colRowDiffs

matrixStats: Benchmark report

---

colDiffs() and rowDiffs() benchmarks

This report benchmark the performance of colDiffs() and rowDiffs() against alternative methods.

Alternative methods

• apply() + diff()
• apply() + diff2()
• diff()

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 3118071 166.6    5709258 305.0  5709258 305.0
Vcells 5912815  45.2   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colDiffs = colDiffs(X), `apply+diff` = apply(X, MARGIN = 2L, FUN = diff), 
+     `apply+diff2` = apply(X, MARGIN = 2L, FUN = diff2), diff = diff(X), unit = "ms")
> X <- t(X)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3116682 166.5    5709258 305.0  5709258 305.0
Vcells 5909095  45.1   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowDiffs = rowDiffs(X), `apply+diff` = apply(X, MARGIN = 1L, FUN = diff), 
+     `apply+diff2` = apply(X, MARGIN = 1L, FUN = diff2), `diff + t` = diff(t(X)), unit = "ms")

Table: Benchmarking of colDiffs(), apply+diff(), apply+diff2() and diff() on integer+10x10 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs	0.001270	0.0018170	0.0021640	0.0020725	0.0022715	0.011992
4	diff	0.006379	0.0071075	0.0086773	0.0082010	0.0086415	0.069156
3	apply+diff2	0.033844	0.0344875	0.0359664	0.0348345	0.0353310	0.119466
2	apply+diff	0.067467	0.0683645	0.0703230	0.0691400	0.0697245	0.128498

	expr	min	lq	mean	median	uq	max
1	colDiffs	1.000000	1.000000	1.00000	1.000000	1.000000	1.000000
4	diff	5.022835	3.911668	4.00976	3.957057	3.804314	5.766845
3	apply+diff2	26.648819	18.980462	16.62004	16.807961	15.554039	9.962141
2	apply+diff	53.123622	37.624931	32.49617	33.360675	30.695356	10.715310

Table: Benchmarking of rowDiffs(), apply+diff(), apply+diff2() and diff + t() on integer+10x10 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	rowDiffs	0.001278	0.0017735	0.0020819	0.0020560	0.002277	0.007329
4	diff + t	0.009225	0.0100965	0.0115328	0.0115210	0.012146	0.029634
3	apply+diff2	0.033501	0.0343785	0.0363943	0.0349085	0.035424	0.139590
2	apply+diff	0.067405	0.0688215	0.0701873	0.0694390	0.070156	0.093201

	expr	min	lq	mean	median	uq	max
1	rowDiffs	1.00000	1.00000	1.000000	1.000000	1.000000	1.000000
4	diff + t	7.21831	5.69298	5.539625	5.603599	5.334212	4.043389
3	apply+diff2	26.21362	19.38455	17.481548	16.978842	15.557312	19.046255
2	apply+diff	52.74257	38.80547	33.713604	33.773833	30.810716	12.716742

Figure: Benchmarking of colDiffs(), apply+diff(), apply+diff2() and diff() on integer+10x10 data as well as rowDiffs(), apply+diff(), apply+diff2() and diff + t() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colDiffs() and rowDiffs() 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	rowDiffs	1.278	1.7735	2.08187	2.0560	2.2770	7.329
1	colDiffs	1.270	1.8170	2.16404	2.0725	2.2715	11.992

	expr	min	lq	mean	median	uq	max
2	rowDiffs	1.0000000	1.000000	1.000000	1.000000	1.0000000	1.00000
1	colDiffs	0.9937402	1.024528	1.039469	1.008025	0.9975845	1.63624

Figure: Benchmarking of colDiffs() and rowDiffs() 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 3115258 166.4    5709258 305.0  5709258 305.0
Vcells 5525911  42.2   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colDiffs = colDiffs(X), `apply+diff` = apply(X, MARGIN = 2L, FUN = diff), 
+     `apply+diff2` = apply(X, MARGIN = 2L, FUN = diff2), diff = diff(X), unit = "ms")
> X <- t(X)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3115249 166.4    5709258 305.0  5709258 305.0
Vcells 5530949  42.2   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowDiffs = rowDiffs(X), `apply+diff` = apply(X, MARGIN = 1L, FUN = diff), 
+     `apply+diff2` = apply(X, MARGIN = 1L, FUN = diff2), `diff + t` = diff(t(X)), unit = "ms")

Table: Benchmarking of colDiffs(), apply+diff(), apply+diff2() and diff() on integer+100x100 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs	0.009920	0.0109060	0.0119752	0.0114910	0.0118260	0.040385
4	diff	0.080247	0.0821780	0.0854313	0.0834555	0.0842980	0.137403
3	apply+diff2	0.241004	0.2459205	0.2529544	0.2478870	0.2517745	0.418276
2	apply+diff	0.675984	0.6930305	0.7169101	0.7018950	0.7154390	1.138782

	expr	min	lq	mean	median	uq	max
1	colDiffs	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
4	diff	8.089415	7.535118	7.134028	7.262684	7.128192	3.402328
3	apply+diff2	24.294758	22.549101	21.123206	21.572274	21.289912	10.357212
2	apply+diff	68.143548	63.545800	59.866280	61.082151	60.497125	28.198143

