Synthetic Benchmarks: Memory
Lavalys Everest Ultimate v5.02
Everest Ultimate is the most useful tool for any and all benchmarkers or overclockers. With the ability to pick up most voltage, temperature, and fan sensors on almost every motherboard available, Everest provides the ability to customize the outputs in a number of forms on your desktop. In addition to this, the memory benchmarking utility provides a useful tool of measuring the changes to your memory sub-system.
Now before anyone asks, we selected DDR3-1066 for the Bloomfield setup because that is the highest officially supported memory speed on that platform. We could have selected DDR3-1600 for both P55 and X58 setups, but that would not have been representative of the platforms in their reference design.
Aside from the i7-975 which boasts very high write and copy speeds due to its 2666Mhz Uncore, all the i5 and i7 performs are roughly in the same ballpark. All the hysteria over the cut-down memory interface was for nothing. Lynnfield's dual-channel DDR3-1333 memory sub-system may only achieve 21.2GB/s of memory bandwidth, compared to Bloomfield's triple-channel DDR3-1066 25.6GB/s interface, but it is inherently more efficient. Those of you who own Core i7/X58 systems should try benchmarking them with only two modules installed, you will be surprised by the results. The i7-870 has a good bit more bandwidth compared to the i5-750, but that is largely due its faster Uncore (2400Mhz vs. 2133Mhz).
As you can see, one of the benefits of a dual-channel memory interface, especially when combined with an integrated memory controller, is low memory latency. Both the Lynnfield and Phenom II AM3 setups feature a dual-channel DDR3-1333 7-7-7-18-1T memory sub-system, which partly explains why their results are so similar. The 870 takes the lead from the 750 by a healthy margin due its faster Uncore.
Now let's see if ScienceMark echoes these results.
ScienceMark v2.0
Although last updated almost 3 years ago, and despite its rudimentary interface, ScienceMark v2.0 remains a favorite for accurately calculating bandwidth on even the newest chipsets.
ScienceMark reveals the i7-975 as having the greatest memory bandwidth, which is true if you average out the read/write/copy bandwidth figures found in the above Everest chart. Once again though, the non-Extreme Edition Core i5 and Core i7 processors are all revealed to have roughly the same memory bandwidth.
ScienceMark is evidently not a fan of AMD's Phenom II AM3 chips, but it definitely loves Lynnfield. With their low latency dual-channel memory interface, Intel's new mainstream offerings easily slides past the Bloomfield models, with the i7-870 even breaking 30 second mark.
Based on these results, we have to conclude that Lynnfield has a very robust memory sub-system. When you combine 21.2GB/s of memory bandwidth with an integrated memory controller's low memory latency, you have a winning solution. There is nothing to criticize here. In fact, we think most consumers will be glad to be able to settle for cheaper and plentiful dual-channel memory kits.
Synthetic Benchmarks: Memory
Lavalys Everest Ultimate v5.02
Everest Ultimate is the most useful tool for any and all benchmarkers or overclockers. With the ability to pick up most voltage, temperature, and fan sensors on almost every motherboard available, Everest provides the ability to customize the outputs in a number of forms on your desktop. In addition to this, the memory benchmarking utility provides a useful tool of measuring the changes to your memory sub-system.

Now before anyone asks, we selected DDR3-1066 for the Bloomfield setup because that is the highest officially supported memory speed on that platform. We could have selected DDR3-1600 for both P55 and X58 setups, but that would not have been representative of the platforms in their reference design.
Aside from the i7-975 which boasts very high write and copy speeds due to its 2666Mhz Uncore, all the i5 and i7 performs are roughly in the same ballpark. All the hysteria over the cut-down memory interface was for nothing. Lynnfield's dual-channel DDR3-1333 memory sub-system may only achieve 21.2GB/s of memory bandwidth, compared to Bloomfield's triple-channel DDR3-1066 25.6GB/s interface, but it is inherently more efficient. Those of you who own Core i7/X58 systems should try benchmarking them with only two modules installed, you will be surprised by the results. The i7-870 has a good bit more bandwidth compared to the i5-750, but that is largely due its faster Uncore (2400Mhz vs. 2133Mhz).

As you can see, one of the benefits of a dual-channel memory interface, especially when combined with an integrated memory controller, is low memory latency. Both the Lynnfield and Phenom II AM3 setups feature a dual-channel DDR3-1333 7-7-7-18-1T memory sub-system, which partly explains why their results are so similar. The 870 takes the lead from the 750 by a healthy margin due its faster Uncore.
Now let's see if ScienceMark echoes these results.
ScienceMark v2.0
Although last updated almost 3 years ago, and despite its rudimentary interface, ScienceMark v2.0 remains a favorite for accurately calculating bandwidth on even the newest chipsets.

ScienceMark reveals the i7-975 as having the greatest memory bandwidth, which is true if you average out the read/write/copy bandwidth figures found in the above Everest chart. Once again though, the non-Extreme Edition Core i5 and Core i7 processors are all revealed to have roughly the same memory bandwidth.

ScienceMark is evidently not a fan of AMD's Phenom II AM3 chips, but it definitely loves Lynnfield. With their low latency dual-channel memory interface, Intel's new mainstream offerings easily slides past the Bloomfield models, with the i7-870 even breaking 30 second mark.
Based on these results, we have to conclude that Lynnfield has a very robust memory sub-system. When you combine 21.2GB/s of memory bandwidth with an integrated memory controller's low memory latency, you have a winning solution. There is nothing to criticize here. In fact, we think most consumers will be glad to be able to settle for cheaper and plentiful dual-channel memory kits.
Last edited by a moderator: