今天给netburst最优化的时候忽然发现其效率对输入数值的大小敏感，因此决定再做一次针对短管的测试
和以前一样，需要64位操作系统
解压后依次运行exe文件，会得到一些txt文件，贴在这里即可
谢谢！

netburst的数据我已经测过了，本次测试主要针对K8（AMD64）和conroe（Core）


BTW
目前netburst的实测结果为：相比32位，每个WU节省11%左右的时间，不算显著，但也不错了

等这个测试完成后，用不了多长时间就可以发布一个BETA版的64位client了

没有操作系统......

又来,重起换系统ing
Batch Test | ShortPipeline Double m2m | V 3.0
CPU: Intel(R) Core(TM)2 CPU          6300  @ 1.86GHz
ID=116 Cycle=38 Count=6
ID=52 Cycle=38 Count=2
ID=55 Cycle=38 Count=14
ID=93 Cycle=38 Count=7
ID=85 Cycle=38 Count=3
ID=125 Cycle=38 Count=4
ID=105 Cycle=38 Count=10
ID=100 Cycle=38 Count=25
ID=99 Cycle=38 Count=5
ID=98 Cycle=38 Count=6
ID=44 Cycle=37 Count=25
ID=106 Cycle=38 Count=6
ID=45 Cycle=37 Count=20
ID=115 Cycle=38 Count=8
ID=51 Cycle=37 Count=20
ID=74 Cycle=38 Count=6
ID=54 Cycle=38 Count=10
ID=78 Cycle=38 Count=3
ID=73 Cycle=38 Count=3
ID=86 Cycle=38 Count=10
ID=48 Cycle=38 Count=3

Batch Test | ShortPipeline Double m2t | V 3.0
CPU: Intel(R) Core(TM)2 CPU          6300  @ 1.86GHz
ID=48 Cycle=38 Count=4
ID=78 Cycle=38 Count=2
ID=85 Cycle=38 Count=3
ID=86 Cycle=38 Count=20
ID=45 Cycle=37 Count=4
ID=99 Cycle=38 Count=3
ID=93 Cycle=38 Count=3
ID=44 Cycle=37 Count=12
ID=100 Cycle=38 Count=21
ID=55 Cycle=38 Count=14
ID=79 Cycle=38 Count=5
ID=20 Cycle=38 Count=2
ID=125 Cycle=38 Count=6
ID=51 Cycle=37 Count=21
ID=116 Cycle=38 Count=23
ID=54 Cycle=38 Count=14
ID=115 Cycle=38 Count=22
ID=105 Cycle=38 Count=5
ID=84 Cycle=38 Count=3
ID=74 Cycle=38 Count=2
ID=94 Cycle=38 Count=2
ID=35 Cycle=38 Count=2

Batch Test | ShortPipeline Normal m2m | V 3.0
CPU: Intel(R) Core(TM)2 CPU          6300  @ 1.86GHz
ID=175 Cycle=70 Count=6
ID=60 Cycle=70 Count=13
ID=107 Cycle=70 Count=5
ID=173 Cycle=70 Count=32
ID=180 Cycle=70 Count=16
ID=171 Cycle=70 Count=8
ID=181 Cycle=70 Count=20
ID=64 Cycle=70 Count=5
ID=182 Cycle=70 Count=23
ID=61 Cycle=70 Count=5
ID=176 Cycle=70 Count=8
ID=166 Cycle=70 Count=9
ID=51 Cycle=70 Count=2
ID=178 Cycle=70 Count=4
ID=105 Cycle=70 Count=3
ID=59 Cycle=70 Count=10
ID=114 Cycle=70 Count=6
ID=110 Cycle=70 Count=11
ID=57 Cycle=70 Count=3
ID=158 Cycle=70 Count=2
ID=112 Cycle=70 Count=2

Batch Test | ShortPipeline Normal m2t | V 3.0
CPU: Intel(R) Core(TM)2 CPU          6300  @ 1.86GHz
ID=207 Cycle=38 Count=13
ID=131 Cycle=38 Count=2
ID=75 Cycle=38 Count=3
ID=83 Cycle=38 Count=120
ID=197 Cycle=38 Count=7
ID=205 Cycle=38 Count=2
ID=43 Cycle=38 Count=13
ID=58 Cycle=38 Count=4
ID=190 Cycle=38 Count=2
ID=82 Cycle=38 Count=21
ID=208 Cycle=38 Count=11

Batch Test | SimpleDIV Normal m2t | V 3.0
CPU: Intel(R) Core(TM)2 CPU          6300  @ 1.86GHz
ID=0 Cycle=96 Count=200

这个是否代表没进步?

