Atom N450 processor: characteristics, photos and reviews. Test results and comparison with analogues
Information about the new processor has leaked to the Internet Intel Atom C3955, which contains 16 computing cores.
The new Intel Atom C3955 processor, codenamed Denverton, contains 16 cores and a clock speed of 2.1 GHz. The processor has 16 MB of second-level cache, i.e. one megabyte per core. With relatively low heat dissipation, the new chip is intended for NAS and other servers. Apparently, this will be one of the fastest processors in the Denverton line.
In the diagnostic and information utility SiSoft Sandra 2015, information was also found on the 16-core Atom C3955 chip. The Serve the Home website compared its performance results with other chips for the same application. The source also notes that the 16-core processor will most likely be delayed for a couple of months due to frequency problems identified in the Intel Atom C2000 series of processors.
Intel updates its Atom line
February 28, 2015To make it easier for people to understand processor performance levels and to better inform customers based on their needs, Intel has decided to rebrand its low-end processors.
Intel Atom processors will now be offered in three different product lines with performance levels of "good", "best" and "best". These chips will be called Atom x3, x5 and x7 respectively. This change will take effect with the new generation of processors.
Atom x3 processors will provide basic but sufficient performance in tablet PCs and smartphones. Intel Atom x5 will have more features and functions and will be aimed at people who need more performance. Flagship models Atom - x7 will provide the highest level of performance of this family.
Atom processors are designed by Intel to provide the longest battery life for mobile devices with increased performance in smartphones, tablets and other gadgets. The company introduced a new slide that explains the position of all model series processors. The slide includes basic Intel Atom, mid-range CPU, which consists of Core M for high-end laptops and more economical Pentium and Celeron, as well as high-performance Core line i.
14 nm Intel Braswell will be released in the third quarter
February 27, 2015Intel's new Atom processors with Braswell microarchitecture should be available in laptops and netbooks in the third quarter of this year. These chips will be released under the Pentium and Celeron brands, and will contain 4 or 2 cores.
The built-in graphics subsystem will be based on Low Power Gen 8. With its 16 execution units and support for DirectX 12 and Open GL 4.2, the new GPU will be capable of displaying images with a resolution of up to 4Kx2K.
The platform will support DDR3L at 1600 MHz in the SODIMM form factor and will be able to address up to 8 GB of memory, which is quite enough for this device segment. The platform will also receive 4x1 PCIe 2.0, 2 SATA 3.0 ports, as well as support for eMMC 4.51 and SD Card 3.01. In total, the platform provides 5 USB ports, 4 of which are USB 3.0 and one USB 2.0. And, of course, there is a high-definition audio processor.
Up to 3 displays can be connected to a Braswell-based system maximum resolution 4Kx2K. First of all, the eDP 1.4 standard will be supported with a resolution of up to 2560x1440 pixels; in addition, it will be possible to connect two more monitors via HDMI or DisplayPort.
Intel won't be able to supply 40 million CPUs for tablets
August 9, 2014Initially, Intel planned to supply 40 million processors for tablet computers in 2014. However, most likely, these plans will never come to fruition, since processors based on the Cherry Trail core have been postponed from November of this year to the first quarter of 2015.
The release of 14 nm Cherry Trail processors was initially scheduled for the third quarter. With this step, Intel wanted to speed up sales of its own CPUs for tablets. However, the company was forced to postpone their release twice, first to November and then to the first quarter of 2015, DigiTimes reports.
To popularize the production of tablets based on x86 processors, Intel decided to subsidize their production for large brand manufacturers. Intel's largest client in the tablet market is currently Asustek Computer. At the same time, Intel did not refuse to support Chinese white-box manufacturers, and a clear confirmation of this is budget tablet Kingsing W8 based on Bay Trail-T for $100.
Cherry Trail processors use 14 nm Airmont architecture and support 32 and 64 bit addressing for Windows and Android OS. Thus, the source notes, devices with new chips will not hit the market before February.
As a result, some observers believe Intel will be able to ship no more than 30 million tablet CPUs this year.
