDiffs_X_S = colDiffs(X_S), `colDiffs(X, rows, cols)` = colDiffs(X, 
+     rows = rows, cols = cols), `colDiffs(X[rows, cols])` = colDiffs(X[rows, cols]), unit = "ms")
> X <- t(X)
> X_S <- t(X_S)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3112137 166.3    5709258 305.0  5709258 305.0
Vcells 5510530  42.1   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowDiffs_X_S = rowDiffs(X_S), `rowDiffs(X, cols, rows)` = rowDiffs(X, 
+     rows = cols, cols = rows), `rowDiffs(X[cols, rows])` = rowDiffs(X[cols, rows]), unit = "ms")

Table: Benchmarking of colDiffs_X_S(), colDiffs(X, rows, cols)() and colDiffs(X[rows, cols])() on integer+1000x10 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	0.005339	0.0055580	0.0058249	0.0056480	0.0057980	0.009813
2	colDiffs(X, rows, cols)	0.009087	0.0093785	0.0097250	0.0094570	0.0095825	0.031486
3	colDiffs(X[rows, cols])	0.012544	0.0130270	0.0132707	0.0131635	0.0132730	0.016739

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	colDiffs(X, rows, cols)	1.702004	1.687388	1.669549	1.674398	1.652725	3.208601
3	colDiffs(X[rows, cols])	2.349504	2.343829	2.278259	2.330648	2.289238	1.705798

Table: Benchmarking of rowDiffs_X_S(), rowDiffs(X, cols, rows)() and rowDiffs(X[cols, rows])() on integer+1000x10 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	rowDiffs_X_S	0.005727	0.0058745	0.0064510	0.0059620	0.0060875	0.039065
2	rowDiffs(X, cols, rows)	0.009787	0.0100330	0.0103754	0.0101645	0.0103020	0.025221
3	rowDiffs(X[cols, rows])	0.014053	0.0145150	0.0148189	0.0146570	0.0148595	0.019932

	expr	min	lq	mean	median	uq	max
1	rowDiffs_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.0000000
2	rowDiffs(X, cols, rows)	1.708923	1.707890	1.608329	1.704881	1.692320	0.6456163
3	rowDiffs(X[cols, rows])	2.453815	2.470849	2.297144	2.458403	2.440986	0.5102265

Figure: Benchmarking of colDiffs_X_S(), colDiffs(X, rows, cols)() and colDiffs(X[rows, cols])() on integer+1000x10 data as well as rowDiffs_X_S(), rowDiffs(X, cols, rows)() and rowDiffs(X[cols, rows])() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colDiffs_X_S() and rowDiffs_X_S() on integer+1000x10 data (original and transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	5.339	5.5580	5.82494	5.648	5.7980	9.813
2	rowDiffs_X_S	5.727	5.8745	6.45101	5.962	6.0875	39.065

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	rowDiffs_X_S	1.072673	1.056945	1.107481	1.055595	1.049931	3.980944

Figure: Benchmarking of colDiffs_X_S() and rowDiffs_X_S() on integer+1000x10 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

10x1000 integer matrix

> X <- data[["10x1000"]]
> rows <- sample.int(nrow(X), size = nrow(X) * 0.7)
> cols <- sample.int(ncol(X), size = ncol(X) * 0.7)
> X_S <- X[rows, cols]
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3112349 166.3    5709258 305.0  5709258 305.0
Vcells 5506252  42.1   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colDiffs_X_S = colDiffs(X_S), `colDiffs(X, rows, cols)` = colDiffs(X, 
+     rows = rows, cols = cols), `colDiffs(X[rows, cols])` = colDiffs(X[rows, cols]), unit = "ms")
> X <- t(X)
> X_S <- t(X_S)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3112343 166.3    5709258 305.0  5709258 305.0
Vcells 5511335  42.1   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowDiffs_X_S = rowDiffs(X_S), `rowDiffs(X, cols, rows)` = rowDiffs(X, 
+     rows = cols, cols = rows), `rowDiffs(X[cols, rows])` = rowDiffs(X[cols, rows]), unit = "ms")

Table: Benchmarking of colDiffs_X_S(), colDiffs(X, rows, cols)() and colDiffs(X[rows, cols])() on integer+10x1000 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	0.005130	0.005291	0.0055358	0.0053835	0.0054755	0.009254
2	colDiffs(X, rows, cols)	0.008535	0.008873	0.0092883	0.0089775	0.0090825	0.033128
3	colDiffs(X[rows, cols])	0.013391	0.013846	0.0142119	0.0139950	0.0141920	0.019400

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	colDiffs(X, rows, cols)	1.663743	1.676999	1.677862	1.667595	1.658753	3.579857
3	colDiffs(X[rows, cols])	2.610331	2.616897	2.567280	2.599610	2.591909	2.096391

