by Ilya Gavrichenkov
12/27/2005 | 08:01 AM
It is not a secret for anyone these days that NetBurst architecture all Pentium 4 processors are based on is living through its last months. The launch of the “progressive” Prescott core manufactured with 90nm production technology revealed a number of serious problems. This was a clear indication that the architecture everyone pinned such great hopes and expectations for, is going through its final lifecycle stage. At first Intel engineers saw NetBurst as architecture with enormous performance potential, which they will be able to fully use with the time by raising the processor clock frequency. However, the noticeable frequency increase resulted into the ever growing power consumption and heat dissipation.
Moreover, the ongoing development of production technologies for semiconductor transistors doesn’t allow eliminating or bypassing the increase in the thermal parameters. As a result, the third generation of processors with NetBurst architecture aka Prescott has been remembered as one of the hottest CPU families. No kidding: the CPUs based on this processor core can consume and as a result dissipate up to 160W, while their clock frequency hasn’t gone over 3.8GHz. I have to stress that this high heat dissipation and power consumption resulted into a great number of other problems. Prescott processors required special mainboards with enhanced voltage regulator and new highly efficient cooling systems.
In fact, the heat dissipation and power consumption problems could be not so noticeable, if it hadn’t been for one thing. Prescott processors failed to demonstrate the expected superior performance that could make us overlook some of the inconvenient weaknesses like that. The performance level set by the competitor’s Athlon 64 solutions turned out unattainably high for the new Intel Prescott, therefore, these CPUs were steadily regarded as Intel’s big failure.
However, the fiasco of NetBurst architecture didn’t discourage Intel. This year the company postulated new priorities in the processor design strategy and started talking about a completely different direction for further CPU architecture development. They had to give up all hopes connected with the NetBurst architecture and get down to finding new solutions that will be highlighted in the middle of the coming year 2006. The major concept of the post-NetBurst era in Intel CPUs evolution will be the end of the megahertz race and primary focus on multi-core support and performance per watt coefficient.
However, a global change of the strategy like that will take some time and require certain preparative work to be done, therefore you don’t see that many new CPUs from Intel in the market today. At the same time, the company has to compete with AMD somehow that is why temporary transitional solutions are needed. This year we witnessed the arrival of the dual-core Intel Pentium D CPUs on the Smithfield core built from two dies with NetBurst architecture. These CPUs were Intel’s response to AMD’s dual-core Athlon 64 X2, which won high scores from the review sites as well as from the regular users very quickly. Frankly speaking, we wouldn’t call Smithfield CPUs a successful solution. They inherited all the major drawbacks of the NetBurst based processors, such as high heat dissipation and power consumption, which forced Intel to even drop their clock speeds down to 3.2GHz and lower. As a result, they ran slower than the Athlon 64 X2 competitors, but were still better than nothing.
So, Intel stopped investing time and effort into further development of the NetBurst architecture, but hadn’t yet completely given up the CPUs based on it, at least not until a worthy alternative is ready to go. Therefore, in early 2006 we will see some changes in the situation with the Pentium 4 and Pentium D CPUs. It will have a lot to do with the introduction of the finer 65nm production technology. In other words, in the beginning of 2006 we will see new Pentium 4 and Pentium D processors based on the new semiconductor dies manufactured with new technology process. Note that the new processors, also known as Cedar Mill and Presler, will not enrich the corresponding processor families with any new features. Sine Intel gave up further NetBurst architecture development, it would be strange to expect any innovations from them at this point. In other words, Cedar Mill and Presler are none other but the same Prescott dies but of smaller physical size thanks to finer production technology.
Cedar Mill core
So, there is actually nothing dramatically new about the upcoming Pentium D and Pentium 4, except for the 65nm production technology. This is primarily true for the Pentium 4 Cedar Mill core, as it will not differ from Prescott at all, except for the electrical and thermal parameters. As far as Presler is concerned, these processors will be slightly different from Smithfield from the architectural standpoint. Moreover, the new production technology will allow Intel to slightly speed up the dual-core CPUs due to higher clock frequencies they will support. Therefore, Presler is going to be the main hero of our today’s article.
