This test is designed to model 'real-world' performance of lossless data compressors. The test set contains a mix of different file types which are chosen with 'What do people use archivers for the most' in mind. The testset should contain data, weighted (in both type and proportion of files in the set) by how often these files are used for compression by normal users using compression software. So for example there will be more txt files then .ocx files in the set (yes, this is arbitrary). The set contains 100's of files and has a total size of over 300 Mb. The idea of a large collection is filtering out the 'noise'. A compressor might perform bad on 1 or 2 filetypes, but on a very large collection it will not hurt as much.

Some programs like CCM and BZIP2 can only compress one file at a time. For these programs a single TAR-file is created containing all files. The files in this TAR-file are ordered alphabetically on suffix, then name. Results of these compressors are marked with an 'Y' in the tarred column.

The testset consists of the following file types :

Considering the fact it's supposed to be a 'real-world' test I will not look at the best possible (command-line or gui) switch combination to use for optimal compression, but only test a limited set as 'normal users' would do. For 7-zip this means for example I will use the GUI and select the Ultra compression method (which can be easily beaten with some good command line switches), WinRar will be tested with max dictionary size and solid archiving etc. Programs are allowed to use a maximum of 800 Mb memory and must finish the compression stage within 24h. Compressed size must be 50% or less compared to the original size to be listed on MFC.

For my single file tests I got lots of requests to add the compression time to the tables. I did not do this for the reasons stated in the single file summary file, but I'm planning to measure compression times for this multiple file test!. I also decided to make this testset 'non public', so it's harder for developers to tune their program towards this specific test. I think this is the most fair way to get 'real life' performance tests.

Scoring system: The program yielding the lowest compressed size is considered the best program. The most efficient (read:use full) program is calculated by multiplying the compression time (in seconds) it took to produce the archive with the power of the archive size divided by the lowest measured archive size. The lower score the better. The basic idea is a compressor X has the same efficiency as compressor Y if X can compress twice as fast as Y and resulting archive size of X is 10% larger than size of Y. (Special thanks to Uwe Herklotz to get this formula right)

score_X = POWER(2; ((size_X / size_TOP) - 1) / 0,1) * time_X

with  score_X     efficiency score for a certain compressor X
      time_X      time elapsed by compressor X
      size_X      archive size achieved with compressor X
      size_TOP    archive size by top archiver (smallest benchmark result)

"0,1" represents 10% and power of 2 ensures that for each 10% worse results (compared
with top) the time is doubled, so any archiver (except top compressor) will get a
penalty on time. The score of top compressor is always equal to its time value.

Results sorted on Compression ratio.

