#processor review

Intel previous processor series Intel Sandy Bridge-E and Intel Ivy-Bridge-E is still not giving the required performance to user, same can be said about the entire platform Socket LGA2011. Intel X79 Express chipset no longer meet modern requirements of high-performance system. Motherboard manufacturers tried to fix the thing in some ways, like party controllers for SATA 6 Gb/s and USB 3.0.

However, so long could not continue, because the needed a new platform and a new micro-architecture. Intel delayed is new processor line – Intel Haswell-E, to allow memory manufacturers to saturate the market standard solutions DDR4.

Main feature of the Intel Haswell-E is its number of the 8-core devices, whereas previously desktop systems include a maximum of 6 cores. The second improvement has touched the memory controller: DDR3 was upgraded to DDR4 with much greater bandwidth. According to the company all this allows multiple threads to achieve growth at the level of 79% compared with the 4-core solutions.

Micro-Architecture Features:

Intel Haswell-E has 8-core semiconductor chip in it. Intel Haswell-E, Intel used the well-established 22-nm process, also they have changed microarchitecture Intel Ivy Bridge with Intel Haswell. Intel has improved its structural level, improved mechanisms for sampling and branch prediction, support for instructions AVX2 and FMA3, increased cache performance of the first and second levels, the presence of voltage on the inverter side CPU.

Increased the maximum number of cores to 8 in Intel Core i7-5960X Extreme Edition vs. 6 core in the previous model Intel Core i7-4960X Extreme Edition. Given the technology of Intel Hyper-Threading, the output we get 16 independent streams instead of 12. Increasing the number of cores affected the amount of cache in the third level (L3). Now its volume can reach as much as 20 MB.

As for the memory controller, this new series will support DDR4 and supported frequency is also increased up to 2133 MHz. The representatives of the generation Intel Ivy Bridge-E was limited to a value of 1866 MHz. Although it is obvious that with overclocking it can be achieved better results. For that some of the companies already RAM sets capable of operating at frequencies – 3000 and 3333 MHz.

Intel Haswell-E 8-core includes 2.6 billion transistors, while in Intel Ivy Bridge-E, there are only 1.86 billion. Die size while also increased: from 257-mm2 to 356-mm2. Only thing that has not changed was the power distribution lines PCI Express. Still their number reaches 40 pieces (x16 + x16 + x8 + x8 or x8 + x8 + x8 + x8). They all conform to PCI Express 3.0.Thus, the solutions Intel Haswell-E is the best suited to create powerful gaming configurations with multiple graphics cards.

Intel Core I 7-5930K Specifications:

We have decided to file a complete specification of Intel Core I 7-5930K in the form of a comparative table to its predecessor Intel Core i7-4930K.

Model Intel Core I 7-5930 K Intel Core i7-4930K Marking (Engineering Sample) SR20R (QFRB) SR1AT (QF81) The Socket Socket LGA2011-V3 Socket LGA2011 Clock Speed Rated 3500 – MHz 3400 – MHz In Turbo Mode 3700 – MHz 3900 – MHz Modifier Rated 35 34 In Turbo Mode 37 39 Base Frequency 100 – MHz 100 – MHz Cache Size Of The First Level L1 6x 32 (Memory Instructions)

6x 32 (Data Memory)

6 x 32 (Memory Instructions)

6 x 32 (Data Memory)

Cache In The Second Level L2 6 x 256 – KB 6 x 256 – KB Cache In The Third Level L3 15 – MB 12 – MB Microarchitecture Intel Haswell Intel Ivy Bridge Codename Intel Haswell-E Intel Ivy Bridge-E Number Of Cores / Threads 6/12 6/12 Support Instructions MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, EM64T, VT-x, AES, AVX, AVX2, FMA3 MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, EM64T, VT-x, AES, AVX Number Of Lines PCI Express 3.0 40 40 Maximum Rated Power (TDP) 140 – W 130 – W Maximum Operating Temperature T Case 66.8 – °C 66.8 – °C Process Technology, Nm 22 22 Support For Enhanced Halt State (C1E)

Enhanced Intel Speedstep

Hyper-Threading

Execute Disable Bit

Intel Virtualization

Turbo Boost 2.0

Enhanced Halt State (C1E)

Enhanced Intel Speedstep

Hyper-Threading

Execute Disable Bit

Intel Virtualization

Turbo Boost 2.0

Built-In Memory Controller Maximum Amount Of Memory 64 – GB 64 – GB Types Of Memory DDR4 DDR3 Maximum Memory Frequency 2133 – Mhz 1866 – Mhz Memory Bandwidth 68 – GB/S 59.7 – GB/S Number Of Memory Channels 4 4 Maximum Number Of Modules Per Channel 2 2

Performance Details:

During normal operation Intel Core i7-5930K core speed was equal to 3500 MHz while the Bus Speed was almost 100 MHz and the multiplier “x35″ and core voltage was 1.037 V. Thus, the hero of the review on the 100 MHz faster than its predecessor (Intel Core i7-4930K) and thus requires a slightly smaller power supply for its correct operation: In 1.037 against 1.080 V.

By using the Intel Turbo Boost 2.0, we have increased multiplier by 2 points to a value of “x37″. CPU clock speed will be increased to the level of 3700 MHz, and the voltage – up to 1,071 V. However, this rate was observed very rarely, and only at low load.

Most of the time the processor core speed was 3600 MHz, and the power supply voltage varies within 1,056 – 1,071 V. As you can see, the mechanism of the Intel Turbo Boost Technology has also undergone some changes.

In idle mode CPU multiplier was reduced to “x12″, core speed was decreased as low as 1200 MHz and  voltage was 0.684 V, whereas in the case of the Intel Core i7-4930K required to 0.824 V.

Cache Memory Intel Core i7-5930K distributed as follows:

First level l1cache memory:

For each of the 6 cores allocated to 32 KB for data with 8 channels of associativity and 32 KB for instructions and with 8 channels of associativity;

Second level l2cache memory:

For each core is given by 256K with 8 channels of associativity;

Cache in the third level L3:

15 MB for all the nuclei with 20 channels associativity.

DDR4 memory controller operates in quad mode and support modules with up to 2133 MHz. Maximum memory capacity is 64 GB.

Testing Performance:

To test Intel Core i7-5930K we hav used GIGABYTE GA-Z97-Gaming G1 WiFi and a set of memory series Corsair Vengeance LPX DDR4-2800 totaling 16 GB (4 x 4 GB). The rest of the configuration of the tested is unchanged.

