by Ilya Gavrichenkov
01/21/2011 | 12:35 PM
Many computer users have mixed feelings about the merger of traditional processor and graphics accelerator functions within new Sandy Bridge microarchitecture. Those who do not intend to use integrated graphics, but believe that the “unnecessary” integrated graphics core has a negative effect on the processors pricing, are particularly unhappy about it. However, their concerns are not totally correct. Even though graphics does increase the die size of the quad-core Sandy Bridge die by about 20%, in reality unification and large production volumes easily make up for Intel’s investments into production of larger dies. However, when you are building a system with integrated graphics inside, having both – computational and graphics – cores inside a single semiconductor die has a ton of advantages. First, it allows to easily share some of the CPU resources, such as L3 cache and memory controller among different cores, which positively affects the integrated graphics performance. Second, by placing all high-frequency and high-performance units inside a single chip, Intel can significantly simplify the overall platform design and give up the use of external high-speed busses.
So, it looks like processors with integrated graphics should suit perfectly for compact computer systems of various purposes. And we have already seen proof of that from the Clarkdale processors released a year ago, which have a graphics core, North Bridge and computational resources inside a single processor packaging. A CPU with an integrated graphics core allows not only giving up “power-hungry” add-on graphics cards and saving some room inside the system case. Once the discrete graphics card and chipset North Bridge are gone, the CPU remains the only chip in need of cooling, which simplifies the design of cooling systems significantly. So, it is quite natural that small LGA1156 mainboards become very popular: dual-core Clarkdale processors inspire users to build miniature computer systems.
New Sandy Bridge processors are ready to go even further. They boast higher level of component integration, and their computational as well as graphics performance have increased. Besides, miniature LGA1155 mainboards will be coming out to the market shortly. Intel also actively participated in expanding this tendency by making sure that they also offer processors for new-generation compact computers. These CPUs have all the advantages of Sandy Bridge microarchitecture, including excellent performance, but boast more modest power consumption and heat dissipation, not exceeding 65 W.
So, if you are planning to put together a small but fast computer system, a special energy-efficiency Sandy Bridge modification may be the way to go. The model lineup of these processors featuring lower power consumption and heat dissipation includes several different models marked with “S” or “T” following the processor model number. “S” stands for 65 W thermal envelope, while “T” indicates 45 W TDP or even lower.
However, low power consumption is achieved at a price, and part of it is slightly lower clock frequency. In other words, energy-efficient processor models are slightly slower than their “fully-fledged” brothers with 95 W TDP. That is why the 65 W S-series seems to be most interesting choice for typical compact systems. Its TDP is low enough to allow putting these processors even inside the smallest Mini-ITX cases, while the performance differences from the regular Sandy Bridge processors should be not very dramatic. We decided to discuss a processor like that in our today’s review, and we picked Core i5-2400S – an average model in terms of speed as well as TDP.
The introduction of Sandy Bridge processors with an “S” suffix in their model name is no big news. Processor modifications with lower power consumption that had an “S” in their model name have been out there before. However, bow this CPU series finally has processors with an integrated graphics core, which makes them even more suitable for compact systems. Especially, since the declared 65 W TDP refers to both: the computational and the graphics core together. And if we take into account that the TDP of Intel H67 Express chipset required for the integrated graphics core to work is only 6.1 W, we can easily picture a fully-functional system built around a modern high-performance quad-core processor, which will only need a 100 W power supply.
It is extremely important that Intel decided not to raise the prices on their energy-efficient Sandy Bridge models: they are only $11 more expensive than their 95 W analogues. Although you shouldn’t forget that S-processors are also a little slower.
In fact, low price of the energy-efficient CPU models comes from the way they are manufactured. Intel doesn’t sort out better semiconductor processor dies, but simply lowers the processor core voltage and clock frequency. For example, the nominal clock speed for Core i5-2400S is 2.5 GHz, while Core i5-2400 works at 3.1 GHz default frequency.