Pos	Program	Switches used	TAR	Compressed	Compression	Comp time	Decomp time	Efficiency
				Size (bytes)	Ratio (%)	(sec)	(sec)	lower is better
001	PAQ8O8	-7	N	64042114	79.76	43660	43807	43660
002	WinRK 3.0.3	Maximum	N	64217766	79.70	30603	29899	31190
003	PAQ8JC	-6 (dir /oen)	N	66191139	79.08	49154	42144	62026
004	PAQAR 4.5	-7e (dir /oen/b)	N	68435715	78.37	55430	55659	89180
005	WinRK 3.0.3	High	N	68947835	78.21	6317	3961	10742
006	PAsQDa 4.4	-7e	N	69109187	78.15	54855	55041	94928
007	UDA 0.301	(none)	Y	71478767	77.41	7457	7472	16677
008	DURILCA 0.5	-m800 -t1	N	72175363	77.19	1997	1794	4816
009	DURILCA 0.5	-m800 -t1 -o32	N	72246808	77.16	1838	1628	4467
010	Slim 0.23d	-m800 -o64	N	73664683	76.71	3665	3605	10384
011	WinUDA 0.291	mode-3	N	73735574	76.69	7291	7263	20817
012	Slim 0.23d	-m800	N	73756703	76.69	3490	3491	9988
013	Slim 0.23d	(none)	N	73781973	76.68	3580	3524	10273
014	KGB 1.1 (PAQ6)	(very good)	N	75341375	76.18	15703	15801	53346
015	CCMx 1.30c	6	Y	77079949	75.64	369	357	1513
016	PPMonstr J	-m800 -o32	N	77699686	75.44	1672	1681	7332
017	CCM 1.30c	6	Y	77926922	75.37	293	281	1317
018	PPMonstr J	-m800 -o16	N	78086417	75.32	1628	1639	7444
019	UHARC 0.6b	-mx -md32768 -mm+	N	78344876	75.24	680	610	3197
020	FreeARC 0.50a	-mx	N	78348737	75.23	285	79	1341
021	PPMonstr J	(none)	N	78443862	75.20	1610	1620	7652
022	CCM 1.30c	(none)	Y	78598980	75.15	277	265	1339
023	LPAQ8	8	Y	78704186	75.12	1312	1321	6414
024	FreeARC 0.50a	(none)	N	79260986	74.95	185	65	961
025	PIMPLE2	(none)	Y	79628157	74.83	5699	5731	30791
026	WinUDA 0.291	mode-0	N	79710669	74.80	2366	2359	12898
027	WinRK 3.0.3	Normal	N	80676067	74.50	614	154	3716
028	COMPRESSIA 1.0b	max sol dict 15 Mb	N	80698340	74.49	2081	99999	12624
029	DURILCA light 0.5	-m800 -t1	N	81666056	74.19	366	173	2465
030	EPM r9	-m800	Y	81934909	74.10	2497	2513	17317
031	CMM4 0.1e	26	Y	82504741	73.92	712	716	5252
032	EPM r9	(none)	Y	82727852	73.85	2328	2350	17592
033	SQUEEZ 5.5	Ultra 32Mb M E Solid	N	82801754	73.83	1262	72	9613
034	Ocamyd 1.66 test1	-m8 -s0	Y	82880385	73.80	12124	12119	93141
035	Quark 0.95r	-l7 -d25	Y	83673567	73.55	5438	90	45521
036	RZM 0.07h	(none)	Y	84030374	73.44	591	42	5142
037	FreeARC 0.50a	-m3	N	84290571	73.36	83	48	743
038	WinRK 3.0.3	Fastest ROLZ3	N	84652024	73.24	252	165	2345
039	SBC 0.970 rev3	-m3 -b62	N	84736176	73.21	253	92	2376
040	Quark 0.95r	(none)	Y	84741314	73.21	1818	89	17083
041	UHARC 0.6b	(none)	N	84954506	73.15	479	66	4606
042	SBC 0.970 rev3	-b9	N	85740265	72.90	134	71	1403
043	StuffIt 12.0	Method=2 L3 E25 bm=on	N	85892062	72.85	367	76	3906
044	SBC 0.970 rev3	(none)	N	86245288	72.74	127	70	1404
045	KGB 1.1 (PAQ6)	(normal)	N	86565142	72.64	2486	2527	28457
046	WinRAR 3.71	Best solid 4096Kb	N	86758171	72.58	172	54	2010
047	7-Zip 4.58b	Ultra solid LZMA 64Mb	N	87230789	72.43	338	17	4158
048	SQUEEZ 5.5	Max 8Mb M E Solid	N	87693302	72.28	347	36	4488
049	ENC 0.15	abies -d256	N	87955590	72.20	9088	99999	120926
050	COMPRESSIA 1.0b	sol dict 15 Mb	N	88152784	72.13	216	154	2936
051	CTXf 0.75	-me	N	88357992	72.07	208	159	2891
052	7-Zip 4.58b	Norm solid LZMA 16Mb ws=32	N	88470477	72.03	274	17	3855
053	WinACE 2.69	Max 4096Kb solid ace20	N	88485235	72.03	225	16	3171
054	ACE 2.04	-m5 -d4096	N	88486043	72.03	227	15	3199