Motherboard GIGABYTE GA-Z97-Gaming G1 WiFi Coolers Scythe Mugen 3 (LGA 1155/1366, AMD Socket AM3 + / FM1 / FM2) ZALMAN CNPS12X (LGA 2011) Memory 2x 4GB DDR3-2400 TwinMOS TwiSTER 9DHCGN4B-HAWP Graphics Card AMD Radeon HD 7970 3GB GDDR5 Hard drive Western Digital Caviar Blue WD10EALX, 1TB, SATA 6 Gb / s, NCQ Power Supply Seasonic X-660 660W, Active PFC, 80 PLUS Gold, 120 mm fan

Note that disabling Intel Turbo Boost 2.0 leads to a drop in system performance by an average of 1-2% depending on the application. These results are quite natural, since the difference in frequency between the nominal and Turbo Mode is only 200 MHz.

We have compared the performance of the Intel Core i7-5930K with Intel Ivy Bridge-E family processors – Intel Core i7-4960X Extreme Edition and Intel Core i7-4930K. Also, we have included in the testing and popular version of the processors Intel Core i7-4770K and AMD FX-9370.

Intel Core i7-5930K demonstrated a very strong performance in the benchmark test, but with respect to its predecessor performance gains turned out not so great – about 5%. Flagship series Intel Ivy Bridge-E new average is ahead by 2%. In other words, a sizeable increase in performance after purchase Intel Core i7-5930K owners top processors for Socket LGA2011 platform will not receive.

However, if the computer is purchased “from scratch” or the modernization of systems, collected on the basis of less productive platform (Socket LGA1155 / 1150, Socket AM3 +, Socket FM2 / FM2 +), then the situation changes radically. Despite the seemingly increased frequency solutions Intel Core i7-4770K and AMD FX-9370, and they cannot compete close to Intel Core i7-5930K in terms of performance. In this case, the average difference in performance of 23% and 62%, respectively (of course, in favor of Intel Core i7-5930K).

Overclocking:

Members of Intel Haswell-E family, all of them have unlocked multiplier, which greatly simplifies the process of overclocking. Although if necessary overclock can produce and by changing the reference frequency.

By increasing the multiplier value to “x46″, we were able to increase the core speed uo to 4600 – MHz, with the base rate has been fixed at around 100 MHz, and the voltage was increased to 1.365 V. In this mode, Intel Core i7-5930K correctly passed the stress test program LinX 0.6.4.

During the experiment, maximum temperature was 93 °C on the hottest core by using a bench cooler Scythe Mugen 3. As a result, the rate of increase was 31.4% relative to the normal frequency was 3500 MHz, which can be considered quite a good result for air cooling.

Performance optimization of Intel Core i7-5930K is list in the table below:

Normal Mode Overclocking Difference Futuremark PCMark 7 PCMark Score 4015 4561 13.60% Computation Suite 8880 10073 13.43% Futuremark 3DMark11 Score 8203 8280 0.94% Physics 12626 13744 8.85% Futuremark 3DMark Vantage CPU Score 40959 50206 22.58% WinRAR 4.20 12168 14419 18.50% Fritz Chess Benchmark 4.2, knodes / s 19789 24101 21.79% TrueCrypt 7.1a (Serpent-Twofish-AES, MB / s) Encryption 347 408 17.58% Decryption 355 433 21.97% Batman Arkham City DirectX 11 126 126 0.00% Resident Evil 5 Benchmark DirectX 10 Smoothing x8 176.2 175.2 -0.57% F1 2012 DirectX 11 97.5 97 -0.51% Average 15.50%

Average performance was increased up to 15.50%. For that cooling system will go to a pretty big load with which to cope, not every air cooler.

Final Words:

Intel Haswell-E is based on the best 22-nm process technology and microarchitecture Intel Haswell. Supported memory DDR4 and new instructions AVX2 and FMA3, and the number of cores in the maximum configuration has been increased from 6 to 8. In short, Intel Haswell-E – it is not re-branding of the old solutions, and is really a new generation of processors with many improvements on the structural and electrical levels.

Intel Core I 7-5930K suggested cost is about $ 583, it demonstrates the phenomenal performance and easily breaks away from the flagship series and Intel Haswell Intel Devil’s Canyon. The average difference in speed is 20-23%, but in multi-threaded applications, it can reach 40-50% and all. Approximately the same results are observed in the games.

The presence of 40 lines PCI Express 3.0 makes Intel Core i7-5930K great choice for building a powerful gaming rig with multiple graphics cards. You can always still improve the performance of your computer by using overclocking by changing the multiplier or by raising the reference frequency. In our case, we have no problems increased his speed to 4600 MHz, which is reflected in the 15% growth compared with the speed at rated frequency (3500 MHz).

Intel Core i7-5930K performance as compared to Intel Core i7-4960X Extreme Edition is only 5% more.

With the release of AMD A10-7800, AMD A8-7600 and AMD A6-7400K situation must change, because among them there are cheaper and more energy-efficient solutions that will significantly expand the scope of the platform Socket FM2 +.

Features, details and characteristics of the AMD Kaveri are given and discus in the following articles;

APU AMD Kaveri: A Revolutionary Step Forward

APU: AMD Kaveri Processor Technology and Features Details

AMD A10-7850K.

For now we have AMD A10-7800 to be review, for that first look at the specifications of the processor.

Specification Details:

Model AMD A 10-7800 Marking AD7800YBI44JA The socket Socket FM2 + Clock Speed 3500 MHz – Normal

3900 MHz – Turbo Mode

Modifier 35 – Normal

39 – Turbo Mode

Base Frequency 100 – MHz Cache size of the first level L1B 2 x 96 – KB (Memory Instructions)

4x 16 – (Data Memory)

Cache in the second level L2 2 x 2048 – KB Microarchitecture AMD Steamroller + AMD GCN Codename AMD Kaveri Number of cores (CPU + GPU) 12 (8 + 4) Support instructions MMX (+), SSE, SSE2, SSE3, SSSE3, SSE4A, SSE4.1, SSE4.2, x86-64, AMD-V, AES, AVX, XOP, FMA3, FMA4 Maximum rated power (TDP) 65 – W Maximum operating temperature 71.3 °C Process Technology 28 – NM Support Software AMD 64-bit

AMD Turbo Core 3.0

AMD PowerNow!