Intel is preparing Cherry Trail Atom by the end of 2014
December 10, 2013The next generation of desktop and mobile processors Atom family will be manufactured using a 14 nm process technology, is called Cherry Trail and is scheduled for release at the end of 2014. Intel is actively working to accelerate development of Atom chips, so the Broadwell and Cherry Trail laptop chips will be released in the same year, both using the 14 nm process.
A series of SoC Cherry View will be prepared for laptops, which is based on the new Airmont core. In turn, Cherry Trail will become processors aimed at tablet PCs. At the end of next year, most likely in September, a Moorefield architecture system-on-chip designed for smartphones will also be released.
Compared to Bay Trail TDP new platform should fall, thanks to the lower electrical losses of the 14 nm process technology, which means that developers will be able to offer more solutions based on Atom with passive cooling. In addition, the 14 nm process will mean another trump card for Intel in the fight against ARM, since next year the leaders of this market, including Qualcomm, Samsung and MediaTek, will only begin to use 20 nm nodes in their chips. However, Intel has yet to integrate its SoCs with LTE modems, which has traditionally been a difficult task. In fact, now only Qualcomm has a processor with a built-in LTE modem. So even the transition to 14 nm production will not make it much easier for Intel to compete in the smartphone market, and only in the future will we be able to find out whether device manufacturers will be interested in new Intel chips. There's still a whole year left to wait.
Intel May Kill Atom Desktop Brand
July 19, 2013Intel has high hopes for its quad-core Bay Trail D platform in terms of sales for the desktop PC market. But it seems that the new SoC may lose the Atom brand name, since according to rumors on the Internet, Intel will use the Celeron brand for all soldered-in BGA processors.
The list of processors includes the Celeron J1750, which will replace the Atom D2550 E, as well as the Celeron J1850, which will replace the 847 and 807 processors based on Sandy Bridge. The Pentium-branded J2850 chip will be faster than the Ivy Bridge Celeron 1007U, and both of these Bay Trail D processors in the BGA socket will appear in the fourth quarter of this year. At the same time, there should also appear mobile versions these processors.
This decision largest manufacturer chips seems quite justified, since Atom processors have long been associated with terribly slow mobile gadgets, like the netbooks of the past, as well as with embedded solutions. Now Intel is counting on the success of its new generation of Atom, and although we will not see such a name again, at least in desktop PCs, the developers have significantly improved the chip, making it quad-core and introducing a graphics core with support for DirectX 11.
AMD Opteron X targets Atom
June 3, 2013AMD doesn't seem to be able to compete successfully with Intel in terms of CPU power consumption, so the company has decided to bring the new Opteron X-series CPUs to market to compete on performance.
Most recently, AMD announced two new 64-bit Opteron processors, the X1150 and X2150 models, designed for microservers. Both models are part of the family codenamed Jaguar architecture, widely known for its presence in the new generation of gaming consoles from Microsoft and Sony.
Intel has a strong presence in the microserver market thanks to sales of the 6-watt Atom S1200 processor, and although AMD's new solutions consume 9 and 11 watts respectively, they have a number of advantages. The company positions its APUs as best solutions in general, thanks to the presence of four computing cores (compared to two for Atom), integrated AMD graphics Radeon HD 8000 in X2150 model, support up to 32 GB random access memory and built-in SATA ports. AMD processors are more expensive, $64 for the X1150 and $99 for the X2150, compared to Intel, which sells the Atom S1200 for $54. And although AMD’s proposal looks very interesting so far, its only competitor is already preparing to release 64-bit Atom SoCs with even lower power consumption, likely once again leaving AMD behind the scenes.
Intel ports Jelly Bean to Atom smartphones
September 26, 2012Intel has long promised to port Jelly Bean on smartphones with Atom processors.
We had absolutely no idea when this might happen, but Mobile Devices Group General Manager Mike Bell recently broke the news to PCWorld that Android 4.1 for Medfield is ready and running on Intel workers' devices. And although this interpretation of the OS is almost ready, its release date is still unknown.
Bell noted that phone manufacturers and suppliers will still have to go through a long process of adaptation and updating. Existing users will undoubtedly be upset to be both so close and so far from the new OS, but it is noted that manufacturers go through the same long journey when releasing ARM-based phones.
