colRowMeans2

matrixStats: Benchmark report

---

colMeans2() and rowMeans2() benchmarks

This report benchmark the performance of colMeans2() and rowMeans2() against alternative methods.

Alternative methods

• apply() + mean()
• .colMeans() and .rowMeans()
• colMeans() and rowMeans()

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"]]
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3139851 167.7    5709258 305.0  5709258 305.0
Vcells 6077827  46.4   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colMeans2 = colMeans2(X, na.rm = FALSE), .colMeans = .colMeans(X, m = nrow(X), 
+     n = ncol(X), na.rm = FALSE), colMeans = colMeans(X, na.rm = FALSE), `apply+mean` = apply(X, MARGIN = 2L, 
+     FUN = mean, na.rm = FALSE), unit = "ms")
> X <- t(X)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3138984 167.7    5709258 305.0  5709258 305.0
Vcells 6075817  46.4   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowMeans2 = rowMeans2(X, na.rm = FALSE), .rowMeans = .rowMeans(X, m = nrow(X), 
+     n = ncol(X), na.rm = FALSE), rowMeans = rowMeans(X, na.rm = FALSE), `apply+mean` = apply(X, MARGIN = 1L, 
+     FUN = mean, na.rm = FALSE), unit = "ms")

Table: Benchmarking of colMeans2(), .colMeans(), colMeans() and apply+mean() 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	0.001118	0.0013945	0.0018182	0.0018415	0.0019805	0.011639
2	.colMeans	0.001602	0.0018650	0.0023100	0.0021000	0.0025185	0.014027
3	colMeans	0.003010	0.0032895	0.0040018	0.0039165	0.0043600	0.013600
4	apply+mean	0.044338	0.0458210	0.0477659	0.0463565	0.0470120	0.144426

	expr	min	lq	mean	median	uq	max
1	colMeans2	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	.colMeans	1.432916	1.337397	1.270511	1.140375	1.271649	1.205172
3	colMeans	2.692308	2.358910	2.200975	2.126799	2.201464	1.168485
4	apply+mean	39.658318	32.858372	26.271116	25.173228	23.737440	12.408798

Table: Benchmarking of rowMeans2(), .rowMeans(), rowMeans() and apply+mean() 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	0.001100	0.0013460	0.0017564	0.0017395	0.0019180	0.010729
2	.rowMeans	0.002473	0.0026470	0.0029007	0.0028050	0.0029790	0.009608
3	rowMeans	0.003673	0.0039865	0.0046466	0.0045145	0.0047140	0.029878
4	apply+mean	0.044563	0.0455930	0.0469140	0.0459645	0.0464085	0.103257

	expr	min	lq	mean	median	uq	max
1	rowMeans2	1.000000	1.000000	1.000000	1.000000	1.000000	1.0000000
2	.rowMeans	2.248182	1.966568	1.651481	1.612532	1.553180	0.8955168
3	rowMeans	3.339091	2.961739	2.645451	2.595286	2.457768	2.7847889
4	apply+mean	40.511818	33.872957	26.709855	26.423972	24.196298	9.6241029

Figure: Benchmarking of colMeans2(), .colMeans(), colMeans() and apply+mean() on integer+10x10 data as well as rowMeans2(), .rowMeans(), rowMeans() and apply+mean() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colMeans2() and rowMeans2() 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	1.100	1.3460	1.75643	1.7395	1.9180	10.729
1	colMeans2	1.118	1.3945	1.81819	1.8415	1.9805	11.639

	expr	min	lq	mean	median	uq	max
2	rowMeans2	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
1	colMeans2	1.016364	1.036033	1.035162	1.058637	1.032586	1.084817

Figure: Benchmarking of colMeans2() and rowMeans2() on integer+10x10 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

100x100 integer matrix

> X <- data[["100x100"]]
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3137564 167.6    5709258 305.0  5709258 305.0
Vcells 5692629  43.5   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colMeans2 = colMeans2(X, na.rm = FALSE), .colMeans = .colMeans(X, m = nrow(X), 
+     n = ncol(X), na.rm = FALSE), colMeans = colMeans(X, na.rm = FALSE), `apply+mean` = apply(X, MARGIN = 2L, 
+     FUN = mean, na.rm = FALSE), unit = "ms")
> X <- t(X)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3137555 167.6    5709258 305.0  5709258 305.0
Vcells 5697667  43.5   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowMeans2 = rowMeans2(X, na.rm = FALSE), .rowMeans = .rowMeans(X, m = nrow(X), 
+     n = ncol(X), na.rm = FALSE), rowMeans = rowMeans(X, na.rm = FALSE), `apply+mean` = apply(X, MARGIN = 1L, 
+     FUN = mean, na.rm = FALSE), unit = "ms")

Table: Benchmarking of colMeans2(), .colMeans(), colMeans() and apply+mean() 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
2	.colMeans	0.009152	0.0095320	0.0100594	0.0099955	0.0103290	0.015906
3	colMeans	0.010689	0.0113320	0.0119437	0.0118010	0.0123520	0.018471
1	colMeans2	0.011058	0.0114345	0.0122556	0.0118535	0.0121290	0.044265
4	apply+mean	0.389271	0.3951415	0.4015145	0.3986820	0.4022945	0.509034

	expr	min	lq	mean	median	uq	max
2	.colMeans	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
3	colMeans	1.167941	1.188838	1.187323	1.180631	1.195856	1.161260
1	colMeans2	1.208260	1.199591	1.218332	1.185884	1.174267	2.782912
4	apply+mean	42.533982	41.454207	39.914514	39.886149	38.948059	32.002640