AMD VT (Virtualization technology)

AMD EVP (Enhanced Virus Protection)

AMD Dual Graphics

AMD TrueAudio

AMD Mantle

AMD Eyefinity

AMD Fluid Motion Video

AMD Configurable TDP

Built-In Memory Controller

Types of memory DDR3 / DDR3L Maximum Frequency 2133 – MHz Number of Memory Channels 2 Maximum Number of Modules Per Channel 2

Integrated graphics AMD Radeon R7 Graphics

Stream Processors 512 Texture units 32 Clock Frequency of GPU 720 – MHz

Appearance:

The processor comes in a small cardboard box in white. The center of the logo can be found with the inscription AMD Kaveri. We note only that the TDP model AMD A10-7800 stated at 65 watts.

APU AMD A10-7800 is no different to the earlier members of the AMD Kaveri: AMD A10-7850K and AMD A10-7700K. On the heat-spreading lid is the name of the series and marking model.

Location of contacts on the back side corresponds to the CPU socket Socket FM2 +. Recall that a family APU AMD Kaveri is not compatible with socket Socket FM2.

Performance Details:

AMD A10-7800 included 12 cores – Maximum configuration for APU AMD Kaveri. Although to a broader and inclusive software distribution, which fully implements the possibility of a heterogeneous architecture (HSA), probably still worth still talk separately about the processor (4 cores) and graphics (8 cores) parts.

During normal tasks and multi operation, AMD A10-7800 clock speed was about 3500 MHz with frequency of 100 MHz and the multiplier “x35″, while the core voltage was 1.352 V. it is 200 MHz slower than the top-end AMD A10-7850K, but also requires less voltage for its correct operation: In 1.352 against 1.392 V.

Using Turbo Core 3.0 software, we have increases multiplier by 4 points to the value of “X39″. CPU clock speed increased up to 3900 MHz, and the voltage is lowered to 1.128 V. This, at first glance, the strange behavior of the processor lies part of the answer to the question: “How can the AMD managed to lower the TDP from 95 to 65 W?” – AMD A10-7800 operates at 3900 MHz with a very small load. Run any process immediately leads to a drop in frequency, sometimes even below the level of 3500 MHz.

In other words, value 3900 MHz, is only on paper, but in reality, the speed of a novelty is from 3000 – 3500 MHz, which is clearly visible on the chart. Supply voltage for this varies from 1.288 V to 1.384 V.

In this mode of operation APU AMD A10-7800 warmed up to 41 °C (using a bench cooler Scythe Mugen 3), which is a pretty good result.

In idle mode the factor decreases to “x14″, thus the frequency is lowered up to 1400 MHz. Voltage in this case is 0.864 V.

Cache AMD A10-7800 distributed in the same manner as in the AMD A10-7850K:

First Level L1 Cache Memory:

For each 4 cores there are 16 KB for data with 4 channels and associativity and for each 2-core module 96 KB for instructions with 3 channels associativity

Second level L2 Cache Memory:

2MB for each 2-core module with 16 channels associativity

Cache memory of the Third level L3 is absent.

DDR3 memory controller works in 2-channel mode and support is frequency up to 2133 MHz.

As at the structural level model is an analogue AMD A10-7800 AMD A10-7850K, it is quite logical that the performance of their graphics cores is the same: 512 unified shader pipelines, 8 ROPs and 32 texture units. It is noteworthy that the frequency of iGPU AMD Radeon R7 Graphics (code name is still a mystery) is not changed and remains at 720 MHz. That is a decrease of the thermal packet processor AMD A10-7800 to the level of 65 watts theoretically should not affect the performance of its graphical part.

Testing Performance:

  As for the testing the new model, we have used both motherboard form AMD and Intel of the same brand – ASUS, with the total memory of 8-GB – 2x 4GB. Other details of the system are listed below:

AMD ASUS F1A75-V PRO (AMD A75, SFM1, DDR3, ATX) GIGABYTE GA-F2A75-D3H (AMD A75, sFM2, DDR3, ATX) AMD ASUS SABERTOOTH 990FX (AMD 990FX, sAM3 +, DDR3, ATX) Intel ASUS P8Z77-V PRO / THUNDERBOLT (Intel Z77, sLGA1155, DDR3, ATX) ASUS P9X79 PRO (Intel X79, sLGA2011, DDR3, ATX) Intel ASRock Z87M OC Formula (Intel Z87, sLGA1150, DDR3, mATX) Cooler Scythe Mugen 3 – Supported both AMD and Intel Motherboards Memory 2 x 4GB – GeIL Frost White EVO POTENZA Dual-Channel 2666MHz Video Card AMD Radeon R9 280X 2 GB GDDR5 Hard Disk Western Digital Caviar Blue WD10EALX, 1TB Power Supply Cooler Master Silent Pro M700 Rs-700-Amba-D3 700w Atx12v V2.3 80 Plus Bronze

First thing you need to know is that activating AMD Turbo Core 3.0 will only increase the performance up to 3%, which cannot be notice so easily, so if you let this option ON or OFF it doesn’t really matter, its all up to you to active or disable it. The price of this processor at this time is about USA $ 155, while the actual price of APU AMD A10-7800 is between $ 165 – 170 Thus, from the camp of “blue” his most Competitors are likely 2-core Intel Core i3 solutions of families Intel Haswell / Haswell Refresh.

As compared to the Intel Core i3-4130, AMD A10 7800 performance is as low as 11 %, while the price of the Intel Core i3-4130 is about 135 $.

And when comparing AMD A10-7800 with AMD FX-6350, which, cost is about $ 135 – 140, performance difference is almost 24% in favor of the latter. From this we can conclude that collect on the basis of AMD A10-7800 gaming PC with a discrete graphics card is impractical, both financially and from a practical point of view.

In our opinion, most realistic use of new items will be mini-PC, HTPC and productive entry-level configuration with a small part of the requirements of 3D-graphics.

As compare with AMD A10-7850K, results have more than good. The advantage of the latter amounted to an average of only 3%. Where is enabled integrated graphics, both models and all were on the same level. But look at this on the difference in energy consumption. System with AMD A10-7800 consumed about 50 watts less than the similar configuration based on AMD A10-7850K. Also do not forget that the flagship AMD Kaveri costs $ 20 more expensive. Therefore, if you do energy efficiency is more important than good overclocking potential, it is still necessary to stop the choice on the model AMD A10-7800.