My next translation material. This time the hero is the Intel Atom C3958 processor, which was tested by the online resource servethehome. But don’t rush to close the page, because we will not be talking about a frail, weak something that is not very suitable for normal use, but about the relatively recently announced 3000th series of these processors (and, in fact, SoCs), aimed at use in storage data, embedded solutions, servers. So, Intel Atom C3958 – review and testing results of the top processor in this family.
Description and characteristics
The 3rd generation of the Atom processor family, codenamed "Denverton", includes quite a few a large number of models. The youngest processor has only 2 cores, but the oldest (which we are talking about now) boasts as many as 16 cores.
To a certain extent, we can say that there are at least 2 top models, these are the C3958 and its close relative - the C3955. I will give the main characteristics of both models.
| CPU | C3955 | C3958 |
| Number of Cores | 16 | |
| Number of threads | 16 | |
| Base frequency(Turbo Boost), GHz | 2.1 | 2.0 |
| Max. frequency, GHz | 2.4 | 2.0 |
| Max. memory capacity, GB | 256 | |
| Number of PCI-Express lines | 8 | |
| Max. quantity SATA | 16 | |
| Built-in LAN support | 4×10/2.5/1 GbE | |
| Intel® QuickAssist support | — | + |
| TDP, W | 32 | 31 |
| Recommended price, $ | 434 | 449 |
Actually, the differences are not very striking. Moreover, the C3955 has support for Turbo Boost, but the older Atom lacks such “turbocharging”. It would seem that it should not be a top model, but still its main difference from the C3955 is the support Intel technologies® QuickAssist.
Briefly about what QuickAssist, or QAT for short, is. This is a set of software and hardware tools to speed up encryption and data compression. QuickAssist is very helpful in cases where it is necessary to compress data on the fly, encrypt data streams, ensure the operation of cryptography, etc. In general, everything related to data protection, authentication, and security. QAT significantly speeds up applications, and quite significantly.
It should be noted that this useful feature is not included in every model. So the C3955 lacks it, although it has its advantages. QuickAssist was also used by Atom C2xxx series processors, but in the new generation the use of the technology has increased high level. So, unlike Atom C2xxx, C3xxx does not require a special driver. In testing, the QAT feature was enabled, although it was not used in the tests below.
Actually, the presence of QAT is perhaps the only argument in favor of the C3958 and not the C3955, although the reason is very compelling. If the tasks performed do not involve the use of encryption, data compression, in general, for what this technology is needed, then there is no point in choosing the C3958.
The fact that this is a server product is indicated by the characteristics of the processor. There is support for a large amount of memory, and the presence of a 16 MB L2 cache (1 MB for each core), ECC, 4 10-Gigabit interfaces, 16 SATA devices, virtualization technologies VT-x, VT-d and etc. By the way, this processor is not supplied to customers as separate component, but only as part of at least the motherboard.
For those who are interested, we present the result of executing the Linux command lscpu, which displays detailed information about the processor and all its features.
Test bench
To carry out the test, the following configuration was assembled:
- Motherboard: Gigabyte MA10-ST0 with Intel Atom C3958 processor soldered on it.
- Memory: 4x 16GB DDR4-2400 RDIMMs (Micron).
- SSD: Intel DC S3710 400GB.
- Boot device: Intel DC S3700 200GB.
A little more about the motherboard. It is very interesting for building data warehouses. “On board” it has 4 slots for installing memory, 32 GB eMMC flash memory manufactured by Kingston, 2 10-Gigabit SFP ports and the same number of Gigabit network ports. There is a PCIe x8 connector, as well as 4 SFF8087 connectors for connecting 16 SATA drives.
A detailed review of this motherboard will be coming soon, but for now we can say that the maximum consumption with two 10Gb SFP+ connections and two connected Gigabit interfaces was 61 W.
Test results
We used our old, tested Linux-Bench scripts. We have a more recent selection of scripts, but in in this case it seemed not so necessary, since the main purpose of this platform is built-in applications. When using such a configuration in data warehouses or in network devices embedded applications do not have a high load, and the use of extended AVX2 and AVX-512 command sets seems unnecessary.