Before we pass over to Presler processors, we would like to talk a little bit more about the 65nm production technology, which Intel finalized and made ready for mass production by the end of 2005, as we have all expected. Especially, since other semiconductor manufacturers are still quite far from reaching these heights. For example, Intel’s main competitor, AMD Company, is planning to introduce 65nm process only in the end of 2006.
What are the major peculiarities of the new Intel P1264 production technology that we should mention in the first place?
First of all I would like to mention the things that remained the same since the times of the 90nm P1262 production technology. Here I am talking about most of the equipment they use in their production lines for semiconductor dies. To cut the long story short, the new Intel 65nm process uses the same UV lithography with 193nm wave combined with phase shift masks technology. However, it didn’t prevent the manufacturer from reducing the effective transistor gate length to 35nm, which is about 30% smaller than in 90nm process.
Nothing has also been changed about the wafer diameter: just as during the 90nm cores manufacturing, they will use 300mm wafers. So, it means that Intel already has all the major equipment for 65nm process, which is installed and works fine. And this is a very important thing. For example, the transition to 130nm technology in 2001 went on very slowly because of the delays with lithographic equipment installation. This time, there are no problems like that. So, there is every reason to hope that new Intel 65nm CPUs will be widely available in the market in early 2006 already.
The materials used for transistor making also remained the same. However, some extra effort was applied to minimize leakage current. Strained silicon technology first introduced in 90nm production process got into its more advanced incarnation in the new 65nm process. While the insulator layer remained at the same 1.2nm, the transistor channels deformation increased by about 15%. As a result, this allowed minimizing leakage currents by about 4 times, thus leading to a 30% increase of transistor responsiveness without any heat dissipation growth.
Note that Intel once again rejected the SOI technology, which has been so successfully used by AMD. According to Intel’s specialists, this technology will not prove efficient for the currently used production technology.
And the last thing I would like to draw your attention to is the larger number of copper metal layers they will be using. There are eight of them in the new CPU, which is one layer more than in 90nm cores. This should help Intel simplify the upcoming dies design.
As we see, there is nothing revolutionary about the new production technology, just as there is nothing revolutionary about the architecture of the new Presler and Cedar Mill CPUs. Nevertheless, there are a few really interesting things about Presler that we should talk more about.
The first generation of Intel’s dual-core CPUs based on 90nm Smithfield core had a lot of drawbacks. This is actually not surprising at all, because the decision about the launch was made without the proper preparation. As a result, Smithfield processors made from Prescott cores turned out not very fast and economical, and at the same time appeared pretty pricy to manufacture. Of course, the die size of the Smithfield based CPU is 206sq.mm, which makes it very hard to achieve high production yields even with the polished-off manufacturing technology.
When working on the Presler design, Intel developers did their best to eliminate at least the major drawbacks of the dual-core Pentium D processors. Especially, since there was every reason for that: the upcoming launch of the finer 65nm production technology. However, you should keep in mind that it is impossible to significantly improve Presler core, because Intel doesn’t have any other processor architecture besides NetBurst at hand right now. As for NetBurst architecture, it cannot be improved as well, as we have already mentioned above. So, you should regard Presler as Intel’s best attempt to improve the Smithfield core without investing a lot of engineering resources into it.
The first simplest means of improvement in this case is certainly the new production technology. The introduction of the more advanced manufacturing process allows reducing the processor die size on the one hand and increasing its clock frequencies (or reducing the heat dissipation and power consumption) on the other. Since the typical heat dissipation of the top Intel Pentium D processors based on 90nm core has already been increased to 130W, it doesn’t make much sense to reduce it back again by shifting to the new production process. The entire infrastructure has already been adjusted to support this pretty high heat dissipation value. So, the new 65nm dual-core Presler processors will boast higher clock frequencies than their predecessors instead. While the maximum clock rate of the Smithfield based CPUs equaled 3.2GHz, Presler processors will be able to reach 3.46GHz bar.