Table: Benchmarking of rowMeans2(), .rowMeans(), rowMeans() and apply+mean() 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	0.011204	0.0115515	0.0121952	0.0120125	0.0124430	0.021796
2	.rowMeans	0.038564	0.0389080	0.0393451	0.0391395	0.0394070	0.049944
3	rowMeans	0.040023	0.0404875	0.0411525	0.0411120	0.0414240	0.048529
4	apply+mean	0.387624	0.3917800	0.3984268	0.3944985	0.3990515	0.500580

	expr	min	lq	mean	median	uq	max
1	rowMeans2	1.000000	1.000000	1.000000	1.000000	1.000000	1.00000
2	.rowMeans	3.441985	3.368221	3.226285	3.258231	3.167002	2.29143
3	rowMeans	3.572206	3.504956	3.374493	3.422435	3.329101	2.22651
4	apply+mean	34.596930	33.915942	32.670873	32.840666	32.070361	22.96660

Figure: Benchmarking of colMeans2(), .colMeans(), colMeans() and apply+mean() on integer+100x100 data as well as rowMeans2(), .rowMeans(), rowMeans() and apply+mean() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colMeans2() and rowMeans2() 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	11.058	11.4345	12.25564	11.8535	12.129	44.265
2	rowMeans2	11.204	11.5515	12.19517	12.0125	12.443	21.796

	expr	min	lq	mean	median	uq	max
1	colMeans2	1.000000	1.000000	1.0000000	1.000000	1.000000	1.0000000
2	rowMeans2	1.013203	1.010232	0.9950659	1.013414	1.025888	0.4923981

Figure: Benchmarking of colMeans2() and rowMeans2() on integer+100x100 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

1000x10 integer matrix

> X <- data[["1000x10"]]
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3138330 167.7    5709258 305.0  5709258 305.0
Vcells 5696415  43.5   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colMeans2 = colMeans2(X, na.rm = FALSE), .colMeans = .colMeans(X, m = nrow(X), 
+     n = ncol(X), na.rm = FALSE), colMeans = colMeans(X, na.rm = FALSE), `apply+mean` = apply(X, MARGIN = 2L, 
+     FUN = mean, na.rm = FALSE), unit = "ms")
> X <- t(X)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3138321 167.7    5709258 305.0  5709258 305.0
Vcells 5701453  43.5   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowMeans2 = rowMeans2(X, na.rm = FALSE), .rowMeans = .rowMeans(X, m = nrow(X), 
+     n = ncol(X), na.rm = FALSE), rowMeans = rowMeans(X, na.rm = FALSE), `apply+mean` = apply(X, MARGIN = 1L, 
+     FUN = mean, na.rm = FALSE), unit = "ms")

Table: Benchmarking of colMeans2(), .colMeans(), colMeans() and apply+mean() 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
2	.colMeans	0.009389	0.0098325	0.0105085	0.0102025	0.0105670	0.037245
3	colMeans	0.010948	0.0113905	0.0123146	0.0120045	0.0125850	0.031726
1	colMeans2	0.012227	0.0128430	0.0133656	0.0132430	0.0134815	0.028296
4	apply+mean	0.100870	0.1024890	0.1056013	0.1033105	0.1048780	0.173595

	expr	min	lq	mean	median	uq	max
2	.colMeans	1.000000	1.000000	1.000000	1.000000	1.000000	1.0000000
3	colMeans	1.166045	1.158454	1.171861	1.176623	1.190972	0.8518190
1	colMeans2	1.302269	1.306178	1.271878	1.298015	1.275812	0.7597261
4	apply+mean	10.743423	10.423493	10.049095	10.125998	9.925050	4.6608941

Table: Benchmarking of rowMeans2(), .rowMeans(), rowMeans() and apply+mean() 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	0.012525	0.0130485	0.0135167	0.0134115	0.0136675	0.027150
4	apply+mean	0.100688	0.1023380	0.1049152	0.1029525	0.1045485	0.209029
2	.rowMeans	0.135357	0.1361160	0.1364440	0.1363250	0.1365440	0.140769
3	rowMeans	0.136966	0.1377290	0.1384011	0.1383265	0.1386560	0.146218

	expr	min	lq	mean	median	uq	max
1	rowMeans2	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
4	apply+mean	8.038962	7.842894	7.761877	7.676434	7.649424	7.699042
2	.rowMeans	10.806946	10.431544	10.094450	10.164784	9.990415	5.184862
3	rowMeans	10.935409	10.555160	10.239241	10.314022	10.144942	5.385562

Figure: Benchmarking of colMeans2(), .colMeans(), colMeans() and apply+mean() on integer+1000x10 data as well as rowMeans2(), .rowMeans(), rowMeans() and apply+mean() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colMeans2() and rowMeans2() 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	colMeans2	12.227	12.8430	13.36558	13.2430	13.4815	28.296
2	rowMeans2	12.525	13.0485	13.51673	13.4115	13.6675	27.150

	expr	min	lq	mean	median	uq	max
1	colMeans2	1.000000	1.000000	1.000000	1.000000	1.000000	1.0000000
2	rowMeans2	1.024372	1.016001	1.011309	1.012724	1.013797	0.9594996

Figure: Benchmarking of colMeans2() and rowMeans2() on integer+1000x10 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

10x1000 integer matrix

> X <- data[["10x1000"]]
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3138552 167.7    5709258 305.0  5709258 305.0
Vcells 5697223  43.5   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colMeans2 = colMeans2(X, na.rm = FALSE), .colMeans = .colMeans(X, m = nrow(X), 
+     n = ncol(X), na.rm = FALSE), colMeans = colMeans(X, na.rm = FALSE), `apply+mean` = apply(X, MARGIN = 2L, 
+     FUN = mean, na.rm = FALSE), unit = "ms")
> X <- t(X)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3138543 167.7    5709258 305.0  5709258 305.0
Vcells 5702261  43.6   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowMeans2 = rowMeans2(X, na.rm = FALSE), .rowMeans = .rowMeans(X, m = nrow(X), 
+     n = ncol(X), na.rm = FALSE), rowMeans = rowMeans(X, na.rm = FALSE), `apply+mean` = apply(X, MARGIN = 1L, 
+     FUN = mean, na.rm = FALSE), unit = "ms")

Table: Benchmarking of colMeans2(), .colMeans(), colMeans() and apply+mean() 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
2	.colMeans	0.009074	0.009620	0.0109615	0.0103155	0.011103	0.023892
1	colMeans2	0.010614	0.011105	0.0127047	0.0119750	0.013109	0.026440
3	colMeans	0.010567	0.011371	0.0130535	0.0125615	0.013851	0.023245
4	apply+mean	3.208715	3.267930	3.4315273	3.3277545	3.455468	8.267523

	expr	min	lq	mean	median	uq	max
2	.colMeans	1.000000	1.000000	1.000000	1.000000	1.000000	1.0000000
1	colMeans2	1.169716	1.154366	1.159031	1.160874	1.180672	1.1066466
3	colMeans	1.164536	1.182017	1.190846	1.217731	1.247501	0.9729198
4	apply+mean	353.616377	339.701715	313.052712	322.597499	311.219355	346.0372928

Table: Benchmarking of rowMeans2(), .rowMeans(), rowMeans() and apply+mean() 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	0.010936	0.0114295	0.0126284	0.012351	0.0133470	0.020203
2	.rowMeans	0.030953	0.0314140	0.0330664	0.031832	0.0331605	0.058767
3	rowMeans	0.032560	0.0330545	0.0351301	0.034382	0.0353905	0.052018
4	apply+mean	3.197030	3.2592550	3.4373507	3.341137	3.4499105	8.425955

	expr	min	lq	mean	median	uq	max
1	rowMeans2	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	.rowMeans	2.830377	2.748502	2.618415	2.577281	2.484491	2.908825
3	rowMeans	2.977323	2.892034	2.781836	2.783742	2.651570	2.574766
4	apply+mean	292.339978	285.161643	272.192098	270.515505	258.478347	417.064545

Figure: Benchmarking of colMeans2(), .colMeans(), colMeans() and apply+mean() on integer+10x1000 data as well as rowMeans2(), .rowMeans(), rowMeans() and apply+mean() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colMeans2() and rowMeans2() 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	10.614	11.1050	12.70472	11.975	13.109	26.440
2	rowMeans2	10.936	11.4295	12.62840	12.351	13.347	20.203

	expr	min	lq	mean	median	uq	max
1	colMeans2	1.000000	1.000000	1.0000000	1.000000	1.000000	1.0000000
2	rowMeans2	1.030337	1.029221	0.9939928	1.031399	1.018155	0.7641074

Figure: Benchmarking of colMeans2() and rowMeans2() on integer+10x1000 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

100x1000 integer matrix

> X <- data[["100x1000"]]
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3138772 167.7    5709258 305.0  5709258 305.0
Vcells 5697806  43.5   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colMeans2 = colMeans2(X, na.rm = FALSE), .colMeans = .colMeans(X, m = nrow(X), 
+     n = ncol(X), na.rm = FALSE), colMeans = colMeans(X, na.rm = FALSE), `apply+mean` = apply(X, MARGIN = 2L, 
+     FUN = mean, na.rm = FALSE), unit = "ms")
> X <- t(X)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3138766 167.7    5709258 305.0  5709258 305.0
Vcells 5747849  43.9   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowMeans2 = rowMeans2(X, na.rm = FALSE), .rowMeans = .rowMeans(X, m = nrow(X), 
+     n = ncol(X), na.rm = FALSE), rowMeans = rowMeans(X, na.rm = FALSE), `apply+mean` = apply(X, MARGIN = 1L, 
+     FUN = mean, na.rm = FALSE), unit = "ms")

Table: Benchmarking of colMeans2(), .colMeans(), colMeans() and apply+mean() 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
2	.colMeans	0.074811	0.0755745	0.0785353	0.0770220	0.0788250	0.108906
3	colMeans	0.076820	0.0776875	0.0822842	0.0796755	0.0830435	0.135364
1	colMeans2	0.099775	0.1002240	0.1053356	0.1023415	0.1066470	0.131193
4	apply+mean	3.770532	3.8167745	4.1021140	3.8867665	4.0373260	15.632112

	expr	min	lq	mean	median	uq	max
2	.colMeans	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
3	colMeans	1.026854	1.027959	1.047736	1.034451	1.053517	1.242943
1	colMeans2	1.333694	1.326162	1.341251	1.328731	1.352959	1.204644
4	apply+mean	50.400770	50.503470	52.232741	50.463069	51.218852	143.537656

