meanOver

matrixStats: Benchmark report

meanOver() benchmarks

This report benchmark the performance of meanOver() against alternative methods.

Alternative methods

mean() + [()
mean.default() + [() - avoids method dispatching

as below

> meanOver_R_v1 <- function(x, na.rm = FALSE, idxs) {
+     mean(x[idxs], na.rm = na.rm)
+ }

and

> meanOver_R_v2 <- function(x, na.rm = FALSE, idxs) {
+     mean.default(x[idxs], na.rm = na.rm)
+ }

Data type "integer"

Data

> rvector <- function(n, mode = c("logical", "double", "integer"), range = c(-100, +100), naProb = 0) {
+     mode <- match.arg(mode)
+     if (mode == "logical") {
+         X <- sample(c(FALSE, TRUE), size = n, replace = TRUE)
+     }     else {
+         x <- runif(n, min = range[1], max = range[2])
+     }
+     storage.mode(x) <- mode
+     if (naProb > 0) 
+         x[sample(n, size = naProb * n)] <- NA
+     x
+ }
> rvectors <- function(scale = 10, seed = 1, ...) {
+     set.seed(seed)
+     data <- list()
+     data[[1]] <- rvector(n = scale * 100, ...)
+     data[[2]] <- rvector(n = scale * 1000, ...)
+     data[[3]] <- rvector(n = scale * 10000, ...)
+     data[[4]] <- rvector(n = scale * 1e+05, ...)
+     data[[5]] <- rvector(n = scale * 1e+06, ...)
+     names(data) <- sprintf("n=%d", sapply(data, FUN = length))
+     data
+ }
> data <- rvectors(mode = mode)

Results

n=1000 vector

All elements

> x <- data[["n=1000"]]
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755108  93.8    2637877 140.9  2637877 140.9
Vcells 18338081 140.0   42812957 326.7 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, refine = TRUE), meanOver_no_refine = meanOver(x, refine = FALSE), 
+     mean = mean(x), mean.default = mean.default(x), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean() and mean.default() on n=1000+all data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
4	mean.default	0.0042	0.0046	0.0052	0.0050	0.0054	0.0181
1	meanOver	0.0058	0.0065	0.0111	0.0069	0.0077	0.3083
2	meanOver_no_refine	0.0058	0.0065	0.0075	0.0069	0.0075	0.0281
3	mean	0.0119	0.0135	0.0172	0.0142	0.0152	0.0789

	expr	min	lq	mean	median	uq	max
4	mean.default	1.000	1.000	1.000	1.000	1.000	1.000
1	meanOver	1.364	1.417	2.109	1.385	1.429	17.042
2	meanOver_no_refine	1.364	1.417	1.431	1.385	1.393	1.553
3	mean	2.818	2.916	3.275	2.846	2.821	4.362
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean() and mean.default() on n=1000+all data. Outliers are displayed as crosses. Times are in milliseconds.

A 20% subset

> x <- data[["n=1000"]]
> subset
[1] 0.2
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755368  93.8    2637877 140.9  2637877 140.9
Vcells 18339882 140.0   42812957 326.7 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=1000+0.2 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	meanOver	0.0046	0.0062	0.0083	0.0069	0.0081	0.1036
2	meanOver_no_refine	0.0046	0.0064	0.0085	0.0077	0.0089	0.0689
4	mean.default+[()	0.0085	0.0100	0.0134	0.0112	0.0123	0.0731
3	mean+[()	0.0185	0.0216	0.0256	0.0243	0.0266	0.0778

	expr	min	lq	mean	median	uq	max
1	meanOver	1.000	1.000	1.000	1.000	1.000	1.0000
2	meanOver_no_refine	1.000	1.031	1.027	1.111	1.095	0.6654
4	mean.default+[()	1.833	1.625	1.613	1.611	1.524	0.7063
3	mean+[()	4.000	3.500	3.081	3.500	3.285	0.7509
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=1000+0.2 data. Outliers are displayed as crosses. Times are in milliseconds.

A 40% subset

> x <- data[["n=1000"]]
> subset
[1] 0.4
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755383  93.8    2637877 140.9  2637877 140.9
Vcells 18340510 140.0   42812957 326.7 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=1000+0.4 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	0.0050	0.0065	0.0073	0.0069	0.0079	0.0135
1	meanOver	0.0046	0.0065	0.0075	0.0073	0.0081	0.0289
4	mean.default+[()	0.0104	0.0123	0.0141	0.0127	0.0135	0.0724
3	mean+[()	0.0212	0.0227	0.0249	0.0239	0.0250	0.0712

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.0000	1.000	1.000	1.000	1.000	1.000
1	meanOver	0.9231	1.000	1.028	1.056	1.024	2.143
4	mean.default+[()	2.0769	1.882	1.927	1.833	1.707	5.371
3	mean+[()	4.2304	3.470	3.399	3.444	3.170	5.285
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=1000+0.4 data. Outliers are displayed as crosses. Times are in milliseconds.

A 80% subset

> x <- data[["n=1000"]]
> subset
[1] 0.8
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755395  93.8    2637877 140.9  2637877 140.9
Vcells 18340718 140.0   42812957 326.7 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=1000+0.8 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	0.0065	0.0077	0.0087	0.0085	0.0092	0.0289
1	meanOver	0.0065	0.0077	0.0089	0.0085	0.0092	0.0254
4	mean.default+[()	0.0154	0.0169	0.0189	0.0177	0.0187	0.0724
3	mean+[()	0.0239	0.0281	0.0300	0.0289	0.0308	0.0724

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.000	1.00	1.000	1.000	1.000	1.000
1	meanOver	1.000	1.00	1.023	1.000	1.000	0.880
4	mean.default+[()	2.353	2.20	2.181	2.091	2.021	2.507
3	mean+[()	3.647	3.65	3.456	3.409	3.333	2.507
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=1000+0.8 data. Outliers are displayed as crosses. Times are in milliseconds.