Thanks to the new production technology Intel managed to increase not only the clock speeds, but also the number of transistors in the CPU. As a result, the L2 cache memory has grown bigger. While Smithfield CPUs featured two L2 caches, 1MB each, i.e. were composed of two regular Prescott cores, the new Presler processors acquired a twice as big L2 cache. So, Presler can actually boast two L2 caches, 2MB each. As a result, it actually means that Presler is in fact made of two Cedar Mill cores, which can also be regarded as 65nm analogs of the Prescott-2M cores used in Pentium 4 600 processor family.
Higher clock frequency and larger cache memory are actually the only two things Intel did to increase the performance of its new-generation dual-core processors manufactured with 65nm process. There are no other in-depth architectural improvements in Presler whatsoever. Therefore, the performance of Intel Smithfield and Presler processors working at the same clock speed may only be different if the application is sensitive to the size of L2 cache memory.
To prove this statement by some practical results, we decided to run a synthetic SiSoft Sandra 2005 benchmark, which is known to use pretty simple algorithms, which do not get affected by the size of the available L2 cache memory. We had a Presler CPU with the unlocked multiplier and enabled Hyper-Threading technology (Intel Pentium Extreme Edition 955), which we decided to compare against Smithfield (Intel Pentium Extreme Edition 840) in the same testing conditions. So, we set the frequency of our Presler CPU to 3.2GHz (16 x 200MHz).
The obtained results clearly indicate that all major execution units of our Presler CPU work exactly like those of the Smithfield manufactured with 90nm technology process.
However, it is still too early to claim that we have looked at all the differences between the Smithfield and Presler processors. One of Presler’s major advantages should become its new production scheme, which will allow Intel to significantly reduce the production cost of these processors. Unlike Smithfield based on a solid core, Presler will be built from two independent processor dies, which will be combined only on the packaging stage.
In other words, Intel doesn’t have to manufacture large dies from the very beginning. There will be two individual Cedar Mill cores under the Presler package, and these cores can either be used for the dual-core CPUs, or for the single-core Pentium 4 600. So, it would be more correct to call Presler not a “dual-core” processor, but a “doubled” processor.
Presler CPU without the packaging lid
Note that with the new opportunities for more cost-effective dual-core CPUs production Intel should very aggressive in delivering the volume to the market in 2006. The desktop segment is expected to be 70% dual-core by the end of next year.
By the way, this forecast makes us believe that Intel regards Cedar Mill as a sort of a side-effect product, with Presler being their flagship solution. However, you shouldn’t forget that NetBurst processor architecture will not last long and if things go as planned, these CPUs will leave the stage in the middle of next year already.
The Presler processor family includes a few Pentium D 900 CPUs and a Pentium Extreme Edition 955 solution for gamers and hardware enthusiasts. These processors differ by their clock frequencies, system bus frequency and availability or absence of the Hyper-Threading technology. Pentium D 900 processors support 2.8-3.4GHz frequencies and use 800MHz system bus. The ratings of the processor models in this family vary from 920 to 950 with the increment of 10. Moreover, Pentium D CPUs do not support Hyper-Threading technology, so the OS sees them as two logical processors.
As for the Pentium Extreme Edition 955, this CPU uses 1066MHz system bus, works at 3.46GHz clock speed and supports Hyper-Threading technology. As a result, the OS sees this process as four logical CPUs. These are exactly the features that determine its positioning for the high-end price segment.
Also it is important to point out that all Presler processors support 64-bit x86 EM64T extensions, EIST power management technology and Intel Virtualization Technology.
The table below sums up all the features of the CPUs in the Presler family:
We received the top model from the family for our tests – Pentium Extreme Edition 955. the processor was formally announced today, on December 27, 2005, however, it is not yet available for sale. The CPU will start shipping in mass quantities to OEMs and system integrators on January 16, 2006. This CPU will be priced at $999, just like any other solutions targeted for wealthy hardware enthusiasts.
