Clarksfield vs Sandy Bridge: Throttlestop + Wprime 1.55
There has been a lot of hype surrounding Sandy Bridge CPUs and their performance jump over the first generation core i7 mobile CPUs (Clarksfield). We opted to use Throttlestop and Wprime 1.55 to put each generation of CPU to the test and see how they stack up. Although we did not use the top-of-the-line Sandy Bridge processor, one should keep in mind that clock for clock, the 2720qm is actually faster than a stock 940xm with 2MB less L3 cache so it balances out.
Intel introduced a new (and unwelcomed) feature with Sandy Bridge processors that many people were not (and still are not unless they read this article) aware of: They implemented a bitlock feature so that every OEM has the option of locking their extreme mobile CPU (2920xm) TDP. In contrast, the first generation extreme core i7 mobility processors did not have this restriction and thus third party programs like Throttlestop allowed the end user to raise the TDP as far as they wanted (within thermal limits) and thus increase performance many times over stock.
If you’re wondering what TDP is and how it relates to Intel Turbo Boost, we’ll give a brief description and advise those that want to learn more about it to visit our Throttlestop forum. TDP is Thermal Design Power and is the maximum amount of power a cooling system in a notebook must dissipate. To understand how TDP and Turbo Boost function for the first generation core i7 extreme processors like 920xm/940xm, we’ll quote UncleWebb’s ThrottleStop guide:
When your CPU is operating below these [TDP power limit] limits, you will get full turbo boost depending on how many cores are operating in the active C0/C1 state. When your CPU goes beyond either of these limits, turbo boost will immediately stop. Turbo boost can rapidly cycle on and off when you are near these limits which allows the CPU to limit its power consumption.
With that out of the way, lets talk about how we conducted our tests: The first set, we compared stock runs of each CPU in WPrime 1.55 using 8 threads and the 1024MB stability test. Since both CPUs have very little difference in base clocks, it could give us an idea of how the two compare. For the second test, we overclocked each CPU as far as we could (without though utilizing any external cooling) and then re-ran the 8 thread 1024MB stability test to compare. To get as accurate results as possible, we did our best to ensure that ambient temperature was similar for both systems.
Note: As you can see the 940XM cannot maintain its initial multipliers (26x) simply because the CPU’s heatsink was not designed to dissipate such amounts of heat and it progressively lowers them in order to remain within thermal limits. In cases like this, ambient temperature can seriously affect the final result.
Here we see the huge impact TDP adjustment has on a CPU – the 940xm which initially lost to the 2720QM at stock clocks was able to soundly beat it by utilizing Throttlestop to raise the TDP to finish nearly a minute faster than it’s stock score and about 9 seconds faster than the overclocked 2720QM. Of course there is a trade-off made when adjusting TDP because the CPU temps also skyrocket but as long as there is proper cooling, then an adjustable TDP is a god send for a mobile CPU. Intel plans to introduce adjustable TDP for Ivy Bridge (and Haswell) but we will have to wait and see if it will be as flexible or as effective as Throttlestop (we doubt it).