Table: Benchmarking of rowDiffs_X_S(), rowDiffs(X, cols, rows)() and rowDiffs(X[cols, rows])() on integer+10x1000 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	rowDiffs_X_S	0.004781	0.004987	0.0054427	0.0050700	0.005226	0.029440
2	rowDiffs(X, cols, rows)	0.007709	0.007960	0.0082193	0.0080545	0.008212	0.012504
3	rowDiffs(X[cols, rows])	0.012309	0.012498	0.0129289	0.0126135	0.012793	0.024966

	expr	min	lq	mean	median	uq	max
1	rowDiffs_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.0000000
2	rowDiffs(X, cols, rows)	1.612424	1.596150	1.510162	1.588659	1.571374	0.4247283
3	rowDiffs(X[cols, rows])	2.574566	2.506116	2.375478	2.487870	2.447952	0.8480299

Figure: Benchmarking of colDiffs_X_S(), colDiffs(X, rows, cols)() and colDiffs(X[rows, cols])() on integer+10x1000 data as well as rowDiffs_X_S(), rowDiffs(X, cols, rows)() and rowDiffs(X[cols, rows])() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colDiffs_X_S() and rowDiffs_X_S() on integer+10x1000 data (original and transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	rowDiffs_X_S	4.781	4.987	5.44266	5.0700	5.2260	29.440
1	colDiffs_X_S	5.130	5.291	5.53577	5.3835	5.4755	9.254

	expr	min	lq	mean	median	uq	max
2	rowDiffs_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.0000000
1	colDiffs_X_S	1.072997	1.060958	1.017107	1.061834	1.047742	0.3143342

Figure: Benchmarking of colDiffs_X_S() and rowDiffs_X_S() on integer+10x1000 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

100x1000 integer matrix

> X <- data[["100x1000"]]
> rows <- sample.int(nrow(X), size = nrow(X) * 0.7)
> cols <- sample.int(ncol(X), size = ncol(X) * 0.7)
> X_S <- X[rows, cols]
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3112548 166.3    5709258 305.0  5709258 305.0
Vcells 5528901  42.2   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colDiffs_X_S = colDiffs(X_S), `colDiffs(X, rows, cols)` = colDiffs(X, 
+     rows = rows, cols = cols), `colDiffs(X[rows, cols])` = colDiffs(X[rows, cols]), unit = "ms")
> X <- t(X)
> X_S <- t(X_S)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3112542 166.3    5709258 305.0  5709258 305.0
Vcells 5578984  42.6   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowDiffs_X_S = rowDiffs(X_S), `rowDiffs(X, cols, rows)` = rowDiffs(X, 
+     rows = cols, cols = rows), `rowDiffs(X[cols, rows])` = rowDiffs(X[cols, rows]), unit = "ms")

Table: Benchmarking of colDiffs_X_S(), colDiffs(X, rows, cols)() and colDiffs(X[rows, cols])() on integer+100x1000 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	0.039308	0.0403705	0.0413522	0.0406230	0.0409395	0.077276
2	colDiffs(X, rows, cols)	0.073999	0.0760795	0.0775657	0.0762720	0.0766425	0.158858
3	colDiffs(X[rows, cols])	0.108884	0.1122845	0.1129651	0.1126845	0.1133515	0.124285

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	colDiffs(X, rows, cols)	1.882543	1.884532	1.875736	1.877557	1.872092	2.055722
3	colDiffs(X[rows, cols])	2.770021	2.781350	2.731783	2.773909	2.768756	1.608326

Table: Benchmarking of rowDiffs_X_S(), rowDiffs(X, cols, rows)() and rowDiffs(X[cols, rows])() on integer+100x1000 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	rowDiffs_X_S	0.039235	0.0397820	0.0419611	0.0404785	0.0410085	0.091386
2	rowDiffs(X, cols, rows)	0.067867	0.0685735	0.0712070	0.0699110	0.0704300	0.111323
3	rowDiffs(X[cols, rows])	0.100585	0.1017305	0.1077241	0.1029630	0.1048435	0.195772

	expr	min	lq	mean	median	uq	max
1	rowDiffs_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	rowDiffs(X, cols, rows)	1.729757	1.723732	1.696977	1.727114	1.717449	1.218163
3	rowDiffs(X[cols, rows])	2.563655	2.557199	2.567236	2.543647	2.556628	2.142254

Figure: Benchmarking of colDiffs_X_S(), colDiffs(X, rows, cols)() and colDiffs(X[rows, cols])() on integer+100x1000 data as well as rowDiffs_X_S(), rowDiffs(X, cols, rows)() and rowDiffs(X[cols, rows])() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colDiffs_X_S() and rowDiffs_X_S() on integer+100x1000 data (original and transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	rowDiffs_X_S	39.235	39.7820	41.96112	40.4785	41.0085	91.386
1	colDiffs_X_S	39.308	40.3705	41.35216	40.6230	40.9395	77.276

	expr	min	lq	mean	median	uq	max
2	rowDiffs_X_S	1.000000	1.000000	1.0000000	1.00000	1.0000000	1.0000
1	colDiffs_X_S	1.001861	1.014793	0.9854875	1.00357	0.9983174	0.8456