Table: Benchmarking of rowDiffs(), apply+diff(), apply+diff2() and diff + t() on integer+100x100 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	rowDiffs	0.009978	0.0108210	0.0112527	0.0111070	0.0114755	0.019667
4	diff + t	0.095520	0.0974725	0.0990526	0.0986085	0.0997935	0.118905
3	apply+diff2	0.242721	0.2464115	0.2495150	0.2485575	0.2513510	0.295441
2	apply+diff	0.684756	0.6932470	0.7055857	0.6986960	0.7083670	1.001963

	expr	min	lq	mean	median	uq	max
1	rowDiffs	1.000000	1.000000	1.000000	1.00000	1.000000	1.000000
4	diff + t	9.573061	9.007717	8.802592	8.87805	8.696222	6.045915
3	apply+diff2	24.325616	22.771601	22.173864	22.37845	21.903272	15.022169
2	apply+diff	68.626578	64.064966	62.703900	62.90592	61.728639	50.946408

Figure: Benchmarking of colDiffs(), apply+diff(), apply+diff2() and diff() on integer+100x100 data as well as rowDiffs(), apply+diff(), apply+diff2() and diff + t() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

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

	expr	min	lq	mean	median	uq	max
2	rowDiffs	9.978	10.821	11.25266	11.107	11.4755	19.667
1	colDiffs	9.920	10.906	11.97519	11.491	11.8260	40.385

	expr	min	lq	mean	median	uq	max
2	rowDiffs	1.0000000	1.000000	1.00000	1.000000	1.000000	1.00000
1	colDiffs	0.9941872	1.007855	1.06421	1.034573	1.030543	2.05344

Figure: Benchmarking of colDiffs() and rowDiffs() 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 3116021 166.5    5709258 305.0  5709258 305.0
Vcells 5529701  42.2   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colDiffs = colDiffs(X), `apply+diff` = apply(X, MARGIN = 2L, FUN = diff), 
+     `apply+diff2` = apply(X, MARGIN = 2L, FUN = diff2), diff = diff(X), unit = "ms")
> X <- t(X)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3116015 166.5    5709258 305.0  5709258 305.0
Vcells 5534744  42.3   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowDiffs = rowDiffs(X), `apply+diff` = apply(X, MARGIN = 1L, FUN = diff), 
+     `apply+diff2` = apply(X, MARGIN = 1L, FUN = diff2), `diff + t` = diff(t(X)), unit = "ms")

Table: Benchmarking of colDiffs(), apply+diff(), apply+diff2() and diff() on integer+1000x10 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs	0.009307	0.0100455	0.0109699	0.0106900	0.0109750	0.036485
4	diff	0.082318	0.0840605	0.0856559	0.0849475	0.0860415	0.127109
3	apply+diff2	0.100588	0.1053980	0.1092378	0.1075710	0.1112395	0.154065
2	apply+diff	0.227661	0.2324470	0.2486088	0.2377635	0.2532850	0.521992

	expr	min	lq	mean	median	uq	max
1	colDiffs	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
4	diff	8.844741	8.367976	7.808281	7.946445	7.839772	3.483870
3	apply+diff2	10.807779	10.492061	9.957978	10.062769	10.135718	4.222694
2	apply+diff	24.461266	23.139416	22.662851	22.241674	23.078360	14.307030

Table: Benchmarking of rowDiffs(), apply+diff(), apply+diff2() and diff + t() on integer+1000x10 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	rowDiffs	0.009678	0.0105085	0.0110044	0.0110370	0.0112580	0.015880
4	diff + t	0.096710	0.0994460	0.1012975	0.1001915	0.1014715	0.181647
3	apply+diff2	0.102084	0.1054135	0.1094620	0.1084810	0.1118010	0.137666
2	apply+diff	0.229195	0.2330665	0.2446532	0.2421940	0.2529835	0.311863

	expr	min	lq	mean	median	uq	max
1	rowDiffs	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
4	diff + t	9.992767	9.463387	9.205176	9.077784	9.013279	11.438728
3	apply+diff2	10.548047	10.031260	9.947104	9.828848	9.930805	8.669144
2	apply+diff	23.682062	22.178855	22.232288	21.943825	22.471442	19.638728

Figure: Benchmarking of colDiffs(), apply+diff(), apply+diff2() and diff() on integer+1000x10 data as well as rowDiffs(), apply+diff(), apply+diff2() and diff + t() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

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

	expr	min	lq	mean	median	uq	max
1	colDiffs	9.307	10.0455	10.96988	10.690	10.975	36.485
2	rowDiffs	9.678	10.5085	11.00441	11.037	11.258	15.880

	expr	min	lq	mean	median	uq	max
1	colDiffs	1.000000	1.00000	1.000000	1.00000	1.000000	1.0000000
2	rowDiffs	1.039862	1.04609	1.003148	1.03246	1.025786	0.4352474