[ Last edited by momomodaTX on 2007-1-14 at 00:08 ]

的确没进步
我只是想看一下输入数据对周期的影响
不过Batch Test | ShortPipeline Normal m2m | V 3.0显示出像conroe这种中等长度管线CPU的两面性...
有点不甘心conroe的表现，明天我再写个新的SSE2版的测试发给你，再测一下
Conroe真是诡异...

Batch Test | ShortPipeline Double m2m | V 3.0
CPU: AMD Athlon(tm) 64 X2 Dual Core Processor 3600+
ID=89 Cycle=21 Count=13
ID=106 Cycle=21 Count=3
ID=117 Cycle=21 Count=82
ID=116 Cycle=21 Count=2
ID=125 Cycle=21 Count=39
ID=122 Cycle=21 Count=5
ID=113 Cycle=21 Count=3
ID=105 Cycle=21 Count=2
ID=124 Cycle=21 Count=6
ID=90 Cycle=21 Count=14
ID=98 Cycle=21 Count=8
ID=66 Cycle=21 Count=3
ID=88 Cycle=21 Count=4
ID=109 Cycle=21 Count=3
ID=97 Cycle=21 Count=3
ID=101 Cycle=21 Count=2

Batch Test | ShortPipeline Double m2t | V 3.0
CPU: AMD Athlon(tm) 64 X2 Dual Core Processor 3600+
ID=122 Cycle=21 Count=4
ID=77 Cycle=21 Count=3
ID=125 Cycle=21 Count=60
ID=90 Cycle=21 Count=69
ID=101 Cycle=21 Count=5
ID=88 Cycle=21 Count=6
ID=97 Cycle=21 Count=4
ID=89 Cycle=21 Count=9
ID=63 Cycle=21 Count=7
ID=98 Cycle=21 Count=7
ID=124 Cycle=21 Count=2
ID=113 Cycle=21 Count=3
ID=66 Cycle=21 Count=2
ID=117 Cycle=21 Count=3
ID=121 Cycle=21 Count=2
ID=96 Cycle=21 Count=2
ID=68 Cycle=21 Count=7

Batch Test | ShortPipeline Normal m2m | V 3.0
CPU: AMD Athlon(tm) 64 X2 Dual Core Processor 3600+
ID=9 Cycle=51 Count=135
ID=40 Cycle=52 Count=4
ID=93 Cycle=52 Count=18
ID=53 Cycle=52 Count=6
ID=17 Cycle=51 Count=13
ID=85 Cycle=52 Count=3
ID=183 Cycle=52 Count=2
ID=42 Cycle=52 Count=5
ID=143 Cycle=52 Count=2
ID=49 Cycle=52 Count=2

Batch Test | ShortPipeline Normal m2t | V 3.0
CPU: AMD Athlon(tm) 64 X2 Dual Core Processor 3600+
ID=183 Cycle=21 Count=4
ID=196 Cycle=21 Count=3
ID=149 Cycle=21 Count=4
ID=208 Cycle=21 Count=2
ID=157 Cycle=21 Count=115
ID=167 Cycle=21 Count=4
ID=154 Cycle=21 Count=14
ID=176 Cycle=21 Count=13
ID=185 Cycle=21 Count=4
ID=177 Cycle=21 Count=13
ID=165 Cycle=21 Count=4
ID=166 Cycle=21 Count=2
ID=199 Cycle=21 Count=2
ID=90 Cycle=21 Count=2
ID=195 Cycle=21 Count=2

Batch Test | SimpleDIV Normal m2t | V 3.0
CPU: AMD Athlon(tm) 64 X2 Dual Core Processor 3600+
ID=0 Cycle=75 Count=200

非K8和conroe的还有用么？不管怎样，贴上来吧。
Batch Test | ShortPipeline Double m2m | V 3.0
CPU: Intel(R) Pentium(R) D CPU 2.80GHz
ID=122 Cycle=58 Count=3
ID=104 Cycle=58 Count=17
ID=105 Cycle=58 Count=4
ID=85 Cycle=57 Count=10
ID=123 Cycle=58 Count=25
ID=106 Cycle=58 Count=9
ID=27 Cycle=59 Count=2
ID=108 Cycle=58 Count=11
ID=89 Cycle=58 Count=18
ID=119 Cycle=58 Count=4
ID=88 Cycle=59 Count=5
ID=118 Cycle=56 Count=6
ID=83 Cycle=59 Count=3
ID=121 Cycle=56 Count=5
ID=84 Cycle=58 Count=18
ID=103 Cycle=59 Count=5
ID=94 Cycle=59 Count=5
ID=86 Cycle=58 Count=8
ID=54 Cycle=57 Count=2
ID=102 Cycle=58 Count=2
ID=81 Cycle=59 Count=8
ID=82 Cycle=58 Count=5
ID=99 Cycle=58 Count=2
ID=74 Cycle=58 Count=2
ID=79 Cycle=59 Count=2
ID=109 Cycle=57 Count=3
ID=101 Cycle=57 Count=2