In our past tests, we were convinced that the best OS for Intel Atom C2000 series processors are Linux and FreeBSD. Windows is not widely used on such platforms and we do not recommend using this platform as a regular computer. There are plenty of other, more advantageous options for this.
Python Linux 4.4.2 Kernel Compile Benchmark
We use this test often. Use standard configuration file, the Linux kernel 4.4.2, taken from kernel.org, and the standard generated configuration loads every thread in the system. The results show the number of compilations per hour.
The results obtained showed very good performance, which is comparable to the results of an 8-core Xeon processor D. Model C3955 showed a little top scores. This is not surprising; nevertheless, differences in microarchitecture should manifest themselves in the operation of processors.
c-ray 1.1
Another ray tracing test we constantly use, it is very popular and shows the difference in performance in multi-threaded systems.
The performance shown is good here too. As expected, the more “nimble” and even turbocharged C3955 showed better results. Interestingly, the Intel Xeon E3 showed similar performance, but it does not have many of the features that the Atom has, and it also consumes more power.
7-zip Compression
A very popular and frequently used cross-platform application for archiving/unzipping data.
The results obtained are very good. Of course, 16 Atom cores are not 16 Xeon D cores, and you won’t be able to compete with the latter. In this case, QAT is not used, and this could significantly change the results, as we will see shortly. If we talk about performance, then in terms of compression speed, the Intel Atom C3958 can be placed somewhere between 6 and 8-core Xeon D. Decompression speed is somewhere between 8 and 12-core Xeon D.
Sysbench CPU test
Another popular test on the Linux platform. We used the CPU test, and not the OLTP test, which is used when testing drives.
We had to remove the results from the C2358 and D525 processors due to the low values, which would have made the graph difficult to read. The test scales well and perfectly loads all available processor cores. It is not surprising that 16 cores came very well.
OpenSSL
A cryptographic package used to encrypt communications between servers. We got the following result.
When we checked again, we got the following (we sorted the results in the same order as in the first test run to make it more convenient).
As we can see, the Intel Atom C3958 competes with the similarly priced Xeon Silver 4108, which is designed for more powerful servers. But more interesting in this case is the comparison with the previous 2000 generation of Atom processors. The top-end C2758 with QAT enabled turned out to be 4 times slower than the C3958, which did not use this function. This is important because OpenSSL is often used in network devices and storage systems.
UnixBench Dhrystone 2 and Whetstone Benchmarks
The tests are old, but for now we continue to use them due to popular demand. UnixBench Dhrystone 2 results.
Whetstone Benchmarks results.
In this case, we see a clear benefit from multi-cores, since in this case it compensates for the compromises in the microarchitecture that had to be made to reduce power consumption. In this case, the option is “by number, not by skill.”
Conclusion
This is not at all the “Atom” that immediately comes to mind when mentioning this family of processors. The base frequency of the Atom C3958 is not so high for our times, there is no support for “Turbo Boost” technology, no third-level cache, no support for the AVX2/AVX-512 instruction set, but 16 cores, 1 MB of L2 cache per core, are significant Improvements in IPC (Inter Process Communications) allow it to compete in performance with Xeon D and Xeon Bronze/Silver.
Naturally, the latter are more suitable for virtualization and general use, but in networking and storage devices, atomic processors are very good.
Nowadays there is a lot of talk about AMD EPYC, but AMD does not have its own solutions that can compete in this segment in terms of the totality of characteristics. Thus, the EPYC 7251 has a TDP of 120 W (compare with Atom), having 8 cores, 16 threads, however, supporting an increase in frequency up to 2.9 GHz. True, AMD has no goals to occupy its niche in this segment, at least not with EPYC.
ARM has been active, but the combination of performance and the use of technologies for accelerating cryptography functions and data compression, which is found in the 3000 series of Atom processors, allow Intel to feel confident in the near future.