055	WinACE 2.69	Norm 4096Kb solid ace20	N	88526291	72.02	191	16	2703
056	TC 5.2 dev2	(none)	Y	89464778	71.72	1107	1112	17344
057	ENC 0.15	-d256	N	89750528	71.63	3735	99999	60355
058	DGCA 1.10	(none)	N	89834304	71.60	209	80	3408
059	CTXf 0.75	(none)	N	90038884	71.54	165	116	2751
060	777 0.04b1	-mg -mu32	N	90229477	71.48	2767	1528	47092
061	WinRAR 3.71	Normal solid 4096Kb	N	90540300	71.38	152	10	2675
062	777 0.04b1	(none)	N	90544382	71.38	3151	1637	55487
063	LZPX(J) 1.2h	-8	Y	91451359	71.09	301	300	5847
064	Bee 0.7.9 bld 0316	-m3 -d8 -s	N	92003980	70.92	1693	1702	34914
065	PIM 2.10	(none)	N	92067900	70.90	259	275	5378
066	LZPX(J) 1.2h	(none)	Y	92075183	70.90	277	280	5757
067	Bee 0.7.9 bld 0316	-m2 -d8	N	93103491	70.57	934	948	21696
068	UFA 0.04b1	-mg -mx -mu32	N	93428292	70.47	1348	352	32433
069	DST 0.91b	-me	Y	93568347	70.42	624	469	15243
070	SQUEEZ 5.5	Norm 4Mb M E Solid	N	94082622	70.26	132	36	3409
071	UFA 0.04b1	(none)	N	94175638	70.23	901	102	23505
072	FreeARC 0.50a	-m2	N	94195811	70.22	33	30	863
073	ACB 2.00c	u	N	95055289	69.95	2738	2870	78561
074	BOA 0.58b	-m15 -s	N	95080468	69.95	2003	2036	57629
075	PPMN 1.00b1 km	-O9 -M:50 -DA	Y	95513817	69.81	675	678	20353
076	BSSC 0.95a	(none)	Y	95811218	69.71	127	57	3955
077	StuffIt 12.0	Method=5 E26 bm=on	N	96069139	69.63	136	152	4355
078	CABARC 1.00.0106	LZX:21	N	96412288	69.52	598	7	19873
079	WINZIP 11.0	Optimize for best comp.	N	96420110	69.52	156	158	5189
080	PPMVC 1.2	-o16 -m256	N	96525071	69.49	325	375	10933
081	DC 0.99.307b	-b4096	N	96592918	69.47	133	99999	4507
082	BMA 1.34b	-m16m -ax	N	96709052	69.43	301	120	10329
083	PPMd var J rev.1	-m256 -o16	N	96737515	69.42	172	192	5921
084	DURILCA light 0.5	(none)	N	96823376	69.39	178	189	6184
085	BOA 0.58b	(none)	N	96857515	69.38	1541	1605	53739
086	BIX 1.00 b7	-mdG -s	N	97079141	69.31	348	10	12430
087	PPMVC 1.2	(none)	N	97152520	69.29	206	256	7417
088	WINIMP 1.21	1 BEST MM 16Mb B=1020Kb	N	97241095	69.26	204	9	7416
089	GRZipII 0.2.4	-b8m -a	Y	97484136	69.19	117	104	4366
090	7-Zip 4.58b	Max PPMD 64Mb ws=16	N	97573749	69.16	207	234	7801
091	WINZIP 11.0	PPMd	N	97754314	69.10	181	197	6955
092	StuffIt 12.0	Method=3 E25 bm=on	N	98017538	69.02	233	122	9212
093	MNZIP	(none)	Y	98488045	68.87	280	64	11649
094	BALZ 1.07	ex	Y	98500078	68.86	602	43	25078
095	QC 0.050	-8	Y	98510933	68.86	1865	576	77783
096	WinHKI 1.74	HKI2 max	N	98771075	68.78	312	319	13384
097	EPC 1.0	-m256 -o18	Y	98797159	68.77	382	99999	16433
098	WINIMP 1.21	1 BEST MM	N	99025199	68.70	137	9	6041
099	QC 0.050	(none)	Y	99055044	68.69	665	250	29417
100	LZPM 0.15	9	Y	99083654	68.68	1615	28	71664
101	UHARC 0.6b	-mz -md32768 -mm+	N	99251157	68.63	72	81	3253
102	BMA 1.34b	(none)	N	99468292	68.56	142	95	6569
103	DC 0.99.307b	(none)	N	99517116	68.54	124	99999	5767
104	BIX 1.00 b7	(none)	N	99694432	68.49	237	10	11235
105	BioArc 1.9	Ultra	N	99699058	68.49	164	191	7779
106	BA 1.01	(none)	Y	99873098	68.43	146	97	7056
107	M03	(none)	Y	99934651	68.41	3519	568	171218
108	Tornado 0.4	-11	Y	99959838	68.40	249	13	12148
109	BioArc 1.9	Normal	N	100012777	68.39	158	185	7753
110	WinHKI 1.74	HKI2 norm	N	100208006	68.32	230	237	11527
111	WINIMP 1.21	Normal MM	N	100399936	68.26	82	9	4196
112	UHBC 1.0	-b4m -m3	Y	100623220	68.19	460	399	24113