Figure: Benchmarking of colDiffs() and rowDiffs() 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 3116246 166.5    5709258 305.0  5709258 305.0
Vcells 5530519  42.2   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colDiffs = colDiffs(X), `apply+diff` = apply(X, MARGIN = 2L, FUN = diff), 
+     `apply+diff2` = apply(X, MARGIN = 2L, FUN = diff2), diff = diff(X), unit = "ms")
> X <- t(X)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3116237 166.5    5709258 305.0  5709258 305.0
Vcells 5535557  42.3   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowDiffs = rowDiffs(X), `apply+diff` = apply(X, MARGIN = 1L, FUN = diff), 
+     `apply+diff2` = apply(X, MARGIN = 1L, FUN = diff2), `diff + t` = diff(t(X)), unit = "ms")

Table: Benchmarking of colDiffs(), apply+diff(), apply+diff2() and diff() on integer+10x1000 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs	0.009235	0.0105635	0.0118427	0.011557	0.012477	0.019136
4	diff	0.077194	0.0802685	0.0858791	0.083122	0.085378	0.143532
3	apply+diff2	1.591354	1.6218525	1.8159368	1.654945	1.714163	6.441504
2	apply+diff	4.965706	5.1157700	5.3844286	5.245360	5.439525	10.323822

	expr	min	lq	mean	median	uq	max
1	colDiffs	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
4	diff	8.358852	7.598665	7.251652	7.192351	6.842831	7.500627
3	apply+diff2	172.317704	153.533630	153.338072	143.198451	137.385790	336.617057
2	apply+diff	537.705035	484.287405	454.662244	453.868608	435.964174	539.497387

Table: Benchmarking of rowDiffs(), apply+diff(), apply+diff2() and diff + t() on integer+10x1000 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	rowDiffs	0.008907	0.0099930	0.0113299	0.0109580	0.0121565	0.017723
4	diff + t	0.091877	0.0947805	0.1032366	0.0990875	0.1057695	0.175348
3	apply+diff2	1.585711	1.6161850	1.6896580	1.6660525	1.7329050	1.980962
2	apply+diff	4.929119	5.0946145	5.4897445	5.2661250	5.4027555	10.473003

	expr	min	lq	mean	median	uq	max
1	rowDiffs	1.00000	1.000000	1.000000	1.00000	1.000000	1.00000
4	diff + t	10.31515	9.484689	9.111847	9.04248	8.700654	9.89381
3	apply+diff2	178.02975	161.731712	149.132254	152.03983	142.549665	111.77351
2	apply+diff	553.39834	509.818323	484.534720	480.57355	444.433472	590.92721

Figure: Benchmarking of colDiffs(), apply+diff(), apply+diff2() and diff() on integer+10x1000 data as well as rowDiffs(), apply+diff(), apply+diff2() and diff + t() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

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

	expr	min	lq	mean	median	uq	max
2	rowDiffs	8.907	9.9930	11.32993	10.958	12.1565	17.723
1	colDiffs	9.235	10.5635	11.84270	11.557	12.4770	19.136

	expr	min	lq	mean	median	uq	max
2	rowDiffs	1.000000	1.00000	1.000000	1.000000	1.000000	1.000000
1	colDiffs	1.036825	1.05709	1.045258	1.054663	1.026364	1.079727

Figure: Benchmarking of colDiffs() and rowDiffs() 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 3116469 166.5    5709258 305.0  5709258 305.0
Vcells 5531104  42.2   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colDiffs = colDiffs(X), `apply+diff` = apply(X, MARGIN = 2L, FUN = diff), 
+     `apply+diff2` = apply(X, MARGIN = 2L, FUN = diff2), diff = diff(X), unit = "ms")
> X <- t(X)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3116460 166.5    5709258 305.0  5709258 305.0
Vcells 5581142  42.6   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowDiffs = rowDiffs(X), `apply+diff` = apply(X, MARGIN = 1L, FUN = diff), 
+     `apply+diff2` = apply(X, MARGIN = 1L, FUN = diff2), `diff + t` = diff(t(X)), unit = "ms")

Table: Benchmarking of colDiffs(), apply+diff(), apply+diff2() and diff() on integer+100x1000 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs	0.087169	0.0925755	0.0985038	0.096522	0.1007805	0.21958
4	diff	0.742494	0.7553605	1.0954965	0.777774	0.8358645	15.65912
3	apply+diff2	2.240516	2.3055240	2.5490086	2.351670	2.4236410	16.39443
2	apply+diff	6.619139	6.8144420	7.5455161	6.957101	7.2693915	22.61019

	expr	min	lq	mean	median	uq	max
1	colDiffs	1.000000	1.00000	1.00000	1.000000	1.000000	1.00000
4	diff	8.517868	8.15940	11.12136	8.057997	8.293911	71.31398
3	apply+diff2	25.703128	24.90426	25.87725	24.364088	24.048710	74.66268
2	apply+diff	75.934552	73.60956	76.60125	72.077873	72.130933	102.97016