Figure: Benchmarking of colDiffs_X_S() and rowDiffs_X_S() on integer+100x1000 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

1000x100 integer matrix

> X <- data[["1000x100"]]
> rows <- sample.int(nrow(X), size = nrow(X) * 0.7)
> cols <- sample.int(ncol(X), size = ncol(X) * 0.7)
> X_S <- X[rows, cols]
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3112764 166.3    5709258 305.0  5709258 305.0
Vcells 5529622  42.2   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colDiffs_X_S = colDiffs(X_S), `colDiffs(X, rows, cols)` = colDiffs(X, 
+     rows = rows, cols = cols), `colDiffs(X[rows, cols])` = colDiffs(X[rows, cols]), unit = "ms")
> X <- t(X)
> X_S <- t(X_S)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3112758 166.3    5709258 305.0  5709258 305.0
Vcells 5579705  42.6   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowDiffs_X_S = rowDiffs(X_S), `rowDiffs(X, cols, rows)` = rowDiffs(X, 
+     rows = cols, cols = rows), `rowDiffs(X[cols, rows])` = rowDiffs(X[cols, rows]), unit = "ms")

Table: Benchmarking of colDiffs_X_S(), colDiffs(X, rows, cols)() and colDiffs(X[rows, cols])() on integer+1000x100 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	0.039726	0.0405095	0.0414681	0.041229	0.0416405	0.054515
2	colDiffs(X, rows, cols)	0.072423	0.0733265	0.0747335	0.074536	0.0749220	0.106975
3	colDiffs(X[rows, cols])	0.101311	0.1028045	0.1059735	0.104548	0.1055730	0.172245

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	colDiffs(X, rows, cols)	1.823063	1.810106	1.802191	1.807854	1.799258	1.962304
3	colDiffs(X[rows, cols])	2.550244	2.537787	2.555541	2.535788	2.535344	3.159589

Table: Benchmarking of rowDiffs_X_S(), rowDiffs(X, cols, rows)() and rowDiffs(X[cols, rows])() on integer+1000x100 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	rowDiffs_X_S	0.040435	0.0408685	0.0414823	0.0410570	0.0415155	0.053527
2	rowDiffs(X, cols, rows)	0.072863	0.0733020	0.0739522	0.0735420	0.0739435	0.081630
3	rowDiffs(X[cols, rows])	0.111725	0.1124385	0.1151624	0.1128715	0.1141905	0.212579

	expr	min	lq	mean	median	uq	max
1	rowDiffs_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	rowDiffs(X, cols, rows)	1.801978	1.793606	1.782741	1.791217	1.781106	1.525025
3	rowDiffs(X[cols, rows])	2.763076	2.751227	2.776185	2.749141	2.750551	3.971435

Figure: Benchmarking of colDiffs_X_S(), colDiffs(X, rows, cols)() and colDiffs(X[rows, cols])() on integer+1000x100 data as well as rowDiffs_X_S(), rowDiffs(X, cols, rows)() and rowDiffs(X[cols, rows])() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colDiffs_X_S() and rowDiffs_X_S() on integer+1000x100 data (original and transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	rowDiffs_X_S	40.435	40.8685	41.48227	41.057	41.5155	53.527
1	colDiffs_X_S	39.726	40.5095	41.46814	41.229	41.6405	54.515

	expr	min	lq	mean	median	uq	max
2	rowDiffs_X_S	1.0000000	1.0000000	1.0000000	1.000000	1.000000	1.000000
1	colDiffs_X_S	0.9824657	0.9912157	0.9996594	1.004189	1.003011	1.018458

Figure: Benchmarking of colDiffs_X_S() and rowDiffs_X_S() on integer+1000x100 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

Data type "double"

Data

> rmatrix <- function(nrow, ncol, mode = c("logical", "double", "integer", "index"), range = c(-100, 
+     +100), na_prob = 0) {
+     mode <- match.arg(mode)
+     n <- nrow * ncol
+     if (mode == "logical") {
+         x <- sample(c(FALSE, TRUE), size = n, replace = TRUE)
+     }     else if (mode == "index") {
+         x <- seq_len(n)
+         mode <- "integer"
+     }     else {
+         x <- runif(n, min = range[1], max = range[2])
+     }
+     storage.mode(x) <- mode
+     if (na_prob > 0) 
+         x[sample(n, size = na_prob * n)] <- NA
+     dim(x) <- c(nrow, ncol)
+     x
+ }
> rmatrices <- function(scale = 10, seed = 1, ...) {
+     set.seed(seed)
+     data <- list()
+     data[[1]] <- rmatrix(nrow = scale * 1, ncol = scale * 1, ...)
+     data[[2]] <- rmatrix(nrow = scale * 10, ncol = scale * 10, ...)
+     data[[3]] <- rmatrix(nrow = scale * 100, ncol = scale * 1, ...)
+     data[[4]] <- t(data[[3]])
+     data[[5]] <- rmatrix(nrow = scale * 10, ncol = scale * 100, ...)
+     data[[6]] <- t(data[[5]])
+     names(data) <- sapply(data, FUN = function(x) paste(dim(x), collapse = "x"))
+     data
+ }
> data <- rmatrices(mode = mode)

