colRowMeans2_subset

matrixStats: Benchmark report

---

colMeans2() and rowMeans2() benchmarks on subsetted computation

This report benchmark the performance of colMeans2() and rowMeans2() 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 3136175 167.5    5709258 305.0  5709258 305.0
Vcells 6002295  45.8   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colMeans2_X_S = colMeans2(X_S, na.rm = FALSE), `colMeans2(X, rows, cols)` = colMeans2(X, 
+     rows = rows, cols = cols, na.rm = FALSE), `colMeans2(X[rows, cols])` = colMeans2(X[rows, cols], 
+     na.rm = FALSE), unit = "ms")
> X <- t(X)
> X_S <- t(X_S)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3135354 167.5    5709258 305.0  5709258 305.0
Vcells 6000266  45.8   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowMeans2_X_S = rowMeans2(X_S, na.rm = FALSE), `rowMeans2(X, cols, rows)` = rowMeans2(X, 
+     rows = cols, cols = rows, na.rm = FALSE), `rowMeans2(X[cols, rows])` = rowMeans2(X[cols, rows], 
+     na.rm = FALSE), unit = "ms")

Table: Benchmarking of colMeans2_X_S(), colMeans2(X, rows, cols)() and colMeans2(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	colMeans2_X_S	0.001012	0.0010510	0.0020686	0.001078	0.0011485	0.095374
2	colMeans2(X, rows, cols)	0.001151	0.0012030	0.0012986	0.001238	0.0013625	0.002528
3	colMeans2(X[rows, cols])	0.001505	0.0016795	0.0019137	0.001777	0.0018630	0.009139

	expr	min	lq	mean	median	uq	max
1	colMeans2_X_S	1.000000	1.000000	1.0000000	1.000000	1.000000	1.0000000
2	colMeans2(X, rows, cols)	1.137352	1.144624	0.6277621	1.148423	1.186330	0.0265062
3	colMeans2(X[rows, cols])	1.487154	1.598002	0.9251009	1.648423	1.622116	0.0958228

Table: Benchmarking of rowMeans2_X_S(), rowMeans2(X, cols, rows)() and rowMeans2(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	rowMeans2_X_S	0.001020	0.0010495	0.0011459	0.0010730	0.0011835	0.003048
2	rowMeans2(X, cols, rows)	0.001145	0.0012030	0.0024958	0.0012345	0.0013545	0.119225
3	rowMeans2(X[cols, rows])	0.001490	0.0016965	0.0019195	0.0017670	0.0019030	0.006960

	expr	min	lq	mean	median	uq	max
1	rowMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	rowMeans2(X, cols, rows)	1.122549	1.146260	2.178058	1.150513	1.144487	39.115814
3	rowMeans2(X[cols, rows])	1.460784	1.616484	1.675161	1.646785	1.607943	2.283465

Figure: Benchmarking of colMeans2_X_S(), colMeans2(X, rows, cols)() and colMeans2(X[rows, cols])() on integer+10x10 data as well as rowMeans2_X_S(), rowMeans2(X, cols, rows)() and rowMeans2(X[cols, rows])() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colMeans2_X_S() and rowMeans2_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
2	rowMeans2_X_S	1.020	1.0495	1.14586	1.073	1.1835	3.048
1	colMeans2_X_S	1.012	1.0510	2.06865	1.078	1.1485	95.374

	expr	min	lq	mean	median	uq	max
2	rowMeans2_X_S	1.0000000	1.000000	1.000000	1.00000	1.0000000	1.00000
1	colMeans2_X_S	0.9921569	1.001429	1.805325	1.00466	0.9704267	31.29068

Figure: Benchmarking of colMeans2_X_S() and rowMeans2_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 3133736 167.4    5709258 305.0  5709258 305.0
Vcells 5667703  43.3   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colMeans2_X_S = colMeans2(X_S, na.rm = FALSE), `colMeans2(X, rows, cols)` = colMeans2(X, 
+     rows = rows, cols = cols, na.rm = FALSE), `colMeans2(X[rows, cols])` = colMeans2(X[rows, cols], 
+     na.rm = FALSE), unit = "ms")
> X <- t(X)
> X_S <- t(X_S)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3133730 167.4    5709258 305.0  5709258 305.0
Vcells 5672786  43.3   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowMeans2_X_S = rowMeans2(X_S, na.rm = FALSE), `rowMeans2(X, cols, rows)` = rowMeans2(X, 
+     rows = cols, cols = rows, na.rm = FALSE), `rowMeans2(X[cols, rows])` = rowMeans2(X[cols, rows], 
+     na.rm = FALSE), unit = "ms")

Table: Benchmarking of colMeans2_X_S(), colMeans2(X, rows, cols)() and colMeans2(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	colMeans2_X_S	0.005694	0.0061670	0.0063227	0.0062875	0.0064465	0.009233
2	colMeans2(X, rows, cols)	0.006572	0.0071280	0.0073177	0.0072620	0.0074115	0.009410
3	colMeans2(X[rows, cols])	0.013319	0.0138425	0.0145661	0.0139635	0.0141635	0.058807

	expr	min	lq	mean	median	uq	max
1	colMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	colMeans2(X, rows, cols)	1.154197	1.155829	1.157359	1.154990	1.149694	1.019170
3	colMeans2(X[rows, cols])	2.339129	2.244608	2.303762	2.220835	2.197084	6.369219

