Content deleted Content added
mention issue with basic on the cray |
m →History: replaced: recently- → recently |
||
(28 intermediate revisions by 11 users not shown) | |||
Line 1:
{{Short description|Type of computer benchmark}}
The '''Creative Computing Benchmark''', also
The Creative Computing Benchmark was one of three common benchmarks of the era. Its primary competition in the early 1980s in the United States was the [[Byte Sieve]], of September 1981, while the earlier [[Rugg/Feldman benchmarks]] of June 1977 were not as well known in the United States, but were widely used in the United Kingdom.
==History==▼
The benchmark first appeared in the November 1983 issue of ''[[Creative Computing]]'' under the title "Benchmark Comparison Test".{{sfn|Ahl|1983|p=259}} In the article, Ahl was careful to state that it tested only a few aspects of the BASIC language, mostly its looping performance. He stated:▼
▲== History ==
{{cquote|...the benchmark program presented here is not representative of the way computers are actually used; it measures only a few aspects of performance, and no one should buy a computer based solely on the results of these measures. Yet, the results provide some interesting comparative data.{{sfn|Ahl|1983|p=259}}}}▼
[[File:Creative_Computing_Benchmark_original_result_set.png|thumb|right|Original result set from November 1983 issue of Creative Computing]]
▲The benchmark first appeared in the November 1983 issue of ''
The initial results were provided for common machines of the era, including the [[Apple II]], [[Commodore 64]] and the recently-released [[IBM PC]]. Most of these machines ran some variation of the stock [[Microsoft BASIC]] and thus provided similar times on the order of two minutes, while the [[16-bit]] PC was near the top of the list at only 24 seconds. the fastest machine in this initial suite was the [[Olivetti M20]] at 13 seconds, and the slowest was [[Atari BASIC]] on the [[Atari 800]] at 6 minutes 58 seconds.{{sfn|Ahl|1983|p=260}}▼
▲{{cquote|... the benchmark program presented here is not representative of the way computers are actually used; it measures only a few aspects of performance, and no one should buy a computer based solely on the results of these measures. Yet, the results provide some interesting comparative data.{{sfn|Ahl|1983|p=259}}}}
In the months following its publication, the magazine was inundated with results for other platforms. It became a regular feature for a time, placed prominently near the front of the magazine with an ever-growing list of results. By March the fastest machine on the list was the [[Cray-1]] at 0.01 seconds, and the slowest was the [[TI SR-50]] [[programmable calculator]] at 12.7 days.{{sfn|Ahl|1984|p=7}}▼
▲The initial results were provided for common machines of the era, including the [[Apple II]], [[Commodore 64]] and the recently
The benchmark had several problems that made it less useful for general purposes. For instance, the system did not test any string manipulation, who's performance varied widely across platforms. It also did not take advantage of any "speedups" available on different platforms, like the possible use of integer variables for loop indexes or turning off video access on machines with shared main memory.{{efn|Most 8-bit machines of the era had a single bank of RAM that was shared between the CPU and display driver, which led to bus contention issues that slowed performance as much as 30%. Turning off the display was a common way to improve compute-bound programs like this benchmark.{{sfn|Wilkinson|1985|p=140}}}} These limitations were widely debated at the time.{{sfn|Wilkinson|1985|p=139}}▼
▲In the months following its publication, the magazine was inundated with results for other platforms. It became a regular feature for a time, placed prominently near the front of the magazine with an ever-growing list of results. By March the fastest machine on the list was the
Its last appearance is in the May 1984 issue, which included values for 183 machines. This issue included a note that the many criticisms of the system had been taken to heart and a new benchmark program was under design.<ref>{{cite magazine |magazine=Creative Computing |date=May 1984 |page=6 |first=David |last=Ahl |title=Creative Computing Benchmark |url=https://archive.org/details/creativecomputing-1984-05/page/n9}}</ref> However, such a program never appeared in the magazine, which increasingly focused on the business market that was coming to dominate [[personal computer]]s in that era.▼
▲The benchmark had several problems that made it less useful for general purposes. For instance, the system did not test any string manipulation,
The benchmark continued to be used as a general-purpose tool after this date,<ref> {{cite magazine |magazine=Creative Computing |date=May 1985 |first=Glenn |last=Hart |pages=46, 48, 51-52 |title=Microsoft BASIC 2.0 for the Mac |url=https://archive.org/stream/creativecomputing-1985-05/Creative_Computing_v11_n05_1985_May#page/n55}}</ref> but as the importance of BASIC dwindled it became less common.▼
▲Its last appearance is in the May 1984 issue, which included values for 183 machines. This issue included a note that the many criticisms of the system had been taken to heart and a new benchmark program was under design.{{sfn|Ahl|1984|p=6}} However, such a program never appeared in the magazine. In the September 1985 issue, David Ahl responded to a Letter to the Editor about the new benchmark program with "Several analysts spent many hours working out three new benchmark tests ... none gave different or better results".<ref>{{
==Code==▼
The following is from later versions of the benchmark code, which reduced the number of compound statements on a line.{{sfn|Ahl|1984|p=7}}▼
▲The benchmark continued to be used as a general-purpose tool after this date,<ref>
10 ' Ahl's Simple Benchmark▼
20 FOR N=1 TO 100: A=N▼
▲==Code==
30 FOR I=1 TO 10▼
This is the original version from the November 1983 edition:{{sfn|Ahl|1983|p=259}}
40 A=SQR(A): R=R+RND(1)▼
{{sxhl|2=basic|1=<nowiki/>
50 NEXT I▼
1 ' Ahl's simple benchmark test
60 FOR I=1 TO 10▼
2 ' In Lines 30 and 40, some computers
70 A=A^2: R=R+RND(1)▼
3 ' may require RND(1) for correct results
80 NEXT I▼
10 PRINT "Accuracy Random"
90 S=S+A: NEXT N▼
20 FOR N=1 TO 100:A=N
100 PRINT ABS(1010-S/5)▼
30 FOR I=1 TO 10:A=SQR(A):R=R+RND(0):NEXT I
110 PRINT ABS(1000-R)▼
40 FOR I=1 TO 10:A=A^2:R=R+RND(0):NEXT I
50 S=S+A:NEXT N
60 PRINT ABS(1010-S/5);ABS(1000-R)
}}
▲The following is from later versions of the benchmark code, which reduced the number of compound statements on a line
{{sxhl|2=basic|1=<nowiki/>
}}
==Notes==
{{
==References==
===Citations===
{{reflist|30em}}
===Bibliography===
* {{
* {{
* {{
[[Category:Benchmarks (computing)]]
[[Category:History of computing]]
[[Category:BASIC programming language]]
|