20% lower clock frequency allowed Intel to reduce the CPU Vcore to 1.05 V for our Core i5-2400S model. Other than that our energy-efficient model and the regular processor with 95 W TDP have identical specifications.
Note that despite significantly different nominal clock rate, Core i5-2400S almost catches up with Core i5-2400 in Turbo mode. In other words, Turbo Boost technology is much more aggressive by the energy-efficient CPU, so it can boost its clock speed quite seriously when not all processor cores are utilized.
Here we see a very interesting picture. When all four cores are loaded with work, our energy-efficient Core i5-2400S will fall far behind his 95 W brother. However, if only one or two processor cores are utilized, Core i5-2400S will perform very close to the regular Core i5-2400 due to Turbo Boost kicking in. it is much more interesting to compare Core i5-2400S against Core i5-2300: it is absolutely impossible to tell which CPU is faster, as it will all depend on the type of operational load at a given moment of time.
In order to illustrate what it looks like in reality, we tested processors in Fritz chess benchmark, which can generate random number of threads to perform calculations.
If we compare the performance of an S-processor against regular LGA1155 CPU models, we will immediately notice a drastic change that happens when we have two or more threads in place. When only one or two cores are working, Core i5-2400S looks quite strong, but when we get to more computationally heavy tasks, it yields even to Core i5-2300. Deteriorated multi-core performance is the ultimate price you pay for lower power consumption and heat dissipation.
During our tests of the energy-efficient Sandy Bridge processor we tried to recreate its actual “habitat”, so we didn’t use any high-performance discrete graphics accelerators in our system, and utilized its integrated graphics core instead. Therefore, Intel H67 Express based mainboard from Gigabyte, GA-H67MA-UD2H, participated very actively in this test session. Unfortunately, this mainboard didn’t allow any overclocking, because the chipset working with the integrated graphics core wouldn’t let us adjust neither the CPU clock frequency multiplier, nor the memory frequency.
Our Core i5-2400S will be competing against Core i5-2400 and Core i5-2300 that do not belong to any specific series. Besides, we also included the results for the top LGA1156 CPU with the integrated graphics core – Core i5-680, which is one of the most popular choices for high-performance miniature systems among previous-generation products.
As a result, our testbeds used the following hardware and software components:
I would like to remind you that we tested all systems using integrated graphics, which makes more sense for energy-efficient systems. The discrete graphics card was used only in one case: in order to estimate the performance of Intel HD Graphics 2000 core integrated into Core i5-2400S processor.
We already have an idea about the performance we could expect from Core i5-2400S. Its practical performance depends directly on the workload type. Therefore, it is very interesting to see how this processor is going to perform in different real-life applications.
As usual, we use SYSmark 2007 suite to estimate the processor performance in general-purpose tasks. It emulates the usage models in popular office and digital content creation and processing applications. The idea behind this test is fairly simple: it produces a single score characterizing the average computer performance.
Although SYSmark 2007 load is not always multi-threaded, energy-efficient Core i5-2400S turns out slower than even Core i5-2300. There is only one strange thing about it and this is not the numbers on the diagram, but the numbers in the CPU model name: we really wonder why Intel decided to mark these processors with such high model names, which will most likely mislead quite a few users. However, it is important to keep in mind that the model numbers of S-series processors, not to mention T-series ones, have very different meaning than that model numbers of the regular Sandy bridge CPUs. Of course, Core i5-2400S didn’t fall behind the previous-generation platform, but it is a completely different CPU, nothing like the Sandy Bridge processors that have already impressed the users in numerous reviews all over the web.
We would like to add a table with detailed results in SYSmark 2007 to the diagram above. All scores are sorted according to the application type:
Now let’s proceed to tests in individual applications.
To test the processors performance during data archiving we resort to WinRAR archiving utility. Using maximum compression rate we archive a folder with multiple files 560 MB in total size.