Table: Benchmarking of rowMeans2_X_S(), rowMeans2(X, cols, rows)() and rowMeans2(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	rowMeans2_X_S	0.005697	0.0061965	0.0064094	0.0063030	0.0064685	0.011850
2	rowMeans2(X, cols, rows)	0.006557	0.0071110	0.0072179	0.0072015	0.0073030	0.008348
3	rowMeans2(X[cols, rows])	0.013179	0.0137340	0.0145504	0.0139095	0.0141270	0.057671

	expr	min	lq	mean	median	uq	max
1	rowMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.0000000
2	rowMeans2(X, cols, rows)	1.150957	1.147583	1.126135	1.142551	1.129010	0.7044726
3	rowMeans2(X[cols, rows])	2.313323	2.216413	2.270157	2.206806	2.183969	4.8667511

Figure: Benchmarking of colMeans2_X_S(), colMeans2(X, rows, cols)() and colMeans2(X[rows, cols])() on integer+100x100 data as well as rowMeans2_X_S(), rowMeans2(X, cols, rows)() and rowMeans2(X[cols, rows])() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colMeans2_X_S() and rowMeans2_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
1	colMeans2_X_S	5.694	6.1670	6.32274	6.2875	6.4465	9.233
2	rowMeans2_X_S	5.697	6.1965	6.40942	6.3030	6.4685	11.850

	expr	min	lq	mean	median	uq	max
1	colMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.00000
2	rowMeans2_X_S	1.000527	1.004784	1.013709	1.002465	1.003413	1.28344

Figure: Benchmarking of colMeans2_X_S() and rowMeans2_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 3134491 167.4    5709258 305.0  5709258 305.0
Vcells 5671775  43.3   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colMeans2_X_S = colMeans2(X_S, na.rm = FALSE), `colMeans2(X, rows, cols)` = colMeans2(X, 
+     rows = rows, cols = cols, na.rm = FALSE), `colMeans2(X[rows, cols])` = colMeans2(X[rows, cols], 
+     na.rm = FALSE), unit = "ms")
> X <- t(X)
> X_S <- t(X_S)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3134482 167.4    5709258 305.0  5709258 305.0
Vcells 5676853  43.4   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowMeans2_X_S = rowMeans2(X_S, na.rm = FALSE), `rowMeans2(X, cols, rows)` = rowMeans2(X, 
+     rows = cols, cols = rows, na.rm = FALSE), `rowMeans2(X[cols, rows])` = rowMeans2(X[cols, rows], 
+     na.rm = FALSE), unit = "ms")

Table: Benchmarking of colMeans2_X_S(), colMeans2(X, rows, cols)() and colMeans2(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	colMeans2_X_S	0.006722	0.0069635	0.0071653	0.0070890	0.0073120	0.009325
2	colMeans2(X, rows, cols)	0.007619	0.0078585	0.0082350	0.0080095	0.0081170	0.029352
3	colMeans2(X[rows, cols])	0.014012	0.0145155	0.0149172	0.0147140	0.0149725	0.022443

	expr	min	lq	mean	median	uq	max
1	colMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	colMeans2(X, rows, cols)	1.133442	1.128527	1.149287	1.129849	1.110093	3.147668
3	colMeans2(X[rows, cols])	2.084499	2.084512	2.081857	2.075610	2.047661	2.406756

Table: Benchmarking of rowMeans2_X_S(), rowMeans2(X, cols, rows)() and rowMeans2(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	rowMeans2_X_S	0.006780	0.0069645	0.0072516	0.0070650	0.007216	0.022338
2	rowMeans2(X, cols, rows)	0.007701	0.0078305	0.0083144	0.0079920	0.008159	0.028389
3	rowMeans2(X[cols, rows])	0.015216	0.0155620	0.0159080	0.0157035	0.016001	0.022441

	expr	min	lq	mean	median	uq	max
1	rowMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	rowMeans2(X, cols, rows)	1.135841	1.124345	1.146560	1.131210	1.130682	1.270884
3	rowMeans2(X[cols, rows])	2.244248	2.234475	2.193735	2.222718	2.217433	1.004611

Figure: Benchmarking of colMeans2_X_S(), colMeans2(X, rows, cols)() and colMeans2(X[rows, cols])() on integer+1000x10 data as well as rowMeans2_X_S(), rowMeans2(X, cols, rows)() and rowMeans2(X[cols, rows])() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colMeans2_X_S() and rowMeans2_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
2	rowMeans2_X_S	6.780	6.9645	7.25157	7.065	7.216	22.338
1	colMeans2_X_S	6.722	6.9635	7.16533	7.089	7.312	9.325

	expr	min	lq	mean	median	uq	max
2	rowMeans2_X_S	1.0000000	1.0000000	1.0000000	1.000000	1.000000	1.0000000
1	colMeans2_X_S	0.9914454	0.9998564	0.9881074	1.003397	1.013304	0.4174501