Results

10x10 double matrix

> X <- data[["10x10"]]
> rows <- sample.int(nrow(X), size = nrow(X) * 0.7)
> cols <- sample.int(ncol(X), size = ncol(X) * 0.7)
> X_S <- X[rows, cols]
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3112982 166.3    5709258 305.0  5709258 305.0
Vcells 5620704  42.9   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colDiffs_X_S = colDiffs(X_S), `colDiffs(X, rows, cols)` = colDiffs(X, 
+     rows = rows, cols = cols), `colDiffs(X[rows, cols])` = colDiffs(X[rows, cols]), unit = "ms")
> X <- t(X)
> X_S <- t(X_S)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3112967 166.3    5709258 305.0  5709258 305.0
Vcells 5620872  42.9   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowDiffs_X_S = rowDiffs(X_S), `rowDiffs(X, cols, rows)` = rowDiffs(X, 
+     rows = cols, cols = rows), `rowDiffs(X[cols, rows])` = rowDiffs(X[cols, rows]), unit = "ms")

Table: Benchmarking of colDiffs_X_S(), colDiffs(X, rows, cols)() and colDiffs(X[rows, cols])() on double+10x10 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	0.001221	0.0013660	0.0017599	0.0014260	0.0015315	0.030316
2	colDiffs(X, rows, cols)	0.001333	0.0015735	0.0017206	0.0016345	0.0017520	0.005095
3	colDiffs(X[rows, cols])	0.001923	0.0021990	0.0023842	0.0022755	0.0024230	0.009834

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	1.000000	1.000000	1.0000000	1.000000	1.000000	1.0000000
2	colDiffs(X, rows, cols)	1.091728	1.151903	0.9776525	1.146213	1.143976	0.1680631
3	colDiffs(X[rows, cols])	1.574939	1.609810	1.3547016	1.595722	1.582109	0.3243832

Table: Benchmarking of rowDiffs_X_S(), rowDiffs(X, cols, rows)() and rowDiffs(X[cols, rows])() on double+10x10 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	rowDiffs_X_S	0.001180	0.0012805	0.0013782	0.0013225	0.0014020	0.003003
2	rowDiffs(X, cols, rows)	0.001353	0.0015270	0.0018422	0.0015790	0.0017075	0.024083
3	rowDiffs(X[cols, rows])	0.001767	0.0019435	0.0020706	0.0019910	0.0021120	0.005571

	expr	min	lq	mean	median	uq	max
1	rowDiffs_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	rowDiffs(X, cols, rows)	1.146610	1.192503	1.336722	1.193951	1.217903	8.019647
3	rowDiffs(X[cols, rows])	1.497458	1.517767	1.502463	1.505482	1.506419	1.855145

Figure: Benchmarking of colDiffs_X_S(), colDiffs(X, rows, cols)() and colDiffs(X[rows, cols])() on double+10x10 data as well as rowDiffs_X_S(), rowDiffs(X, cols, rows)() and rowDiffs(X[cols, rows])() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colDiffs_X_S() and rowDiffs_X_S() on double+10x10 data (original and transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	rowDiffs_X_S	1.180	1.2805	1.37817	1.3225	1.4020	3.003
1	colDiffs_X_S	1.221	1.3660	1.75993	1.4260	1.5315	30.316

	expr	min	lq	mean	median	uq	max
2	rowDiffs_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.00000
1	colDiffs_X_S	1.034746	1.066771	1.277005	1.078261	1.092368	10.09524

Figure: Benchmarking of colDiffs_X_S() and rowDiffs_X_S() on double+10x10 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

100x100 double matrix

> X <- data[["100x100"]]
> rows <- sample.int(nrow(X), size = nrow(X) * 0.7)
> cols <- sample.int(ncol(X), size = ncol(X) * 0.7)
> X_S <- X[rows, cols]
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3113184 166.3    5709258 305.0  5709258 305.0
Vcells 5626639  43.0   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colDiffs_X_S = colDiffs(X_S), `colDiffs(X, rows, cols)` = colDiffs(X, 
+     rows = rows, cols = cols), `colDiffs(X[rows, cols])` = colDiffs(X[rows, cols]), unit = "ms")
> X <- t(X)
> X_S <- t(X_S)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3113178 166.3    5709258 305.0  5709258 305.0
Vcells 5636722  43.1   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowDiffs_X_S = rowDiffs(X_S), `rowDiffs(X, cols, rows)` = rowDiffs(X, 
+     rows = cols, cols = rows), `rowDiffs(X[cols, rows])` = rowDiffs(X[cols, rows]), unit = "ms")

Table: Benchmarking of colDiffs_X_S(), colDiffs(X, rows, cols)() and colDiffs(X[rows, cols])() on double+100x100 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	0.004421	0.0049180	0.0059535	0.006255	0.0066635	0.010450
2	colDiffs(X, rows, cols)	0.007536	0.0080215	0.0085251	0.008248	0.0085180	0.019900
3	colDiffs(X[rows, cols])	0.019850	0.0205255	0.0224120	0.021927	0.0224355	0.067467

