SiSoftware Sandra是一款可运行在32bit和64bit Windows操作系统上的分析软件,这款软件可以对于系统进行方便、快捷的基准测试,还可以用于查看系统的软件、硬件等信息。从Sandra 2007开始支持SSE4指令集。SiSoftware Sandra所有的基准测试都针对SMP和SMT进行了优化,最高可支持32/64路平台,这也是我们选择这款软件的原因之一。
SiSoftware Sandra Pro Business 2009 | |||
测试对象 | Snoop Filter Enable | Snoop Filter Disable | |
Processor Arithmetic Benchmark | |||
Dhrystone ALU | 97947MIPS | 97925MIPS | |
Dhrystone ALU vs SPEED | 36.82MIPS/MHz | 36.81MIPS/MHz | |
Whetstone iSSE3 | 72459MFLOPS | 70139MFLOPS | |
Dhrystone iSSE3 vs SPEED | 27.24MFLOPS/MHz | 26.37MFLOPS/MHz | |
Processor Multi-Media Benchmark | |||
Multi-Media Int x8 iSSE4.1 | 252.87MPixel/s | 252.04MPixel/s | |
Multi-Media Int x8 iSSE4.1 vs SPEED | 95.07kPixels/s/MHz | 94.75kPixels/s/MHz | |
Multi-Media Float x4 iSSE2 | 168.54MPixel/s | 168.41MPixel/s | |
Multi-Media Float x4 iSSE2 vs SPEED | 63.36kPixels/s/MHz | 63.31kPixels/s/MHz | |
Multi-Media Double x2 iSSE2 | 89.54MPixel/s | 89.44MPixel/s | |
Multi-Media Double x2 iSSE2 vs SPEED | 33.66kPixels/s/MHz | 33.62kPixels/s/MHz | |
Multi-Core Efficiency Benchmark | |||
Inter-Core Bandwidth | 21.00GB/s | 21.25GB/s | |
Inter-Core Bandwidth vs SPEED | 8.08MB/s/MHz | 8.14MB/s/MHz | |
Inter-Core Latency | 89ns | 89ns | |
Inter-Core Latency vs SPEED | 0.03ns/MHz | 0.03ns/MHz | |
Memory Bandwidth Benchmark | |||
Int Buff'd iSSE2 Memory Bandwidth | 4.22GB/s | 4.18GB/s | |
Int Buff'd iSSE2 Memory Bandwidth vs SPEED | 6.48MB/s/MHz | 6.42MB/s/MHz | |
Float Buff'd iSSE2 Memory Bandwidth | 4.16GB/s | 4.18GB/s | |
Float Buff'd iSSE2 Memory Bandwidth vs SPEED | 6.40MB/s/MHz | 6.42MB/s/MHz | |
Memory Latency Benchmark | |||
Memory(Random Access) Latency | 114ns | 113ns | |
Memory(Random Access) Latency vs SPEED | 0.17ns/MHz | 0.17ns/MHz | |
Speed Factor | 100.90 | 99.50 | |
Internal Data Cache Latency | 3clocks | 3clocks | |
L2 On-board Cache Latency | 18clocks | 18clocks | |
Cache and Memory Benchmark | |||
Cache/Memory Bandwidth | 71.76GB/s | 76.03GB/s | |
Cache/Memory Bandwidth vs SPEED | 27.63MB/s/MHz | 29.27MB/s/MHz | |
Speed Factor | 102.50 | 97.70 | |
Internal Data Cache | 421.73GB/s | 422.81GB/s | |
L2 On-board Cache | 120.50GB/s | 140.41GB/s | |
.NET Arithmetic Benchmark | |||
Dhrystone .NET | 27928MIPS | 28282MIPS | |
Dhrystone .NET vs SPEED | 10.50MIPS/MHz | 10.63MIPS/MHz | |
Whetstone .NET | 38915MFLOPS | 45356MFLOPS | |
Whetstone .NET vs SPEED | 14.63MFLOPS/MHz | 15.92MFLOPS/MHz | |
.NET Multi-Media Benchmark | |||
Multi-Media Int x1 .NET | 40.49MPixel/s | 39.99MPixel/s | |
Multi-Media Int x1 .NET vs SPEED | 15.22kPixels/s/MHz | 15.03kPixels/s/MHz | |
Multi-Media Float x1 .NET | 11.39MPixel/s | 11.27MPixel/s | |
Multi-Media Float x1 .NET vs SPEED | 4.28kPixels/s/MHz | 4.24kPixels/s/MHz | |
Multi-Media Double x1 .NET | 20.23MPixel/s | 19.98MPixel/s | |
Multi-Media Double x1 .NET vs SPEED | 7.60kPixels/s/MHz | 7.51kPixels/s/MHz |
在Sisoftware测试中,Snoop Filter的作用似乎没有刚才体现得那么明显。在处理器架构、内存带宽和.NET多媒体等三方面的测试中,Snoop Filter开启状态下有微弱的优势。不过在另外四个项目——处理器多媒体、.NET架构、内存延迟、缓存及内存等项目的测试中,Snoop Filter开启状态的性能还不如关闭状态,性能反而有些下降。