Configurable AMD TDP:

Technology AMD Configurable TDP allows you to change a TDP of the processor depending on the type of tasks. Just put the desired component in the motherboard BIOS, after which the system automatically optimizes its clock frequency and other parameters. In other words, generate automatic daunkloking – the inverse of overclocking utility. In most cases, it involves lowering speed of the device to minimize its power consumption and heat.

As part of desktop PC and this idea does not look very attractive, because nowadays you can easily acquire productive cooler at an affordable price. And for today’s power supplies extra 20-30 watts are not some impossible task. But everything changes dramatically when it comes to nettops, HTPC and other types of compact computers. In this case, as they say, once every watt counts.

Ability to change power consumption and heat with just one parameter will be welcomed by many. After all, not everyone has enough knowledge and patience to manually select values, while still every time testing the entire system for stability. That is, in certain situations, technology AMD Configurable TDP can be quite useful and simplify the configuration of the system.

Overclocking:

One of the differences between models – AMD A10-7800 and AMD A10-7850K is no unlocked multiplier. Thus, its acceleration is performed by changing the reference frequency, which is a more difficult process, which may not always succeed.

We have able to increase the clock speed up to 4233 MHz, which is 21 % more than the normal frequency and 8.5% in the turbo mode. Reference frequency was raised from 100 to 121 MHz, and CPU voltage increased up to 1.48 V, and chipset voltage increased up to 1.2 V. In this mode, the processor correctly passed the stress test program LinX 0.6.4. This result again confirms the high overclocking potential solutions from AMD, even with a locked multiplier. Temperature of the APU AMD A10-7800 under maximum load is only 48 °C.

System performance with the different task and software is given in the table below:

Normal Mode With Overclocking Difference Futuremark PCMark 7 PCMark Score 3254 3387 4.09% Computation Suite 6939 7215 3.98% Futuremark 3DMark11 Score 6439 6663 3.48% Physics 4108 4546 10.66% CINEBENCH R11.5 OpenGL, fps 63.44 67.04 5.67% CPU, pts 3.37 4 18.69% CPU (Single Core), pts 0.99 1.09 10.10% WinRAR 4.20 3348 3359 0.33% Fritz Chess Benchmark 4.2, knodes / s 7177 8260 15.09% Batman Arkham City DirectX 11 (fps) 97 101.5 4.64% Resident Evil 5 Benchmark DirectX 10 Smoothing x8 (fps) 88.9 92.1 3.60% F1 2012 DirectX 11, fps 61 65 6.56% RUSE DirectX 9, fps 26 28.2 8.46% Average 13 %

Average performance aws increased almost up to 13 %. It is interesting to note that the increase in the transition 400 MHz CPU turbo virtually no impact on PC performance. Whereas a further 333 MHz after the dissolution allowed increasing its speed is almost 13%.

Changing the base frequency directly affects the other parameters, in particular the speed of the memory subsystem. In our case it is a 2133 MHz to 2258 MHz increased.

Also increased, and the speed of the integrated graphics core AMD Radeon R7 Graphics: from 720 MHz to 886 MHz (+ 23% compared to the nominal value).

Graphics performance is affected as follows:

Normal Mode With Overclocking Difference Futuremark 3DMark11 Score 2502 2891 15.55% Graphics score 2344 2771 18.22% Physics score 4088 4154 1.61% Combined score 2328 2559 9.92% Futuremark 3DMark Vantage Score 8694 9591 10.32% GPU score 8199 9324 13.72% CPU Score 10620 10492 -1.21% Warhammer 40,000: Dawn of War II – Retribution DirectX 10, fps 57.9 61.6 6.39% Resident Evil 5 Benchmark DirectX 10, fps 65.3 70.1 7.35% DiRT: Showdown DirectX 11, fps 67.1 75.1 11.92% Average 14 %

As you can see the average performance difference of iGPU AMD Radeon R7 Graphics without increasing the clock frequency of processor cores is almost 14 %. In game it will go for up to 5 -7 FPS.

Final Words:

AMD A10-7800 can be called the “lite” version of the processor AMD A10-7850K. Reducing normal frequency to 200 MHz and adjust accordingly work with AMD Turbo Core 3.0, the manufacturer was able to reduce the TDP trends from 95 to 65 watts.

It is worth mentioning that AMD A10-7800 lost unlocked multiplier. However, overclocking and desire to achieve maximum energy efficiency – are incompatible concepts. In this case, on the contrary, many even think about daunklokinge. Here is just useful technology AMD Configurable TDP, which allows for even lower the TDP, up to the level of 45 watts.

According to the test result, to build a gaming system or with intermediate or high levels use, there is nothing more suitable available in the market. However, when it comes to the performance of computers for study, work, entertainment and running undemanding games, this 4-core APU is an excellent choice.

In most cases, buying AMD A10-7800 would be preferable, and most importantly, a balanced one, in comparison with a bunch of “budget processor + discrete graphics card.”

Now today we have new model from the AMD Kabini – AMD Athlon 5150. You can say it a lite verision of the AMD Athlon 5350, because the only difference between them is clock speed. Now let’s start with the specifications of the processor.

Specification Details:

Model AMD Athlon 5150 Labeling AD5150JAH44HM CPU Socket Socket AM1 Clock Speed 1600 – MHz Factor 16 Base Frequency 100 – MHz Cache Memory Of The First Level L1 4x 32 – KB (Memory Instructions)

4x 32 – KB (Data Memory)

Cache Memory Of The Second Level L2 2048 Microarchitecture AMD Jaguar + AMD GCN Codename AMD Kabini Number Of Cores / Threads 4/4 Support Document MMX (+), SSE, SSE2, SSE3, SSSE3, SSE4A, SSE4.1, SSE4.2, x86-64, AMD-V, AES, AVX Supply Voltage 0.5 – 1.4 – V TDP 25 – W Critical Temperature 90 – °C Tehprotsess 28 -NM Technology Support AMD 64-bit

AMD Virtualization

AMD UVD (Universal Video Decoder)

AMD VCE (Video Codec Engine)

Built-In Memory Controller Max Memory 16 – GB Types Of Memory DDR3 / DDR3L Maximum Frequency Support 1600 – MHz Number Of Memory Channels 1 Maximum Number Of Modules Per Channel 2 AMD Radeon HD 8400 (AMD Radeon R3 Graphics) Stream Processors 128 Texture Units 8 Clock Speed GPU 600 – MHz

Support Document

DirectX 11.2

OpenGL 4.3

OpenCL 1.2

SHADER Model 5.0

Packaging:

Processors from the APU AMD Kabini, including AMD Athlon 5150 comes in the same box with the white and red color. But with the difference of the logo and information on the label on which manufacturer has traditionally placed only basic specifications: Clock Frequency (1.6 GHz), Cache Size L2 (2 MB) and Number of Processor Cores (4).