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	colDiffs(X, rows, cols)	1.704592	1.631049	1.431950	1.318625	1.278307	1.904306
3	colDiffs(X[rows, cols])	4.489934	4.173546	3.764506	3.505516	3.366924	6.456172

Table: Benchmarking of rowDiffs_X_S(), rowDiffs(X, cols, rows)() and rowDiffs(X[cols, rows])() on double+100x100 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	rowDiffs_X_S	0.004366	0.0049175	0.0059636	0.0062895	0.006589	0.008911
2	rowDiffs(X, cols, rows)	0.007813	0.0083855	0.0088068	0.0086520	0.008881	0.013165
3	rowDiffs(X[cols, rows])	0.011526	0.0122080	0.0139787	0.0137565	0.014331	0.045332

	expr	min	lq	mean	median	uq	max
1	rowDiffs_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	rowDiffs(X, cols, rows)	1.789510	1.705236	1.476767	1.375626	1.347852	1.477388
3	rowDiffs(X[cols, rows])	2.639945	2.482562	2.344003	2.187217	2.174989	5.087196

Figure: Benchmarking of colDiffs_X_S(), colDiffs(X, rows, cols)() and colDiffs(X[rows, cols])() on double+100x100 data as well as rowDiffs_X_S(), rowDiffs(X, cols, rows)() and rowDiffs(X[cols, rows])() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colDiffs_X_S() and rowDiffs_X_S() on double+100x100 data (original and transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	4.421	4.9180	5.95349	6.2550	6.6635	10.450
2	rowDiffs_X_S	4.366	4.9175	5.96359	6.2895	6.5890	8.911

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	1.0000000	1.0000000	1.000000	1.000000	1.0000000	1.0000000
2	rowDiffs_X_S	0.9875594	0.9998983	1.001696	1.005516	0.9888197	0.8527273

Figure: Benchmarking of colDiffs_X_S() and rowDiffs_X_S() on double+100x100 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

1000x10 double matrix

> X <- data[["1000x10"]]
> rows <- sample.int(nrow(X), size = nrow(X) * 0.7)
> cols <- sample.int(ncol(X), size = ncol(X) * 0.7)
> X_S <- X[rows, cols]
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3113393 166.3    5709258 305.0  5709258 305.0
Vcells 5628027  43.0   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colDiffs_X_S = colDiffs(X_S), `colDiffs(X, rows, cols)` = colDiffs(X, 
+     rows = rows, cols = cols), `colDiffs(X[rows, cols])` = colDiffs(X[rows, cols]), unit = "ms")
> X <- t(X)
> X_S <- t(X_S)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3113387 166.3    5709258 305.0  5709258 305.0
Vcells 5638110  43.1   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowDiffs_X_S = rowDiffs(X_S), `rowDiffs(X, cols, rows)` = rowDiffs(X, 
+     rows = cols, cols = rows), `rowDiffs(X[cols, rows])` = rowDiffs(X[cols, rows]), unit = "ms")

Table: Benchmarking of colDiffs_X_S(), colDiffs(X, rows, cols)() and colDiffs(X[rows, cols])() on double+1000x10 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	0.004376	0.0051410	0.0078655	0.0066685	0.0074575	0.029527
2	colDiffs(X, rows, cols)	0.007692	0.0084225	0.0107188	0.0087285	0.0096460	0.034947
3	colDiffs(X[rows, cols])	0.020139	0.0207075	0.0258190	0.0222230	0.0237175	0.070039

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	colDiffs(X, rows, cols)	1.757770	1.638300	1.362754	1.308915	1.293463	1.183561
3	colDiffs(X[rows, cols])	4.602148	4.027913	3.282552	3.332534	3.180355	2.372032

Table: Benchmarking of rowDiffs_X_S(), rowDiffs(X, cols, rows)() and rowDiffs(X[cols, rows])() on double+1000x10 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	rowDiffs_X_S	0.004147	0.005472	0.0067631	0.0063880	0.0067495	0.038996
2	rowDiffs(X, cols, rows)	0.009847	0.010400	0.0109062	0.0106120	0.0107630	0.029539
3	rowDiffs(X[cols, rows])	0.013736	0.015166	0.0158284	0.0159175	0.0162925	0.021415

	expr	min	lq	mean	median	uq	max
1	rowDiffs_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.0000000
2	rowDiffs(X, cols, rows)	2.374488	1.900585	1.612601	1.661240	1.594637	0.7574879
3	rowDiffs(X[cols, rows])	3.312274	2.771564	2.340399	2.491782	2.413883	0.5491589