113	DARK 0.51	-b128m	N	100785450	68.14	152	93	8109
114	DARK 0.51	(none)	N	100877062	68.11	149	91	8028
115	BruteCM 0.1d	(none)	Y	100902600	68.10	271	407	14642
116	UHBC 1.0	(none)	Y	100990062	68.08	363	296	19799
117	WinRAR 3.71	Fast solid 4096Kb	N	101008351	68.07	125	12	6831
118	JAR 1.02	(none)	N	101169071	68.02	89	16	4949
119	PPMN 1.00b1 km	(none)	Y	101181370	68.02	501	509	27898
120	QLFC 6.6w	BS=8388608	N	101217544	68.01	1793	78	100233
121	EPC 1.0	(none)	Y	101258666	67.99	209	230	11736
122	GRZipII 0.2.4	(none)	Y	101686370	67.86	118	110	6940
123	ZZIP 0.36c	-mx -a -6m	N	102392643	67.63	4146	100	263206
124	CABARC 1.00.0106	LZX:17	N	102561882	67.58	175	7	11315
125	PPMd var J rev.1	(none)	N	102743525	67.52	122	141	8045
126	QLFC 6.6w	(none)	N	102938276	67.46	883	71	59467
127	WinTurtle 1.60	Solid	N	103129672	67.40	73	79	5019
128	DACT 0.8.41	-b1000000	Y	103176088	67.39	727	37	50237
129	DACT 0.8.41	(none)	Y	103378718	67.32	735	41	51916
130	DST 0.91b	(none)	Y	103386376	67.32	287	118	20289
131	WinTurtle 1.60	(none)	N	103567658	67.26	64	71	4614
132	ABC 2.4	(none)	Y	103669744	67.23	1200	194	87474
133	ZZIP 0.36c	(none)	N	103845404	67.17	1827	100	135735
134	Bicom 1.01	(none)	Y	103940513	67.14	2140	2172	160634
135	BIT 0.1	(none)	Y	104789891	66.88	177	41	14895
136	QazaR 0.0 pre5	-d9 -l7	Y	104867105	66.85	922	257	76509
137	YBS 0.03f	(none)	Y	105074083	66.79	140	70	11881
138	ASD 0.2.0	-m5 -mdg	N	105109740	66.77	6296	415	536351
139	RZIP 2.1	-9	Y	105305418	66.71	161	61	14009
140	RZIP 2.1	(none)	Y	105417723	66.68	137	57	12067
141	UC II 3.05	(none)	N	105608406	66.62	313	10	28143
142	M99 2.1	-m	Y	105711137	66.58	224	99999	20366
143	Tornado 0.4	-5	Y	105742922	66.57	27	13	2463
144	SZIP 1.12	-b41 -o40	Y	105793183	66.56	963	193	88336
145	RINGS 1.5c	5	Y	105800703	66.56	72	74	6610
146	CHILE 0.3d	(none)	Y	106072542	66.47	890	157	84145
147	CHILE 0.3d	-b=2048	Y	106165564	66.44	1092	160	104288
148	UC II 3.05	-TT	N	106353574	66.38	136	11	13255
149	BBB ver1	m512	Y	106364761	66.38	31957	1321	3118475
150	WinHKI 1.74	HKI1 max	N	106391218	66.37	158	39	15462
151	SZIP 1.12	(none)	Y	106567179	66.31	134	70	13366
152	ShipInBottle 1.0 b17	alg:ppm :mem:128	N	106742422	66.26	4204	4243	427357
153	QUAD 1.12	x	Y	106968291	66.19	176	55	18334
154	AI 1.1	-mt	N	107365151	66.06	70444	98105	7660262
155	YBS 0.03f	-m16m	Y	107462346	66.03	181	89	19891
156	HA 0.999b	a21	N	107637414	65.98	961	360	107627
157	BZIP2 1.0.5	(none)	Y	107720923	65.95	138	35	15596
158	QazaR 0.0 pre5	(none)	Y	108165979	65.81	155	109	18381
159	QUAD 1.12	(none)	Y	108215434	65.79	106	55	12638
160	ASD 0.2.0	(none)	N	109101360	65.51	404	137	53014
161	KZIP 14-APR-2007	(none)	N	110694932	65.01	4719	8	735810
162	TarsaLZP 21Aug2007	(none)	Y	110994614	64.91	94	114	15140
163	HOOK 1.3	800	Y	111275351	64.83	244	253	40512
164	SEMONE 0.6 b2	(none)	Y	111442686	64.77	204	28	34490
165	WINZIP 11.0	Enhanced Deflate	N	111503290	64.75	112	9	19060
166	LHARK 0.4d	-tob	N	111870915	64.64	83	8	14698
167	THOR 0.96	e5	Y	112275112	64.51	24	6	4440
168	MAR	-g	N	112554786	64.42	252	10	48055
169	TURTLE 0.07	(none)	Y	112646749	64.39	88	83	16949
170	LHARK 0.4d	(none)	N	113439063	64.14	37	8	7764
171	LZ2A	-Cb12 -Szd26	Y	113588897	64.09	350	41	74647
172	PKZIP 2.50	-exx	N	114022347	63.96	28	8	6259