In the box of the AMD Athlon 5150, you will find the following content:

Processor packaged in plastic blister

Cooler

User Manual Guide

Logo sticker of APU AMD Athlon Series

There is now difference between AMD Athlon 5150  and other AMD Kabini Processors. On the heat-spreading lid is the series title and labeling model, also name of countries where processor was made up (Germany) and where the final assembly of processor (Taiwan).

Cooling System:

Cooler comes with the processor chip is also the same as the other AMD Kabibi processor, also with the same TDP level  (25 – W). Moreover, such flexibility can save money in the development of processors, as is not required for each group of models re-calculate the parameters of the cooling system.

Its design is very simple: A small aluminum radiator which is consisting of four sections of aluminum fins cooled by a low-profile 50-mm fan.

Height of the cooling system is only 40 mm, which allows to use it in a very compact case, which often serve as a basis for multimedia PC (HTPC). Recall that coolers with fastening for boards equipped with processor sockets Socket AM3 / AM3 + / FM2 / FM2 +, AMD AM1 platform will not work.

In the normal use of the system the cooling system was quite well. During the long passage of a stress test for processor cores and integrated graphics core AMD Athlon 5150 temperature did not rise above 45 °C, and the computer is idle temperature was 30 °C.

Fan speed was varied within 1300 – 2600 rev / min. Same maximum value is 4000 rev / min, which can be achieved by activating the corresponding profile in the BIOS of the motherboard.

Performance Analysis:

With the normal test and performance AMD Athlon 5150 clock speed was almost at  1600 – MHz, while the multiplier was x16 and Bus Speed was 100 – MHz, at this time the core voltage was 1.296 V.

In the idle form, AMD Athlon 5150 clock speed decreased up to 800 – Mhz, while the multiplier was x8 and voltage is 1.092 V.

AMD Athlon 5150 cache is distributed in the same manner as in previous model – 4-core models of family AMD Kabini:

First Level Cache L1:

For each of the 4 cores allocated to 32 KB data with 8 channels of associativity and 32 KB for instructions with 2 channels associativity;

Second Level Cache L2:

2 MB for all nuclei with 16 channels associativity;

Memory controller works in single channel mode and maximum frequency speed support is up to 1600 – Mhz, while the maximum memory supported by the AMD Athlon 5150 is up to 16 – GB.

To defining characteristics of iGPU, we have used AIDA64m and we have got the following results.

In AMD Athlon 5150 they have used AMD Radeon R3 graphics core, AMD Radeon HD 8400, which is built on an advanced microarchitecture AMD GCN. It includes 128 stream processors, 4 rasterization module and 8 texture units, and clock frequency speed is 600 MHz. To save energy during the absence of a large load on iGPU, frequency automatically reduced to 266 MHz.

Testing Performance:

As for the testing the new model, we have used both motherboard form AMD and Intel of the same brand – ASUS, with the total memory of 8-GB – 2x 4GB. Other details of the system are listed below:

AMD ASUS F1A75-V PRO (AMD A75, SFM1, DDR3, ATX) GIGABYTE GA-F2A75-D3H (AMD A75, sFM2, DDR3, ATX) AMD ASUS SABERTOOTH 990FX (AMD 990FX, sAM3 +, DDR3, ATX) Intel ASUS P8Z77-V PRO / THUNDERBOLT (Intel Z77, sLGA1155, DDR3, ATX) ASUS P9X79 PRO (Intel X79, sLGA2011, DDR3, ATX) Intel ASRock Z87M OC Formula (Intel Z87, sLGA1150, DDR3, mATX) Cooler Scythe Mugen 3 – Supported both AMD and Intel Motherboards Memory 2 x 4GB – GeIL Frost White EVO POTENZA Dual-Channel 2666MHz Video Card AMD Radeon R9 280X 2 GB GDDR5 Hard Disk Western Digital Caviar Blue WD10EALX, 1TB Power Supply Cooler Master Silent Pro M700 Rs-700-Amba-D3 700w Atx12v V2.3 80 Plus Bronze

Basically to compare the performance against the INTEL we should have use – Intel Bay Trail-D, but these processor are not available yet, so we will compare the performance of AMD Athlon 5150 against AMD Trinity, AMD Richland and Intel Haswell.

But first let’s compare the performance of the AMD Athlon 5150 its own family AMD Kabini – the performance of AMD Athlon 5150 against the AMD Athlon 5350 was lower with the average of 20 to 23%, while against AMD Athlon 3850, it was higher with the average of 16- 19%.

It is due do the difference between the clock speed which is 300 – MHz lower than AMD Athlon 5150.

As compared to AMD A4-4000 – AMD Athlon 5150 have better performance in the normal tasks, while in the heave benchmarks such as CINEBENCH R11.5, its performance was low.

And at the end as compare to Intel Celeron G1820, it has a distinct advantage over AMD Athlon 5150, which on average is 90%.

Now let’s analyze the performance of the graphics processor AMD Athlon 5150. As we expected, this indicator hero angle is only slightly inferior flagship series: the difference is an average of only 4%.Roughly the same gap observed in 5150 and AMD Athlon from AMD A4-4000. But Intel Celeron G1820 again demonstrates a solid advantage – at 22%.

Overclocking:

Raise the multiplier on the APU AMD Athlon 5150 is not possible, but it was possible to increase the reference frequency of 5 MHz (from 100 to 105 MHz). After such manipulation processor speed 1600 MHz to 1680 MHz has increased, that is an increase of 5%. It also increased and effective memory frequency – from 1600 to 1680 MHz.

Computer performance optimization of the parameters of the system is reflected as follows:

Normal Mode Overclocking Difference Futuremark 3DMark11 Score 4114 4266 3.69% Physics 1945 2027 4.22% TrueCrypt 7.1a (Serpent-Twofish-AES, MB / s) Encryption 63 67.1 6.51% Decryption 63.1 67.1 6.34% Batman Arkham City DirectX 11 (fps) 40.5 44.5 9.88% Rezident Evil 5 Benchmark DirectX 10 Smoothing x8 (fps) 21.4 22.6 5.61% RUSE DirectX 9, fps 11 11.8 7.27%

Average performance increase in acceleration equal 6%. It will make no differ on your daily work, while the game performance may get a little bit change.