Table: Benchmarking of rowMeans2(), .rowMeans(), rowMeans() and apply+mean() 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	0.105809	0.1063705	0.1106376	0.1072835	0.1092985	0.204664
2	.rowMeans	0.272818	0.2733865	0.2794856	0.2739900	0.2760480	0.367536
3	rowMeans	0.274481	0.2752600	0.2806756	0.2763950	0.2789175	0.347585
4	apply+mean	3.788474	3.8303300	4.0666559	3.8958810	4.0174210	15.718836

	expr	min	lq	mean	median	uq	max
1	rowMeans2	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	.rowMeans	2.578401	2.570135	2.526135	2.553888	2.525634	1.795802
3	rowMeans	2.594118	2.587747	2.536891	2.576305	2.551888	1.698320
4	apply+mean	35.804837	36.009326	36.756538	36.313888	36.756415	76.803131

Figure: Benchmarking of colMeans2(), .colMeans(), colMeans() and apply+mean() on integer+100x1000 data as well as rowMeans2(), .rowMeans(), rowMeans() and apply+mean() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colMeans2() and rowMeans2() 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	99.775	100.2240	105.3356	102.3415	106.6470	131.193
2	rowMeans2	105.809	106.3705	110.6376	107.2835	109.2985	204.664

	expr	min	lq	mean	median	uq	max
1	colMeans2	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	rowMeans2	1.060476	1.061328	1.050335	1.048289	1.024862	1.560022

Figure: Benchmarking of colMeans2() and rowMeans2() on integer+100x1000 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

1000x100 integer matrix

> X <- data[["1000x100"]]
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3139005 167.7    5709258 305.0  5709258 305.0
Vcells 5698593  43.5   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colMeans2 = colMeans2(X, na.rm = FALSE), .colMeans = .colMeans(X, m = nrow(X), 
+     n = ncol(X), na.rm = FALSE), colMeans = colMeans(X, na.rm = FALSE), `apply+mean` = apply(X, MARGIN = 2L, 
+     FUN = mean, na.rm = FALSE), unit = "ms")
> X <- t(X)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3138999 167.7    5709258 305.0  5709258 305.0
Vcells 5748636  43.9   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowMeans2 = rowMeans2(X, na.rm = FALSE), .rowMeans = .rowMeans(X, m = nrow(X), 
+     n = ncol(X), na.rm = FALSE), rowMeans = rowMeans(X, na.rm = FALSE), `apply+mean` = apply(X, MARGIN = 1L, 
+     FUN = mean, na.rm = FALSE), unit = "ms")

Table: Benchmarking of colMeans2(), .colMeans(), colMeans() and apply+mean() 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
2	.colMeans	0.079366	0.0796445	0.0803061	0.0800025	0.0803930	0.085562
3	colMeans	0.081143	0.0817155	0.0829258	0.0823115	0.0829675	0.129585
1	colMeans2	0.105009	0.1053790	0.1062905	0.1058115	0.1062135	0.124981
4	apply+mean	0.838260	0.8452160	0.8647765	0.8540085	0.8741790	1.017310

	expr	min	lq	mean	median	uq	max
2	.colMeans	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
3	colMeans	1.022390	1.026003	1.032622	1.028862	1.032024	1.514516
1	colMeans2	1.323098	1.323117	1.323566	1.322602	1.321179	1.460707
4	apply+mean	10.561953	10.612359	10.768500	10.674773	10.873820	11.889741

Table: Benchmarking of rowMeans2(), .rowMeans(), rowMeans() and apply+mean() 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	0.114807	0.1166280	0.1211237	0.1175665	0.1195020	0.219929
2	.rowMeans	0.371756	0.3728350	0.3784171	0.3732835	0.3743715	0.456133
3	rowMeans	0.373874	0.3746180	0.3790754	0.3753410	0.3780370	0.460845
4	apply+mean	0.852882	0.8707085	0.9019386	0.8792360	0.8988245	1.298366

	expr	min	lq	mean	median	uq	max
1	rowMeans2	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	.rowMeans	3.238095	3.196788	3.124220	3.175084	3.132763	2.074001
3	rowMeans	3.256544	3.212076	3.129655	3.192585	3.163437	2.095426
4	apply+mean	7.428833	7.465690	7.446426	7.478627	7.521418	5.903569

Figure: Benchmarking of colMeans2(), .colMeans(), colMeans() and apply+mean() on integer+1000x100 data as well as rowMeans2(), .rowMeans(), rowMeans() and apply+mean() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colMeans2() and rowMeans2() 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	105.009	105.379	106.2905	105.8115	106.2135	124.981
2	rowMeans2	114.807	116.628	121.1237	117.5665	119.5020	219.929

	expr	min	lq	mean	median	uq	max
1	colMeans2	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	rowMeans2	1.093306	1.106748	1.139554	1.111094	1.125111	1.759699