173	ShipInBottle 1.0 b17	alg:lz	N	114051091	63.95	1244	49	278928
174	SLUG 1.27b	(none)	Y	114187051	63.91	12	11	2731
175	LZXQ 0.4	normal	Y	114492987	63.81	177	273	41631
176	PKZIP 2.50	(none)	N	114501263	63.81	22	8	5179
177	AMG 2.2	(Max Compression)	N	114662758	63.76	244	99999	58454
178	VuZip 1.8	Max	N	114679608	63.75	89	11	21360
179	DZip 2.90	-9	N	114688490	63.75	76	8	18258
180	EAZEL 1.0	(best)	N	114728370	63.73	99	12	23886
181	CODEC 3.21	(none)	N	114767275	63.72	62	44	15022
182	GZIP 1.2.4	-9	Y	114857977	63.69	67	7	16394
183	AIN 2.32	-m1	N	115160947	63.60	94	7	23767
184	ESP 1.92	(none)	N	115164577	63.60	32	8	8094
185	LZTurbo 0.92	-43	Y	115219801	63.58	34	18	8651
186	VuZip 1.8	Default	N	115261129	63.57	41	11	10479
187	DZip 2.90	(none)	N	115269882	63.56	33	8	8443
188	LZ2A	(none)	Y	115298983	63.55	305	44	78276
189	File2Pack 2.0	(none)	N	115305901	63.55	32	8	8219
190	EAZEL 1.0	(normal)	N	115309762	63.55	44	11	11305
191	GZIP 1.2.4	(none)	Y	115442964	63.51	28	7	7299
192	AIN 2.32	(none)	N	115576936	63.47	25	7	6612
193	M99 2.1	(none)	Y	115632111	63.45	154	51	40974
194	BCArchive 1.08.7	(none)	N	115821124	63.39	41	28	11134
195	ARJ32 3.14	(none)	N	116105225	63.30	41	9	11482
196	ARJ 2.84	(none)	N	116107572	63.30	53	10	14846
197	WINZIP 10.0	Normal	N	116113423	63.30	31	9	8689
198	WinXP (Built-in)	(none)	N	116337271	63.23	28	30	8041
199	aPLib 0.43	(none)	N	117596762	62.83	650	10	213919
200	SR3a	(none)	Y	118183199	62.64	53	59	18586
201	ALZip 7.0b1	(normal)	N	118492877	62.54	36	14	13055
202	HPACK 0.79a	(none)	N	120702014	61.85	973	74	448138
203	FlashZIP 0.03	x5	Y	121118395	61.71	157	35	75643
204	SAR 1.0	(none)	N	122067423	61.41	236	20	126006
205	SYMBRA 0.2	-m5	Y	122120558	61.40	2879	1733	1546035
206	AR 1.0	(none)	Y	122147342	61.39	242	20	130332
207	DeepFreezer 1.06	(none)	N	123350351	61.01	45	15	27606
208	Chaos Comp 3.0	(none)	N	123681560	60.90	73	14	46417
209	FlashZIP 0.03	3	Y	124223714	60.73	50	34	33714
210	Ocamyd 1.66 test1	(none)	Y	124262690	60.72	519	518	351428
211	THOR 0.96	e3	Y	124588656	60.62	6	7	4209
212	LCSSR 0.2	-l9 -b6	Y	126378787	60.05	7578	7233	6451940
213	HYPER 2.5	(none)	N	128439870	59.40	55	32	58530
214	LZOP 1.02rc1	-9	Y	129601733	59.03	132	7	159296
215	ARX 1.0	(none)	N	133464199	57.81	84	21	153980
216	QuickLZ 1.40b9	mode3	N	133973909	57.65	21	9	40678
217	CTW 0.1	(none)	Y	134687380	57.43	6911	7405	14461790
218	Secura Archiver 1.7	(none)	N	139876348	55.79	334	7	1225561
219	CA-ZIP 3.4	(none)	N	140247762	55.67	61	8	233011
220	AI 1.1	(none)	N	140259735	55.66	9153	15193	35008476
221	LZTurbo 0.92	-21	Y	144135815	54.44	6	7	34911
222	PACKET 0.01	(none)	Y	144686589	54.26	16	23	98813
223	ASH 04a	-m800 -s80 -o60	Y	144885507	54.20	3229	2182	20375685
224	LZC 0.08	2	Y	149651909	52.70	33	14	348821
225	BriefLZ 1.04	(none)	N	151611876	52.08	19	6	248297
226	THOR 0.96	e1	Y	151936954	51.97	3	6	40609
227	LZW 0.2	(none)	Y	153289520	51.55	645	15	10107286
228	LZOP 1.02rc1	-1	Y	154993814	51.01	7	6	131912
229	LZOP 1.02rc1	(none)	Y	155389089	50.88	7	6	137678
230	QuickLZ 1.40b9	mode1	N	156840006	50.42	3	9	69038
231	BigCrunch 0.4a1	(none)	N	157502600	50.21	1411	3829	34884919
232	CODER 1.1	-ew 4194304	N	158788621	49.81	849	274	24125028
233	SRANK 1.0	(none)	Y	164110929	48.12	13	13	657171
234	TESTSET		N	316355757	0.00	0	0	0

