Testing Part One: Stock Clock Speeds
The following system was used to test this memory:
| CPU: | Intel Core i7 3770k @ 4GHz |
| RAM: | Crucial Ballistix Tactical 2x8GB 1600-8-8-8 1.35v |
| Motherboard: | Gigabyte Z77x-UP4-TH |
| GPU: | MSI 4670 GDDR3 |
| Storage: | OCZ Vertex 3 3.5″ 120GB |
| PSU: | Antec HCP-750 |
| Case: | Motherboard tray benching station, ripped out of an old case |
For benchmarks I’ll be using the following:
- SuperPi Mod1.5 XS’s 32M test.
- WPrime 1.55’s 32m and 1024M tests.
- MaxxPi’s Maxxmem test.
We’ll go in that order too! First though, CPUz’s SPD tab, and the settings I get when I load the XMP profile:
Note the dual XMP profiles for 1.35v and 1.5v. All 1.35v DDR3 RAM must be able to run at 1.5v, or it won’t get the JEDEC seal of approval. Also note the 1.28v spec for lower speeds!
When I left everything on AUTO I got 1.5v and 1333-9-9-9, so unlike the last Crucial RAM I tested you’ll need to manually set the timings, or at least enable XMP.
Here we have XMP. The Command Rate in the SPD profile is 2T, most or all Z77 boards and IB chips can run 1T, many boards will default to 1T even when the SPD table says otherwise. If you can run 1T it’s worth doing so.
Moving right along, benchmark results at the clocks we just saw.
SuperPi Mod1.5 32M:
Solid. Slower than the last kit due to using windows 7 rather then windows XP, but good for 1600MHz ram and a 4GHz 3770k. Given identical OS’s, this kit is faster.
WPrime 1.55 32M and 1024M
Same deal as above, really. These numbers will be more relevant when we get to the OCing section.
MaxxPi’s Maxxmem
This is almost exactly the same speed, the tighter timings managed to offset the OS difference. There’s a lot to be said for tight timings, especially in the benchmarking world.
All told for this page, the RAM really does what it says it will, 1600MHz at 8-8-8 timings and 1.35v. Those are solid numbers.
On the next page, I’ll explore how far the RAM will go at various voltages, this is where things get interesting.