Figure: Benchmarking of colMeans2_X_S() and rowMeans2_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 3134694 167.5    5709258 305.0  5709258 305.0
Vcells 5672624  43.3   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colMeans2_X_S = colMeans2(X_S, na.rm = FALSE), `colMeans2(X, rows, cols)` = colMeans2(X, 
+     rows = rows, cols = cols, na.rm = FALSE), `colMeans2(X[rows, cols])` = colMeans2(X[rows, cols], 
+     na.rm = FALSE), unit = "ms")
> X <- t(X)
> X_S <- t(X_S)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3134688 167.5    5709258 305.0  5709258 305.0
Vcells 5677707  43.4   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowMeans2_X_S = rowMeans2(X_S, na.rm = FALSE), `rowMeans2(X, cols, rows)` = rowMeans2(X, 
+     rows = cols, cols = rows, na.rm = FALSE), `rowMeans2(X[cols, rows])` = rowMeans2(X[cols, rows], 
+     na.rm = FALSE), unit = "ms")

Table: Benchmarking of colMeans2_X_S(), colMeans2(X, rows, cols)() and colMeans2(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	colMeans2_X_S	0.005980	0.0061935	0.0064048	0.0063075	0.006527	0.008118
2	colMeans2(X, rows, cols)	0.008556	0.0088300	0.0092065	0.0089355	0.009063	0.032344
3	colMeans2(X[rows, cols])	0.014404	0.0147800	0.0151480	0.0149365	0.015171	0.022208

	expr	min	lq	mean	median	uq	max
1	colMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	colMeans2(X, rows, cols)	1.430769	1.425688	1.437450	1.416647	1.388540	3.984233
3	colMeans2(X[rows, cols])	2.408696	2.386373	2.365113	2.368054	2.324345	2.735649

Table: Benchmarking of rowMeans2_X_S(), rowMeans2(X, cols, rows)() and rowMeans2(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	rowMeans2_X_S	0.006087	0.0063315	0.0068189	0.0064500	0.0066295	0.036717
2	rowMeans2(X, cols, rows)	0.008336	0.0084690	0.0086735	0.0085570	0.0087445	0.010745
3	rowMeans2(X[cols, rows])	0.013369	0.0138670	0.0143932	0.0140035	0.0143590	0.032030

	expr	min	lq	mean	median	uq	max
1	rowMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.0000000
2	rowMeans2(X, cols, rows)	1.369476	1.337598	1.271985	1.326667	1.319029	0.2926437
3	rowMeans2(X[cols, rows])	2.196320	2.190160	2.110785	2.171085	2.165925	0.8723480

Figure: Benchmarking of colMeans2_X_S(), colMeans2(X, rows, cols)() and colMeans2(X[rows, cols])() on integer+10x1000 data as well as rowMeans2_X_S(), rowMeans2(X, cols, rows)() and rowMeans2(X[cols, rows])() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colMeans2_X_S() and rowMeans2_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
1	colMeans2_X_S	5.980	6.1935	6.40476	6.3075	6.5270	8.118
2	rowMeans2_X_S	6.087	6.3315	6.81890	6.4500	6.6295	36.717

	expr	min	lq	mean	median	uq	max
1	colMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	rowMeans2_X_S	1.017893	1.022281	1.064661	1.022592	1.015704	4.522912

Figure: Benchmarking of colMeans2_X_S() and rowMeans2_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 3134893 167.5    5709258 305.0  5709258 305.0
Vcells 5695287  43.5   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colMeans2_X_S = colMeans2(X_S, na.rm = FALSE), `colMeans2(X, rows, cols)` = colMeans2(X, 
+     rows = rows, cols = cols, na.rm = FALSE), `colMeans2(X[rows, cols])` = colMeans2(X[rows, cols], 
+     na.rm = FALSE), unit = "ms")
> X <- t(X)
> X_S <- t(X_S)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3134887 167.5    5709258 305.0  5709258 305.0
Vcells 5745370  43.9   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowMeans2_X_S = rowMeans2(X_S, na.rm = FALSE), `rowMeans2(X, cols, rows)` = rowMeans2(X, 
+     rows = cols, cols = rows, na.rm = FALSE), `rowMeans2(X[cols, rows])` = rowMeans2(X[cols, rows], 
+     na.rm = FALSE), unit = "ms")

Table: Benchmarking of colMeans2_X_S(), colMeans2(X, rows, cols)() and colMeans2(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	colMeans2_X_S	0.047693	0.0480815	0.0483832	0.048213	0.0484440	0.053227
2	colMeans2(X, rows, cols)	0.058319	0.0587005	0.0599412	0.058904	0.0591065	0.148362
3	colMeans2(X[rows, cols])	0.118974	0.1194300	0.1202710	0.119697	0.1203885	0.139045

	expr	min	lq	mean	median	uq	max
1	colMeans2_X_S	1.00000	1.000000	1.000000	1.000000	1.000000	1.000000
2	colMeans2(X, rows, cols)	1.22280	1.220854	1.238886	1.221745	1.220099	2.787345
3	colMeans2(X[rows, cols])	2.49458	2.483907	2.485803	2.482671	2.485107	2.612302

Table: Benchmarking of rowMeans2_X_S(), rowMeans2(X, cols, rows)() and rowMeans2(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	rowMeans2_X_S	0.048577	0.0490745	0.0502173	0.0492735	0.049849	0.101283
2	rowMeans2(X, cols, rows)	0.074924	0.0754840	0.0761068	0.0757975	0.076091	0.085869
3	rowMeans2(X[cols, rows])	0.111337	0.1122705	0.1144994	0.1129130	0.114409	0.173477

	expr	min	lq	mean	median	uq	max
1	rowMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.0000000
2	rowMeans2(X, cols, rows)	1.542376	1.538151	1.515551	1.538301	1.526430	0.8478126
3	rowMeans2(X[cols, rows])	2.291969	2.287756	2.280080	2.291556	2.295111	1.7127948

Figure: Benchmarking of colMeans2_X_S(), colMeans2(X, rows, cols)() and colMeans2(X[rows, cols])() on integer+100x1000 data as well as rowMeans2_X_S(), rowMeans2(X, cols, rows)() and rowMeans2(X[cols, rows])() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colMeans2_X_S() and rowMeans2_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
1	colMeans2_X_S	47.693	48.0815	48.38318	48.2130	48.444	53.227
2	rowMeans2_X_S	48.577	49.0745	50.21728	49.2735	49.849	101.283

	expr	min	lq	mean	median	uq	max
1	colMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.00000
2	rowMeans2_X_S	1.018535	1.020652	1.037908	1.021996	1.029003	1.90285

Figure: Benchmarking of colMeans2_X_S() and rowMeans2_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 3135112 167.5    5709258 305.0  5709258 305.0
Vcells 5696080  43.5   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colMeans2_X_S = colMeans2(X_S, na.rm = FALSE), `colMeans2(X, rows, cols)` = colMeans2(X, 
+     rows = rows, cols = cols, na.rm = FALSE), `colMeans2(X[rows, cols])` = colMeans2(X[rows, cols], 
+     na.rm = FALSE), unit = "ms")
> X <- t(X)
> X_S <- t(X_S)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3135103 167.5    5709258 305.0  5709258 305.0
Vcells 5746158  43.9   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowMeans2_X_S = rowMeans2(X_S, na.rm = FALSE), `rowMeans2(X, cols, rows)` = rowMeans2(X, 
+     rows = cols, cols = rows, na.rm = FALSE), `rowMeans2(X[cols, rows])` = rowMeans2(X[cols, rows], 
+     na.rm = FALSE), unit = "ms")

Table: Benchmarking of colMeans2_X_S(), colMeans2(X, rows, cols)() and colMeans2(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	colMeans2_X_S	0.052139	0.0526350	0.0530807	0.0528350	0.0531475	0.058491
2	colMeans2(X, rows, cols)	0.056382	0.0568355	0.0571660	0.0571275	0.0573860	0.058743
3	colMeans2(X[rows, cols])	0.115089	0.1157060	0.1178964	0.1160285	0.1168615	0.198967

	expr	min	lq	mean	median	uq	max
1	colMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	colMeans2(X, rows, cols)	1.081379	1.079804	1.076964	1.081244	1.079750	1.004308
3	colMeans2(X[rows, cols])	2.207350	2.198271	2.221078	2.196054	2.198815	3.401669

Table: Benchmarking of rowMeans2_X_S(), rowMeans2(X, cols, rows)() and rowMeans2(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	rowMeans2_X_S	0.053237	0.0538700	0.0545218	0.0541565	0.0546360	0.067469
2	rowMeans2(X, cols, rows)	0.074706	0.0754095	0.0760186	0.0756445	0.0759925	0.102267
3	rowMeans2(X[cols, rows])	0.124307	0.1255445	0.1288937	0.1267330	0.1286625	0.222305

	expr	min	lq	mean	median	uq	max
1	rowMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	rowMeans2(X, cols, rows)	1.403272	1.399842	1.394279	1.396776	1.390887	1.515763
3	rowMeans2(X[cols, rows])	2.334974	2.330509	2.364076	2.340125	2.354903	3.294921

Figure: Benchmarking of colMeans2_X_S(), colMeans2(X, rows, cols)() and colMeans2(X[rows, cols])() on integer+1000x100 data as well as rowMeans2_X_S(), rowMeans2(X, cols, rows)() and rowMeans2(X[cols, rows])() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colMeans2_X_S() and rowMeans2_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
1	colMeans2_X_S	52.139	52.635	53.08073	52.8350	53.1475	58.491
2	rowMeans2_X_S	53.237	53.870	54.52182	54.1565	54.6360	67.469

	expr	min	lq	mean	median	uq	max
1	colMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	rowMeans2_X_S	1.021059	1.023464	1.027149	1.025012	1.028007	1.153494

Figure: Benchmarking of colMeans2_X_S() and rowMeans2_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 3135330 167.5    5709258 305.0  5709258 305.0
Vcells 5787185  44.2   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colMeans2_X_S = colMeans2(X_S, na.rm = FALSE), `colMeans2(X, rows, cols)` = colMeans2(X, 
+     rows = rows, cols = cols, na.rm = FALSE), `colMeans2(X[rows, cols])` = colMeans2(X[rows, cols], 
+     na.rm = FALSE), unit = "ms")
> X <- t(X)
> X_S <- t(X_S)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3135312 167.5    5709258 305.0  5709258 305.0
Vcells 5787348  44.2   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowMeans2_X_S = rowMeans2(X_S, na.rm = FALSE), `rowMeans2(X, cols, rows)` = rowMeans2(X, 
+     rows = cols, cols = rows, na.rm = FALSE), `rowMeans2(X[cols, rows])` = rowMeans2(X[cols, rows], 
+     na.rm = FALSE), unit = "ms")

Table: Benchmarking of colMeans2_X_S(), colMeans2(X, rows, cols)() and colMeans2(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	colMeans2_X_S	0.001048	0.0010715	0.0014291	0.0010960	0.0011680	0.028541
2	colMeans2(X, rows, cols)	0.001187	0.0012495	0.0013424	0.0012710	0.0013140	0.004488
3	colMeans2(X[rows, cols])	0.001631	0.0018775	0.0020696	0.0019435	0.0020545	0.009945

	expr	min	lq	mean	median	uq	max
1	colMeans2_X_S	1.000000	1.000000	1.0000000	1.000000	1.00000	1.0000000
2	colMeans2(X, rows, cols)	1.132634	1.166122	0.9393604	1.159671	1.12500	0.1572475
3	colMeans2(X[rows, cols])	1.556298	1.752217	1.4481562	1.773266	1.75899	0.3484461

Table: Benchmarking of rowMeans2_X_S(), rowMeans2(X, cols, rows)() and rowMeans2(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	rowMeans2_X_S	0.001037	0.0010640	0.0011604	0.0010835	0.0011735	0.003148
2	rowMeans2(X, cols, rows)	0.001208	0.0012345	0.0016028	0.0012510	0.0013245	0.031066
3	rowMeans2(X[cols, rows])	0.001510	0.0017415	0.0019289	0.0018165	0.0019705	0.008154

	expr	min	lq	mean	median	uq	max
1	rowMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	rowMeans2(X, cols, rows)	1.164899	1.160244	1.381306	1.154592	1.128675	9.868488
3	rowMeans2(X[cols, rows])	1.456123	1.636748	1.662283	1.676511	1.679165	2.590216

Figure: Benchmarking of colMeans2_X_S(), colMeans2(X, rows, cols)() and colMeans2(X[rows, cols])() on double+10x10 data as well as rowMeans2_X_S(), rowMeans2(X, cols, rows)() and rowMeans2(X[cols, rows])() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colMeans2_X_S() and rowMeans2_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	rowMeans2_X_S	1.037	1.0640	1.16038	1.0835	1.1735	3.148
1	colMeans2_X_S	1.048	1.0715	1.42910	1.0960	1.1680	28.541

	expr	min	lq	mean	median	uq	max
2	rowMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.0000000	1.000000
1	colMeans2_X_S	1.010608	1.007049	1.231579	1.011537	0.9953132	9.066391

Figure: Benchmarking of colMeans2_X_S() and rowMeans2_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 3135529 167.5    5709258 305.0  5709258 305.0
Vcells 5793138  44.2   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colMeans2_X_S = colMeans2(X_S, na.rm = FALSE), `colMeans2(X, rows, cols)` = colMeans2(X, 
+     rows = rows, cols = cols, na.rm = FALSE), `colMeans2(X[rows, cols])` = colMeans2(X[rows, cols], 
+     na.rm = FALSE), unit = "ms")
> X <- t(X)
> X_S <- t(X_S)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3135523 167.5    5709258 305.0  5709258 305.0
Vcells 5803221  44.3   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowMeans2_X_S = rowMeans2(X_S, na.rm = FALSE), `rowMeans2(X, cols, rows)` = rowMeans2(X, 
+     rows = cols, cols = rows, na.rm = FALSE), `rowMeans2(X[cols, rows])` = rowMeans2(X[cols, rows], 
+     na.rm = FALSE), unit = "ms")

Table: Benchmarking of colMeans2_X_S(), colMeans2(X, rows, cols)() and colMeans2(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	colMeans2_X_S	0.005872	0.0064425	0.0065669	0.0065555	0.0066285	0.010052
2	colMeans2(X, rows, cols)	0.007360	0.0079870	0.0083096	0.0081505	0.0083455	0.019401
3	colMeans2(X[rows, cols])	0.021687	0.0221865	0.0231218	0.0223320	0.0226375	0.067764

	expr	min	lq	mean	median	uq	max
1	colMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	colMeans2(X, rows, cols)	1.253406	1.239736	1.265392	1.243307	1.259033	1.930064
3	colMeans2(X[rows, cols])	3.693290	3.443772	3.520977	3.406605	3.415177	6.741345