Final Words:

According to the results, AMD Athlon 5150 is in between AMD Athlon 3850 and AMD Sempron AMD Athlon 5350. AMD Athlon 5150 was better performance with an average of almost 18% and the second ahead by almost 22%.

Both models are equipped with the same graphics core AMD Radeon HD 8400. In this case, the performance difference is only 4%. At the same time AMD Athlon 5150 worth U.S. $ 10 cheaper than AMD Athlon 5350.

If compare with desktop processors, APU AMD A4-4000, demonstrating even better results in many everyday applications office applications, surfing the Internet, playing multimedia. In addition, AMD Athlon 5150 will be less hot during operation, and the whole system will consume less energy.

With the release of AMD AM1 platform equipped with a separate processor socket, with which you have batter chance to update your system, you can purchase a separate processor with the required performance and a motherboard with an appropriate functional, and if necessary, replace them easily.

AMD submit the entire lineup of the new CPU, which includes four models:

AMD Sempron 2650

AMD Sempron 3850

AMD Athlon 5150

AMD Athlon 5350

We already have reviewed the AMD Athlon 5350, and today we have CPU: AMD Sempron 3850 to be reviewed, let’s start with the specifications of the CPU.

Specification Details:

Model AMD Sempron 3850 Labeling SD3850JAH44HM CPU Socket Socket AM1 Clock Speed 1300 – MHz Factor 13 Base Frequency 100 – MHz Cache Memory Of The First Level L1 4 x 32 (Memory Instructions) – KB

4 x 32 (Data Memory) – KB

Cache Memory Of The Second Level L2 2048 – KB Microarchitecture AMD Jaguar + AMD GCN Codename AMD Kabini Number Of Cores / Threads 4/4 Support Document MMX (+), SSE, SSE2, SSE3, SSSE3, SSE4A, SSE4.1, SSE4.2, x86-64, AMD-V, AES, AVX Voltage 0.5 – 1.4 – V Maximum Rated Power (TDP) 25 – W Critical Temperature 90 Tehprotsess 28 – NM Technology Support AMD 64-bit

AMD Virtualization

AMD UVD (Universal Video Decoder)

Built-In Memory Controller Max Memory Support 16 -GB Types Of Memory DDR3 / DDR3L Maximum Frequency Support 1600 – MHz Number Of Memory Channels 1 Maximum Number Of Modules Per Channel 2 AMD Radeon R3 Graphics (AMD Radeon HD 8280) Stream Processors 128 Modules Rasterization 4 Texture Units 8 Clock Speed GPU 450 – MHz Support Document DirectX 11.2

OpenGL 4.3

OpenCL 1.2

Shader Model 5.0

Packaging and Cooling System:

AMD Sempron 3850 comes in a small box in white and red colors, remember this model belongs to the APU family,  that is part of the processor and graphics engine in it merged into one. Label can be found all the essential information regarding the processor, clock frequency (1.3 GHz), the size of the cache L2 (2 MB) and the number of processor cores (4).

In the box along the processor chip, you will find the

Processor packaged in plastic blister

Cooler

User Manual Guide

APU AMD Sempron Logo sticker

AMD Sempron 3850 is no different from the previously models from AMD Kabini, AMD Athlon 5350. On heat-spreading lid is the name of the series and marking model. Location of contacts on the back side corresponds to CPU socket Socket AM1.

Cooler Design:

APU AMD Sempron 3850 cooler is the same as the previously model, with the same level of TDP (25W). Normal cooling system is compact (55 x 55 x 40 mm) and two spring-loaded plastic clips instead of the usual clips.

Cooler radiator has standard aluminum core, which runs four sections of thin aluminum fins. For their low profile blower used 50mm fan labeled 1A213LQ00. Power is supplied via a 3-pin connector with support for monitoring the rotational speed of blades.

In the testing cooling system performance was quite good. At maximum load on the processor cores and integrated graphics core temperature did not rise above 46 °C. Depending on the profile of the motherboard BIOS, the fan speed can be dynamically changed within the 1300 – 4000 rev / min. If select Silent this indicator will remain permanently at the level of 1300 rev / min, regardless of the load on the processor.

Performance Analysis:

In the normal operation, the core speed of the AMD Sempron 3850 is almost equal to 1300 MHz with the 100 –MHZ Bus speed and multiplier was x13, core voltage was 1.280 V at that time.

In the idle mode, core speed was decreases to 800 – MHz, while the multiplier was x8 and voltage was 1.168 V. while power-saving mode of AMD Athlon 5350 this figure dropped to the level of 1.024 V at the same clock frequency.

Cache AMD Sempron 3850 is distributed as follows:

First level L1 Cache:

For each of the 4 cores allocated to 32 KB data with 8 channels of associativity and 32 KB for instructions with 2 channels associativity;

Second level L2 Cache:

2 MB for all nuclei with 16 channels associativity;

The cache memory of the third level L3 is absent.

Memory operates in single channel mode and maximum memory frequency support is up to 1600 MHz and maximum memory support is 16 GB.

To defining the characteristics of iGPU, we have used AIDA64.

In AMD Sempron 3850 graphics core set of series AMD Radeon R3 Graphics codenamed is AMD Radeon HD 8280. AMD Athlon 5350 (AMD Radeon HD 8400) has 128 stream processors, 4 modules and 8 texture units. The difference lies only in the frequency formula. Rated speed of the graphics core AMD Radeon HD 8280 is 450 MHz, while the AMD Radeon HD 8400, it reaches 600 MHz. To save energy during the absence of heavy load on the graphics core, its frequency is automatically reduced from 450 to 266 MHz.

Testing Performance:

As for the testing the new model, we have used both motherboard form AMD and Intel of the same brand – ASUS, with the total memory of 8-GB – 2x 4GB. Other details of the system are listed below:

AMD ASUS F1A75-V PRO (AMD A75, SFM1, DDR3, ATX) GIGABYTE GA-F2A75-D3H (AMD A75, sFM2, DDR3, ATX) AMD ASUS SABERTOOTH 990FX (AMD 990FX, sAM3 +, DDR3, ATX) Intel ASUS P8Z77-V PRO / THUNDERBOLT (Intel Z77, sLGA1155, DDR3, ATX) ASUS P9X79 PRO (Intel X79, sLGA2011, DDR3, ATX) Intel ASRock Z87M OC Formula (Intel Z87, sLGA1150, DDR3, mATX) Cooler Scythe Mugen 3 – Supported both AMD and Intel Motherboards Memory 2 x 4GB – GeIL Frost White EVO POTENZA Dual-Channel 2666MHz Video Card AMD Radeon R9 280X 2 GB GDDR5 Hard Disk Western Digital Caviar Blue WD10EALX, 1TB

Power Supply

Cooler Master Silent Pro M700 Rs-700-Amba-D3 700w Atx12v V2.3 80 Plus Bronze

Basically to compare the performance against the INTEL we should have use – Intel Bay Trail-D, but these processor are not available yet, so to compare the performance, we have AMD processor, AMD Athlon 5350 and AMD A4-4000.

