Update 03/07/2007 – Today we are updating this article to include the recently released Conroe-L architecture upon which the Intel Celeron processors are based. The Conroe-L is a single core design featuring a small 512KB L2 cache, furthermore this series uses an 800MHz FSB. The early version we have obtained of this processor does not have its multiplier locked but future versions will have a locked multiplier. As is traditional with Celerons these new processors do not have Intel VT-x instruction support or SpeedStep. While the thermal design of the Core 2 Duo processors is 65w, these new Celeron processors have a thermal design of just 35w. As a result Intel is shipping these Celeron processors with a low-profile heatsink.
While we are updating this article to accommodate for the Conroe-L architecture, it is no longer an “Intel Conroe Cache Performance” article as previously titled. We have decided to rename the article to “Intel Desktop Processor Series Performance” as this article now provides an apples to apples comparison between the Conroe (4MB), Allendale (2MB), Allendale (1MB) and now the Conroe-L (512KB) cores. The idea is to compare these architectures at the same frequency (2.4GHz) to provide overclockers with an accurate means to determine which processor series will offer them the best performance when overclocked.
Currently Intel has a stack of Core 2 Duo dual-core desktop processors on offer and what was once just a family of four, has now become many more. The problem is, while they vary in price, there is also a very big difference between them all in terms of performance. This is because Intel has not only created a range of processors that operate at different frequencies, they also feature different front side bus speeds and L2 cache capacities.
The original group of Core 2 Processors includes the E6700, E6600, E6400 and E6300. All four processors operate on the quad-pumped 1066MHz FSB, yet it is only the E6600 and E6700 that feature the larger 4MB L2 cache. The E6300 and E6400 come with a smaller 2MB L2 cache which impacts performance. However, with the introduction of several new processors Intel has axed the E6300 and E6400. They are currently being replaced by the E6320 and E6420 processors, which will utilize the larger 4MB L2 cache. Therefore the 4MB L2 cache processors are now the E6700, E6600, E6420 and E6320.
Intel then went and released a new product family, which would all feature the smaller 2MB L2 cache. While the original E6700, E6600, E6420 and E6320 use the “Conroe” architecture, these new 2MB L2 cache processors are based on the “Allendale” architecture. This is actually a bit confusing really, as the Allendale core only differs from the Conroe by using the smaller 2MB L2 cache and an 800MHz quad-pumped FSB. Therefore, you could argue that the original E6400 and E6300 processors were in fact Allendale parts rather than members of the Conroe family.
Nevertheless, to keep things simple let’s just say that these new processors are part of the Allendale family. Featuring the 2MB L2 cache we have the E4300, E4400 and the upcoming E4500. Now to make things even more confusing, Intel has released another new processor series based on the Allendale core, but this time they have even gone away from the Core 2 Duo naming. Just recently Intel has quietly released the E2140 and E2160 processors featuring even less L2 cache, this time just 1MB.
However, these processors are not called the Core 2 Duo E2140 for example, rather it is the Pentium E2140 or as the box says Pentium Dual-Core E2140. So the Pentium is back at mild clock speeds and with very little L2 cache. The E2140 is clocked at 1.6GHz, while the E2160 comes at 1.8GHz, with the latter priced at just $97
US. This is interesting because the E4300 is also clocked at 1.8GHz, yet because of the larger cache it costs $117
US. Then there is the E6320 which operates at 1.86GHz and costs $165 US.
Update: Adding to the seemingly endless line up on Intel desktop processors we now also have the Celeron 420 (1.6GHz) which is priced at just $50 US, then at $59 US we have the 430 (1.8GHz). Finally for a little more the $65 US 440 comes clocked at 2.0GHz matching that of the E4400 which costs $139 US. So the new Celeron 400 series of processors are cheaper than the also recently released Pentium E2000 series, but not much cheaper!
So essentially what we have are four Intel desktop processors, all clocked around 1.8-2.0GHz, featuring four different L2 cache sizes (512KB, 1MB, 2MB and 4MB). Furthermore you will pay around 80% more for the E6320 than you would for the E2160, but with four times more L2 cache how much better is the E6320? Well this is something we really wanted to know and thought you guys might like to find out as well. Therefore we have taken the 440 (512KB), E2160 (1MB L2), E4300 (2MB L2) and E6600 (4MB L2) and clocked them all at 2.4GHz for a fair comparison.
Now keep in mind that all four processors have been selected because they use a 9x or greater clock multiplier. By running all four processors on the same 1066MHz FSB we get an even 2.4GHz out of each processor, meaning the only difference between each processor is the L2 cache size. This is the best way to do this comparison we feel for a few reasons. First of all we know that the Celeron 400, Core 2 Duo E4000 and E2000 series do not feature a 2.4GHz processor and that’s not the point.
Most of you reading this would more than likely plan to overclock these processors if you bought one. Now since a processor from the E2000 series is capable of overclocking to similar frequencies as an E6000 series processor, it is important to know how much the L2 cache will impact your performance once overclocked. For example, if you are chasing serious performance and want to reach 3.2GHz, there would be little point to save money on the E2160 when the E6320 kills it at 3.2GHz due to the larger L2 cache. That said, we do not know just yet how much difference the L2 cache makes when cutting it down to just 1MB, so let’s move on to find out...