Intel’s CPU brand names can be confusing (part 1)
During the days of Intel’s Pentium 4 and AMD’s Athlon 64 CPUs, it was common practice to compare CPUs by looking at only one aspect: their clock rate. Just like nowadays, this was not really appropriate as there are many other characteristics which need to be taken into account to determine the performance of a CPU. In such a direct comparison, AMD’s Athlon always "lost" to the Pentium 4 since the Pentium was clocked at much higher rates. Despite the lower clock rate, the performance of the Athlon was close to that of the Pentium, sometimes also surpassing it. Motivated by this discrepancy, the marketing department of AMD renamed the Athlon brand CPUs. With the new naming conventions, AMD’s CPUs did not carry the clock rate in their name. They rather had a number assigned to them, such as "3800+" for a model clocked at 2.4 GHz. By doing so, AMD marketing wanted to express that this CPU was equal or better (hence the plus sign) than a Pentium 4 clocked at 3.8 GHz.
This rather aggressive marketing move is a perfect example of the naming intransparency that sometimes "shrouds" the CPU market and makes it difficult to get a clear overview.
In this article, I would like to have a look at Intel’s current range of server and desktop CPU brands, especially at wrong conclusions that one might draw due to the naming. Since most of the new brands have only recently been introduced, this is a good time to do so.
There is a myriad of things that can be considered when comparing CPUs. This article is intended to give a compact overview of the most important aspects, especially for people who cannot invest the time to constantly follow the rapidly evolving CPU market.
This is Intel’s top-of-the line brand. Still, there are three types of CPUs named "Core i7". The Bloomfield Core i7 was the first to enter the market. It uses socket 1366 and has clock ranges from 2.66 GHz to 3.33 GHz. The Lynnfield Core i7 uses socket 1156 and clocks at 2.8 GHz and 2.93 GHz. Both Bloomfield and Lynnfield have four physical cores and can handle four additional threads through HT. In tests, the turbo mode of the Lynnfield has stood out in a very positive way. The third Core i7 CPU has been released only recently, it’s the Gulftown line which is using socket 1336, has a clock rate of 3,33 GHz and six physical cores as well as the capacity to handle six additional threads through HT. While there are differences, any Core i7 CPU delivers extremely high performance and is a good choice.
Interestingly, some CPUs of the Lynnfield line are also sold as Core i5 CPUs. At this point of time, there is one clocked at 2.67 GHz and a high-efficiency version at 2.4 GHz. The remainder of the Core i5s are Clarkdale CPUs which, although sharing the same socket, differ massively from the Lynnfield line. The first difference is the number of cores, since there are only two. More interestingly, Clarkdale CPUs contain a majority of components that previously were included in the north bridge of a mainboard. The most important change is that Clarkdale delivers integrated graphics as well as the memory controller with the CPU. Whereas its 32 nm process and its thereby decreased power consumption might make the Clarkdale line look interesting for servers, there are arguments against that. First of all, for a CPU newly introduced into the market, two cores is a rather low number. Also, the memory controller is connected via QPI which causes higher latency times when accessing the RAM as compared to the Lynnfield line. In conclusion, the Clarkdale might be interesting for a certain niche server type, but it rather is intended for slim home and office computers and that is where it excels.
Currently, there are only two Core i3 CPUs and both of them are Clarkdale CPUs.
This is where things get really confusing. These two brands have been around since 1993 (1998 respectively) and thus comprise a great variety of CPUs from different lines and even architectures. There are Pentiums and Celerons from the Clarkdale line, but also Core- and Netburst-based models. It is important to know that Intel is using these two brands to offer well-priced CPUs and to distinguish them from the other, more expensive brands. This is interesting when one is looking for a bargain, because there sometimes are Pentium and Celeron CPUs which are not far from the expensive brands but which cost a lot less. For new servers, these shouldn’t be taken into consideration, though.
There is a strong discrepancy between the brand name and the actual line the CPU belongs to. While this is important for marketing reasons (future lines must "fit into" a brand), the current allocation of the different lines to brands is not intuitive.
The Core i7 brand contains three different lines, one of them also being sold as Core i5. Luckily, any Core i7 is a good choice, so this grouping creates no risk for the customer.
The same is not true for the Core i5. When one purchases a machine with a Core i5 CPU, one must really look into the details not to end up with a Clarkfield CPU which is much weaker than its Lynnfield brother.
The existence of the Core i3 brand seems not too reasonable, either. At this point of time, one might wonder why there is a need for a third brand when it only contains CPUs of a line which is also sold as Core i5. From a long term marketing perspective, it certainly is required, though, as there most likely will be lines only sold as Core i3.
Finally, when buying a new Pentium or Celeron CPU, one really needs to make sure what one is purchasing since the brand name pretty much is a wild card.
My personal bottom line (and this might not be the right thing for everybody) was to get a Core i7, which is the best choice at the moment and one can’t do anything wrong.
Next Part: Core 2 Duo, Core 2 Quad, Xeons, Atom and virtualization