Figure: Benchmarking of colMeans2() and rowMeans2() 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"]]
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3139245 167.7    5709258 305.0  5709258 305.0
Vcells 5814528  44.4   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colMeans2 = colMeans2(X, na.rm = FALSE), .colMeans = .colMeans(X, m = nrow(X), 
+     n = ncol(X), na.rm = FALSE), colMeans = colMeans(X, na.rm = FALSE), `apply+mean` = apply(X, MARGIN = 2L, 
+     FUN = mean, na.rm = FALSE), unit = "ms")
> X <- t(X)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3139227 167.7    5709258 305.0  5709258 305.0
Vcells 5814651  44.4   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowMeans2 = rowMeans2(X, na.rm = FALSE), .rowMeans = .rowMeans(X, m = nrow(X), 
+     n = ncol(X), na.rm = FALSE), rowMeans = rowMeans(X, na.rm = FALSE), `apply+mean` = apply(X, MARGIN = 1L, 
+     FUN = mean, na.rm = FALSE), unit = "ms")

Table: Benchmarking of colMeans2(), .colMeans(), colMeans() and apply+mean() 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	0.001137	0.0014075	0.0018672	0.0018675	0.002051	0.012019
2	.colMeans	0.001644	0.0018460	0.0022731	0.0020850	0.002520	0.011391
3	colMeans	0.002991	0.0034025	0.0039936	0.0038965	0.004390	0.011282
4	apply+mean	0.045506	0.0463995	0.0481880	0.0470175	0.047545	0.145097

	expr	min	lq	mean	median	uq	max
1	colMeans2	1.000000	1.000000	1.000000	1.000000	1.000000	1.0000000
2	.colMeans	1.445910	1.311545	1.217387	1.116466	1.228669	0.9477494
3	colMeans	2.630607	2.417407	2.138835	2.086479	2.140419	0.9386804
4	apply+mean	40.022867	32.965897	25.807903	25.176707	23.181375	12.0723022

Table: Benchmarking of rowMeans2(), .rowMeans(), rowMeans() and apply+mean() 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
2	.rowMeans	0.001270	0.0014695	0.0017708	0.0016930	0.0019405	0.005853
1	rowMeans2	0.001105	0.0013700	0.0017901	0.0017550	0.0019500	0.010126
3	rowMeans	0.002554	0.0028565	0.0035454	0.0033785	0.0037425	0.022294
4	apply+mean	0.044823	0.0460120	0.0478368	0.0464970	0.0470385	0.100305

	expr	min	lq	mean	median	uq	max
2	.rowMeans	1.0000000	1.0000000	1.000000	1.000000	1.000000	1.000000
1	rowMeans2	0.8700787	0.9322899	1.010899	1.036621	1.004896	1.730053
3	rowMeans	2.0110236	1.9438585	2.002163	1.995570	1.928627	3.808987
4	apply+mean	35.2937008	31.3113304	27.014683	27.464265	24.240402	17.137366

Figure: Benchmarking of colMeans2(), .colMeans(), colMeans() and apply+mean() on double+10x10 data as well as rowMeans2(), .rowMeans(), rowMeans() and apply+mean() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colMeans2() and rowMeans2() 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	1.105	1.3700	1.79007	1.7550	1.950	10.126
1	colMeans2	1.137	1.4075	1.86718	1.8675	2.051	12.019

	expr	min	lq	mean	median	uq	max
2	rowMeans2	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
1	colMeans2	1.028959	1.027372	1.043076	1.064103	1.051795	1.186945

Figure: Benchmarking of colMeans2() and rowMeans2() on double+10x10 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

100x100 double matrix

> X <- data[["100x100"]]
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3139462 167.7    5709258 305.0  5709258 305.0
Vcells 5815557  44.4   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colMeans2 = colMeans2(X, na.rm = FALSE), .colMeans = .colMeans(X, m = nrow(X), 
+     n = ncol(X), na.rm = FALSE), colMeans = colMeans(X, na.rm = FALSE), `apply+mean` = apply(X, MARGIN = 2L, 
+     FUN = mean, na.rm = FALSE), unit = "ms")
> X <- t(X)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3139453 167.7    5709258 305.0  5709258 305.0
Vcells 5825595  44.5   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowMeans2 = rowMeans2(X, na.rm = FALSE), .rowMeans = .rowMeans(X, m = nrow(X), 
+     n = ncol(X), na.rm = FALSE), rowMeans = rowMeans(X, na.rm = FALSE), `apply+mean` = apply(X, MARGIN = 1L, 
+     FUN = mean, na.rm = FALSE), unit = "ms")

Table: Benchmarking of colMeans2(), .colMeans(), colMeans() and apply+mean() 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
2	.colMeans	0.008221	0.0087925	0.0096838	0.0092225	0.0096805	0.019484
3	colMeans	0.009916	0.0105470	0.0118815	0.0112720	0.0117985	0.036548
1	colMeans2	0.011310	0.0116740	0.0129054	0.0122155	0.0125745	0.046964
4	apply+mean	0.443032	0.4478055	0.4671414	0.4521345	0.4629810	0.650094

	expr	min	lq	mean	median	uq	max
2	.colMeans	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
3	colMeans	1.206179	1.199545	1.226943	1.222228	1.218790	1.875795
1	colMeans2	1.375745	1.327722	1.332675	1.324532	1.298952	2.410388
4	apply+mean	53.890281	50.930395	48.239418	49.025156	47.826145	33.365531