Table: Benchmarking of rowMeans2_X_S(), rowMeans2(X, cols, rows)() and rowMeans2(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	rowMeans2_X_S	0.005919	0.0064740	0.0067625	0.0066570	0.0067815	0.014183
2	rowMeans2(X, cols, rows)	0.007441	0.0079700	0.0081432	0.0081145	0.0082450	0.010104
3	rowMeans2(X[cols, rows])	0.013215	0.0139585	0.0149644	0.0142105	0.0144640	0.061126

	expr	min	lq	mean	median	uq	max
1	rowMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.0000000
2	rowMeans2(X, cols, rows)	1.257138	1.231078	1.204183	1.218943	1.215808	0.7124022
3	rowMeans2(X[cols, rows])	2.232641	2.156086	2.212866	2.134670	2.132862	4.3098075

Figure: Benchmarking of colMeans2_X_S(), colMeans2(X, rows, cols)() and colMeans2(X[rows, cols])() on double+100x100 data as well as rowMeans2_X_S(), rowMeans2(X, cols, rows)() and rowMeans2(X[cols, rows])() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colMeans2_X_S() and rowMeans2_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	colMeans2_X_S	5.872	6.4425	6.56686	6.5555	6.6285	10.052
2	rowMeans2_X_S	5.919	6.4740	6.76247	6.6570	6.7815	14.183

	expr	min	lq	mean	median	uq	max
1	colMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	rowMeans2_X_S	1.008004	1.004889	1.029787	1.015483	1.023082	1.410963

Figure: Benchmarking of colMeans2_X_S() and rowMeans2_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 3135741 167.5    5709258 305.0  5709258 305.0
Vcells 5794560  44.3   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colMeans2_X_S = colMeans2(X_S, na.rm = FALSE), `colMeans2(X, rows, cols)` = colMeans2(X, 
+     rows = rows, cols = cols, na.rm = FALSE), `colMeans2(X[rows, cols])` = colMeans2(X[rows, cols], 
+     na.rm = FALSE), unit = "ms")
> X <- t(X)
> X_S <- t(X_S)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3135732 167.5    5709258 305.0  5709258 305.0
Vcells 5804638  44.3   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowMeans2_X_S = rowMeans2(X_S, na.rm = FALSE), `rowMeans2(X, cols, rows)` = rowMeans2(X, 
+     rows = cols, cols = rows, na.rm = FALSE), `rowMeans2(X[cols, rows])` = rowMeans2(X[cols, rows], 
+     na.rm = FALSE), unit = "ms")

Table: Benchmarking of colMeans2_X_S(), colMeans2(X, rows, cols)() and colMeans2(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	colMeans2_X_S	0.006849	0.0070510	0.0072699	0.0071870	0.0073375	0.009987
2	colMeans2(X, rows, cols)	0.007977	0.0083060	0.0090472	0.0084335	0.0086530	0.050012
3	colMeans2(X[rows, cols])	0.022526	0.0228485	0.0236583	0.0230470	0.0233030	0.040380

	expr	min	lq	mean	median	uq	max
1	colMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	colMeans2(X, rows, cols)	1.164696	1.177989	1.244474	1.173438	1.179285	5.007710
3	colMeans2(X[rows, cols])	3.288947	3.240462	3.254284	3.206762	3.175877	4.043256

Table: Benchmarking of rowMeans2_X_S(), rowMeans2(X, cols, rows)() and rowMeans2(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	rowMeans2_X_S	0.006955	0.0071505	0.0076115	0.0072860	0.0074295	0.036883
2	rowMeans2(X, cols, rows)	0.007956	0.0082800	0.0086619	0.0084415	0.0086805	0.022799
3	rowMeans2(X[cols, rows])	0.015898	0.0163690	0.0169838	0.0165925	0.0168355	0.029619

	expr	min	lq	mean	median	uq	max
1	rowMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.0000000
2	rowMeans2(X, cols, rows)	1.143925	1.157961	1.138006	1.158592	1.168383	0.6181439
3	rowMeans2(X[cols, rows])	2.285837	2.289210	2.231342	2.277313	2.266034	0.8030529

Figure: Benchmarking of colMeans2_X_S(), colMeans2(X, rows, cols)() and colMeans2(X[rows, cols])() on double+1000x10 data as well as rowMeans2_X_S(), rowMeans2(X, cols, rows)() and rowMeans2(X[cols, rows])() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colMeans2_X_S() and rowMeans2_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
1	colMeans2_X_S	6.849	7.0510	7.26990	7.187	7.3375	9.987
2	rowMeans2_X_S	6.955	7.1505	7.61149	7.286	7.4295	36.883

	expr	min	lq	mean	median	uq	max
1	colMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	rowMeans2_X_S	1.015477	1.014112	1.046987	1.013775	1.012538	3.693101