Table: Benchmarking of rowDiffs(), apply+diff(), apply+diff2() and diff + t() on integer+100x1000 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	rowDiffs	0.080714	0.085084	0.2295152	0.088314	0.0927070	13.38035
4	diff + t	0.856953	0.890669	1.2247585	0.915878	0.9646175	15.19377
3	apply+diff2	2.246759	2.315821	2.7450913	2.377005	2.4885400	17.32280
2	apply+diff	6.627130	6.754207	9.8319702	6.929648	7.4608230	252.47289

	expr	min	lq	mean	median	uq	max
1	rowDiffs	1.00000	1.00000	1.000000	1.00000	1.00000	1.000000
4	diff + t	10.61715	10.46811	5.336284	10.37070	10.40501	1.135528
3	apply+diff2	27.83605	27.21805	11.960389	26.91538	26.84306	1.294645
2	apply+diff	82.10633	79.38281	42.837988	78.46602	80.47745	18.868932

Figure: Benchmarking of colDiffs(), apply+diff(), apply+diff2() and diff() on integer+100x1000 data as well as rowDiffs(), apply+diff(), apply+diff2() and diff + t() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

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

	expr	min	lq	mean	median	uq	max
2	rowDiffs	80.714	85.0840	229.51522	88.314	92.7070	13380.35
1	colDiffs	87.169	92.5755	98.50383	96.522	100.7805	219.58

	expr	min	lq	mean	median	uq	max
2	rowDiffs	1.000000	1.000000	1.0000000	1.000000	1.000000	1.0000000
1	colDiffs	1.079974	1.088048	0.4291821	1.092941	1.087086	0.0164106

Figure: Benchmarking of colDiffs() and rowDiffs() 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 3116702 166.5    5709258 305.0  5709258 305.0
Vcells 5531789  42.3   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colDiffs = colDiffs(X), `apply+diff` = apply(X, MARGIN = 2L, FUN = diff), 
+     `apply+diff2` = apply(X, MARGIN = 2L, FUN = diff2), diff = diff(X), unit = "ms")
> X <- t(X)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3116693 166.5    5709258 305.0  5709258 305.0
Vcells 5581827  42.6   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowDiffs = rowDiffs(X), `apply+diff` = apply(X, MARGIN = 1L, FUN = diff), 
+     `apply+diff2` = apply(X, MARGIN = 1L, FUN = diff2), `diff + t` = diff(t(X)), unit = "ms")

Table: Benchmarking of colDiffs(), apply+diff(), apply+diff2() and diff() on integer+1000x100 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs	0.081767	0.0857730	0.0929655	0.0910595	0.0988515	0.114236
4	diff	0.738713	0.7458715	0.8026512	0.7582425	0.8596720	1.034397
3	apply+diff2	0.837349	0.8475045	0.9754013	0.8719135	0.9433095	7.770486
2	apply+diff	2.097026	2.1284360	2.6722813	2.2429140	2.5575960	9.666034

	expr	min	lq	mean	median	uq	max
1	colDiffs	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
4	diff	9.034366	8.695877	8.633864	8.326891	8.696601	9.054913
3	apply+diff2	10.240672	9.880784	10.492082	9.575206	9.542693	68.021342
2	apply+diff	25.646361	24.814755	28.744880	24.631301	25.873113	84.614605

Table: Benchmarking of rowDiffs(), apply+diff(), apply+diff2() and diff + t() on integer+1000x100 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	rowDiffs	0.086141	0.0880170	0.0940816	0.0900150	0.096867	0.128768
4	diff + t	0.871176	0.8827525	1.0105405	0.9042490	1.000484	7.358402
3	apply+diff2	0.870173	0.8794700	1.0184523	0.9163815	1.006802	7.316249
2	apply+diff	2.125665	2.1449170	2.6449411	2.1888180	2.536551	9.075185

	expr	min	lq	mean	median	uq	max
1	rowDiffs	1.00000	1.000000	1.00000	1.00000	1.00000	1.00000
4	diff + t	10.11337	10.029341	10.74111	10.04554	10.32842	57.14465
3	apply+diff2	10.10173	9.992047	10.82520	10.18032	10.39365	56.81729
2	apply+diff	24.67658	24.369349	28.11327	24.31615	26.18591	70.47702

Figure: Benchmarking of colDiffs(), apply+diff(), apply+diff2() and diff() on integer+1000x100 data as well as rowDiffs(), apply+diff(), apply+diff2() and diff + t() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

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

	expr	min	lq	mean	median	uq	max
2	rowDiffs	86.141	88.017	94.08158	90.0150	96.8670	128.768
1	colDiffs	81.767	85.773	92.96547	91.0595	98.8515	114.236

	expr	min	lq	mean	median	uq	max
2	rowDiffs	1.0000000	1.0000000	1.0000000	1.000000	1.000000	1.0000000
1	colDiffs	0.9492228	0.9745049	0.9881368	1.011604	1.020487	0.8871459