Table: Benchmarking of rowMeans2(), .rowMeans(), rowMeans() and apply+mean() 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	0.011374	0.0117905	0.0123781	0.0122205	0.0126775	0.024969
2	.rowMeans	0.018457	0.0203275	0.0207877	0.0204935	0.0206905	0.043692
3	rowMeans	0.021277	0.0217830	0.0223839	0.0223900	0.0227155	0.030016
4	apply+mean	0.404403	0.4088945	0.4151887	0.4106470	0.4145620	0.532848

	expr	min	lq	mean	median	uq	max
1	rowMeans2	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	.rowMeans	1.622736	1.724058	1.679394	1.676977	1.632065	1.749850
3	rowMeans	1.870670	1.847504	1.808346	1.832167	1.791797	1.202131
4	apply+mean	35.555038	34.679997	33.542230	33.603126	32.700611	21.340382

Figure: Benchmarking of colMeans2(), .colMeans(), colMeans() and apply+mean() on double+100x100 data as well as rowMeans2(), .rowMeans(), rowMeans() and apply+mean() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colMeans2() and rowMeans2() 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	11.310	11.6740	12.90537	12.2155	12.5745	46.964
2	rowMeans2	11.374	11.7905	12.37809	12.2205	12.6775	24.969

	expr	min	lq	mean	median	uq	max
1	colMeans2	1.000000	1.000000	1.0000000	1.000000	1.000000	1.0000000
2	rowMeans2	1.005659	1.009979	0.9591426	1.000409	1.008191	0.5316626

Figure: Benchmarking of colMeans2() and rowMeans2() on double+100x100 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

1000x10 double matrix

> X <- data[["1000x10"]]
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3139688 167.7    5709258 305.0  5709258 305.0
Vcells 5816683  44.4   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colMeans2 = colMeans2(X, na.rm = FALSE), .colMeans = .colMeans(X, m = nrow(X), 
+     n = ncol(X), na.rm = FALSE), colMeans = colMeans(X, na.rm = FALSE), `apply+mean` = apply(X, MARGIN = 2L, 
+     FUN = mean, na.rm = FALSE), unit = "ms")
> X <- t(X)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3139679 167.7    5709258 305.0  5709258 305.0
Vcells 5826721  44.5   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowMeans2 = rowMeans2(X, na.rm = FALSE), .rowMeans = .rowMeans(X, m = nrow(X), 
+     n = ncol(X), na.rm = FALSE), rowMeans = rowMeans(X, na.rm = FALSE), `apply+mean` = apply(X, MARGIN = 1L, 
+     FUN = mean, na.rm = FALSE), unit = "ms")

Table: Benchmarking of colMeans2(), .colMeans(), colMeans() and apply+mean() 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
2	.colMeans	0.009459	0.0097845	0.0102891	0.010161	0.0104700	0.024951
3	colMeans	0.010823	0.0114160	0.0121760	0.011978	0.0125580	0.025641
1	colMeans2	0.012415	0.0129345	0.0134168	0.013192	0.0135985	0.025389
4	apply+mean	0.152337	0.1540075	0.1570167	0.155667	0.1576625	0.216901

	expr	min	lq	mean	median	uq	max
2	.colMeans	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
3	colMeans	1.144201	1.166743	1.183383	1.178821	1.199427	1.027654
1	colMeans2	1.312507	1.321938	1.303984	1.298297	1.298806	1.017554
4	apply+mean	16.104979	15.739946	15.260493	15.320047	15.058500	8.693078

Table: Benchmarking of rowMeans2(), .rowMeans(), rowMeans() and apply+mean() 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	0.012531	0.0130100	0.0135245	0.0134345	0.0137750	0.021325
2	.rowMeans	0.020336	0.0205390	0.0208271	0.0207940	0.0209900	0.025154
3	rowMeans	0.021891	0.0222575	0.0228277	0.0227350	0.0230475	0.029208
4	apply+mean	0.110166	0.1129025	0.1155217	0.1139625	0.1153240	0.196645

	expr	min	lq	mean	median	uq	max
1	rowMeans2	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	.rowMeans	1.622855	1.578709	1.539953	1.547806	1.523775	1.179555
3	rowMeans	1.746948	1.710799	1.687874	1.692285	1.673140	1.369660
4	apply+mean	8.791477	8.678132	8.541647	8.482824	8.371978	9.221336

Figure: Benchmarking of colMeans2(), .colMeans(), colMeans() and apply+mean() on double+1000x10 data as well as rowMeans2(), .rowMeans(), rowMeans() and apply+mean() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colMeans2() and rowMeans2() 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	12.415	12.9345	13.41682	13.1920	13.5985	25.389
2	rowMeans2	12.531	13.0100	13.52452	13.4345	13.7750	21.325

	expr	min	lq	mean	median	uq	max
1	colMeans2	1.000000	1.000000	1.000000	1.000000	1.000000	1.0000000
2	rowMeans2	1.009343	1.005837	1.008027	1.018382	1.012979	0.8399307

Figure: Benchmarking of colMeans2() and rowMeans2() on double+1000x10 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