Figure: Benchmarking of colMeans2_X_S() and rowMeans2_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 3135944 167.5    5709258 305.0  5709258 305.0
Vcells 5794692  44.3   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colMeans2_X_S = colMeans2(X_S, na.rm = FALSE), `colMeans2(X, rows, cols)` = colMeans2(X, 
+     rows = rows, cols = cols, na.rm = FALSE), `colMeans2(X[rows, cols])` = colMeans2(X[rows, cols], 
+     na.rm = FALSE), unit = "ms")
> X <- t(X)
> X_S <- t(X_S)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3135938 167.5    5709258 305.0  5709258 305.0
Vcells 5804775  44.3   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowMeans2_X_S = rowMeans2(X_S, na.rm = FALSE), `rowMeans2(X, cols, rows)` = rowMeans2(X, 
+     rows = cols, cols = rows, na.rm = FALSE), `rowMeans2(X[cols, rows])` = rowMeans2(X[cols, rows], 
+     na.rm = FALSE), unit = "ms")

Table: Benchmarking of colMeans2_X_S(), colMeans2(X, rows, cols)() and colMeans2(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	colMeans2_X_S	0.007723	0.007972	0.0081978	0.0080790	0.0082330	0.012079
2	colMeans2(X, rows, cols)	0.010596	0.010898	0.0116180	0.0110600	0.0112635	0.060178
3	colMeans2(X[rows, cols])	0.017175	0.017608	0.0183274	0.0178085	0.0180270	0.031155

	expr	min	lq	mean	median	uq	max
1	colMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	colMeans2(X, rows, cols)	1.372006	1.367035	1.417200	1.368981	1.368092	4.982035
3	colMeans2(X[rows, cols])	2.223877	2.208731	2.235649	2.204295	2.189603	2.579270

Table: Benchmarking of rowMeans2_X_S(), rowMeans2(X, cols, rows)() and rowMeans2(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	rowMeans2_X_S	0.007709	0.0079320	0.0082596	0.0080340	0.008173	0.023288
2	rowMeans2(X, cols, rows)	0.009782	0.0101190	0.0103566	0.0103075	0.010485	0.011927
3	rowMeans2(X[cols, rows])	0.015188	0.0157655	0.0162749	0.0159710	0.016128	0.033182

	expr	min	lq	mean	median	uq	max
1	rowMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.0000000
2	rowMeans2(X, cols, rows)	1.268906	1.275719	1.253891	1.282985	1.282883	0.5121522
3	rowMeans2(X[cols, rows])	1.970165	1.987582	1.970422	1.987926	1.973327	1.4248540

Figure: Benchmarking of colMeans2_X_S(), colMeans2(X, rows, cols)() and colMeans2(X[rows, cols])() on double+10x1000 data as well as rowMeans2_X_S(), rowMeans2(X, cols, rows)() and rowMeans2(X[cols, rows])() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colMeans2_X_S() and rowMeans2_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
2	rowMeans2_X_S	7.709	7.932	8.25960	8.034	8.173	23.288
1	colMeans2_X_S	7.723	7.972	8.19782	8.079	8.233	12.079

	expr	min	lq	mean	median	uq	max
2	rowMeans2_X_S	1.000000	1.000000	1.0000000	1.000000	1.000000	1.0000000
1	colMeans2_X_S	1.001816	1.005043	0.9925202	1.005601	1.007341	0.5186791

Figure: Benchmarking of colMeans2_X_S() and rowMeans2_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 3136143 167.5    5709258 305.0  5709258 305.0
Vcells 5840147  44.6   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colMeans2_X_S = colMeans2(X_S, na.rm = FALSE), `colMeans2(X, rows, cols)` = colMeans2(X, 
+     rows = rows, cols = cols, na.rm = FALSE), `colMeans2(X[rows, cols])` = colMeans2(X[rows, cols], 
+     na.rm = FALSE), unit = "ms")
> X <- t(X)
> X_S <- t(X_S)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3136137 167.5    5709258 305.0  5709258 305.0
Vcells 5940230  45.4   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowMeans2_X_S = rowMeans2(X_S, na.rm = FALSE), `rowMeans2(X, cols, rows)` = rowMeans2(X, 
+     rows = cols, cols = rows, na.rm = FALSE), `rowMeans2(X[cols, rows])` = rowMeans2(X[cols, rows], 
+     na.rm = FALSE), unit = "ms")

Table: Benchmarking of colMeans2_X_S(), colMeans2(X, rows, cols)() and colMeans2(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	colMeans2_X_S	0.049093	0.0495460	0.0501448	0.0497405	0.0501215	0.062054
2	colMeans2(X, rows, cols)	0.067140	0.0675770	0.0712509	0.0678340	0.0680755	0.247151
3	colMeans2(X[rows, cols])	0.196305	0.1978025	0.2038208	0.1988080	0.2003185	0.516251

	expr	min	lq	mean	median	uq	max
1	colMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	colMeans2(X, rows, cols)	1.367608	1.363924	1.420903	1.363758	1.358210	3.982838
3	colMeans2(X[rows, cols])	3.998635	3.992300	4.064645	3.996904	3.996658	8.319383