We measured the performance in Adobe Photoshop using our own benchmark made from Retouch Artists Photoshop Speed Test that has been creatively modified. It includes typical editing of four 10-megapixel images from a digital photo camera.
We use Apple iTunes utility to test audio transcoding speed. It transcodes the contents of a CD disk into AAC format. Note that the typical peculiarity of this utility is its ability to utilize only a pair of processor cores.
In order to measure how fast our testing participants can transcode a video into H.264 format we used x264 HD benchmark. It works with an original MPEG-2 video recorded in 720p resolution with 4 Mbps bitrate. I have to say that the results of this test are of great practical value, because the x264 codec is also part of numerous popular transcoding utilities, such as HandBrake, MeGUI, VirtualDub, etc.
We use special Cinebench test to measure the final rendering speed in Maxon Cinema 4D.
The results of all tests in individual resource-hungry applications unanimously show that Core i5-2400S is not as fast as 95 W Core i5 processors from the new Sandy Bridge generation. And the higher is the operational load on the cores, the greater is the performance difference between them. And we are talking not about a slight difference but about pretty significant one: 65 W Core i5-2400S may be up to 10-20% slower than the 95 W Core i5-2400 CPU.
The clock frequency of new energy-efficient processors has been lowered significantly, but so far we have only been discussing the computational cores. Although Sandy Bridge graphics core is located on the same semiconductor die, it uses its own power lines and is clocked independently from the rest of the processor. Intel took advantage of this peculiarity and decided not to slow it down. As a result, even energy-efficient Sandy Bridge models boast a fully-fledged HD Graphics 2000 graphics core with six execution units and 850-1100 MHz clock speed. So, looks like the graphics performance of Core i5-2400S should be just as good as that of regular 95 W CPUs.
In fact, this is exactly what we see on the graphs above. Core i5-2400S gets very close to regular Core i5-2400 in games. Intel engineers reasoned soundly that graphics core is one of the key units of an energy-efficient processor and therefore its performance is a critical parameter. Most miniature systems use integrated graphics instead of discrete graphics accelerators specifically because for reasons of lowering power consumption and heat dissipation. That is why high graphics performance is an important trump for the S-series processors. And frankly, this performance is pretty good compared to the previous-generation graphics. Although the GPU inside Core i5-2400S is a junior Intel HD Graphics 2000 core with six execution units that cannot compete in performance with today’s discrete low-end graphics accelerators, it should be sufficient for miniature systems that are rarely used for gaming.
Energy-efficient Core i5-2400S turned out to be not as fast as we had expected it to be, considering that it is a Sandy Bridge product. For the sake of lower heat dissipation, its clock frequency has been significantly reduced, so there is nothing to be surprised about here. It is a different question, whether it was worth it. Is a slower Sandy Bridge processor a real energy-efficient CPU? Power consumption tests will help us to answer this question.
The graphs below show the full power draw of the computer (without the monitor) measured after the power supply. It is the total of the power consumption of all the system components. The PSU's efficiency is not taken into account. The CPUs are loaded by running the 64-bit LinX 0.6.4 utility. Graphics cores were loaded using FurMark 1.8.2 utility. Moreover, we enabled C1E and Enhanced Intel SpeedStep power-saving technologies to ensure that computer power draw in idle mode is measured correctly.
In idle mode there is no real difference between energy-efficient and regular processors. It is quite logical as in this case all power-saving technologies kick in and the clock frequencies of all Sandy Bridge processors drop to 1.6 GHz. However, Sandy Bridge platform consumes considerably less power in this mode than an LGA1156 platform: new processors distribute power better in idle mode by disconnecting the idling units and lowering the graphics core frequency, too.
However, when the computational cores are loaded to the full extent, we can clearly see the power advantage of Core i5-2400S processor. Lower clock speed and core voltage work their magic and the system with a Core i5-2400S inside needs 13 W less power than the system with a regular Core i5-2400. Of course, it is nothing like 30 W of difference in thermal characteristics between the two new processors, but still it is tangible.