Let’s take a look at the formal CPU specifications:
Pentium Extreme Edition 955
Typical heat dissipation
Max typical packaging temperature
2MB + 2MB
Hyper-Threading Technology support
Intel Extended Memory 64 Technology (EM64T)
Execute Disable Bit Feature (NX)
Enhanced Intel SpeedStep (EIST)
Intel Virtualization Technology (VT)
It catches your eye right away that the new 65nm manufacturing technology didn’t really affect the thermal and electrical parameters of the CPU. The working core voltage got only 5% smaller, while heat dissipation and maximum die temperature remained almost the same as those of the similar processor models manufactured with 90nm process. However, you shouldn’t forget that the new Pentium Extreme Edition 955 features a twice as large L2 cache as that of the Pentium Extreme Edition 840 and works at 8% higher clock frequency. So, there still is some effect from the 65nm production technology.
I would also like to point out that the default clock frequency multiplier of the Pentium Extreme Edition 955 CPU is set to 13x. It is a very remarkable thing, because Pentium 4 and Pentium D processors manufactured with 90nm technology supported clock multipliers over 14x. With the new production technology, both: Presler and Cedar Mill acquired 12x+ multipliers support. Besides the 3.46GHz CPU supporting 1066MHz bus, they have also managed to implement Intel Enhanced SpeedStep technology in all new processor models, including processors starting at 2.8GHz and up to the top solution we are testing today. As a result, when power saving technology gets activated, the CPUs with 800MHz bus drop their clock speed down to 2.4GHz, while the CPU with 1066MHz bus slows down to 3.2GHz.
The diagnostic CPU-Z tool reports the following about the Pentium Extreme Edition 955 processor:
Nothing unexpected here, the utility does recognize the CPU absolutely correctly.
I would like to say a few words about the compatibility of the Presler based processors with the contemporary mainboards, since there are a few really important things about it. First of all I would like to point out that the CPUs supporting 800MHz bus will work just fine in mainboards supporting 90nm Pentium D. As for the Pentium Extreme Edition 955 support, things are a little bit more tricky here. This processor works with a 1066MHz bus, which hasn’t been implied when they designed those chipsets for dual-core CPUs. Therefore, the launch of the new Intel Pentium Extreme Edition 955 CPU is also accompanied by the launch of the Intel 975X Express chipset.
I have to stress that there is nothing new about this chipset. There are only two features that distinguish it from the Intel 955X Express. Firstly, it officially supports Pentium Extreme Edition 955, and secondly, it allows splitting the PCI Express x16 graphics bus into two PCI Express x8 busses. As a result you can use the new mainboards based on the Intel 975X Express chipset to build multi-monitor configurations or SLI and CrossFire systems (in case there is proper video driver support, which is currently available only by ATI).
However, an i975X based mainboard is not a must for the Pentium Extreme Edition 955 CPU. According to our tests, this CPU can work perfectly fine in mainboards based on the older i955X chipset. For example, ASUS P5WD2 Premium worked with this CPU without any issues at all, so it is not always necessary to replace the mainboard if you are getting a new Pentium Extreme Edition 955 CPU.
Intel set the same typical heat dissipation level for the new Pentium Extreme Edition 955 as they did for the Pentium Extreme Edition 840: 130W. So, both these CPUs should have similar power consumption rate. In other words, all the recommendations we have previously given you regarding the powerful power supply units and cooling systems for the systems built with dual-core Intel processors, are also valid for the systems with Intel Pentium Extreme Edition 955. But this is just the theory.
In practice, it would be very interesting to measure the power consumption of this CPU. It is based on the core manufactured with new technological process, which should make it less power-hungry anyway. So, we had to run these tests.
The diagram below shows the power consumption rates of the Pentium Extreme Edition 955 side by side with the results taken for the two top-end AMD and Intel processors manufactured with the 90nm technology. As always, we used a special S&M utility to measure the maximum power consumption (you can download this utility here ). We measured the current that goes through the CPU power circuitry. So, the numbers given below do not take into account the efficiency of the CPU voltage regulator laid out on the mainboard.