Figure: Benchmarking of colDiffs() and rowDiffs() 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 3116941 166.5    5709258 305.0  5709258 305.0
Vcells 5648432  43.1   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colDiffs = colDiffs(X), `apply+diff` = apply(X, MARGIN = 2L, FUN = diff), 
+     `apply+diff2` = apply(X, MARGIN = 2L, FUN = diff2), diff = diff(X), unit = "ms")
> X <- t(X)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3116923 166.5    5709258 305.0  5709258 305.0
Vcells 5648555  43.1   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowDiffs = rowDiffs(X), `apply+diff` = apply(X, MARGIN = 1L, FUN = diff), 
+     `apply+diff2` = apply(X, MARGIN = 1L, FUN = diff2), `diff + t` = diff(t(X)), unit = "ms")

Table: Benchmarking of colDiffs(), apply+diff(), apply+diff2() and diff() on double+10x10 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs	0.001332	0.0017685	0.0021764	0.002117	0.0023290	0.012753
4	diff	0.006683	0.0074415	0.0087978	0.008487	0.0088460	0.063720
3	apply+diff2	0.033904	0.0349200	0.0356623	0.035278	0.0356905	0.046555
2	apply+diff	0.068321	0.0694755	0.0710144	0.069875	0.0705875	0.130362

	expr	min	lq	mean	median	uq	max
1	colDiffs	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
4	diff	5.017267	4.207803	4.042382	4.008975	3.798197	4.996471
3	apply+diff2	25.453453	19.745547	16.386045	16.664147	15.324388	3.650514
2	apply+diff	51.292042	39.284987	32.629591	33.006613	30.308072	10.222065

Table: Benchmarking of rowDiffs(), apply+diff(), apply+diff2() and diff + t() on double+10x10 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	rowDiffs	0.001196	0.0016925	0.0020263	0.0020565	0.0022110	0.006898
4	diff + t	0.008912	0.0101180	0.0112274	0.0113460	0.0118790	0.028226
3	apply+diff2	0.033201	0.0339955	0.0352648	0.0343690	0.0348130	0.109465
2	apply+diff	0.067129	0.0680925	0.0695140	0.0686705	0.0695905	0.092455

	expr	min	lq	mean	median	uq	max
1	rowDiffs	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
4	diff + t	7.451505	5.978139	5.540714	5.517141	5.372682	4.091911
3	apply+diff2	27.760033	20.085967	17.403200	16.712375	15.745364	15.869093
2	apply+diff	56.127926	40.231906	34.305186	33.391928	31.474672	13.403160

Figure: Benchmarking of colDiffs(), apply+diff(), apply+diff2() and diff() on double+10x10 data as well as rowDiffs(), apply+diff(), apply+diff2() and diff + t() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

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

	expr	min	lq	mean	median	uq	max
2	rowDiffs	1.196	1.6925	2.02634	2.0565	2.211	6.898
1	colDiffs	1.332	1.7685	2.17638	2.1170	2.329	12.753

	expr	min	lq	mean	median	uq	max
2	rowDiffs	1.000000	1.000000	1.000000	1.000000	1.00000	1.000000
1	colDiffs	1.113712	1.044904	1.074045	1.029419	1.05337	1.848797

Figure: Benchmarking of colDiffs() and rowDiffs() 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 3117158 166.5    5709258 305.0  5709258 305.0
Vcells 5648569  43.1   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colDiffs = colDiffs(X), `apply+diff` = apply(X, MARGIN = 2L, FUN = diff), 
+     `apply+diff2` = apply(X, MARGIN = 2L, FUN = diff2), diff = diff(X), unit = "ms")
> X <- t(X)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3117149 166.5    5709258 305.0  5709258 305.0
Vcells 5658607  43.2   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowDiffs = rowDiffs(X), `apply+diff` = apply(X, MARGIN = 1L, FUN = diff), 
+     `apply+diff2` = apply(X, MARGIN = 1L, FUN = diff2), `diff + t` = diff(t(X)), unit = "ms")

Table: Benchmarking of colDiffs(), apply+diff(), apply+diff2() and diff() on double+100x100 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs	0.007826	0.0089330	0.0109103	0.0093670	0.0101715	0.105528
4	diff	0.072652	0.0745505	0.0822667	0.0759545	0.0807860	0.293838
3	apply+diff2	0.281168	0.2854040	0.2985432	0.2890935	0.2951325	0.475516
2	apply+diff	0.683353	0.6971835	0.8540302	0.7070430	0.7260250	12.893743

	expr	min	lq	mean	median	uq	max
1	colDiffs	1.000000	1.000000	1.00000	1.000000	1.000000	1.000000
4	diff	9.283414	8.345517	7.54030	8.108733	7.942388	2.784455
3	apply+diff2	35.927421	31.949401	27.36350	30.862976	29.015632	4.506065
2	apply+diff	87.318298	78.045841	78.27764	75.482332	71.378361	122.183146