By the way, speaking specifically about the CPU power consumption, we can offer you the following data from the System Agent for your reference:
In other words, 95 or 65 W are not even close to what we see. The calculated TDP is obviously declared with a substantial reserve, so even a regular Core i5-2400 consumes less than 65 W under heavy load, and Core i5-2400S can actually be assigned the lowest 45 W TDP just fine.
However, we haven’t yet checked out graphics load. Let’s see what happens when the integrated graphics core is busy processing 3D graphics.
Core i5-2400S and Core i5-2400 demonstrate almost the same power consumption in this test. So, it looks like the S-processors can only be called energy-efficient in relation to their computational cores, but not the integrated graphics. Which is exactly what we could expect taking into account that Intel HD Graphics 2000 core is the same in a regular and energy-efficient CPUs. By the way, both graphics cores, in Core i5-2400S and Core i5-2400, use the same GU voltage setting.
The results obtained under complex load involving all CPU resources are quite expected:
Core i5-2400S based system needs less power due to lower power consumption of the computational cores. Moreover, total maximum power consumption of the system including a mainboard with the CPU and memory, an SSD and a boxed CPU cooler fits within 70 W if we have an S-series processor, and within 80 W if we have a regular Sandy Bridge CPU. These numbers speak for themselves: the new LGA1155 platform poses as a great candidate for a compact and quiet computer system.
Another type of complex workload is HD video playback. They managed to almost completely unload the computational cores due to a special hardware video coder in the new Sandy Bridge processors.
Moreover, the LGA1155 system is overall surprisingly energy-efficient. It is not without reason that Intel claims Sandy Bridge to be a great choice for HTPC.
Power consumption during HD video content playback is only 6 W higher than power consumption in idle mode on any Sandy Bridge modification.
More than half of all desktop processors on Sandy Bridge microarchitecture boast lower power consumption. However, so far we haven’t seen any real benefit in such a strong focus from Intel on energy-efficient models.
Of course, we won’t deny that Core i5-2400S processor we tested today consumes less power than standard CPUs with 95 W TDP. And it has every chance to win those users who care a lot about saving energy and lowering system noise. Especially, since it is only $11 more expensive than the regular CPU.
However, if you decide to go with Core i5-2400S or any other similar processor from the S-series, you should keep in mind a few things. Most importantly, Intel doesn’t preselect better semiconductor dies for their energy-efficient models, but simply lowers the clock speeds, as simple as that. As a result, Core i5-2400S runs slower than the regular Core i5-2400, and sometimes this difference is substantial and can reach up to 20% under heavy load. However, the energy savings in this case are not that noticeable. Theoretically, the TDP difference between energy-efficient and regular processors is 30 W, but in reality we saw only 12 W at the most in a very limited number of usage scenarios. Moreover, there is simply no difference of any kind during HD video playback or 3D graphics processing.
Frankly speaking, Core i5-2400S not only performs worse than Core i5-2400, but also can’t boast the best performance-per-watt ratio, which is pretty strange for an energy-efficient model. On the other hand, even regular Sandy Bridge processors do not consume that much power to begin with and may suit perfectly fine for small and quiet systems. So, solutions like Core i5-2400S are mostly niche products rather than an appealing option for home and HTPC systems, because you can always go with a faster and cheaper Core i5-2300 instead of Core i5-2400S without any harm to the system power consumption.
In conclusion I would like to stress that you shouldn’t base your choice of an energy-efficient processor on its model number. As we have seen today, Core i5-2400S works slower that Core i5-2300, which means that Intel uses a different base for their S-series numbering. Obviously, T-series processors are marked using the whole different algorithm, so looks like the formerly crystal-clear model numbers hierarchy has been seriously disturbed, and definitely not in users’ favor.