n=10000 vector

All elements

> x <- data[["n=10000"]]
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755342  93.8    2637877 140.9  2637877 140.9
Vcells 18340463 140.0   42812957 326.7 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, refine = TRUE), meanOver_no_refine = meanOver(x, refine = FALSE), 
+     mean = mean(x), mean.default = mean.default(x), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean() and mean.default() on n=10000+all data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
4	mean.default	0.0108	0.0116	0.0156	0.0119	0.0214	0.0504
1	meanOver	0.0150	0.0158	0.0195	0.0166	0.0243	0.0543
2	meanOver_no_refine	0.0150	0.0162	0.0190	0.0166	0.0243	0.0269
3	mean	0.0166	0.0177	0.0236	0.0185	0.0308	0.1436

	expr	min	lq	mean	median	uq	max
4	mean.default	1.000	1.000	1.000	1.000	1.000	1.0000
1	meanOver	1.393	1.367	1.252	1.387	1.135	1.0763
2	meanOver_no_refine	1.393	1.400	1.217	1.387	1.135	0.5344
3	mean	1.536	1.533	1.514	1.548	1.441	2.8473
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean() and mean.default() on n=10000+all data. Outliers are displayed as crosses. Times are in milliseconds.

A 20% subset

> x <- data[["n=10000"]]
> subset
[1] 0.2
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755419  93.8    2637877 140.9  2637877 140.9
Vcells 18341794 140.0   42812957 326.7 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=10000+0.2 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	meanOver	0.0062	0.0081	0.0115	0.0116	0.0135	0.0316
2	meanOver_no_refine	0.0065	0.0085	0.0128	0.0127	0.0146	0.0493
4	mean.default+[()	0.0177	0.0283	0.0383	0.0335	0.0354	0.4947
3	mean+[()	0.0258	0.0316	0.0636	0.0456	0.0485	1.5518

	expr	min	lq	mean	median	uq	max
1	meanOver	1.000	1.000	1.000	1.00	1.000	1.000
2	meanOver_no_refine	1.062	1.048	1.107	1.10	1.086	1.561
4	mean.default+[()	2.875	3.500	3.317	2.90	2.628	15.670
3	mean+[()	4.187	3.904	5.507	3.95	3.600	49.157
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=10000+0.2 data. Outliers are displayed as crosses. Times are in milliseconds.

A 40% subset

> x <- data[["n=10000"]]
> subset
[1] 0.4
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755431  93.8    2637877 140.9  2637877 140.9
Vcells 18343105 140.0   42812957 326.7 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=10000+0.4 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	0.0092	0.0108	0.0517	0.0115	0.0127	3.8580
1	meanOver	0.0096	0.0108	0.0133	0.0116	0.0131	0.0674
4	mean.default+[()	0.0320	0.0331	0.0372	0.0339	0.0350	0.1216
3	mean+[()	0.0397	0.0416	0.0505	0.0427	0.0443	0.2244

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.000	1.000	1.0000	1.000	1.000	1.0000
1	meanOver	1.042	1.000	0.2580	1.000	1.030	0.0175
4	mean.default+[()	3.458	3.071	0.7207	2.933	2.758	0.0315
3	mean+[()	4.291	3.857	0.9766	3.700	3.485	0.0582
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=10000+0.4 data. Outliers are displayed as crosses. Times are in milliseconds.

A 80% subset

> x <- data[["n=10000"]]
> subset
[1] 0.8
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755443  93.8    2637877 140.9  2637877 140.9
Vcells 18345477 140.0   42812957 326.7 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=10000+0.8 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	0.0162	0.0189	0.0262	0.0273	0.0316	0.0427
1	meanOver	0.0158	0.0198	0.0266	0.0279	0.0325	0.0427
4	mean.default+[()	0.0604	0.0637	0.0929	0.1039	0.1153	0.1574
3	mean+[()	0.0701	0.0728	0.1014	0.1072	0.1257	0.2156

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.0000	1.000	1.000	1.000	1.000	1.000
1	meanOver	0.9762	1.051	1.017	1.021	1.030	1.000
4	mean.default+[()	3.7379	3.377	3.547	3.803	3.652	3.685
3	mean+[()	4.3332	3.857	3.871	3.922	3.982	5.045
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=10000+0.8 data. Outliers are displayed as crosses. Times are in milliseconds.

n=100000 vector

All elements

> x <- data[["n=100000"]]
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755390  93.8    2637877 140.9  2637877 140.9
Vcells 18345009 140.0   42812957 326.7 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, refine = TRUE), meanOver_no_refine = meanOver(x, refine = FALSE), 
+     mean = mean(x), mean.default = mean.default(x), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean() and mean.default() on n=100000+all data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
4	mean.default	0.1024	0.1419	0.1695	0.1777	0.1952	0.5736
2	meanOver_no_refine	0.1363	0.1807	0.2007	0.2048	0.2181	0.3376
1	meanOver	0.1359	0.1804	0.1985	0.2048	0.2188	0.3149
3	mean	0.1113	0.1671	0.1884	0.2052	0.2146	0.2960

	expr	min	lq	mean	median	uq	max
4	mean.default	1.000	1.000	1.000	1.000	1.000	1.0000
2	meanOver_no_refine	1.331	1.274	1.184	1.153	1.117	0.5886
1	meanOver	1.327	1.271	1.171	1.153	1.121	0.5490
3	mean	1.087	1.178	1.111	1.155	1.100	0.5161
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean() and mean.default() on n=100000+all data. Outliers are displayed as crosses. Times are in milliseconds.

A 20% subset

> x <- data[["n=100000"]]
> subset
[1] 0.2
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755467  93.8    2637877 140.9  2637877 140.9
Vcells 18351928 140.1   42812957 326.7 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=100000+0.2 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

expr	min	lq	mean	median	uq	max
meanOver	0.0408	0.0458	0.0584	0.0535	0.0708	0.0997
meanOver_no_refine	0.0404	0.0458	0.0602	0.0599	0.0724	0.1213
mean+[()	0.1605	0.1680	0.2210	0.1971	0.2702	0.3141
mean.default+[()	0.1494	0.1536	0.2129	0.2254	0.2529	0.6625

expr	min	lq	mean	median	uq	max
meanOver	1.0000	1.000	1.000	1.000	1.000	1.000
meanOver_no_refine	0.9906	1.000	1.030	1.119	1.022	1.216
mean+[()	3.9339	3.668	3.783	3.683	3.815	3.151
mean.default+[()	3.6603	3.353	3.644	4.212	3.571	6.645
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=100000+0.2 data. Outliers are displayed as crosses. Times are in milliseconds.