As our tests showed, the latter was faster on average by 53% due to a significant difference in clock speed – 1300 MHz versus 2050 MHz.

As compare to the AMD APUs – AMD A4-4000 and AMD A4-6300, the performance of the AMD Sempron 3850 is not so good, performance of AMD A4-4000 was 25% better and performance of AMD A4-6300 was almost 50% more than the AMD Sempron 3850. It is due to the difference of processor cores between these processors, AMD Sempron 3850 four of them, whereas the comparable APU AMD half.

New GPU also unable to compete with the above model. Average advantage of their built-in graphics cores on the AMD Radeon HD 8280 (AMD Sempron processor 3850) were as follows:

AMD Radeon HD 8400 (AMD Athlon processor 5350) – 30%

AMD Radeon HD 7480D (processor AMD A4-4000) – 26%

AMD Radeon HD 8370D (processor AMD A4-6300) – 37%

Intel HD Graphics (Intel Celeron G1820) – 20%

Only indicator of where AMD Sempron 3850 including the leader, is the level of energy consumption. System based on AMD Sempron 3850 will not only consume little power, but also cooler. So, you can try to create a computer with a completely passive cooling in a compact package. With representatives of the families of AMD Trinity / Richland and Intel Haswell crank is unlikely.

Overclocking:

AMD Kabini processors are great energy-efficient solutions so that they have great overclocking abilities. When testing AMD Athlon 5350 we were able to speed it up to 50 MHz by increasing the multiplier. In the case of AMD Sempron 3850 this possibility was unavailable, but it was possible to optimize at the level of the reference frequency.

Because of its lifting, 5 MHz (from 100 to 105 MHz) from the nominal rate trends 1300 MHz to 1365 MHz increased, i.e. 5%. Note that this does not have to perform any manipulation of the supply voltage.

Average performance increase in acceleration equal slightly more than 7%. This increase in speed in most cases will not be noticeable by the user during work in everyday applications. However, in games increase was 11-12%, which is at low graphics settings will provide a couple of extra FPS.

Final Words:

With the release of the AMD Kabini, energy-efficient processors, AMD decided to reproduce the AMD Sempron, because at one time AMD Sempron personification of budget systems. But for now AMD Sempron will not regain that place but with the low power consumption (25 W), which makes it a good base for a multimedia PC and it is possible to think about building a completely silent configuration.

4-cores allow you to feel confident when dealing with multiple open documents or Internet pages, and relatively powerful graphics core does not constrain you when working with applications that use 3D-graphics.

If efficiency is not a priority goal, then we would recommend to look to the other processors for entry-level platforms, or Intel Socket LGA1150 AMD Socket FM2. At approximately the same cost AMD Sempron 3850 they show better performance processor and graphics parts. And in some cases there may be almost two-fold advantage.

AMD new APU AMD Sempron based on AMD Jaguar, with the design of SoC (System-on-Chip), they combine the computational and graphics cores, memory controller, and chipset. For now there are only 2 models are available of this new APU series: AMD Sempron 2650 and AMD Sempron 3850:

Model APU AMD Sempron 2650 AMD Sempron 3850 Number Of Cores / Threads 2/2 4/4 CPU Frequency 1.45 – GHz 1.3 – GHz Cache Memory Of Second Level (L2) 1 -MB 2 – MB Graphics Core AMD Radeon R3 AMD Radeon R3 GPU Frequency 400 – MHz 450 – MHz Number Of Stream Processors 128 128 Maximum Supported Memory DDR3 1333 – MHz 1600 – MHz TDP 25 -W 25 -W

This time we have AMD Sempron 2650 to review, let’s start with it.

Packaging and Cooling System:

The box of the new model is in white color, from the front you can see the processor chip in the box, there is nothing on the front of the box but just a logo of AMD. One side of the box have the hologram and serial number of the product, also a line written – Solutions to everyday problems, in other words – work with documents and multimedia files, as well as surfing the internet.

In the box you will find the following items;

Processor Chip

Cooling System

Instructions on Installing CPU

Sticker for Computer Case

With the help of instructions on installing CPU you can easily install the CPU in the motherboard, also you can correctly fix the cooling system .

The cooler with the processor chip as a small radiator with the two spring-loaded clips, with which you can fixed it on the motherboard. One fan on the radiator of the Foxconn PVA050E12L 50 mm, operating voltage of 12 V and a current of 0.16 A.

Fan speed in automatic mode was between 1300 and 4000 rev/min. Up to 3000 rev/min it is virtually silent, and only at 4000 rev / min appears faint background noise. In normal mode GPU temperature was about 26 °C and processor cores temperatures was 40 °C.

Specifications Detail:

Front side of the chip have the name of the series “AMD Sempron 2600 series FSlb” also the name of the countries in which this model is made (Germany)  and final assembly took place in Taiwan. Back side contains a set of contacts that are compatible with the latest socket – Socket AM1.

Remember you should be very careful when installing the APU on the board, not to damage long enough and thin copper contacts.

Model AMD Sempron 2650 Labeling SD2650JAH23HM CPU Socket Socket AM1 Base Clock Speed – Normal 1.45 – MHz Factor 14.5 Basic System Bus Frequency 100 – MHz Cache Memory Of The First Level L1 2 x 32 (Data Memory) -KB

2 x 32 (Memory Instructions) – KB

Cache Memory Of The Second Level L2 1024 – KB Microarchitecture AMD Jaguar + AMD GCN Codename AMD Kabini Number Of Cores / Threads 2/2 Support Document MMX (+), SSE, SSE2, SSE3, SSSE3, SSE4A, SSE4.1, SSE4.2, x86-64, AMD-V, AES, AVX Supply Voltage 0.5 – 1.4 –V TDP 25 – W Critical Temperature 90 – °C Tehprotsess 28 – NM Technology Support AMD 64-bit

AMD Virtualization

AMD UVD (Universal Video Decoder)

AMD VCE (Video Codec Engine)

Built-In Memory Controller Max Memory 16 –GB Types Of Memory DDR3 Maximum Frequency 1333 – MHz Number Of Memory Channels 1 Integrated Graphics Core AMD Radeon R3 (Radeon HD 8240) Stream Processors 128 Texture Units 8 Clock Speed GPU 400 – MHz Support Document DirectX 11.2

OpenGL 4.3

OpenCL 1.2

Shader Model 5.0

As you can see in the spec, AMD Sempron 2650 have low TDP (25-W). This makes it possible to use not only compact and low noise active cooler, but also completely passive cooling system.