Remarks:
Ten programs so far crashed during compressing:
- Emilcont 0.3a
- RKC 1.02
- PPMZ2 0.81
- HIPP 0.5819
- OrangeArchiver 1.05
- PPMT 1.0
- HIP beta1
- LZ 1.0
- HuffComp 1.3
- Chile 0.5

- Some programs fail to correctly decompressing the archive created. Those programs are marked with a red penalty time of 99999 seconds. Do *not* use these programs to backup your data!.

- (De)compression time of ACB 2.00c is an approximation as ACB only runs under Win98 (which is installed on the 1800+ machine only). Measured time on the 1800+ is devided by the approximate ratio of compression speed on both machines.

- The command-line version of Squeez 5 seems to get different compression then the GUI version. The switch combination '-m5 -MD32768 -uxx9 -ME -MM -s' gets it down to 83.384.794 bytes...
- Stuffit 12.0 contains recompression techniques for several file types like JPG, GIF, BMP and PDF. For the MFC test I disabled PDF recompression as it appears to be "pixel perfect" rather then "bit-perfect" for most PDF files. This hurts compression a bit, but prevents the decompressed file to be slightly different than the original. This would be fine for normal users, but this is a lossless compression test. It would be unfair towards the lossless "bit perfect" compressors to include these result.