A 40% subset

> x <- data[["n=100000"]]
> subset
[1] 0.4
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755479  93.8    2637877 140.9  2637877 140.9
Vcells 18361936 140.1   42812957 326.7 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=100000+0.4 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	meanOver	0.1116	0.1265	0.1393	0.1315	0.1432	0.2845
2	meanOver_no_refine	0.1093	0.1278	0.1559	0.1322	0.1444	0.8923
4	mean.default+[()	0.3961	0.4929	0.5282	0.4997	0.5174	1.3185
3	mean+[()	0.4820	0.5251	0.5558	0.5420	0.5605	0.7707

	expr	min	lq	mean	median	uq	max
1	meanOver	1.0000	1.000	1.000	1.000	1.000	1.000
2	meanOver_no_refine	0.9793	1.011	1.119	1.006	1.008	3.137
4	mean.default+[()	3.5483	3.898	3.792	3.801	3.613	4.635
3	mean+[()	4.3172	4.152	3.990	4.123	3.914	2.709
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=100000+0.4 data. Outliers are displayed as crosses. Times are in milliseconds.

A 80% subset

> x <- data[["n=100000"]]
> subset
[1] 0.8
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755491  93.8    2637877 140.9  2637877 140.9
Vcells 18382471 140.3   42812957 326.7 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=100000+0.8 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	meanOver	0.1405	0.1592	0.2278	0.1798	0.2816	0.5320
2	meanOver_no_refine	0.1386	0.1603	0.2157	0.1809	0.2668	0.6683
4	mean.default+[()	0.6017	0.6198	0.8459	0.7023	1.0120	2.7547
3	mean+[()	0.6202	0.6513	0.8479	0.7172	1.0742	1.5279

	expr	min	lq	mean	median	uq	max
1	meanOver	1.0000	1.000	1.0000	1.000	1.0000	1.000
2	meanOver_no_refine	0.9863	1.007	0.9469	1.006	0.9474	1.256
4	mean.default+[()	4.2822	3.894	3.7139	3.907	3.5940	5.178
3	mean+[()	4.4137	4.092	3.7229	3.989	3.8147	2.872
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=100000+0.8 data. Outliers are displayed as crosses. Times are in milliseconds.

n=1000000 vector

All elements

> x <- data[["n=1000000"]]
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755439  93.8    2637877 140.9  2637877 140.9
Vcells 18382011 140.3   42812957 326.7 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, refine = TRUE), meanOver_no_refine = meanOver(x, refine = FALSE), 
+     mean = mean(x), mean.default = mean.default(x), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean() and mean.default() on n=1000000+all data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
4	mean.default	1.183	1.262	1.851	2.019	2.113	6.343
3	mean	1.226	1.647	1.922	2.056	2.156	3.263
2	meanOver_no_refine	1.496	1.600	2.176	2.196	2.421	4.950
1	meanOver	1.501	1.811	2.146	2.210	2.445	3.594

	expr	min	lq	mean	median	uq	max
4	mean.default	1.000	1.000	1.000	1.000	1.000	1.0000
3	mean	1.037	1.304	1.038	1.019	1.020	0.5144
2	meanOver_no_refine	1.264	1.268	1.176	1.088	1.146	0.7805
1	meanOver	1.268	1.435	1.160	1.095	1.157	0.5666
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean() and mean.default() on n=1000000+all data. Outliers are displayed as crosses. Times are in milliseconds.

A 20% subset

> x <- data[["n=1000000"]]
> subset
[1] 0.2
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755516  93.8    2637877 140.9  2637877 140.9
Vcells 18443126 140.8   42812957 326.7 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=1000000+0.2 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	0.8946	1.097	1.298	1.191	1.294	3.228
1	meanOver	0.8176	1.068	1.254	1.195	1.279	3.391
4	mean.default+[()	2.0175	2.953	3.710	3.281	4.158	17.903
3	mean+[()	2.0418	3.133	3.712	3.471	4.202	9.454

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.0000	1.0000	1.0000	1.000	1.0000	1.000
1	meanOver	0.9139	0.9733	0.9662	1.003	0.9885	1.051
4	mean.default+[()	2.2552	2.6917	2.8579	2.755	3.2139	5.547
3	mean+[()	2.2823	2.8559	2.8590	2.915	3.2480	2.929
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=1000000+0.2 data. Outliers are displayed as crosses. Times are in milliseconds.

A 40% subset

> x <- data[["n=1000000"]]
> subset
[1] 0.4
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755528  93.8    2637877 140.9  2637877 140.9
Vcells 18543134 141.5   42812957 326.7 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=1000000+0.4 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	meanOver	1.048	1.404	1.616	1.562	1.753	4.473
2	meanOver_no_refine	1.059	1.377	1.649	1.594	1.768	4.703
4	mean.default+[()	3.602	5.168	6.149	5.928	6.573	22.382
3	mean+[()	3.592	5.265	6.269	6.066	6.901	22.139

	expr	min	lq	mean	median	uq	max
1	meanOver	1.000	1.0000	1.000	1.000	1.000	1.000
2	meanOver_no_refine	1.011	0.9807	1.021	1.020	1.008	1.051
4	mean.default+[()	3.439	3.6805	3.806	3.794	3.749	5.003
3	mean+[()	3.429	3.7498	3.880	3.882	3.936	4.949
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=1000000+0.4 data. Outliers are displayed as crosses. Times are in milliseconds.