Table: Benchmarking of rowDiffs(), apply+diff(), apply+diff2() and diff + t() on double+100x100 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	rowDiffs	0.007596	0.0088750	0.0105745	0.0094270	0.0118065	0.022192
4	diff + t	0.053961	0.0564360	0.2286574	0.0583420	0.0693790	16.516990
3	apply+diff2	0.240055	0.2433520	0.2666842	0.2464280	0.2586960	0.449597
2	apply+diff	0.642614	0.6501085	0.6950446	0.6604355	0.6884510	1.091673

	expr	min	lq	mean	median	uq	max
1	rowDiffs	1.000000	1.000000	1.00000	1.000000	1.000000	1.00000
4	diff + t	7.103871	6.358986	21.62343	6.188819	5.876339	744.27677
3	apply+diff2	31.602817	27.419944	25.21951	26.140660	21.911320	20.25942
2	apply+diff	84.599000	73.251662	65.72824	70.057866	58.311185	49.19219

Figure: Benchmarking of colDiffs(), apply+diff(), apply+diff2() and diff() on double+100x100 data as well as rowDiffs(), apply+diff(), apply+diff2() and diff + t() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

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

	expr	min	lq	mean	median	uq	max
1	colDiffs	7.826	8.933	10.91027	9.367	10.1715	105.528
2	rowDiffs	7.596	8.875	10.57452	9.427	11.8065	22.192

	expr	min	lq	mean	median	uq	max
1	colDiffs	1.0000000	1.0000000	1.0000000	1.000000	1.000000	1.0000000
2	rowDiffs	0.9706108	0.9935072	0.9692262	1.006406	1.160743	0.2102949

Figure: Benchmarking of colDiffs() and rowDiffs() 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 3117384 166.5    5709258 305.0  5709258 305.0
Vcells 5649631  43.2   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colDiffs = colDiffs(X), `apply+diff` = apply(X, MARGIN = 2L, FUN = diff), 
+     `apply+diff2` = apply(X, MARGIN = 2L, FUN = diff2), diff = diff(X), unit = "ms")
> X <- t(X)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3117375 166.5    5709258 305.0  5709258 305.0
Vcells 5659669  43.2   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowDiffs = rowDiffs(X), `apply+diff` = apply(X, MARGIN = 1L, FUN = diff), 
+     `apply+diff2` = apply(X, MARGIN = 1L, FUN = diff2), `diff + t` = diff(t(X)), unit = "ms")

Table: Benchmarking of colDiffs(), apply+diff(), apply+diff2() and diff() on double+1000x10 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs	0.007905	0.0086945	0.0095011	0.0090655	0.0097005	0.021953
4	diff	0.073266	0.0750120	0.0781225	0.0760670	0.0769490	0.111926
3	apply+diff2	0.145058	0.1481450	0.1536575	0.1499690	0.1534765	0.201126
2	apply+diff	0.230347	0.2374750	0.3072987	0.2439350	0.2505075	5.969996

	expr	min	lq	mean	median	uq	max
1	colDiffs	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
4	diff	9.268311	8.627523	8.222441	8.390822	7.932478	5.098438
3	apply+diff2	18.350158	17.038933	16.172552	16.542827	15.821504	9.161664
2	apply+diff	29.139405	27.313244	32.343387	26.908058	25.824184	271.944427

Table: Benchmarking of rowDiffs(), apply+diff(), apply+diff2() and diff + t() on double+1000x10 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	rowDiffs	0.007774	0.0087175	0.0093711	0.009075	0.0095330	0.018593
4	diff + t	0.054855	0.0570195	0.1176786	0.058473	0.0606920	5.704849
3	apply+diff2	0.104058	0.1081360	0.1146844	0.110227	0.1172655	0.178295
2	apply+diff	0.191114	0.1957605	0.2085546	0.205078	0.2111210	0.291587

	expr	min	lq	mean	median	uq	max
1	rowDiffs	1.000000	1.000000	1.00000	1.000000	1.000000	1.000000
4	diff + t	7.056213	6.540809	12.55755	6.443306	6.366516	306.827785
3	apply+diff2	13.385387	12.404474	12.23804	12.146226	12.301007	9.589362
2	apply+diff	24.583741	22.456037	22.25499	22.598127	22.146334	15.682623

Figure: Benchmarking of colDiffs(), apply+diff(), apply+diff2() and diff() on double+1000x10 data as well as rowDiffs(), apply+diff(), apply+diff2() and diff + t() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colDiffs() and rowDiffs() 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	colDiffs	7.905	8.6945	9.50113	9.0655	9.7005	21.953
2	rowDiffs	7.774	8.7175	9.37114	9.0750	9.5330	18.593

	expr	min	lq	mean	median	uq	max
1	colDiffs	1.0000000	1.000000	1.0000000	1.000000	1.0000000	1.0000000
2	rowDiffs	0.9834282	1.002645	0.9863185	1.001048	0.9827328	0.8469457