Figure: Benchmarking of colDiffs_X_S(), colDiffs(X, rows, cols)() and colDiffs(X[rows, cols])() on double+1000x10 data as well as rowDiffs_X_S(), rowDiffs(X, cols, rows)() and rowDiffs(X[cols, rows])() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colDiffs_X_S() and rowDiffs_X_S() on double+1000x10 data (original and transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	rowDiffs_X_S	4.147	5.472	6.76313	6.3880	6.7495	38.996
1	colDiffs_X_S	4.376	5.141	7.86552	6.6685	7.4575	29.527

	expr	min	lq	mean	median	uq	max
2	rowDiffs_X_S	1.000000	1.0000000	1.000	1.00000	1.000000	1.0000000
1	colDiffs_X_S	1.055221	0.9395102	1.163	1.04391	1.104897	0.7571802

Figure: Benchmarking of colDiffs_X_S() and rowDiffs_X_S() on double+1000x10 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

10x1000 double matrix

> X <- data[["10x1000"]]
> rows <- sample.int(nrow(X), size = nrow(X) * 0.7)
> cols <- sample.int(ncol(X), size = ncol(X) * 0.7)
> X_S <- X[rows, cols]
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3113599 166.3    5709258 305.0  5709258 305.0
Vcells 5628164  43.0   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colDiffs_X_S = colDiffs(X_S), `colDiffs(X, rows, cols)` = colDiffs(X, 
+     rows = rows, cols = cols), `colDiffs(X[rows, cols])` = colDiffs(X[rows, cols]), unit = "ms")
> X <- t(X)
> X_S <- t(X_S)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3113593 166.3    5709258 305.0  5709258 305.0
Vcells 5638247  43.1   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowDiffs_X_S = rowDiffs(X_S), `rowDiffs(X, cols, rows)` = rowDiffs(X, 
+     rows = cols, cols = rows), `rowDiffs(X[cols, rows])` = rowDiffs(X[cols, rows]), unit = "ms")

Table: Benchmarking of colDiffs_X_S(), colDiffs(X, rows, cols)() and colDiffs(X[rows, cols])() on double+10x1000 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	0.004063	0.0045650	0.0057744	0.0055980	0.0060260	0.036479
2	colDiffs(X, rows, cols)	0.008793	0.0092485	0.0100056	0.0094615	0.0096425	0.052644
3	colDiffs(X[rows, cols])	0.013318	0.0143585	0.0154045	0.0153940	0.0159020	0.022930

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.0000000
2	colDiffs(X, rows, cols)	2.164164	2.025958	1.732763	1.690157	1.600149	1.4431317
3	colDiffs(X[rows, cols])	3.277874	3.145345	2.667724	2.749911	2.638898	0.6285808

Table: Benchmarking of rowDiffs_X_S(), rowDiffs(X, cols, rows)() and rowDiffs(X[cols, rows])() on double+10x1000 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	rowDiffs_X_S	0.003720	0.0054570	0.0060726	0.005778	0.006033	0.033113
2	rowDiffs(X, cols, rows)	0.007415	0.0079015	0.0085686	0.008188	0.008419	0.040076
3	rowDiffs(X[cols, rows])	0.011631	0.0135290	0.0142688	0.013870	0.014210	0.034452

	expr	min	lq	mean	median	uq	max
1	rowDiffs_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	rowDiffs(X, cols, rows)	1.993280	1.447957	1.411026	1.417099	1.395492	1.210280
3	rowDiffs(X[cols, rows])	3.126613	2.479201	2.349689	2.400485	2.355379	1.040437

Figure: Benchmarking of colDiffs_X_S(), colDiffs(X, rows, cols)() and colDiffs(X[rows, cols])() on double+10x1000 data as well as rowDiffs_X_S(), rowDiffs(X, cols, rows)() and rowDiffs(X[cols, rows])() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colDiffs_X_S() and rowDiffs_X_S() on double+10x1000 data (original and transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	4.063	4.565	5.77439	5.598	6.026	36.479
2	rowDiffs_X_S	3.720	5.457	6.07263	5.778	6.033	33.113

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	1.0000000	1.0000	1.000000	1.000000	1.000000	1.0000000
2	rowDiffs_X_S	0.9155796	1.1954	1.051649	1.032154	1.001162	0.9077277

Figure: Benchmarking of colDiffs_X_S() and rowDiffs_X_S() on double+10x1000 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

100x1000 double matrix

> X <- data[["100x1000"]]
> rows <- sample.int(nrow(X), size = nrow(X) * 0.7)
> cols <- sample.int(ncol(X), size = ncol(X) * 0.7)
> X_S <- X[rows, cols]
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3113798 166.3    5709258 305.0  5709258 305.0
Vcells 5673577  43.3   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colDiffs_X_S = colDiffs(X_S), `colDiffs(X, rows, cols)` = colDiffs(X, 
+     rows = rows, cols = cols), `colDiffs(X[rows, cols])` = colDiffs(X[rows, cols]), unit = "ms")
> X <- t(X)
> X_S <- t(X_S)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3113792 166.3    5709258 305.0  5709258 305.0
Vcells 5773660  44.1   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowDiffs_X_S = rowDiffs(X_S), `rowDiffs(X, cols, rows)` = rowDiffs(X, 
+     rows = cols, cols = rows), `rowDiffs(X[cols, rows])` = rowDiffs(X[cols, rows]), unit = "ms")

