SiSoftware Sandra缓存内存测试主要包括内存带宽、内存延迟等性能的测试。
SiSoftware Sandra Pro Business 2010 | ||||
|---|---|---|---|---|
测试对象 | 双路Intel Nehalem-EP Xeon X5570 | 双路Intel Westmere-EP Xeon X5670 | 双路Intel Westmere-EP Xeon X5680 | DELL PowerEdge M910 四路Intel Nehalem-EX Xeon E7540 |
Memory Bandwidth Benchmark 内存带宽测试 | ||||
Aggregate Memory Performance | 33.86GB/s | |||
Int Buff'd iSSE2 Memory Bandwidth | 38GB/s | 35GB/s | 35.2GB/s | 33.86GB/s |
Float Buff'd iSSE2 Memory Bandwidth | 38GB/s | 35GB/s | 35.18GB/s | 33.85GB/s |
Memory Latency Benchmark(Random) 内存延迟测试(随机) | ||||
Memory(Random Access) Latency (越小越好) | 80ns | 83ns | 82ns | 192ns |
Speed Factor (越小越好) | 55.50 | 57.00 | 64.60 | 98.10 |
Internal Data Cache | 4clocks | 4clocks | 4clocks | 4clocks |
L2 On-board Cache | 11clocks | 10clocks | 10clocks | 10clocks |
L3 On-board Cache | 49clocks | 57clocks | 60clocks | 84clocks |
Memory Latency Benchmark(Linear) 内存延迟测试(线性) | ||||
Memory(Linear Access) Latency (越小越好) | 7ns | 7ns | 7ns | 41ns |
Speed Factor (越小越好) | 4.80 | 5.10 | 5.50 | 20.70 |
Internal Data Cache | 4clocks | 4clocks | 4clocks | 4clocks |
L2 On-board Cache | 10clocks | 11clocks | 11clocks | 10clocks |
L3 On-board Cache | 13clocks | 13clocks | 13clocks | 34clocks |
Cache and Memory Benchmark 缓存及内存测试 | ||||
Cache/Memory Bandwidth | 142GB/s | 183.26GB/s | 195.6GB/s | 315GB/s |
Speed Factor (越小越好) | 21.20 | 31.00 | 35.20 | 34.80 |
Internal Data Cache | 471GB/s | 663.51GB/s | 744.49GB/s | 919.66GB/s |
L2 On-board Cache | 295.4GB/s | 537.88GB/s | 611GB/s | 749GB/s |
L3 On-board Cache | 336.6GB/s | |||
从结果来看，内存带宽和内存延迟上M910都不占优势，这是由于：主频较低、内存存取通路较长、内存频率较低以及M910进采用了单内存控制器配置，Cache/Memory带宽表现良好则是因为新的环形总线架构。无法解释的是L3缓存的延迟，甚高，不太符合笔者之前的猜测，是否是M910的缘故？这要等笔者测完另外两台Nehalem-EX机器才能知晓了。