If we consider top solutions with QAT support, we can see significant progress compared to the previous generation (Atom C2758). The only thing that has decreased is the clock speed (by about 17%). The rest are continuous improvements. Judge for yourself, the number of cores has doubled (from 8 to 16), the cache and maximum memory have quadrupled (to 16 MB and 256 GB, respectively), PCIe has updated the generation, and support for a 10-gigabit network has appeared. But for the significantly increased performance we had to pay for increased TDP.
Unfortunately, prices have increased, and significantly. True, a wide range of models allows you to choose an option (for example, Atom C3758) that is cheaper and can successfully replace the previous top-end processor in relevant areas of application.
The good picture of significantly increased performance is only spoiled by the price, because at $449, the Atom C3958 competes with the Intel Xeon Silver 4108 and Xeon D lines, and these, whatever one may say, are birds of a slightly different feather.
Intel Atom are processors for inexpensive and small laptops, netbooks, nettops and tablets/smartphones. Their architecture made them energy efficient and not at all expensive.
The Atom series initially includes two families: the Z series (codenamed Silverthorne) for tablets and some nettops, and the N series (codenamed Diamondville) for more traditional netbooks and nettops. Both families are manufactured on the 45nm process and include support for MMX, SSE, SSE2, SSE3, SSSE3, Intel 64, XD-Bit and IVT. High-end models also support Hyper-Threading.
The fastest Intel Atom processors outperform Celeron processors. For example, Atom 1.6 GHz is quite comparable to Pentium M 1.2 GHz.
Towards the end of 2009, Intel introduced the second generation of Atom processors - Pineview. They were equipped with GMA 3150 graphics and a DDR2 memory controller. The 45nm Atom N450 and N470 were very popular back in the day, as was the N280 before it. The most latest models lines include support for DDR3 memory (such as N455) and dual-core options.
The Oak Trail platform (32nm process) was introduced in 2011 and is directly descended from Silverthorne. It is intended for tablets and netbooks, its index is Z600. The core is very similar to the Pineview series, but the system-on-chip now includes GMA 600 graphics from PowerVR.
Modern Intel Atom processors
Saltwell (32 nm), 2012-2013
Penwell (32 nm), 2013-2014
Cloverview (32 nm), 2013
Cloverview (32 nm), 2013
Cedarview (32 nm), 2011-1012
They are part of the Cedar Trail platform. Built-in graphics provide 1080p video playback, screen resolution up to 2560x1600 pixels.
Cedarview-M (32 nm), 2011
Supports up to 2 GB of DDR3-800 RAM.
Merrifield (22 nm), 2014
Energy consumption is 4.7 times less than Saltwell. Two Silvermont cores, graphics core - PowerVR G6400. Memory controller LPDDR3-533 up to 4 GB.
Bay Trail-T (22 nm), 2014
The performance increase compared to Clover Trail is 50-60%. Have low power consumption. Graphics (Gen 7) in chips without the D index support a resolution of 2560x1600 pixels, with a D index - 1920x1200. Memory controller - LPDDR3-1066 up to 4 GB. All processors are quad-core. No Hyper-Threading support.