Conclusions and observations:
- Top performers (based on compression) in this test are PAQ8 and WinRK (PWCM). They are able to compress the 300+ Mb testset to under 62 Mb (80% reduction in size) but take a minimum of 8,5 hour to complete the test. The number one program (PAQ8O8) takes 11+ hours and number four (PAQAR) even 17 hours to complete the test. WinRK, the program with the 2nd best compression (79.7%) takes about 8,5 hours. Not surprisingly all mentioned programs make use of a PAQ(-like) engine for compression. If you have files with embedded images (e.g. Word DOC files) use PAQ8, it will recognize them and separately compress them, boosting compression significantly. All mentioned programs (except the current number 2, WinRK) are free of charge.

- WinRAR, WinRK (Fastest ROLZ3) but especially FreeARC, SBC 0.970 and CCM score very well on efficiency, they have good compression and are reasonably fast. Both SBC and WinRK are capable of compressing the testset down to 82 Mb in times just over 4 minutes, FreeARC is even able to do this in 1m40. People looking for good (but not ultimate) and fast compression should have a look at the programs FreeARC (e.g. -m3 mode), WinRAR and SBC. The fastest of the 'high power' compressors in this test are CCM and FreeARC, they compress the testset down to 79 Mb(!) in just over 5 minutes. CCM and FreeARC have similar good compression, but FreeARC is faster.

- The speed-kings in this test are THOR, QuickLZ, SLUG and LZOP, the first two compress the total testset of 300+ MB in under 3.4 seconds (which generates a total data stream of 300 MB input + 150 MB output / 3.5 sec = 128 MB/s). At this speed IO/harddisk performance has a large impact on the compression speed, when compression from-to memory probably even higher compression speeds are possible. As a consequence of this speed compression ratio is not very high (barely 50%). Other fast compressors, with a bit better compression, are the oldies Pkzip, Arj and Gzip. If you are looking for a fast compressor with still good compression, consider UHARC 0.6b using LZP-mode (-mz) or FreeARC in m1 or m2 mode.

- Decompression speeds are usually a lot faster then compression speeds except for the compressors using a PPM or PAQ-engine. Decompression times of 5-7 seconds seem to be the systems (IO/hardisk) limit; compressing to ramdisk (or /dev/null) should probably be even faster...

- Normally I would expect 7-Zip to outperform WinRAR, but in this case I suspect WinRAR beats 7-Zip because there are (embedded) multimedia files (wav,bmp etc) which can be handled by WinRAR's multimedia filter. 7-Zip still lacks such a filter. I included multimedia files because many games are distributed in compressed form (games typically have lots of bmp and wav-like files).

- The higher the fraction of pre-compressed files the smaller the difference between the top and the other compressors. Because the testset contains files like chm,gif,jpeg the difference between the compressors looks rather small. I will probably make a 2nd testset with easy to compress non-multimedia data later on.