10x1000 double matrix

> X <- data[["10x1000"]]
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3139910 167.7    5709258 305.0  5709258 305.0
Vcells 5816830  44.4   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colMeans2 = colMeans2(X, na.rm = FALSE), .colMeans = .colMeans(X, m = nrow(X), 
+     n = ncol(X), na.rm = FALSE), colMeans = colMeans(X, na.rm = FALSE), `apply+mean` = apply(X, MARGIN = 2L, 
+     FUN = mean, na.rm = FALSE), unit = "ms")
> X <- t(X)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3139901 167.7    5709258 305.0  5709258 305.0
Vcells 5826868  44.5   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowMeans2 = rowMeans2(X, na.rm = FALSE), .rowMeans = .rowMeans(X, m = nrow(X), 
+     n = ncol(X), na.rm = FALSE), rowMeans = rowMeans(X, na.rm = FALSE), `apply+mean` = apply(X, MARGIN = 1L, 
+     FUN = mean, na.rm = FALSE), unit = "ms")

Table: Benchmarking of colMeans2(), .colMeans(), colMeans() and apply+mean() 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
2	.colMeans	0.009119	0.0096180	0.0108982	0.0103445	0.0111785	0.020629
3	colMeans	0.010592	0.0115365	0.0132383	0.0128655	0.0143210	0.023064
1	colMeans2	0.012080	0.0125470	0.0143993	0.0133060	0.0143800	0.065356
4	apply+mean	3.233188	3.3239640	3.5124924	3.4266600	3.5333660	8.546768

	expr	min	lq	mean	median	uq	max
2	.colMeans	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
3	colMeans	1.161531	1.199470	1.214727	1.243704	1.281120	1.118038
1	colMeans2	1.324707	1.304533	1.321260	1.286287	1.286398	3.168161
4	apply+mean	354.555105	345.598253	322.300532	331.254290	316.085879	414.308401

Table: Benchmarking of rowMeans2(), .rowMeans(), rowMeans() and apply+mean() 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	0.012227	0.0127810	0.0146545	0.0137915	0.0147080	0.062302
2	.rowMeans	0.019566	0.0202355	0.0215739	0.0209840	0.0225900	0.026842
3	rowMeans	0.021130	0.0218835	0.0246065	0.0231680	0.0245015	0.077906
4	apply+mean	3.246822	3.3370775	3.5229248	3.4122410	3.5453170	8.447450

	expr	min	lq	mean	median	uq	max
1	rowMeans2	1.000000	1.000000	1.000000	1.000000	1.000000	1.0000000
2	.rowMeans	1.600229	1.583249	1.472164	1.521517	1.535899	0.4308369
3	rowMeans	1.728143	1.712190	1.679107	1.679875	1.665862	1.2504574
4	apply+mean	265.545269	261.096745	240.398345	247.416235	241.046845	135.5887451

Figure: Benchmarking of colMeans2(), .colMeans(), colMeans() and apply+mean() on double+10x1000 data as well as rowMeans2(), .rowMeans(), rowMeans() and apply+mean() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colMeans2() and rowMeans2() 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	colMeans2	12.080	12.547	14.39934	13.3060	14.380	65.356
2	rowMeans2	12.227	12.781	14.65453	13.7915	14.708	62.302

	expr	min	lq	mean	median	uq	max
1	colMeans2	1.000000	1.00000	1.000000	1.000000	1.000000	1.0000000
2	rowMeans2	1.012169	1.01865	1.017722	1.036487	1.022809	0.9532713

Figure: Benchmarking of colMeans2() and rowMeans2() on double+10x1000 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

100x1000 double matrix

> X <- data[["100x1000"]]
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3140130 167.8    5709258 305.0  5709258 305.0
Vcells 5818143  44.4   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colMeans2 = colMeans2(X, na.rm = FALSE), .colMeans = .colMeans(X, m = nrow(X), 
+     n = ncol(X), na.rm = FALSE), colMeans = colMeans(X, na.rm = FALSE), `apply+mean` = apply(X, MARGIN = 2L, 
+     FUN = mean, na.rm = FALSE), unit = "ms")
> X <- t(X)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3140124 167.8    5709258 305.0  5709258 305.0
Vcells 5918186  45.2   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowMeans2 = rowMeans2(X, na.rm = FALSE), .rowMeans = .rowMeans(X, m = nrow(X), 
+     n = ncol(X), na.rm = FALSE), rowMeans = rowMeans(X, na.rm = FALSE), `apply+mean` = apply(X, MARGIN = 1L, 
+     FUN = mean, na.rm = FALSE), unit = "ms")

Table: Benchmarking of colMeans2(), .colMeans(), colMeans() and apply+mean() 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
2	.colMeans	0.064567	0.065199	0.0720234	0.0684560	0.0778060	0.114816
3	colMeans	0.066360	0.067253	0.0743774	0.0703435	0.0786795	0.161419
1	colMeans2	0.100799	0.101367	0.1175601	0.1050505	0.1103315	0.726208
4	apply+mean	4.343545	4.405486	4.6693956	4.4575465	4.5427805	18.057693

	expr	min	lq	mean	median	uq	max
2	.colMeans	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
3	colMeans	1.027770	1.031504	1.032684	1.027572	1.011227	1.405893
1	colMeans2	1.561154	1.554732	1.632250	1.534570	1.418033	6.324972
4	apply+mean	67.271904	67.569840	64.831691	65.115498	58.385992	157.275058