| Model | Cache | Clock frequency - Turbo, GHz | Cores/threads |
| Intel Atom Z3795 | 2 MB | 1,59-2,39 | 4/4 |
| Intel Atom Z3785 | 2 MB | 1,49-2,41 | 4/4 |
| Intel Atom Z3775 | 2 MB | 1,46-2,39 | 4/4 |
| Intel Atom Z3775D | 2 MB | 1,49-2,41 | 4/4 |
| Intel Atom Z3770 | 2 MB | 1,46-2,4 | 4/4 |
| Intel Atom Z3770D | 2 MB | 1,5-2,41 | 4/4 |
| Intel Atom Z3736F | 2 MB | 1,33-2,16 | 4/4 |
| Intel Atom Z3736G | 2 MB | 1,33-2,16 | 4/4 |
| Intel Atom Z3745 | 2 MB | 1,33-1,86 | 4/4 |
| Intel Atom Z3745D | 2 MB | 1,33-1,83 | 4/4 |
| Intel Atom Z3740 | 2 MB | 1,33-1,86 | 4/4 |
| Intel Atom Z3740D | 2 MB | 1,33-1,83 | 4/4 |
| Intel Atom Z3735D | 2 MB | 1,33-1,83 | 4/4 |
| Intel Atom Z3735E | 2 MB | 1,33-1,83 | 4/4 |
| Intel Atom Z3735F | 2 MB | 1,33-1,83 | 4/4 |
| Intel Atom Z3735G | 2 MB | 1,33-1,83 | 4/4 |
| Intel Atom Z3680 | 1 MB | 1,33-2,0 | 2/2 |
| Intel Atom Z3680D | 1 MB | 1,33-2,0 | 2/2 |
A year ago, Intel announced the release of a new series of processors - Atom. The new CPUs are intended exclusively for mobile computers, and their characteristics fully comply with all the requirements of this type of device. This primarily applies to power consumption, which does not exceed 4 W (TDP). Such low performance is achieved due to the new architecture, which is not similar to any of the previous ones Intel architectures, although it includes their individual features. The core consists of 47 million transistors, and since they are manufactured using a 45-nm process technology, it becomes clear why the Atom is such a compact and cost-effective processor. Currently, Intel has two series of Atom processors. The first is called Z (Z500-Z540 processors), it is based on the Silverthorne core and is intended for mobile systems class MID (Mobile Internet Devices). The second series based on the Diamondville core was announced relatively recently (in March of this year) and includes two models (N270 and 230). It is intended for desktop systems(Nettops) and budget laptops (Netbooks).
| Core | Frequency, GHz | FSB, MHz | L2, kb | TDP, W | Technical process, nm | Core area, mm 2 | Number of transactions (million) | |
| Atom Z500 | Silverthorne | 0,8 | 400 | 512 | 0,65 | 45 | 25 | 47 |
| Atom Z510 | Silverthorne | 1,1 | 400 | 512 | 2 | 45 | 25 | 47 |
| Atom Z520 | Silverthorne | 1,33 | 533 | 512 | 2 | 45 | 25 | 47 |
| Atom Z530 | Silverthorne | 1,6 | 533 | 512 | 2 | 45 | 25 | 47 |
| Atom Z540 | Silverthorne | 1,86 | 533 | 512 | 2,4 | 45 | 25 | 47 |
| Atom N270 | Diamondville | 1,6 | 533 | 512 | 2,5 | 45 | 25 | 47 |
| Atom 230 | Diamondville | 1,6 | 533 | 512 | 4 | 45 | 25 | 47 |
All Atom processors have a 56 KB L1 cache, of which 32 KB is allocated for instruction cache and 24 KB for data. All processors can also execute 32-bit code and support additional instruction sets MMX, SSE, SSE2, SSE3 and SSSE3. As for 64-bit code (x86-64), it is supported only by the Diamondville core and only on the Atom 230 model. currently All Atom processors are single-core. At the same time, they support Hyper-Threading technology, which allows you to execute two parallel threads of commands. Towards the end of 2008, Intel plans to release the first dual-core Atom processors. Rumors are circulating online about the Atom 330 model, which will operate at a frequency of 1.6 GHz (FSB frequency - 533 MHz), and each core will have 512 KB of L2 cache. Atom Z series processors support virtualization technology as well as C1E Speedstep power saving technology. In addition to the Z series, the C1E Speedstep supports the Atom N270 processor, built on the Diamondville core. The range of Atom processors is quite large, and includes two cores for different systems. To avoid confusion, it is important to note that processors work with specific chipsets, and these are what determine the purpose of the final product. Along with the new processors, Intel has released a series of chipsets - UL11L, US15L, US15W - which are also designed to work with the Atom Z series (Silverthorne core).