A 80% subset

> x <- data[["n=1000000"]]
> subset
[1] 0.8
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755540  93.8    2637877 140.9  2637877 140.9
Vcells 18743142 143.0   42812957 326.7 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=1000000+0.8 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.811	2.399	2.848	2.705	3.081	7.971
1	meanOver	1.775	2.502	2.873	2.765	3.112	7.638
4	mean.default+[()	7.361	10.995	13.259	12.238	13.915	45.206
3	mean+[()	8.352	11.258	13.512	12.358	13.764	32.473

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.0000	1.000	1.000	1.000	1.000	1.0000
1	meanOver	0.9798	1.043	1.009	1.022	1.010	0.9582
4	mean.default+[()	4.0642	4.582	4.656	4.524	4.517	5.6711
3	mean+[()	4.6115	4.692	4.744	4.568	4.468	4.0737
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=1000000+0.8 data. Outliers are displayed as crosses. Times are in milliseconds.

n=10000000 vector

All elements

> x <- data[["n=10000000"]]
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755487  93.8    2637877 140.9  2637877 140.9
Vcells 18743428 143.1   42812957 326.7 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, refine = TRUE), meanOver_no_refine = meanOver(x, refine = FALSE), 
+     mean = mean(x), mean.default = mean.default(x), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean() and mean.default() on n=10000000+all data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
3	mean	11.66	13.64	16.53	16.70	19.04	29.00
4	mean.default	11.59	14.79	18.01	18.03	20.26	71.44
1	meanOver	14.62	16.80	20.21	19.92	22.71	46.23
2	meanOver_no_refine	14.61	17.62	20.35	20.29	21.61	62.55

	expr	min	lq	mean	median	uq	max
3	mean	1.0000	1.000	1.000	1.000	1.000	1.000
4	mean.default	0.9936	1.085	1.090	1.079	1.064	2.464
1	meanOver	1.2532	1.232	1.223	1.193	1.193	1.595
2	meanOver_no_refine	1.2524	1.292	1.232	1.215	1.135	2.157
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean() and mean.default() on n=10000000+all data. Outliers are displayed as crosses. Times are in milliseconds.

A 20% subset

> x <- data[["n=10000000"]]
> subset
[1] 0.2
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755564  93.8    2637877 140.9  2637877 140.9
Vcells 19343912 147.6   42812957 326.7 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=10000000+0.2 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	8.242	9.584	10.30	10.44	10.99	18.00
1	meanOver	8.308	9.909	10.53	10.57	11.13	13.65
4	mean.default+[()	25.131	30.762	37.96	33.80	36.23	361.61
3	mean+[()	24.721	30.790	35.30	33.84	36.27	56.97

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.000	1.000	1.000	1.000	1.000	1.0000
1	meanOver	1.008	1.034	1.022	1.012	1.012	0.7581
4	mean.default+[()	3.049	3.210	3.685	3.237	3.296	20.0854
3	mean+[()	2.999	3.212	3.427	3.241	3.300	3.1644
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=10000000+0.2 data. Outliers are displayed as crosses. Times are in milliseconds.

A 40% subset

> x <- data[["n=10000000"]]
> subset
[1] 0.4
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755576  93.8    2637877 140.9  2637877 140.9
Vcells 20343920 155.3   42812957 326.7 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=10000000+0.4 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
1	meanOver	10.48	13.91	14.71	14.69	15.92	21.39
2	meanOver_no_refine	10.60	13.40	14.69	14.70	16.47	19.79
4	mean.default+[()	44.24	58.43	68.27	64.11	70.99	392.93
3	mean+[()	43.35	58.82	68.62	64.34	73.51	348.81

	expr	min	lq	mean	median	uq	max
1	meanOver	1.000	1.0000	1.0000	1.000	1.000	1.0000
2	meanOver_no_refine	1.012	0.9633	0.9983	1.001	1.034	0.9251
4	mean.default+[()	4.222	4.2008	4.6408	4.364	4.459	18.3653
3	mean+[()	4.137	4.2286	4.6646	4.379	4.617	16.3036
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=10000000+0.4 data. Outliers are displayed as crosses. Times are in milliseconds.

A 80% subset

> x <- data[["n=10000000"]]
> subset
[1] 0.8
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755588  93.8    2637877 140.9  2637877 140.9
Vcells 22344829 170.5   42812957 326.7 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=10000000+0.8 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	17.28	18.17	21.92	20.85	25.31	30.88
1	meanOver	17.25	18.22	21.84	21.46	25.21	30.55
3	mean+[()	81.87	94.19	117.78	106.58	123.62	421.72
4	mean.default+[()	82.26	96.75	121.88	115.64	126.60	428.60

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.0000	1.000	1.000	1.000	1.0000	1.0000
1	meanOver	0.9984	1.003	0.996	1.030	0.9959	0.9893
3	mean+[()	4.7369	5.184	5.372	5.113	4.8841	13.6561
4	mean.default+[()	4.7600	5.324	5.559	5.548	5.0016	13.8788
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on integer+n=10000000+0.8 data. Outliers are displayed as crosses. Times are in milliseconds.

Data type "double"

Data

> rvector <- function(n, mode = c("logical", "double", "integer"), range = c(-100, +100), naProb = 0) {
+     mode <- match.arg(mode)
+     if (mode == "logical") {
+         X <- sample(c(FALSE, TRUE), size = n, replace = TRUE)
+     }     else {
+         x <- runif(n, min = range[1], max = range[2])
+     }
+     storage.mode(x) <- mode
+     if (naProb > 0) 
+         x[sample(n, size = naProb * n)] <- NA
+     x
+ }
> rvectors <- function(scale = 10, seed = 1, ...) {
+     set.seed(seed)
+     data <- list()
+     data[[1]] <- rvector(n = scale * 100, ...)
+     data[[2]] <- rvector(n = scale * 1000, ...)
+     data[[3]] <- rvector(n = scale * 10000, ...)
+     data[[4]] <- rvector(n = scale * 1e+05, ...)
+     data[[5]] <- rvector(n = scale * 1e+06, ...)
+     names(data) <- sprintf("n=%d", sapply(data, FUN = length))
+     data
+ }
> data <- rvectors(mode = mode)