Table: Benchmarking of colDiffs_X_S(), colDiffs(X, rows, cols)() and colDiffs(X[rows, cols])() on double+100x1000 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	0.033815	0.035542	0.0800326	0.0492765	0.1263535	0.169962
2	colDiffs(X, rows, cols)	0.066597	0.068433	0.1108244	0.0825050	0.1708720	0.205080
3	colDiffs(X[rows, cols])	0.187242	0.192313	0.3433570	0.2327500	0.3806580	7.369489

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	colDiffs(X, rows, cols)	1.969451	1.925412	1.384741	1.674327	1.352333	1.206623
3	colDiffs(X[rows, cols])	5.537247	5.410866	4.290212	4.723347	3.012643	43.359627

Table: Benchmarking of rowDiffs_X_S(), rowDiffs(X, cols, rows)() and rowDiffs(X[cols, rows])() on double+100x1000 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	rowDiffs_X_S	0.033111	0.0363000	0.0736204	0.0416885	0.1268655	0.142161
2	rowDiffs(X, cols, rows)	0.067007	0.0684435	0.1836151	0.0820270	0.1715915	7.180782
3	rowDiffs(X[cols, rows])	0.107105	0.1115635	0.1933004	0.1298830	0.3001800	0.325740

	expr	min	lq	mean	median	uq	max
1	rowDiffs_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	rowDiffs(X, cols, rows)	2.023708	1.885496	2.494080	1.967617	1.352547	50.511617
3	rowDiffs(X[cols, rows])	3.234726	3.073375	2.625638	3.115559	2.366128	2.291346

Figure: Benchmarking of colDiffs_X_S(), colDiffs(X, rows, cols)() and colDiffs(X[rows, cols])() on double+100x1000 data as well as rowDiffs_X_S(), rowDiffs(X, cols, rows)() and rowDiffs(X[cols, rows])() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colDiffs_X_S() and rowDiffs_X_S() on double+100x1000 data (original and transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	rowDiffs_X_S	33.111	36.300	73.62036	41.6885	126.8655	142.161
1	colDiffs_X_S	33.815	35.542	80.03263	49.2765	126.3535	169.962

	expr	min	lq	mean	median	uq	max
2	rowDiffs_X_S	1.000000	1.0000000	1.000000	1.000000	1.0000000	1.00000
1	colDiffs_X_S	1.021262	0.9791185	1.087099	1.182017	0.9959642	1.19556

Figure: Benchmarking of colDiffs_X_S() and rowDiffs_X_S() on double+100x1000 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

1000x100 double matrix

> X <- data[["1000x100"]]
> rows <- sample.int(nrow(X), size = nrow(X) * 0.7)
> cols <- sample.int(ncol(X), size = ncol(X) * 0.7)
> X_S <- X[rows, cols]
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3114014 166.4    5709258 305.0  5709258 305.0
Vcells 5673724  43.3   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colDiffs_X_S = colDiffs(X_S), `colDiffs(X, rows, cols)` = colDiffs(X, 
+     rows = rows, cols = cols), `colDiffs(X[rows, cols])` = colDiffs(X[rows, cols]), unit = "ms")
> X <- t(X)
> X_S <- t(X_S)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3114008 166.4    5709258 305.0  5709258 305.0
Vcells 5773807  44.1   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowDiffs_X_S = rowDiffs(X_S), `rowDiffs(X, cols, rows)` = rowDiffs(X, 
+     rows = cols, cols = rows), `rowDiffs(X[cols, rows])` = rowDiffs(X[cols, rows]), unit = "ms")

Table: Benchmarking of colDiffs_X_S(), colDiffs(X, rows, cols)() and colDiffs(X[rows, cols])() on double+1000x100 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	0.033956	0.0358095	0.0724575	0.0428235	0.125650	0.157237
2	colDiffs(X, rows, cols)	0.062317	0.0636225	0.1053283	0.0694575	0.164298	0.170479
3	colDiffs(X[rows, cols])	0.108404	0.1147840	0.2376048	0.1349520	0.299808	4.780825

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	colDiffs(X, rows, cols)	1.835228	1.776693	1.453656	1.621948	1.307585	1.084217
3	colDiffs(X[rows, cols])	3.192484	3.205406	3.279227	3.151354	2.386057	30.405216

Table: Benchmarking of rowDiffs_X_S(), rowDiffs(X, cols, rows)() and rowDiffs(X[cols, rows])() on double+1000x100 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	rowDiffs_X_S	0.033388	0.0363885	0.1431267	0.0483815	0.1278610	6.522680
2	rowDiffs(X, cols, rows)	0.073139	0.0742570	0.1271230	0.0907515	0.1766875	0.481092
3	rowDiffs(X[cols, rows])	0.115327	0.1184000	0.1909909	0.1368305	0.3002710	0.402991

	expr	min	lq	mean	median	uq	max
1	rowDiffs_X_S	1.000000	1.000000	1.0000000	1.000000	1.000000	1.0000000
2	rowDiffs(X, cols, rows)	2.190577	2.040672	0.8881854	1.875748	1.381872	0.0737568
3	rowDiffs(X[cols, rows])	3.454145	3.253775	1.3344183	2.828158	2.348417	0.0617830