The chipsets have similar characteristics, and each consists of one chip that implements the functionality, and a “north” and “south bridge”. New chipsets support Intel Atom processors with frequency system bus 100 or 133 MHz (400/533 MHz QPB), have a built-in single-channel 400 or 533 MHz DDR2 memory controller (maximum memory capacity is 1 GB). Also, the new series chipsets have built-in graphics Intel core GMA500, which in addition to three-dimensional graphics provides hardware decoding of video formats H.264, MPEG2, VC1 and WMV9. D-SUB and DVI-I outputs, as well as TV-Out, are supported. In addition, a bus controller is provided PCI Express spec 1.0. A few words about the expansion capabilities of UL and US chipsets - they support one IDE channel, eight USB 2.0 ports, as well as an HD audio subsystem. UL11L, US15L, US15W chipsets are part of the Centrino Atom 2 platform, which also includes Atom processors and modules wireless communication Wi-Fi, WiMAX and 3G. It should be noted that the heat dissipation of the UL11L chipset is 1.6 W, and the US series chipsets are no more than 2.3 W. As a result, the total heat dissipation between the UL11L chipset and the Atom processor is 2.25 W! This is exactly what you need mobile devices, because it is unprecedented low level energy consumption ensures long operating time. As for the Atom N270 and Atom 230 processors based on the Diamondville core, they are designed for cheap, economical and small-sized systems (Nettops and Netbooks) with the 945GC chipset. It is precisely such a system, or rather, motherboard Today we will test:
Please note that a massive heatsink with a fan is designed to cool the chipset, while the processor itself is content with a modest low-profile heatsink (in the background). Externally, the processor looks like this:
You will notice that the Atom 230 is directly soldered onto the board, so it will not be possible to upgrade the system. And if you burn out the processor during overclocking (more on that a little later), then you will have to replace the entire motherboard. The CPU-Z utility provides the following information:
This version of the utility incorrectly detects the processor core (Silverthorne instead of the correct Diamondville). Below are the specifications motherboard Gigabyte GC230D:
| CPU | Intel Atom 230 (Diamondville) |
| Chipset | Northbridge Intel 945GC |
| System memory | One 240-pin DDR-II SDRAM DIMM slot |
| Graphic arts | Built-in GMA950 graphics core |
| Expansion options | One 32-bit PCI Bus Master slot |
| Overclocking options | HTT frequency change from 100 to 700 MHz |
| Disk subsystem | One channel UltraDMA133/100/66/33 Bus Master IDE (supports up to two ATAPI devices & RAID 0, 1) |
| BIOS | 4 MBit Flash ROM |
| Miscellaneous | One FDD port, one serial and one parallel port, PS/2 mouse and keyboard ports |
| Power management | Wake from modem, mouse, keyboard, network, timer and USB |
| Monitoring | Monitoring processor temperature, monitoring voltages, determining the rotation speed of two fans |
| Size | ATX form factor, 170x170 mm (6.68" x 6.68") |
When Atom is faster than Core?
- Intel Blog
Stuck in a traffic jam behind the wheel of a car theoretically capable of reaching speeds of more than 200 km/h, and watching cyclists on tricycles overtake me, I thought... no, not about how to get everyone on bicycles, and not about solving transport problems humanity through teleportation, and... oh Intel processors Core and Intel Atom. Namely - Atom compared to Core is, in fact, a scooter compared to a car. It consumes less fuel and costs significantly less. But on the other hand, the speed of a scooter is just as noticeably inferior to a car (despite even the ways to “accelerate” the scooter above the factory settings). But, nevertheless, in traffic jams or on narrow streets the scooter is faster. No wonder the scooter got its name from the English “ to scoot" - to run away, as it was successfully used by English teenagers to escape from the police.
Now let's get back to the CPU. Let’s replace “fuel” with “electricity” and “speed” with “performance”, and we will get a complete analogy of the behavior of Inel Atom and Intel Core. But then it is reasonable to assume that there are “traffic jams” and “nooks and crannies” in which Atom will overtake Core. Let's look for them.
