笔者还进行了开关超线程的测试，以对比在Nehalem-EX平台上超线程起到的影响力度：
SiSoftware Sandra Pro Business 2010 | ||||
|---|---|---|---|---|
测试对象 | 双路Intel Westmere-EP Xeon X5680 3.33GHz | DELL PowerEdge M910 四路Intel Nehalem-EX Xeon E7540 2.0GHz | 四路Intel Nehalem-EX | 四路Intel Nehalem-EX Xeon X7560 2.27GHz |
Processor Arithmetic Benchmark 处理器算术性能测试 | ||||
Aggregate Arithmetic Performance | 249.2GOPS | 298.6GOPS | 337.12GOPS | 451.7GOPS |
Dhrystone iSSE4.2 | 291.36GIPS | 350.47GIPS | 440GIPS | 518.5GIPS |
Whetstone iSSE3 | 207GFLOPS | 246.73GFLOPS | 234.22GFLOPS | 385GFLOPS |
Processor Multi-Media Benchmark 处理器多媒体性能测试 | ||||
Aggregate Multi-Media Performance | 465.8MPixel/s | 562.11MPixel/s | 752.77MPixel/s | 852.7MPixel/s |
Multi-Media Int x16 iSSE4.1 | 534.13MPixel/s | 646.86MPixel/s | 860.65MPixel/s | 979.21MPixel/s |
Multi-Media Float x8 iSSE2 | 397.47MPixel/s | 477.37MPixel/s | 644.9MPixel/s | 726.17MPixel/s |
Multi-Media Double x4 iSSE2 | 216.17MPixel/s | 260MPixel/s | 324.72MPixel/s | 394.66MPixel/s |
Multi-Core Efficiency Benchmark 多核效率测试 | ||||
Inter-Core Bandwidth | 84GB/s | 126.76GB/s | ||
Inter-Core Latency (越小越好) | 16ns | 76ns | ||
Cryptography Benchmark 加密解密性能测试 | ||||
Cryptographic Bandwidth | 11.72GB/s | 3GB/s | 4.46GB/s | 4.51GB/s |
AES128-ECB iAES Cryptographic Bandwidth | 21GB/s | 2.88GB/s | 4GB/s | 4.36GB/s |
SHA256 iSSE4 Hashing Bandwidth | 2.52GB/s | 3GB/s | 4.88GB/s | 4.66GB/s |
.NET Arithmetic Benchmark .NET算术性能测试 | ||||
Aggregate .NET Performance | 80.22GOPS | 89.47GOPS | 128.11GOPS | 134.76GOPS |
Dhrystone .NET | 37GIPS | 31.67GIPS | 101.25GIPS | 49.18GIPS |
Whetstone .NET | 123.43GFLOPS | 147.26GFLOPS | 155GFLOPS | 220.35GFLOPS |
.NET Multi-Media Benchmark .NET多媒体性能测试 | ||||
Aggregate .NET Multi-Media Performance | 71.4MPixel/s | 90MPixel/s | 107.7MPixel/s | 136.17MPixel/s |
Multi-Media Int x1 .NET | 100.36MPixel/s | 127.38MPixel/s | 167.4MPixel/s | 191.7MPixel/s |
Multi-Media Float x1 .NET | 42.42MPixel/s | 52.7MPixel/s | 48MPixel/s | 80.64MPixel/s |
Multi-Media Double x1 .NET | 78.48MPixel/s | 102.48MPixel/s | 89.36MPixel/s | 156.37MPixel/s |
总运算性能明显地下降了，不过多核效率测试神奇地可以运行了。少数测试在关闭超线程状态下运行得更好：SHA256 iSSE4 Hashing Bandwidth和Dhrystone .NET Arithmetic Benchmark。