Figure: Benchmarking of colDiffs_X_S(), colDiffs(X, rows, cols)() and colDiffs(X[rows, cols])() on double+1000x100 data as well as rowDiffs_X_S(), rowDiffs(X, cols, rows)() and rowDiffs(X[cols, rows])() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colDiffs_X_S() and rowDiffs_X_S() on double+1000x100 data (original and transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	33.956	35.8095	72.45754	42.8235	125.650	157.237
2	rowDiffs_X_S	33.388	36.3885	143.12670	48.3815	127.861	6522.680

	expr	min	lq	mean	median	uq	max
1	colDiffs_X_S	1.0000000	1.000000	1.000000	1.000000	1.000000	1.00000
2	rowDiffs_X_S	0.9832725	1.016169	1.975318	1.129789	1.017597	41.48311

Figure: Benchmarking of colDiffs_X_S() and rowDiffs_X_S() on double+1000x100 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

Appendix

Session information

R version 3.6.1 Patched (2019-08-27 r77078)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Ubuntu 18.04.3 LTS

Matrix products: default
BLAS:   /home/hb/software/R-devel/R-3-6-branch/lib/R/lib/libRblas.so
LAPACK: /home/hb/software/R-devel/R-3-6-branch/lib/R/lib/libRlapack.so

locale:
 [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C              
 [3] LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8    
 [5] LC_MONETARY=en_US.UTF-8    LC_MESSAGES=en_US.UTF-8   
 [7] LC_PAPER=en_US.UTF-8       LC_NAME=C                 
 [9] LC_ADDRESS=C               LC_TELEPHONE=C            
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C       

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] microbenchmark_1.4-6    matrixStats_0.55.0-9000 ggplot2_3.2.1          
[4] knitr_1.24              R.devices_2.16.0        R.utils_2.9.0          
[7] R.oo_1.22.0             R.methodsS3_1.7.1       history_0.0.0-9002     

loaded via a namespace (and not attached):
 [1] Biobase_2.45.0       bit64_0.9-7          splines_3.6.1       
 [4] network_1.15         assertthat_0.2.1     highr_0.8           
 [7] stats4_3.6.1         blob_1.2.0           robustbase_0.93-5   
[10] pillar_1.4.2         RSQLite_2.1.2        backports_1.1.4     
[13] lattice_0.20-38      glue_1.3.1           digest_0.6.20       
[16] colorspace_1.4-1     sandwich_2.5-1       Matrix_1.2-17       
[19] XML_3.98-1.20        lpSolve_5.6.13.3     pkgconfig_2.0.2     
[22] genefilter_1.66.0    purrr_0.3.2          ergm_3.10.4         
[25] xtable_1.8-4         mvtnorm_1.0-11       scales_1.0.0        
[28] tibble_2.1.3         annotate_1.62.0      IRanges_2.18.2      
[31] TH.data_1.0-10       withr_2.1.2          BiocGenerics_0.30.0 
[34] lazyeval_0.2.2       mime_0.7             survival_2.44-1.1   
[37] magrittr_1.5         crayon_1.3.4         statnet.common_4.3.0
[40] memoise_1.1.0        laeken_0.5.0         R.cache_0.13.0      
[43] MASS_7.3-51.4        R.rsp_0.43.1         tools_3.6.1         
[46] multcomp_1.4-10      S4Vectors_0.22.1     trust_0.1-7         
[49] munsell_0.5.0        AnnotationDbi_1.46.1 compiler_3.6.1      
[52] rlang_0.4.0          grid_3.6.1           RCurl_1.95-4.12     
[55] cwhmisc_6.6          rappdirs_0.3.1       labeling_0.3        
[58] bitops_1.0-6         base64enc_0.1-3      boot_1.3-23         
[61] gtable_0.3.0         codetools_0.2-16     DBI_1.0.0           
[64] markdown_1.1         R6_2.4.0             zoo_1.8-6           
[67] dplyr_0.8.3          bit_1.1-14           zeallot_0.1.0       
[70] parallel_3.6.1       Rcpp_1.0.2           vctrs_0.2.0         
[73] DEoptimR_1.0-8       tidyselect_0.2.5     xfun_0.9            
[76] coda_0.19-3         

Total processing time was 22.4 secs.

Reproducibility

To reproduce this report, do:

html <- matrixStats:::benchmark('colRowDiffs_subset')

---

Copyright Dongcan Jiang. Last updated on 2019-09-10 20:40:15 (-0700 UTC). Powered by RSP.

<script> var link = document.createElement('link'); link.rel = 'icon'; link.href = "data:image/png;base64,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" document.getElementsByTagName('head')[0].appendChild(link); </script>