Table: Benchmarking of rowMeans2_X_S(), rowMeans2(X, cols, rows)() and rowMeans2(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	rowMeans2_X_S	0.051898	0.0523625	0.0549039	0.0526590	0.053032	0.152804
2	rowMeans2(X, cols, rows)	0.089185	0.0895640	0.0907285	0.0897950	0.090092	0.144052
3	rowMeans2(X[cols, rows])	0.122574	0.1287530	0.1354444	0.1316945	0.137083	0.259074

	expr	min	lq	mean	median	uq	max
1	rowMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	rowMeans2(X, cols, rows)	1.718467	1.710461	1.652496	1.705217	1.698823	0.942724
3	rowMeans2(X[cols, rows])	2.361825	2.458878	2.466937	2.500892	2.584911	1.695466

Figure: Benchmarking of colMeans2_X_S(), colMeans2(X, rows, cols)() and colMeans2(X[rows, cols])() on double+100x1000 data as well as rowMeans2_X_S(), rowMeans2(X, cols, rows)() and rowMeans2(X[cols, rows])() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colMeans2_X_S() and rowMeans2_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
1	colMeans2_X_S	49.093	49.5460	50.14479	49.7405	50.1215	62.054
2	rowMeans2_X_S	51.898	52.3625	54.90390	52.6590	53.0320	152.804

	expr	min	lq	mean	median	uq	max
1	colMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	rowMeans2_X_S	1.057136	1.056846	1.094907	1.058674	1.058069	2.462436

Figure: Benchmarking of colMeans2_X_S() and rowMeans2_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 3136353 167.5    5709258 305.0  5709258 305.0
Vcells 5840284  44.6   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colMeans2_X_S = colMeans2(X_S, na.rm = FALSE), `colMeans2(X, rows, cols)` = colMeans2(X, 
+     rows = rows, cols = cols, na.rm = FALSE), `colMeans2(X[rows, cols])` = colMeans2(X[rows, cols], 
+     na.rm = FALSE), unit = "ms")
> X <- t(X)
> X_S <- t(X_S)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3136353 167.5    5709258 305.0  5709258 305.0
Vcells 5940377  45.4   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowMeans2_X_S = rowMeans2(X_S, na.rm = FALSE), `rowMeans2(X, cols, rows)` = rowMeans2(X, 
+     rows = cols, cols = rows, na.rm = FALSE), `rowMeans2(X[cols, rows])` = rowMeans2(X[cols, rows], 
+     na.rm = FALSE), unit = "ms")

Table: Benchmarking of colMeans2_X_S(), colMeans2(X, rows, cols)() and colMeans2(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	colMeans2_X_S	0.052271	0.0527665	0.0535432	0.0529310	0.0535930	0.061478
2	colMeans2(X, rows, cols)	0.059460	0.0599315	0.0619651	0.0601750	0.0605510	0.106062
3	colMeans2(X[rows, cols])	0.119594	0.1202450	0.1292758	0.1211125	0.1230175	0.405650

	expr	min	lq	mean	median	uq	max
1	colMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	colMeans2(X, rows, cols)	1.137533	1.135787	1.157292	1.136857	1.129830	1.725202
3	colMeans2(X[rows, cols])	2.287961	2.278813	2.414420	2.288120	2.295402	6.598295

Table: Benchmarking of rowMeans2_X_S(), rowMeans2(X, cols, rows)() and rowMeans2(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	rowMeans2_X_S	0.053730	0.0542410	0.0551803	0.0545730	0.0548390	0.101148
2	rowMeans2(X, cols, rows)	0.086766	0.0871835	0.0875073	0.0873960	0.0876775	0.091674
3	rowMeans2(X[cols, rows])	0.128705	0.1336225	0.1393993	0.1367735	0.1396765	0.286298

	expr	min	lq	mean	median	uq	max
1	rowMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.0000000
2	rowMeans2(X, cols, rows)	1.614852	1.607336	1.585843	1.601451	1.598817	0.9063353
3	rowMeans2(X[cols, rows])	2.395403	2.463496	2.526253	2.506248	2.547029	2.8304860

Figure: Benchmarking of colMeans2_X_S(), colMeans2(X, rows, cols)() and colMeans2(X[rows, cols])() on double+1000x100 data as well as rowMeans2_X_S(), rowMeans2(X, cols, rows)() and rowMeans2(X[cols, rows])() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colMeans2_X_S() and rowMeans2_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	colMeans2_X_S	52.271	52.7665	53.54319	52.931	53.593	61.478
2	rowMeans2_X_S	53.730	54.2410	55.18028	54.573	54.839	101.148

	expr	min	lq	mean	median	uq	max
1	colMeans2_X_S	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	rowMeans2_X_S	1.027912	1.027944	1.030575	1.031022	1.023249	1.645271

Figure: Benchmarking of colMeans2_X_S() and rowMeans2_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 22.76 secs.

Reproducibility

To reproduce this report, do:

html <- matrixStats:::benchmark('colRowMeans2_subset')

---

Copyright Henrik Bengtsson. Last updated on 2019-09-10 20:43:35 (-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>