System Specs & Memory Bandwidth

AMD Trinity System Specs
  • AMD A10-5700 (3.4GHz - 4.0GHz)
  • Asrock FM2A85X Extreme6
  • 8GB DDR3 RAM
  • Radeon HD 7970 GHz Edition
  • OCZ Vector 256GB SSD
  • OCZ Mk III Silencer 750w
  • Windows 8 64-bit
AMD Vishera System Specs
  • AMD FX-8350 (4.2GHz - 4.40GHz)
  • AMD FX-6350 (3.9GHz - 4.20GHz)
  • AMD FX-4350 (4.0GHz - 4.20GHz)
  • Asrock Fatal1ty 990FX Professional
  • 8GB DDR3 RAM
  • Radeon HD 7970 GHz Edition
  • OCZ Vector 256GB SSD
  • OCZ Mk III Silencer 750w
  • Windows 8 64-bit
Intel Ivy Bridge System Specs
  • Intel Core i7-3770K (3.5GHz - 3.9GHz)
  • Intel Core i5-3570K (3.4GHz - 3.8GHz)
  • Intel Core i5-3450 (3.1GHz - 3.5GHz)
  • Asrock Z77 Extreme11
  • 8GB DDR3 RAM
  • Radeon HD 7970 GHz Edition
  • OCZ Vector 256GB SSD
  • OCZ Mk III Silencer 750w
  • Windows 8 64-bit
Haswell System Specs
  • Intel Core i7-4770K (3.5GHz - 3.9GHz)
  • Asrock Z87 Professional
  • 8GB DDR3 RAM
  • Radeon HD 7970 GHz Edition
  • OCZ Vector 256GB SSD
  • OCZ Mk III Silencer 750w
  • Windows 8 64-bit
Intel Sandy Bridge-E System Specs
  • Intel Core i7-3960X (3.3GHz - 3.9GHz)
  • Asrock X79 Extreme11
  • 16GB DDR3 RAM
  • Radeon HD 7970 GHz Edition
  • OCZ Vector 256GB SSD
  • OCZ Mk III Silencer 750w
  • Windows 8 64-bit

The 9% increase in clock speed paired with the 17% increase in memory frequency, plus the other efficiency improvements the Ivy Bridge architecture provided over Sandy Bridge, allowed the Core i7-4960X to provide 17% more memory bandwidth than the i7-3960X. With a total bandwidth figure of 39.79GB/s, the Core i7-4960X blew everything out of the water and almost doubled the bandwidth of the i7-4770K.

While we saw a massive improvement in memory bandwidth performance, the cache performance gains were only very minor. With around 20MB/s more performance for the L1 and L2 caches the Core i7-4960X was around 4% faster. That gain can be mostly attributed to the increase in clock speed. When compared to the Core i7-4770K the i7-4960X provided significant improvement for the L2 cache, while the L1 cache was worlds slower.