So, according to all generally accepted measurements Intel performance Core is significantly ahead of Atom. In the "Performance" section of the Wikipedia article about Intel Atom, a harsh verdict is read: " about half the performance Pentium processor M of the same frequency"
If we compare Atom specifically with Core, then according to tomshardware tests, the Intel Core i3-530 defeats the Intel Atom D510 with a crushing score: 
At the same time, it should be noted that tomshardware is clearly biased towards Atom. So, for example, if the running time of some task on Core-i3 is 1:38, then this is exactly how it is reported - “one minute, 38 seconds.” And if Atom performs something in 7:26, then, according to the authors, this is “about eight minutes.” But the main thing is to compare processors with different clock frequencies (2.93 GHz Core i3 and 1.66 GHz Atom) and not make allowances for wind. That is, the Core result must be divided by 2.93/1.66~1.76, which gives the final result of Atom losing from 2.15 to 2.6 times.
Why is Atom slower?
Quick answer: because it is cheaper and more energy efficient, which is incompatible with high performance.Correct answer: Firstly, because Atom retains the FSB bus, while Core i3 has a memory controller integrated into the CPU, which speeds up data access. In addition, Atom has a cache size that is four times smaller, and if the data does not fit in the cache, then slower memory access affects performance in full.
And secondly, the Atom microarchitecture is not Core2, used in Core i3, but Bonnell. In short, Bonnell is a continuation of the Pentium ideas, it has only 2 integer ALUs (versus three in the Core), and most importantly, there is no instruction reordering, register renaming, or speculative execution inherent in Core ).
How is it clear that in order to help Atom overtake Core, you need to:
- Take a nanoset, a small set of data, so that it fits in the cache.
- Try using float data to load the FPU rather than the ALU
- If possible, deprive Core of the benefits of out-of-order execution.
They were carried out on my existing Intel Core i5 2.53 GHz and the already mentioned Atom D510, and were a set of mathematical function calls for float data with a built-in performance assessment “number of functions per second”, i.e. the bigger, the better.
The tests included the calculation of trigonometric functions both directly (C runtime, “x87” test) and by series expansion; using the Cephes library code; as well as vector implementation through SSE intrinsic functions (tests ending in _ps). At the same time, taking into account the difference in clock frequencies, the results were scaled by 2.53/1.66~1.524
Tests compiled by Microsoft Visual Studio 2008 with release optimization by default.
The data obtained fully confirms the first place of Intel Atom from the end. That is, the goal has not been achieved, let's move on to the next point - we will complicate the work of the Out-of-order CPU.
Making the task more difficult
Let's create an artificial test that will contain unpredictable branches containing computationally heavy functions, so that the result of Core's speculative calculations is constantly discarded, i.e. turned out to be unnecessary work.Like that:
int rnd= rand()/(RAND_MAX + 1.) * 3; if (rnd%3==0) fn0(); if (rnd%3==1) fn1(); if (rnd%3==2) fn2();
Moreover, the functions will consist of chained calculations, so that Core cannot, by reordering instructions and renaming registers, calculate any of such expressions in advance, “out of turn.” Here is a simple example of such code
for (i=0; i< N; ++i) {
y+=((x[i]*x[i]+ A)/B[i]*x[i]+C[i])*D[i];
}
By the way, similar functions are used in the above tests cephes_logf and cephes_expf, where the advantage of Core is minimal.
But, despite all the obstacles, Core still turned out to be faster. The minimum gap between Core and Atom that I managed to get using various combinations of calculations and randomness is as much as two times! That is, Atom is still lagging behind.
But if I had stopped there, you simply would not have known about it - the post would not have taken place.
The next step was to compile the tests using Intel Compiler. The version used was Composer XE 2011 update 9 (12.1) with default Release optimization settings - similar to the Microsoft compiler.
The graph below shows the results of the above tests, including the rand I added, compiled by both VS2008 and Intel Compiler.
Look carefully. This is not an optical illusion. For four tests, the green line points showing the Atom result for tests compiled by Intel Compiler are higher than the burgundy line points showing the i5 result for tests compiled by VS2008. That is, Atom actually turns out to be more than twice as fast on the same code as Core i5.
Do you think this is an advertisement for an Intel compiler?
Absolutely not. I don't work in the advertising department or in the compilation group.
This is simply a statement that your optimized code can run much faster on Atom than unoptimized code on Core. Or - unoptimized on Core will be slower than optimized on Atom.
These are exactly the same bumps and crannies that prevent the car from accelerating.
You can draw your own conclusions.