Figure: Benchmarking of colDiffs() and rowDiffs() 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 3117606 166.5    5709258 305.0  5709258 305.0
Vcells 5650870  43.2   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colDiffs = colDiffs(X), `apply+diff` = apply(X, MARGIN = 2L, FUN = diff), 
+     `apply+diff2` = apply(X, MARGIN = 2L, FUN = diff2), diff = diff(X), unit = "ms")
> X <- t(X)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3117597 166.5    5709258 305.0  5709258 305.0
Vcells 5660908  43.2   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowDiffs = rowDiffs(X), `apply+diff` = apply(X, MARGIN = 1L, FUN = diff), 
+     `apply+diff2` = apply(X, MARGIN = 1L, FUN = diff2), `diff + t` = diff(t(X)), unit = "ms")

Table: Benchmarking of colDiffs(), apply+diff(), apply+diff2() and diff() on double+10x1000 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs	0.007948	0.0106020	0.0129899	0.0123225	0.013964	0.029984
4	diff	0.039178	0.0426205	0.0475171	0.0455745	0.048588	0.090872
3	apply+diff2	1.579078	1.6598810	1.8660162	1.7064235	1.770217	6.770882
2	apply+diff	4.886083	5.3134105	5.5818545	5.5080760	5.673311	10.608300

	expr	min	lq	mean	median	uq	max
1	colDiffs	1.00000	1.000000	1.000000	1.000000	1.000000	1.000000
4	diff	4.92929	4.020043	3.658018	3.698478	3.479519	3.030683
3	apply+diff2	198.67614	156.563007	143.651758	138.480300	126.770087	225.816502
2	apply+diff	614.75629	501.170581	429.708594	446.993386	406.281223	353.798693

Table: Benchmarking of rowDiffs(), apply+diff(), apply+diff2() and diff + t() on double+10x1000 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	rowDiffs	0.007831	0.0105750	0.0123780	0.0119180	0.0135355	0.023236
4	diff + t	0.054646	0.0583525	0.0655831	0.0626105	0.0692040	0.140861
3	apply+diff2	1.576733	1.6441280	1.7074512	1.6826055	1.7476215	1.966085
2	apply+diff	4.970668	5.2324175	5.6254805	5.3840850	5.6055665	11.002043

	expr	min	lq	mean	median	uq	max
1	rowDiffs	1.000000	1.000000	1.000000	1.00000	1.000000	1.000000
4	diff + t	6.978164	5.517967	5.298349	5.25344	5.112777	6.062188
3	apply+diff2	201.345039	155.473097	137.942194	141.18187	129.113923	84.613746
2	apply+diff	634.742434	494.791253	454.473370	451.76078	414.138118	473.491264

Figure: Benchmarking of colDiffs(), apply+diff(), apply+diff2() and diff() on double+10x1000 data as well as rowDiffs(), apply+diff(), apply+diff2() and diff + t() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colDiffs() and rowDiffs() 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	rowDiffs	7.831	10.575	12.37802	11.9180	13.5355	23.236
1	colDiffs	7.948	10.602	12.98986	12.3225	13.9640	29.984

	expr	min	lq	mean	median	uq	max
2	rowDiffs	1.000000	1.000000	1.00000	1.00000	1.000000	1.000000
1	colDiffs	1.014941	1.002553	1.04943	1.03394	1.031657	1.290411

Figure: Benchmarking of colDiffs() and rowDiffs() 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 3117829 166.6    5709258 305.0  5709258 305.0
Vcells 5651018  43.2   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colDiffs = colDiffs(X), `apply+diff` = apply(X, MARGIN = 2L, FUN = diff), 
+     `apply+diff2` = apply(X, MARGIN = 2L, FUN = diff2), diff = diff(X), unit = "ms")
> X <- t(X)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3117820 166.6    5709258 305.0  5709258 305.0
Vcells 5751056  43.9   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowDiffs = rowDiffs(X), `apply+diff` = apply(X, MARGIN = 1L, FUN = diff), 
+     `apply+diff2` = apply(X, MARGIN = 1L, FUN = diff2), `diff + t` = diff(t(X)), unit = "ms")

Table: Benchmarking of colDiffs(), apply+diff(), apply+diff2() and diff() on double+100x1000 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs	0.068753	0.0809590	0.2186692	0.0936815	0.1062405	11.97728
4	diff	0.671257	0.6902585	0.8713896	0.7059570	0.8116290	12.64426
3	apply+diff2	2.657512	2.7096860	3.3058297	2.7729975	3.2179400	15.45456
2	apply+diff	6.709998	6.8670360	8.1240188	7.1100785	7.9345530	20.51747

	expr	min	lq	mean	median	uq	max
1	colDiffs	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
4	diff	9.763312	8.526025	3.984967	7.535714	7.639544	1.055687
3	apply+diff2	38.653033	33.469855	15.117947	29.600268	30.289202	1.290323
2	apply+diff	97.595712	84.821156	37.152091	75.896292	74.684824	1.713032