Results

n=1000 vector

All elements

> x <- data[["n=1000"]]
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755535  93.8    2637877 140.9  2637877 140.9
Vcells 27899861 212.9   51455548 392.6 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, refine = TRUE), meanOver_no_refine = meanOver(x, refine = FALSE), 
+     mean = mean(x), mean.default = mean.default(x), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean() and mean.default() on n=1000+all data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
4	mean.default	0.0046	0.0054	0.0056	0.0054	0.0058	0.0065
2	meanOver_no_refine	0.0050	0.0058	0.0063	0.0062	0.0065	0.0112
1	meanOver	0.0062	0.0073	0.0084	0.0081	0.0085	0.0608
3	mean	0.0123	0.0135	0.0153	0.0146	0.0154	0.0747

	expr	min	lq	mean	median	uq	max
4	mean.default	1.000	1.000	1.000	1.000	1.000	1.000
2	meanOver_no_refine	1.083	1.071	1.122	1.143	1.133	1.706
1	meanOver	1.333	1.357	1.501	1.500	1.466	9.292
3	mean	2.667	2.500	2.725	2.714	2.666	11.409
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean() and mean.default() on n=1000+all data. Outliers are displayed as crosses. Times are in milliseconds.

A 20% subset

> x <- data[["n=1000"]]
> subset
[1] 0.2
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755612  93.8    2637877 140.9  2637877 140.9
Vcells 23900445 182.4   51455548 392.6 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=1000+0.2 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	0.0023	0.0042	0.0055	0.0058	0.0065	0.0089
1	meanOver	0.0031	0.0046	0.0064	0.0062	0.0071	0.0543
4	mean.default+[()	0.0054	0.0069	0.0108	0.0100	0.0108	0.1736
3	mean+[()	0.0123	0.0142	0.0192	0.0200	0.0208	0.0793

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.000	1.000	1.000	1.000	1.000	1.000
1	meanOver	1.334	1.091	1.171	1.067	1.088	6.130
4	mean.default+[()	2.333	1.636	1.967	1.733	1.647	19.609
3	mean+[()	5.333	3.363	3.495	3.466	3.176	8.957
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=1000+0.2 data. Outliers are displayed as crosses. Times are in milliseconds.

A 40% subset

> x <- data[["n=1000"]]
> subset
[1] 0.4
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755624  93.8    2637877 140.9  2637877 140.9
Vcells 23900553 182.4   51455548 392.6 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=1000+0.4 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	0.0050	0.0058	0.0065	0.0062	0.0073	0.0085
1	meanOver	0.0058	0.0065	0.0080	0.0075	0.0085	0.0601
4	mean.default+[()	0.0108	0.0119	0.0128	0.0127	0.0131	0.0404
3	mean+[()	0.0212	0.0231	0.0243	0.0239	0.0246	0.0712

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.000	1.000	1.000	1.000	1.000	1.000
1	meanOver	1.154	1.133	1.240	1.219	1.158	7.090
4	mean.default+[()	2.154	2.066	1.978	2.062	1.790	4.772
3	mean+[()	4.230	4.000	3.749	3.874	3.368	8.408
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=1000+0.4 data. Outliers are displayed as crosses. Times are in milliseconds.

A 80% subset

> x <- data[["n=1000"]]
> subset
[1] 0.8
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755636  93.8    2637877 140.9  2637877 140.9
Vcells 23901897 182.4   51455548 392.6 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=1000+0.8 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	0.0058	0.0067	0.0081	0.0073	0.0085	0.0635
1	meanOver	0.0073	0.0085	0.0092	0.0089	0.0104	0.0115
4	mean.default+[()	0.0150	0.0167	0.0177	0.0173	0.0181	0.0458
3	mean+[()	0.0262	0.0281	0.0299	0.0296	0.0308	0.0801

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.000	1.000	1.000	1.000	1.000	1.0000
1	meanOver	1.267	1.257	1.137	1.210	1.227	0.1818
4	mean.default+[()	2.600	2.485	2.192	2.368	2.136	0.7212
3	mean+[()	4.533	4.171	3.699	4.052	3.636	1.2606
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=1000+0.8 data. Outliers are displayed as crosses. Times are in milliseconds.

n=10000 vector

All elements

> x <- data[["n=10000"]]
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755583  93.8    2637877 140.9  2637877 140.9
Vcells 23901429 182.4   51455548 392.6 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, refine = TRUE), meanOver_no_refine = meanOver(x, refine = FALSE), 
+     mean = mean(x), mean.default = mean.default(x), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean() and mean.default() on n=10000+all data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	0.0131	0.0179	0.0191	0.0185	0.0192	0.0697
4	mean.default	0.0208	0.0269	0.0274	0.0277	0.0281	0.0470
1	meanOver	0.0227	0.0327	0.0337	0.0339	0.0350	0.0377
3	mean	0.0296	0.0383	0.0397	0.0397	0.0404	0.0770

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.000	1.000	1.000	1.000	1.00	1.0000
4	mean.default	1.588	1.505	1.437	1.500	1.46	0.6740
1	meanOver	1.735	1.828	1.769	1.833	1.82	0.5414
3	mean	2.265	2.140	2.085	2.146	2.10	1.1050
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean() and mean.default() on n=10000+all data. Outliers are displayed as crosses. Times are in milliseconds.

A 20% subset

> x <- data[["n=10000"]]
> subset
[1] 0.2
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755660  93.8    2637877 140.9  2637877 140.9
Vcells 23902513 182.4   51455548 392.6 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=10000+0.2 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	0.0104	0.0116	0.0131	0.0131	0.0142	0.0181
1	meanOver	0.0158	0.0166	0.0186	0.0177	0.0191	0.0832
4	mean.default+[()	0.0343	0.0381	0.0393	0.0393	0.0404	0.0462
3	mean+[()	0.0435	0.0512	0.0528	0.0520	0.0531	0.1193

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.000	1.000	1.000	1.000	1.000	1.000
1	meanOver	1.519	1.433	1.422	1.353	1.338	4.596
4	mean.default+[()	3.296	3.300	3.010	3.000	2.838	2.553
3	mean+[()	4.185	4.433	4.037	3.970	3.729	6.595
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=10000+0.2 data. Outliers are displayed as crosses. Times are in milliseconds.