Batch Test | ShortPipeline Double m2t | V 3.0
CPU: Intel(R) Pentium(R) D CPU 2.80GHz
ID=108 Cycle=58 Count=16
ID=104 Cycle=58 Count=6
ID=119 Cycle=59 Count=10
ID=121 Cycle=56 Count=8
ID=74 Cycle=58 Count=3
ID=94 Cycle=57 Count=7
ID=86 Cycle=59 Count=14
ID=54 Cycle=58 Count=2
ID=123 Cycle=55 Count=24
ID=89 Cycle=58 Count=14
ID=106 Cycle=58 Count=27
ID=105 Cycle=58 Count=8
ID=109 Cycle=57 Count=10
ID=85 Cycle=58 Count=7
ID=84 Cycle=58 Count=13
ID=90 Cycle=59 Count=3
ID=122 Cycle=59 Count=3
ID=107 Cycle=60 Count=2
ID=101 Cycle=59 Count=2
ID=117 Cycle=59 Count=5
ID=52 Cycle=58 Count=3
ID=53 Cycle=58 Count=2

Batch Test | ShortPipeline Normal m2m | V 3.0
CPU: Intel(R) Pentium(R) D CPU 2.80GHz
ID=6 Cycle=142 Count=23
ID=21 Cycle=143 Count=10
ID=13 Cycle=143 Count=22
ID=18 Cycle=142 Count=8
ID=27 Cycle=143 Count=9
ID=3 Cycle=138 Count=15
ID=23 Cycle=139 Count=4
ID=78 Cycle=143 Count=13
ID=77 Cycle=140 Count=5
ID=74 Cycle=141 Count=2
ID=67 Cycle=142 Count=8
ID=28 Cycle=141 Count=5
ID=184 Cycle=144 Count=18
ID=68 Cycle=140 Count=2
ID=16 Cycle=141 Count=5
ID=66 Cycle=141 Count=3
ID=24 Cycle=143 Count=11
ID=8 Cycle=139 Count=5
ID=71 Cycle=141 Count=6
ID=65 Cycle=146 Count=3
ID=179 Cycle=142 Count=2
ID=76 Cycle=141 Count=3
ID=26 Cycle=145 Count=2
ID=73 Cycle=142 Count=2
ID=20 Cycle=137 Count=2
ID=70 Cycle=141 Count=2

Batch Test | ShortPipeline Normal m2t | V 3.0
CPU: Intel(R) Pentium(R) D CPU 2.80GHz
ID=164 Cycle=59 Count=38
ID=201 Cycle=61 Count=4
ID=187 Cycle=61 Count=8
ID=99 Cycle=61 Count=2
ID=193 Cycle=62 Count=4
ID=176 Cycle=61 Count=2
ID=156 Cycle=59 Count=12
ID=130 Cycle=62 Count=5
ID=161 Cycle=60 Count=6
ID=157 Cycle=61 Count=7
ID=209 Cycle=61 Count=4
ID=206 Cycle=60 Count=4
ID=184 Cycle=60 Count=26
ID=203 Cycle=59 Count=8
ID=124 Cycle=60 Count=2
ID=108 Cycle=61 Count=3
ID=207 Cycle=62 Count=3
ID=177 Cycle=60 Count=2
ID=202 Cycle=60 Count=2
ID=173 Cycle=62 Count=8
ID=192 Cycle=62 Count=2
ID=162 Cycle=60 Count=2
ID=195 Cycle=62 Count=2
ID=198 Cycle=60 Count=6
ID=205 Cycle=61 Count=4
ID=191 Cycle=60 Count=2
ID=136 Cycle=59 Count=5
ID=182 Cycle=63 Count=2
ID=208 Cycle=61 Count=4
ID=196 Cycle=62 Count=3
ID=200 Cycle=62 Count=2

Batch Test | SimpleDIV Normal m2t | V 3.0
CPU: Intel(R) Pentium(R) D CPU 2.80GHz
ID=0 Cycle=159 Count=200

是不是同时得有64位CPU和系统才能测试？

恩，不用再测这个了，数据够了