Table: Benchmarking of rowMeans2(), .rowMeans(), rowMeans() and apply+mean() 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	0.107488	0.1080635	0.1176995	0.1089730	0.117304	0.241564
2	.rowMeans	0.180992	0.1825000	0.1893512	0.1835685	0.185773	0.251952
3	rowMeans	0.180206	0.1834630	0.1892023	0.1851485	0.188669	0.246761
4	apply+mean	3.915669	4.0004425	4.3035005	4.0729265	4.219132	17.113837

	expr	min	lq	mean	median	uq	max
1	rowMeans2	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	.rowMeans	1.683834	1.688822	1.608767	1.684532	1.583689	1.043003
3	rowMeans	1.676522	1.697733	1.607502	1.699031	1.608377	1.021514
4	apply+mean	36.428894	37.019368	36.563443	37.375556	35.967508	70.845975

Figure: Benchmarking of colMeans2(), .colMeans(), colMeans() and apply+mean() on double+100x1000 data as well as rowMeans2(), .rowMeans(), rowMeans() and apply+mean() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colMeans2() and rowMeans2() 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	100.799	101.3670	117.5601	105.0505	110.3315	726.208
2	rowMeans2	107.488	108.0635	117.6995	108.9730	117.3040	241.564

	expr	min	lq	mean	median	uq	max
1	colMeans2	1.00000	1.000000	1.000000	1.000000	1.000000	1.0000000
2	rowMeans2	1.06636	1.066062	1.001186	1.037339	1.063196	0.3326375

Figure: Benchmarking of colMeans2() and rowMeans2() on double+100x1000 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

1000x100 double matrix

> X <- data[["1000x100"]]
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3140363 167.8    5709258 305.0  5709258 305.0
Vcells 5818301  44.4   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colMeans2 = colMeans2(X, na.rm = FALSE), .colMeans = .colMeans(X, m = nrow(X), 
+     n = ncol(X), na.rm = FALSE), colMeans = colMeans(X, na.rm = FALSE), `apply+mean` = apply(X, MARGIN = 2L, 
+     FUN = mean, na.rm = FALSE), unit = "ms")
> X <- t(X)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3140357 167.8    5709258 305.0  5709258 305.0
Vcells 5918344  45.2   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowMeans2 = rowMeans2(X, na.rm = FALSE), .rowMeans = .rowMeans(X, m = nrow(X), 
+     n = ncol(X), na.rm = FALSE), rowMeans = rowMeans(X, na.rm = FALSE), `apply+mean` = apply(X, MARGIN = 1L, 
+     FUN = mean, na.rm = FALSE), unit = "ms")

Table: Benchmarking of colMeans2(), .colMeans(), colMeans() and apply+mean() 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
2	.colMeans	0.077399	0.0777785	0.0811696	0.0784920	0.0821365	0.117364
3	colMeans	0.079115	0.0797960	0.0836713	0.0808430	0.0838830	0.115651
1	colMeans2	0.104898	0.1053610	0.1102334	0.1063325	0.1139030	0.133721
4	apply+mean	0.938313	0.9515165	1.0568774	0.9737810	1.0373125	6.995496

	expr	min	lq	mean	median	uq	max
2	.colMeans	1.000000	1.000000	1.000000	1.000000	1.000000	1.0000000
3	colMeans	1.022171	1.025939	1.030821	1.029952	1.021263	0.9854044
1	colMeans2	1.355289	1.354629	1.358062	1.354692	1.386753	1.1393698
4	apply+mean	12.123064	12.233670	13.020611	12.406118	12.629130	59.6051259

Table: Benchmarking of rowMeans2(), .rowMeans(), rowMeans() and apply+mean() 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	0.107895	0.1088235	0.1138603	0.1095450	0.1119620	0.206995
2	.rowMeans	0.187725	0.1883625	0.1950287	0.1887395	0.1905115	0.328856
3	rowMeans	0.189587	0.1902130	0.1967597	0.1910290	0.1935570	0.320987
4	apply+mean	0.965276	0.9926910	1.1065505	1.0498105	1.0774335	6.995676

	expr	min	lq	mean	median	uq	max
1	rowMeans2	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2	.rowMeans	1.739886	1.730899	1.712878	1.722940	1.701573	1.588715
3	rowMeans	1.757143	1.747904	1.728080	1.743840	1.728774	1.550699
4	apply+mean	8.946439	9.122028	9.718496	9.583372	9.623207	33.796353

Figure: Benchmarking of colMeans2(), .colMeans(), colMeans() and apply+mean() on double+1000x100 data as well as rowMeans2(), .rowMeans(), rowMeans() and apply+mean() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colMeans2() and rowMeans2() 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	104.898	105.3610	110.2334	106.3325	113.903	133.721
2	rowMeans2	107.895	108.8235	113.8603	109.5450	111.962	206.995

	expr	min	lq	mean	median	uq	max
1	colMeans2	1.000000	1.000000	1.000000	1.000000	1.0000000	1.000000
2	rowMeans2	1.028571	1.032863	1.032902	1.030212	0.9829592	1.547962

Figure: Benchmarking of colMeans2() and rowMeans2() 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 28.82 secs.

Reproducibility

To reproduce this report, do:

html <- matrixStats:::benchmark('colMeans2')

---

Copyright Henrik Bengtsson. Last updated on 2019-09-10 20:44:05 (-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>