Presler processors are really more economical than their predecessors. Of course, there was no significant improvement, but the new Presler processor working at a higher frequency than Smithfield and featuring larger L2 cache still demonstrated about 13% lower power consumption. From the thermal and electrical prospective, this puts the new Pentium Extreme Edition 955 processor on the same level with the top single-core Intel CPUs, which feature lower claimed typical heat dissipation of 115W. So, Intel has ensured that there is some extra reserve for maneuvers, namely that they do not have to be very strict about the dies selection for the top of the line dual-core CPUs.
On the other hand, all this enthusiasm about the lower power consumption of the 65nm processor cores vanishes when we try to compare it against the competitor from AMD. Well, this is the peculiarity of the NetBurst architecture: it is not economical and there is nothing we can do about it.
I would also like to say a few words about the power consumption of Presler processor in idle mode. The thing is that Intel guys specifically stress some innovations introduced in the 65nm manufacturing technology that allow reducing the power consumption in idle mode. Namely, there are special sleep transistors, which stop sending power to those parts of the die that are not working at the moment. However, our measurements show that the power consumption in idle mode is about the same by Presler and Smithfield and equals about 50W.
The new manufacturing technology will undoubtedly be of interest to overclockers. Especially, since it implies higher clock frequency potential of the new processor cores, as we have already mentioned above. So, even if Intel is not going to increase the clock rates of the CPUs with NetBurst architecture, we will do that for them.
I have to say that when I was preparing to start our overclocking experiments on Presler processors this time, we didn’t aim at reaching the maximum clock speed at any rate. it is evident that the new CPUs on 65nm cores will be able to reach the highest frequencies with appropriate cooling solution, of course. However, one of the major reasons why Intel decided to give up the NetBurst architecture is exactly the increasing heat dissipation of the CPUs as their frequency goes up. In other words, extreme overclocking fans should definitely go for Presler CPUs for their experiments. With highly efficient cooling solutions you can overclock Presler based CPUs to 5.5GHz, which we know from the first reports made by some lucky computer enthusiasts. We were trying to solve a slightly different task in our lab. Our main goal was to find that maximum frequency, which will be attainable for any Presler based CPU owner without superior cooling involved. So, all our overclocking attempts of the new Pentium Extreme Edition 955 were carried out with the regular default CPU cooler.
First of all, I would like to say that dual-core Pentium Extreme Edition CPUs are quite convenient for overclocking. Intel doesn’t lock their nominal clock frequency multiplier that is why you can overclock these processors not only by increasing the bus frequency, but also by raising the clock frequency multiplier. This justifies partially for the high price of these processors and makes overclocking much simpler.
We assembled a system on ASUS P5WD2 Premium mainboard for our tests. It was equipped with a pair of Corsair CM2X1024-6400PRO memory modules. Besides, there was an NVIDIA GeForce 6800 GT graphics card and a Western Digital WD740GD HDD installed.
First we tried to increase the clock the frequency multiplier. It turned out that even with the nominal cooler, which is not that dramatically efficient, our Pentium Extreme Edition 955 works just fine without increasing the Vcore at up to 15x multiplier setting. Note that the processor frequency grew up to 4GHz in this case, which is 15% above the nominal. I have to admit that dual-core processors on 90nm cores couldn’t boast the same success.
As for the further frequency increase, unfortunately the system would lose stability once the multiplier was set to 16x. However, we haven’t yet tried to adjust the Vcore. Once we increased the Vcore to 1.375V, the CPU easily went over 4.26GHz.
Further experiments showed that the CPU can work stably without throttling at up to 4.3GHz frequency. So, it looks like 4.26GHz appears a pretty nice variant for mainstream systems equipped with no special cooling solutions.
The important thing about overclocking the CPU by 25% is the fact that all other system knots continued working in the nominal mode, because the CPU clock frequency was reached by simply raising the frequency multiplier from 13x to 16x. And since we managed to achieve these results so easily, we decided it would make sense to include them into the performance analysis as well (Pentium Extreme Edition 955 working at 4.26GHz (16 x 266MHz).