Table: Benchmarking of rowDiffs(), apply+diff(), apply+diff2() and diff + t() on double+100x1000 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	rowDiffs	0.068515	0.0820635	0.1000359	0.091831	0.101646	0.266283
4	diff + t	0.454952	0.5172565	0.6786516	0.546358	0.598187	12.250272
3	apply+diff2	2.284115	2.3557905	3.1236122	2.572386	2.977993	15.632607
2	apply+diff	6.343725	6.5136880	8.2095237	7.187853	8.147941	22.801852

	expr	min	lq	mean	median	uq	max
1	rowDiffs	1.000000	1.000000	1.00000	1.000000	1.000000	1.00000
4	diff + t	6.640181	6.303125	6.78408	5.949603	5.885003	46.00471
3	apply+diff2	33.337444	28.706922	31.22491	28.012169	29.297690	58.70674
2	apply+diff	92.588849	79.373753	82.06578	78.272620	80.159972	85.63015

Figure: Benchmarking of colDiffs(), apply+diff(), apply+diff2() and diff() on double+100x1000 data as well as rowDiffs(), apply+diff(), apply+diff2() and diff + t() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

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

	expr	min	lq	mean	median	uq	max
2	rowDiffs	68.515	82.0635	100.0359	91.8310	101.6460	266.283
1	colDiffs	68.753	80.9590	218.6692	93.6815	106.2405	11977.281

	expr	min	lq	mean	median	uq	max
2	rowDiffs	1.000000	1.0000000	1.000000	1.000000	1.000000	1.00000
1	colDiffs	1.003474	0.9865409	2.185907	1.020151	1.045201	44.97952

Figure: Benchmarking of colDiffs() and rowDiffs() 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 3118062 166.6    5709258 305.0  5709258 305.0
Vcells 5652492  43.2   22267496 169.9 56666022 432.4
> colStats <- microbenchmark(colDiffs = colDiffs(X), `apply+diff` = apply(X, MARGIN = 2L, FUN = diff), 
+     `apply+diff2` = apply(X, MARGIN = 2L, FUN = diff2), diff = diff(X), unit = "ms")
> X <- t(X)
> gc()
          used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells 3118053 166.6    5709258 305.0  5709258 305.0
Vcells 5752530  43.9   22267496 169.9 56666022 432.4
> rowStats <- microbenchmark(rowDiffs = rowDiffs(X), `apply+diff` = apply(X, MARGIN = 1L, FUN = diff), 
+     `apply+diff2` = apply(X, MARGIN = 1L, FUN = diff2), `diff + t` = diff(t(X)), unit = "ms")

Table: Benchmarking of colDiffs(), apply+diff(), apply+diff2() and diff() on double+1000x100 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	colDiffs	0.067611	0.0821295	0.0927182	0.088574	0.1024565	0.134425
4	diff	0.341218	0.3933645	0.5556137	0.421854	0.4479860	7.588598
3	apply+diff2	0.891035	1.0466215	1.2835500	1.109388	1.1563550	7.021451
2	apply+diff	1.739943	2.1044035	2.6775548	2.251618	2.4016450	14.261253

	expr	min	lq	mean	median	uq	max
1	colDiffs	1.000000	1.000000	1.000000	1.000000	1.000000	1.00000
4	diff	5.046782	4.789564	5.992501	4.762729	4.372451	56.45228
3	apply+diff2	13.178847	12.743551	13.843564	12.524985	11.286302	52.23322
2	apply+diff	25.734614	25.622992	28.878426	25.420750	23.440631	106.09078

Table: Benchmarking of rowDiffs(), apply+diff(), apply+diff2() and diff + t() on double+1000x100 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	rowDiffs	0.069929	0.0835605	0.1677974	0.0925420	0.1084595	6.896821
4	diff + t	0.460530	0.5702410	0.6175363	0.5985205	0.6342225	1.395840
3	apply+diff2	0.902764	1.0965220	1.4181183	1.1716210	1.2517770	7.471189
2	apply+diff	1.802847	2.2327080	2.8208954	2.3852495	2.5657390	10.253641

	expr	min	lq	mean	median	uq	max
1	rowDiffs	1.00000	1.000000	1.000000	1.000000	1.000000	1.0000000
4	diff + t	6.58568	6.824289	3.680250	6.467555	5.847551	0.2023889
3	apply+diff2	12.90972	13.122492	8.451373	12.660424	11.541423	1.0832801
2	apply+diff	25.78111	26.719658	16.811320	25.774778	23.656194	1.4867199

Figure: Benchmarking of colDiffs(), apply+diff(), apply+diff2() and diff() on double+1000x100 data as well as rowDiffs(), apply+diff(), apply+diff2() and diff + t() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

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

	expr	min	lq	mean	median	uq	max
1	colDiffs	67.611	82.1295	92.71817	88.574	102.4565	134.425
2	rowDiffs	69.929	83.5605	167.79738	92.542	108.4595	6896.821

	expr	min	lq	mean	median	uq	max
1	colDiffs	1.000000	1.000000	1.000000	1.000000	1.000000	1.00000
2	rowDiffs	1.034284	1.017424	1.809757	1.044799	1.058591	51.30609

Figure: Benchmarking of colDiffs() and rowDiffs() 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 35.3 secs.

Reproducibility

To reproduce this report, do:

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

---

Copyright Henrik Bengtsson. Last updated on 2019-09-10 20:40:52 (-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>