A 40% subset

> x <- data[["n=10000"]]
> subset
[1] 0.4
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755672  93.8    2637877 140.9  2637877 140.9
Vcells 23903521 182.4   51455548 392.6 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=10000+0.4 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	0.0089	0.0104	0.0135	0.0123	0.0152	0.0246
1	meanOver	0.0135	0.0150	0.0190	0.0169	0.0210	0.0462
4	mean.default+[()	0.0370	0.0397	0.0464	0.0416	0.0487	0.1120
3	mean+[()	0.0462	0.0504	0.0604	0.0537	0.0710	0.1001

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.000	1.000	1.000	1.000	1.000	1.000
1	meanOver	1.522	1.444	1.408	1.375	1.380	1.875
4	mean.default+[()	4.174	3.814	3.429	3.375	3.202	4.547
3	mean+[()	5.217	4.851	4.462	4.359	4.670	4.062
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=10000+0.4 data. Outliers are displayed as crosses. Times are in milliseconds.

A 80% subset

> x <- data[["n=10000"]]
> subset
[1] 0.8
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755684  93.8    2637877 140.9  2637877 140.9
Vcells 23906836 182.4   51455548 392.6 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=10000+0.8 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	0.0154	0.0164	0.0191	0.0185	0.0192	0.0450
1	meanOver	0.0235	0.0264	0.0284	0.0269	0.0293	0.0504
4	mean.default+[()	0.0712	0.0737	0.0759	0.0751	0.0762	0.1097
3	mean+[()	0.0808	0.0839	0.0883	0.0851	0.0874	0.1936

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.000	1.000	1.000	1.000	1.00	1.000
1	meanOver	1.525	1.612	1.488	1.458	1.52	1.120
4	mean.default+[()	4.625	4.506	3.983	4.062	3.96	2.436
3	mean+[()	5.250	5.129	4.633	4.604	4.54	4.299
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=10000+0.8 data. Outliers are displayed as crosses. Times are in milliseconds.

n=100000 vector

All elements

> x <- data[["n=100000"]]
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755631  93.8    2637877 140.9  2637877 140.9
Vcells 23906368 182.4   51455548 392.6 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, refine = TRUE), meanOver_no_refine = meanOver(x, refine = FALSE), 
+     mean = mean(x), mean.default = mean.default(x), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean() and mean.default() on n=100000+all data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	0.1036	0.1372	0.1999	0.2379	0.2444	0.2622
4	mean.default	0.2013	0.2514	0.2982	0.3157	0.3189	0.7853
3	mean	0.2541	0.2937	0.3286	0.3328	0.3540	0.7622
1	meanOver	0.2025	0.2847	0.3708	0.4254	0.4362	0.5917

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.000	1.000	1.000	1.000	1.000	1.000
4	mean.default	1.944	1.832	1.492	1.327	1.305	2.996
3	mean	2.454	2.140	1.644	1.399	1.448	2.908
1	meanOver	1.955	2.074	1.855	1.788	1.784	2.257
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean() and mean.default() on n=100000+all data. Outliers are displayed as crosses. Times are in milliseconds.

A 20% subset

> x <- data[["n=100000"]]
> subset
[1] 0.2
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755708  93.8    2637877 140.9  2637877 140.9
Vcells 23912852 182.5   51455548 392.6 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=100000+0.2 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	0.0416	0.0479	0.0614	0.0631	0.0685	0.0935
1	meanOver	0.0751	0.1097	0.1115	0.1126	0.1207	0.2225
4	mean.default+[()	0.1790	0.2666	0.2599	0.2737	0.2779	0.3449
3	mean+[()	0.1940	0.2868	0.2949	0.2930	0.3218	0.3761

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.000	1.000	1.000	1.000	1.000	1.000
1	meanOver	1.806	2.289	1.815	1.784	1.761	2.379
4	mean.default+[()	4.306	5.562	4.231	4.335	4.056	3.687
3	mean+[()	4.667	5.984	4.801	4.640	4.697	4.021
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=100000+0.2 data. Outliers are displayed as crosses. Times are in milliseconds.

A 40% subset

> x <- data[["n=100000"]]
> subset
[1] 0.4
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755720  93.8    2637877 140.9  2637877 140.9
Vcells 23922860 182.6   51455548 392.6 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=100000+0.4 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	0.0720	0.0905	0.1131	0.1097	0.1267	0.2264
1	meanOver	0.1213	0.1809	0.1928	0.1929	0.2079	0.3549
4	mean.default+[()	0.3430	0.3628	0.4825	0.5079	0.5335	0.6513
3	mean+[()	0.3576	0.4970	0.5381	0.5463	0.5784	0.7399

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.000	1.000	1.000	1.000	1.000	1.000
1	meanOver	1.685	2.000	1.705	1.758	1.641	1.568
4	mean.default+[()	4.765	4.011	4.266	4.630	4.213	2.878
3	mean+[()	4.968	5.494	4.758	4.979	4.567	3.269
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=100000+0.4 data. Outliers are displayed as crosses. Times are in milliseconds.

