AES NI Acceleration

• Intel Core i7 5960X / 5930K / 5820K Haswell-E review: 8 cores at last
  1. Tested AES-NI Acceleration in Intel CPU can give ~5.3 times boost in AES
  2. Tested AES-NI Acceleration in AMD CPU can give 3.8-4 times boost

Hyperthreading

• Hyper-Threading may be Killing your Parallel PerformanceHT test on Windows 2008 R2 POV-Ray actually reduce performance
• Maximized performance: Comparing the effects of Hyper-Threading, software updates3D software benefit by turning HT off
• Hyper-threading on VMware vSphere (2010 on ESX4, ESX Server already aware of Hyperthreading and can balance load)
• Enabling HyperThreading for vSphere Servers (vSphere 4 & 5 can better use of Hyperthreading CPU resources. VCPU0, VCPU1 already on the 1st CPU’s 1st Core. VCPU2,VCPU3 use 2nd Core. etc…)
• VMworld 2012: vSphere Performance Best Practices (Hyperthreading add ~30% CPU performance?)

Application benchmark

• Photoshop Lightroom 5.3 benchmark on CPU raw conversion
  • i5 4670K OC (3.8Ghz base?) – 161
  • i5 4670K (3.4-3.8Ghz) – 181 (75% faster than X50 BE)
  • AMD X4 980 BE (3.7Ghz) – 266
  • * X4 X50 BE (3.1Ghz) – ~317
• Lightroom 4.4 Benchmark scalability test (CPU clock is more important than RAM speed)

Passmark CPUmark

• Passmark CPU High End
  • Intel Core i7-4790 4C8T @ 3.60-4.00GHz (10100) [Haswell-D]
  • Intel Core i7-4771 4C8T @ 3.50-3.90GHz (9941) [Haswell-D]
  • Intel Core i5-4690 4C @ 3.50-3.90GHz (7622) [Haswell-D]
  • Intel Core i7-5500U 2C4T @ 2.40-3.00Ghz (4261) [Broadwell-U]
  • Intel Core i5-4200M 2C4T @ 2.50-3.10Ghz (4094) [Haswell-M]
  • Intel Core i5-3230M 2C4T @ 2.60-3.20GHz (3936) [Ivy Bridge]
  • Intel Core i5-5200U 2C4T @ 2.20-2.70GHz (3717) [Broadwell-U, 15w TDP, 7.5w TDP-down]
• Passmark CPU High-Mid Range
  • Intel Core M-5Y71 2C4T @ 1.20-2.90GHz (3678) [Broadwell, Tablet]
  • Intel Core i3-4100M 2C4T @ 2.50GHz (3675) [Haswell-M]
  • AMD Phenom II X4 B55 4C @ 3.2Ghz (3479) [X2 555 BE unlocked]
  • Intel Core i5-4210U 2C4T @ 1.70-2.70GHz (3439) [Haswell-U, 15w TDP]
  • Intel Core i3-4000M 2C4T @ 2.40GHz (3271) [Haswell-M]
  • Intel Core M-5Y10 2C4T @ 0.80-2.00GHz (2752) [Broadwell, Tablet, 4.5w TDP]
  • Intel Core M-5Y10c 2C4T @ 0.80-2.00GHz (2568) [Same as 5Y10, Graphic clock base 100Mhz become 300Mhz]
  • Intel Core i3-4030U 2C4T @ 1.90GHz (2711) [Haswell-U]
  • Intel Core i3-2330M 2C4T @ 2.20GHz (2540) [Sandy Bridge]
  • Intel Core i3-4005U 2C4T @ 1.70GHz (2476) [Haswell-U]
  • AMD Phenom II X2 550 2C @ 3.1Ghz (2050)
• Passmark CPU Low-Mid Range
  • Atom Z8500 4C @ 1.44-2.24Ghz (1716) [Cherry Trail, Broadwell Intel HD Graphics]
  • Atom Z8350 4C @ 1.44-1.92Ghz (1500?) [Cherry Trail, Broadwell Intel HD Graphics]
  • Intel Pentium P6300 2C @ 2.27GHz (1346) [Westmere]
  • Atom Z3775 4C @ 1.46-2.39Ghz (1326) [Bay Trail-T, 2w Scenario]
  • Atom Z3740 4C @ 1.33-1.86Ghz (1069) [Bay Trail-T]
  • Atom Z3736 4C (922)
  • Celeron N2840 2C @ 2.16-2.58Ghz (1058) [Bay Trail-M, 4.5-7.5w TDP]
• Passmark CPU Low End
  
  • Celeron SU2300 1.2Ghz (914)
  • Atom Z2760 1.8Ghz (679)
  • Atom N570 1.66Ghz (579)
  • Pentium M 1.73Ghz (444)
  • Anandtech: Tegra 3 1.3Ghz ~33% slower than Atom Z2760 at benchmark in Windows RT
• Passmark Single Thread CPU benchmark
  • Celeron SU2300 1.2Ghz (486) – 100%
  • Atom Z3740 1.33Ghz (364) – (66% of SU2300 clock to clock)