As for the thermal mode, the maximum temperature of the Pentium Extreme Edition 955 processor overclocked to 4.26GHz (under full workload) reached 79o C. When we measured the CPU temperature in the nominal mode, that is at 3.46GHz, it never went over 65o C. So, this indicates clearly that further frequency increase will require more efficient cooling solution onboard.
During this test session we will compare the performance of the new dual-core Intel Pentium Extreme Edition 955 CPU against the top-end dual-core and single-core processors from AMD and Intel. In other words, Pentium Extreme Edition 955 will compete against AMD Athlon 64 FX, Athlon 64 X2, Intel Pentium 4, Pentium 4 Extreme Edition and Pentium Extreme Edition processors.
For our tests we assembled a few systems from the components listed below:
The BIOS Setup of the mainboards was adjusted for maximum performance.
As usual, we will test the performance of our systems in “general-purpose” application with the SYSMark 2004 SE testing suite. This benchmark emulates the user’s work in popular applications with multi-tasking of the operating system actively involved.
As we see, the new dual-core Intel processors work much faster than the older ones based on Smithfield core. It even allows Pentium Extreme Edition 955 to outperform Athlon 64 X2 4800+, which has always been unattainably fast for Intel dual-core processors.
PCMark05 is another popular test suite for “general” system performance. Since it creates multi-threaded workload during work, the CPUs with two cores perform the best of all in it.
First of all I would like to point out the high performance of the overclocked Pentium Extreme Edition 955. By raising the clock frequency of this processor to 4.26GHz, we put this CPU into the leading position in many benchmarks. And PCMark05 is no exception. However, even when Presler based CPU works at its nominal speed, the results appear quite high, so that it even surpasses the top competitor’s solution – AMD Athlon 64 X2 4800+. In fact, there is nothing surprising about it, as this test is optimize for NetBurst architecture.
We gave up the mp3 file speed encoding tests performed with the Lame codec in favor of Apple iTunes, because this application boasts quality optimization for multi-threading. Therefore, dual-core CPUs show their best here. If we do not take into account he results shown by the overclocked Presler processor, the leader here is AMD Athlon 64 X2 4800+. By the way, I would also like to stress that Pentium D 950 turns out faster than Pentium Extreme Edition 955, which means that Hyper-Threading technology has a negative effect on iTunes performance.
The recently released version 6.1 of the popular DivX codec has been optimized nicely for multi-threaded environments. As a result, if the single-core processors could compete with the dual-core ones on equal terms before, now the situation has got completely different. The advantage of dual-core processors is indisputable. As for the performance of our today’s newcomer, Pentium Extreme Edition 955, we cannot help giving due credit to it. This processor managed to outperform AMD Athlon 64 X2 4800+ by about 7%, and this baby is the today’s fastest dual-core offer from AMD. The overclocked Pentium Extreme Edition 955 gets another 19% faster.
The current XviD codec version is not so well-optimized for multi-threading support. Therefore, only the overclocked Pentium Extreme Edition 955 can compete with single-core testing participants, because its working frequency is higher than that of Intel’s top-of-the-line single-core Pentium 4.
Since Windows Media Encoder can use dual-core quite efficiently by creating two computational streams, the corresponding processors can boast much better results in this test than their single-core fellows. The best result here belongs to Athlon 64 X2 4800+ and even the overclocked Pentium Extreme Edition 955 cannot catch up with it. By the way, in Windows Media Encoder, just as in iTunes, Pentium D 950 appears faster than Pentium Extreme Edition 955, which once again indicates the negative effect of the Hyper-Threading on the result.
If the dual-core AMD Athlon 64 X2 4800+ has always been a leader in Photoshop, now the situation has changed. Pentium Extreme Edition 955 wins the title of the fastest CPU here. Even when running at its nominal speed, it outperforms AMD Athlon 64 X2 4800+ by about 5%.
The same is true for the performance in Adobe Premiere.