A 80% subset

> x <- data[["n=100000"]]
> subset
[1] 0.8
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755732  93.8    2637877 140.9  2637877 140.9
Vcells 23942868 182.7   51455548 392.6 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=100000+0.8 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	0.1444	0.1979	0.2238	0.2125	0.2371	0.3415
1	meanOver	0.2306	0.3563	0.3830	0.3653	0.4254	0.5401
4	mean.default+[()	0.7206	0.9990	1.1829	1.1248	1.4053	1.8043
3	mean+[()	0.7699	1.1427	1.2426	1.1760	1.4480	1.7947

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.000	1.000	1.000	1.000	1.000	1.000
1	meanOver	1.597	1.801	1.711	1.719	1.794	1.582
4	mean.default+[()	4.992	5.049	5.286	5.293	5.926	5.284
3	mean+[()	5.333	5.775	5.552	5.534	6.106	5.256
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=100000+0.8 data. Outliers are displayed as crosses. Times are in milliseconds.

n=1000000 vector

All elements

> x <- data[["n=1000000"]]
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755679  93.8    2637877 140.9  2637877 140.9
Vcells 23942400 182.7   51455548 392.6 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, refine = TRUE), meanOver_no_refine = meanOver(x, refine = FALSE), 
+     mean = mean(x), mean.default = mean.default(x), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean() and mean.default() on n=1000000+all data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.534	1.626	1.997	1.787	2.487	2.823
4	mean.default	2.913	3.011	3.494	3.458	4.015	4.730
3	mean	2.963	3.022	3.526	3.512	3.797	4.941
1	meanOver	3.026	3.114	3.865	3.682	4.667	6.525

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.000	1.000	1.000	1.000	1.000	1.000
4	mean.default	1.899	1.851	1.750	1.935	1.614	1.675
3	mean	1.931	1.858	1.765	1.965	1.526	1.750
1	meanOver	1.972	1.914	1.935	2.060	1.876	2.311
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean() and mean.default() on n=1000000+all data. Outliers are displayed as crosses. Times are in milliseconds.

A 20% subset

> x <- data[["n=1000000"]]
> subset
[1] 0.2
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755756  93.8    2637877 140.9  2637877 140.9
Vcells 24004460 183.2   51455548 392.6 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=1000000+0.2 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.490	1.503	1.548	1.519	1.607	1.763
1	meanOver	2.892	2.910	3.050	2.970	3.137	4.390
3	mean+[()	3.321	3.415	3.838	3.846	3.966	5.386
4	mean.default+[()	3.248	3.638	4.238	3.857	4.267	23.142

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.000	1.000	1.000	1.000	1.000	1.000
1	meanOver	1.941	1.935	1.970	1.955	1.952	2.490
3	mean+[()	2.229	2.271	2.479	2.532	2.468	3.055
4	mean.default+[()	2.179	2.420	2.738	2.539	2.655	13.129
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=1000000+0.2 data. Outliers are displayed as crosses. Times are in milliseconds.

A 40% subset

> x <- data[["n=1000000"]]
> subset
[1] 0.4
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755768  93.8    2637877 140.9  2637877 140.9
Vcells 24104468 184.0   51455548 392.6 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=1000000+0.4 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.618	1.849	2.013	2.010	2.188	2.754
1	meanOver	3.130	3.244	3.761	3.679	4.176	5.046
4	mean.default+[()	5.147	6.478	7.745	7.294	8.102	26.860
3	mean+[()	5.175	6.289	7.461	7.433	8.063	26.511

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.000	1.000	1.000	1.000	1.000	1.000
1	meanOver	1.935	1.754	1.868	1.831	1.908	1.832
4	mean.default+[()	3.182	3.504	3.847	3.629	3.703	9.752
3	mean+[()	3.199	3.402	3.706	3.698	3.685	9.625
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=1000000+0.4 data. Outliers are displayed as crosses. Times are in milliseconds.

A 80% subset

> x <- data[["n=1000000"]]
> subset
[1] 0.8
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755780  93.8    2637877 140.9  2637877 140.9
Vcells 24304476 185.5   51455548 392.6 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=1000000+0.8 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	2.036	2.439	2.734	2.525	3.164	5.757
1	meanOver	3.898	4.558	4.845	4.659	5.370	6.740
4	mean.default+[()	10.575	12.585	14.210	13.882	14.753	34.128
3	mean+[()	9.552	11.966	15.163	14.100	15.738	35.201

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.000	1.000	1.000	1.000	1.000	1.000
1	meanOver	1.914	1.868	1.772	1.845	1.697	1.171
4	mean.default+[()	5.193	5.159	5.197	5.499	4.662	5.928
3	mean+[()	4.691	4.905	5.546	5.585	4.974	6.115
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=1000000+0.8 data. Outliers are displayed as crosses. Times are in milliseconds.

n=10000000 vector

All elements

> x <- data[["n=10000000"]]
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755727  93.8    2637877 140.9  2637877 140.9
Vcells 24304008 185.5   51455548 392.6 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, refine = TRUE), meanOver_no_refine = meanOver(x, refine = FALSE), 
+     mean = mean(x), mean.default = mean.default(x), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean() and mean.default() on n=10000000+all data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	13.93	17.13	17.52	17.37	17.62	22.24
3	mean	26.77	32.68	33.47	33.19	33.97	40.75
4	mean.default	26.47	32.56	33.43	33.22	33.95	39.02
1	meanOver	26.87	34.11	34.84	34.59	35.44	42.24

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.000	1.000	1.000	1.000	1.000	1.000
3	mean	1.921	1.908	1.910	1.911	1.927	1.832
4	mean.default	1.900	1.901	1.908	1.913	1.927	1.755
1	meanOver	1.928	1.992	1.988	1.991	2.011	1.899
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean() and mean.default() on n=10000000+all data. Outliers are displayed as crosses. Times are in milliseconds.

A 20% subset

> x <- data[["n=10000000"]]
> subset
[1] 0.2
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755804  93.8    2637877 140.9  2637877 140.9
Vcells 24904492 190.1   51455548 392.6 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=10000000+0.2 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	12.09	14.68	15.10	14.97	15.54	20.88
1	meanOver	23.86	28.77	29.09	29.17	30.19	41.48
3	mean+[()	32.97	39.01	42.89	40.24	43.98	73.90
4	mean.default+[()	32.41	38.62	46.43	40.40	45.38	331.60

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.000	1.000	1.000	1.000	1.000	1.000
1	meanOver	1.974	1.959	1.926	1.949	1.942	1.987
3	mean+[()	2.728	2.657	2.840	2.689	2.830	3.539
4	mean.default+[()	2.682	2.630	3.074	2.699	2.920	15.881
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=10000000+0.2 data. Outliers are displayed as crosses. Times are in milliseconds.

