CPU Intel E7200 (Wolfdale)

In 2007, Intel successfully migrated to the 45-nm process technology and early in 2008 presented first products. Those wer 4-core Yorkfield CPU, and they were just the first to arrive in the retail. A bit later, Intel announces a release of a whole series of E8ххх CPUs based on the Wolfdale core. In particular, announced were the models E8200, E8400, and E8500 with the clock speed varying within 2.66 GHz to 3.16 GHz. All these processors run at the 333 MHz (1333 QPB) bus, and it is easy to see that the E8500 uses the fractional multiplier (9.5). The reason for emerging such multipliers is straightforward: due to its undisputable technological and architectural advantages over AMD's products, Intel does not need at all to take part in the races of clock speeds. Frankly, 65-nm products offer quite a substantial capability for clock speeds, and Intel might have rested on its laurels at least until late 2008. So, why is the "IT industry engine" speeding up a migration towards the finer process technology? The answer is straightforward: use of the 45-nm process allows Intel manufacturing more cores from a single wafer, thus reducing the prime cost and finally earning more profits.

However, quite a long time has passed until we were able to find the new E8ххх series in the retail. In fact, the new dual-core CPUs immediately turned a hard-to-get thing. Anyway, the situation is getting better, and the key factor was the release of the E7ххх series of 45-nm process which currently includes only one model - E7200. This series is being positioned as a "replacement" of the E4xxx series and is thus aimed at affordably priced middle-end systems.

Intel is going on with the division of processors over the market sectors in its traditional manner through reducing the clock speeds, the FSB speed, and also cutting down the L2 cache size. In particular, E7200 offers the clock speed 2.53 GHz, FSB = 266 MHz, the multiplier 9.5, and the 3 MB L2 cache size. In fact, this core is a Wolfdale with the L2 cache size cut down by half. To make the differences among the cores more amenable for perception, we have drawn up the following table:

Note that the 3 MB version of the Wolfdale core offers the same area and the number of transistors as the 6 MB version. That means both versions are cut from the same wafers and therefore offer the same capability for clock speeds. Therefore, physically the E7200 offers a 6 MB L2 cache, with its half "disabled". We also note that in the 7th series support for virtualization has been disabled. This is another proof that the E7xxx series is aimed at replacement of the E4ххх series in which this feature is also disabled.

The visual differences of the new processors are about the number and layout of capacitors:

The CPU-Z utility displays the following information:

With the specifications for E7200 included, this table looks like this:

We purposely called the release of the 7th series a key factor. As we can see, Intel is expanding its assortment and introducing 45-nm processors into the middle-end market. In particular, the recommended price for E7200 is set to $133. It looks like Intel has enough resources to fill this vast market sector. By the way, that should be a good news for AMD which has got so little time left to improve the attraction of its products. Because, if Intel keeps advancing towards the 45-nm process at this pace, then by the end of the year users will get the value series E5xxx as a replacement of the super popular E2xxx series, followed by a possible migration of dual-core Celeron processors to a new process early next year.

Overclocking

A few words on the overclocking. In theory, processors of the E8xxx and E7xxx series are highly attractive for overclocking. The finer process technology provides a higher clock speed limit and at the same time reduces the heat emission. If we look at E7200, it appears to be more preferable for overclocking because it offers a relatively low rated FSB speed (266 MHz) and, therefore, a high multiplier - 9.5. If we assume that the clock speed limit for the Wolfdale core is at about 4 GHz, then we attain this limit with the FSB speed set to 422 MHz. In case of using DDR2 memory, it will run at 844 MHz (with the minimum multiplier 1:1 selected). In other words, these conditions are feasible for most modern motherboards and DDR2 memory modules.

Now let's try the overclocking capability of the test processor.

With the core voltage (Vcore) slightly raised, we were able to attain a stable operation at the clock speed ~3.9 GHz. To overcome the 4 GHz bar, we had to raise the Vcore to 1.55V. However, we believe these results are good enough and quite promising. The thing is, we tested an "engineering" sample with an early stepping. Normally, later steppings (found in the retail versions of processors) provide higher overclocking capability under the lower Vcore voltage. Therefore, we believe the technological limit for the Wolfdale 3 MB is at about 4.3-4.5 GHz. Therefore, to attain the limit we would have to use a higher quality motherboard (which provides an operational stability at FSB = 500 MHz) with the "overclocker-friendly" DDR2 memory running at a frequency higher than 1000 MHz at least.

In fact, all this reasoning holds true for E7200 only. For models of higher rating (E7300, E7400), the requirements to the other components are not so rigid, whereas for process of low rating (E7100) they are going down.

Performance tests

Since the graphs are vast enough, we advise that you should note the following points. First, we tested the performance of E7200 versus E6700 which offers closest clock speed. Secondly, we examined the effect of difference in L2 cache sizes in Wolfdale - 3 MB versus Conroe - 4 MB. For that, we configured E7200 in a way so that it match the E6550. That is, we reduced the multiplier to 7 and raised the FSB to 333 MHz. Therefore, both the processors were running at the same clock speed 2.33 GHz, and the difference in speed depends solely on the architectural differences of the cores (primarily by the different sizes of the L2 cache). Finally, we compared the performance at overclocking.

In our test setup, we used the following hardware:

Let's first take a look at the results of synthetic benchmarks.

Now on to the gaming benchmarks (performance in fps).

Tests of application software.

Video encoding (DivX, Xvid) was measured in seconds, i.e. the less the better.

Data compression (WinRAR) was measured in Kb/sec, i.e. the more, the better.

Final Words

From the viewpoint of the regular user, the new 45-nm dual-core processors make a minor difference in terms of attraction as compared to the 65-nm predecessors. Their major advantages are about the lower heat emission level (at that, nor 65-nm products had anything to complain about), as well as support for the SSE 4.1 instruction set which should add to the performance in the optimized software. Similarly, users will get a performance gain in those applications whose performance depends on the L2 cache size.

If we come back to the 4-core processors, here 45-nm products are more attractive since with this number of cores heat emission plays a more important part.

As regards the PC enthusiasts, 45-nm processors appear to be the best choice for them. The thing is, the clock speed 4 GHz becomes quite attainable without extreme overclocking and without using high-quality components. For instance, the highest clock speed for the 65-nm processor (retail) E6550 is 3.85 GHz (at our test lab), while overclocking of the 45-nm E7200 engineering sample allowed us to attain the 4 GHz mark. Therefore, retail processors of the 7th series will overclock even better than that!