Although the final rendering in 3ds max can be easily split into a few parallel threads, we cannot say the same about the work in viewports. So, we see two completely different results in one and the same application. Dual-core processors are leading the race during final rendering, while in viewports the results is just the opposite. However, the overclocked Pentium Extreme Edition 955 working at 4.26GHz wins in both cases anyway due to its higher working frequency.
The tests in gaming applications turned into hell for us this time. The results we obtained during the test session didn’t have any logical explanation. Later one we found out that it was all happening because of the NVIDIA’s Forceware 81.95 driver. When NVIDIA introduced multi-threading support in their 80-series drivers, there emerged numerous issues causing unexplainable performance instability in 3D applications. In particular, these drivers do not always enabled multi-threading support, and the most remarkable thing is that enabling multi-threading doesn’t always result into performance increase in dual-core platforms. Therefore, NVIDIA recommends to disable multi-threading support for the current Forceware driver versions. Especially, since many games start acquiring multi-threading support of their own little by little, which leads to better performance than the corresponding support in the drivers can provide.
Therefore, we decided to run all gaming benchmarks with disabled multi-threading support in the driver.
The situation here is quite common. AMD CPUs provide better gaming performance than their Intel competitors. And even the Presler processor overclocked to 4.26GHz cannot outperform the single-core AMD CPUs. Although the overclocked Pentium Extreme Edition 955 managed to defeat the top dual-core Athlon 64 X2 4800+ model.
Note that the results of Quake 4 test on the diagram above refer to the first version of the game. However, the recently released Quake 4 patch version 1.0.5 added multi-threading support to this game. So, we decided to run one more test.
As we see, enabled multi-threading support had a really significant influence on the results. Athlon 64 X2 4800+ becomes 43% faster, Pentium Extreme Edition 955 – 13% faster, and Pentium D 950 - 39% faster. This is actually a clear indication that Hyper-Threading technology doesn’t do the dual-core processor any good. While in a single-core CPU it is quite efficient and helps improve the performance by about 9%.
Pentium Extreme Edition 955 (Presler) processor we have reviewed today left a very favorable impression. Especially against the background of its less successful predecessor – Pentium Extreme Edition 840 based on the 90nm Smithfield core.
First of all I would like to say that Intel improved the performance of its dual-core solution quite significantly. Thanks to the finer 65nm production technology, Intel increased the clock frequency of its new CPU and equipped it with the larger L2 cache. As a result, Pentium Extreme Edition 955 is not an eternal loser in the dual-core duel anymore. There are a lot of applications where Pentium Extreme Edition 955 managed to defeat AMD Athlon 64 X2 4800+, the top dual-core solution from Intel’s competitor. In fact, AMD processor remains the leader only in gaming applications, in professional OpenGL tasks and a few codecs. However, Athlon 64 X2 4800+ cannot be called a fully-fledged rival to Pentium Extreme Edition 955, because in about two weeks from now AMD will launch its new faster dual-core Athlon 64 FX-60 processor. Then we will talk again about the today’s fastest dual-core solutions.
Besides higher performance, the new Presler core of the dual-core Intel CPUs can also boast very high overclocking potential. We managed to reach 4.26GHz clock speed on Pentium Extreme Edition 955 easily without any additional cooling involved. And the experiments carried out by some hardware enthusiasts proved that even 5.5GHz is not the maximum for the new Intel Pentium Extreme Edition 955 on the Presler core. So, you can speed it up quite tangibly by simply overclocking to higher frequency.
However, we shouldn’t also disregard some drawbacks of the NetBurst architecture, which remained even after the transition to 65nm process. I am talking about high heat dissipation and power consumption of the new Presler based processors. Although more advanced production technology allowed to bring these parameters down a little bit, they still remained as high as those for the top-end single-core Prescott based CPUs. The typical heat dissipation of the new Pentium Extreme Edition 955 is even set at 130W by default.
In conclusion I would like to say that the major advantage of the Presler CPU for the manufacturer is its internal core structure composed of two individual dies. As a result, Intel managed to significantly reduce the production cost for these processors, which will help inexpensive dual-core Pentium D processors invade the market in 2006.