A 40% subset

> x <- data[["n=10000000"]]
> subset
[1] 0.4
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755816  93.8    2637877 140.9  2637877 140.9
Vcells 25904500 197.7   51455548 392.6 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=10000000+0.4 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	14.26	17.32	18.16	17.77	18.99	24.91
1	meanOver	27.51	32.82	35.25	34.42	36.43	52.39
4	mean.default+[()	58.12	69.69	85.13	76.38	88.16	423.77
3	mean+[()	56.98	67.69	84.85	76.62	88.58	375.36

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.000	1.000	1.000	1.000	1.000	1.000
1	meanOver	1.929	1.895	1.940	1.937	1.919	2.103
4	mean.default+[()	4.075	4.023	4.687	4.299	4.643	17.012
3	mean+[()	3.995	3.908	4.671	4.312	4.665	15.069
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=10000000+0.4 data. Outliers are displayed as crosses. Times are in milliseconds.

A 80% subset

> x <- data[["n=10000000"]]
> subset
[1] 0.8
> idxs <- sort(sample(length(x), size = subset * length(x), replace = FALSE))
> gc()
           used  (Mb) gc trigger  (Mb) max used  (Mb)
Ncells  1755828  93.8    2637877 140.9  2637877 140.9
Vcells 27904508 212.9   51455548 392.6 68120027 519.8
> stats <- microbenchmark(meanOver = meanOver(x, idxs = idxs, refine = TRUE), meanOver_no_refine = meanOver(x, 
+     idxs = idxs, refine = FALSE), `mean+[()` = meanOver_R_v1(x, idxs = idxs), `mean.default+[()` = meanOver_R_v2(x, 
+     idxs = idxs), unit = "ms")

Table: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=10000000+0.8 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	20.40	22.76	23.74	23.23	24.16	31.10
1	meanOver	39.48	43.97	45.77	44.74	46.49	60.21
4	mean.default+[()	116.79	131.13	169.81	148.02	161.03	589.74
3	mean+[()	117.55	138.76	163.17	149.91	162.95	465.92

	expr	min	lq	mean	median	uq	max
2	meanOver_no_refine	1.000	1.000	1.000	1.000	1.000	1.000
1	meanOver	1.935	1.932	1.928	1.926	1.924	1.936
4	mean.default+[()	5.725	5.761	7.152	6.373	6.666	18.963
3	mean+[()	5.762	6.096	6.872	6.454	6.746	14.982
Figure: Benchmarking of meanOver(), meanOver_no_refine(), mean+[()() and mean.default+[()() on double+n=10000000+0.8 data. Outliers are displayed as crosses. Times are in milliseconds.

Appendix

Session information

R Under development (unstable) (2015-02-27 r67909)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 7 x64 (build 7601) Service Pack 1

locale:
[1] LC_COLLATE=English_United States.1252 
[2] LC_CTYPE=English_United States.1252   
[3] LC_MONETARY=English_United States.1252
[4] LC_NUMERIC=C                          
[5] LC_TIME=English_United States.1252    

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

other attached packages:
[1] markdown_0.7.7          microbenchmark_1.4-2    matrixStats_0.14.0-9000
[4] ggplot2_1.0.0           knitr_1.9.3             R.devices_2.13.0       
[7] R.utils_2.0.0           R.oo_1.19.0             R.methodsS3_1.7.0      

loaded via a namespace (and not attached):
 [1] Rcpp_0.11.4           GenomeInfoDb_1.3.13   formatR_1.0.3        
 [4] plyr_1.8.1            base64enc_0.1-3       tools_3.2.0          
 [7] digest_0.6.8          RSQLite_1.0.0         annotate_1.45.2      
[10] evaluate_0.5.7        gtable_0.1.2          R.cache_0.11.1-9000  
[13] lattice_0.20-30       DBI_0.3.1             parallel_3.2.0       
[16] mvtnorm_1.0-2         proto_0.3-10          R.rsp_0.20.0         
[19] genefilter_1.49.2     stringr_0.6.2         IRanges_2.1.41       
[22] S4Vectors_0.5.21      stats4_3.2.0          grid_3.2.0           
[25] Biobase_2.27.2        AnnotationDbi_1.29.17 XML_3.98-1.1         
[28] survival_2.38-1       multcomp_1.3-9        TH.data_1.0-6        
[31] reshape2_1.4.1        scales_0.2.4          MASS_7.3-39          
[34] splines_3.2.0         BiocGenerics_0.13.6   xtable_1.8-0         
[37] mime_0.2.1            colorspace_1.2-4      labeling_0.3         
[40] sandwich_2.3-2        munsell_0.4.2         Cairo_1.5-6          
[43] zoo_1.7-12

Total processing time was 4.16 mins.

Reproducibility

To reproduce this report, do:

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

Copyright Henrik Bengtsson. Last updated on 2015-03-02 17:33:38 (-0800 UTC). Powered by RSP.

[Benchmark reports](Benchmark reports)

meanOver

meanOver() benchmarks

Alternative methods

Data type "integer"

Data

Results

n=1000 vector

All elements

A 20% subset

A 40% subset

A 80% subset

n=10000 vector

All elements

A 20% subset

A 40% subset

A 80% subset

n=100000 vector

All elements

A 20% subset

A 40% subset

A 80% subset

n=1000000 vector

All elements

A 20% subset

A 40% subset

A 80% subset

n=10000000 vector

All elements

A 20% subset

A 40% subset

A 80% subset

Data type "double"

Data

Results

n=1000 vector

All elements

A 20% subset

A 40% subset

A 80% subset

n=10000 vector

All elements

A 20% subset

A 40% subset

A 80% subset

n=100000 vector

All elements

A 20% subset

A 40% subset

A 80% subset

n=1000000 vector

All elements

A 20% subset

A 40% subset

A 80% subset

n=10000000 vector

All elements

A 20% subset

A 40% subset

A 80% subset

Appendix

Session information

Reproducibility

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!