With the normal performance APU multiplier was x 14, and clock speed was at 1447 MHz. Core voltage was almost 1.3 V.

In the idle, the clock speed was dropped up to 798 MHz with a multiplier value of x 8 and voltage was 1.072 V.

Cache memory in AMD Sempron 2650 is distributed as follows. Caching data from 32 KB L1 cache per core with 8 channels of associativity, and 32 KB L1 cache per core with 2 channels associativity allocated for instructions. There is still a general 1024KB L2 cache with 16 channels of associativity.

Memory controller works in single channel mode and support DDR3 modules with up to 1333 MHz. in the normal mode supported memory speed is 1333 MHz and with the overclocking you can use up to 1600 MHz.

GPU-Z utility is not able to correctly identify the characteristics of the integrated graphics core AMD Radeon R3, this information was obtained by using AIDA64. This video card has a 128-th universal shader pipelines with 8 texture units and 4 ROP units, as well as hardware support for DirectX 11.2. Basic GPU frequency is 400 MHz, which is automatically lowered to 200 MHz in idle mode. The maximum amount of memory allocated for the needs of its operations is 512 MB.

Testing Performance:

As for the testing the new model, we have used both motherboard form AMD and Intel of the same brand – ASUS, with the total memory of 8-GB – 2x 4GB. Other details of the system are listed below:

AMD ASUS F1A75-V PRO (AMD A75, SFM1, DDR3, ATX) GIGABYTE GA-F2A75-D3H (AMD A75, sFM2, DDR3, ATX) AMD ASUS SABERTOOTH 990FX (AMD 990FX, sAM3 +, DDR3, ATX) Intel ASUS P8Z77-V PRO / THUNDERBOLT (Intel Z77, sLGA1155, DDR3, ATX) ASUS P9X79 PRO (Intel X79, sLGA2011, DDR3, ATX) Intel ASRock Z87M OC Formula (Intel Z87, sLGA1150, DDR3, mATX) Cooler Scythe Mugen 3 – Supported both AMD and Intel Motherboards Memory 2 x 4GB – GeIL Frost White EVO POTENZA Dual-Channel 2666MHz Video Card AMD Radeon R9 280X 2 GB GDDR5 Hard Disk Western Digital Caviar Blue WD10EALX, 1TB Power Supply COOLER MASTER Silent Pro M700 RS-700-AMBA-D3 700W ATX12V V2.3 80 PLUS BRONZE

Basically to compare the performance against the INTEL we should have use – Intel Bay Trail-D, but these processor are not available yet, so to compare the performance, we have AMD processor, AMD Sempron 3850 and AMD A4-4000.

As compare to the AMD Sempron 3850, the performance of the AMD Sempron 2650, was as low as 70%, which is because the difference of value in the clock speed and cache.

The performance of the AMD A4-4000 is 95% higher than the AMD Sempron 2650, not only due to features of the microarchitecture, but also because of twice the clock frequency.

As for the performance of the integrated graphics adapter model AMD Sempron 2650, it is also substantially lower than the selected opponents. Model for AMD Sempron 3850 the difference is 26%.Versions for AMD A4-4000, Intel Celeron G1820 and Intel Celeron G1830 it has already reached 60-65%.

Overclocking:

By increasing the multiplier up o x 15, and core speed was almost 1575-MHZ, while the clock voltage was does not exceed from 1.284 V. we have got the following results in the different tasks and games performance.

Normal Mode Overclocking Difference Futuremark 3DMark Vantage CPU Score 2004 2164 7.98% WinRAR 4.20 900 985 9.44%  Fritz Chess Benchmark 4.2 1432 1521 6.22% TrueCrypt 7.1a (Serpent-Twofish-AES) Encryption 29.1 29.4 1.03% Decryption 29.1 30.7 5.50% Batman Arkham City DirectX 11 16 21 31.25% Rezident Evil 5 DirectX 10 15.5 16.6 7.10% Average: 10.91%

Average performance was increased up to almost 11%, which is quite a good result, but still significantly reduce the backlog of competitors it is not possible.

Final Words:

Main advantage of this model is Low TDP (25 – W), also it will help you to build a low-cost system, for the normal home use or office work. Second characteristic enable the use of sufficiently compact housing that takes up little space on your desktop. Also, due to the low TDP trends, possible replacement for the standard cooling system for quiet passive version.

Integrated graphics core AMD Radeon R3 will not only comfortable to work with standard applications, but also give an opportunity to familiarize with some undemanding computer games on low graphics settings.

AMD FX-6350 has high TDP (125W), which significantly increases the demands on standard cooling system.

Packaging and Cooling System:

The box of the processor is in red and white color, with the AMD FX processor, at the bottom CPU refers to a series AMD FX, and prefix Black Edition says its unlocked multiplier.

From the side view you can see the processor chip in the box, and at bottom you can see the base frequency of the CPU and the overall size of its cache, protective hologram and serial number.

After opening the box, you will find:

Processor Chip

Cooling System

Sticker For Computer Case

Installation Manual

This set is quite sufficient to assemble the computer based on this CPU.

Cooling system (Heat-sink) of AMD FX-6350 has fan AVC DV07020B12U based rolling bearing, which is the operating voltage of 12 V and current – 0.7 A. And a massive radiator based on 48 plates. Heat transferred from a large copper base contact with the processor through a thin layer of thermal paste, and then transmitted to the four 6 mm heat-pipes. Heat pipes are soldered to both fins and copper base, which should positively affect the efficiency of the cooling system.

One thing you should keep in mind that processor has enough high TDP (125W), so you need to take care of head blowing out through case fans. This will facilitate the release of the heated air outside the chassis. You should buy a high-performance cooler and replacing them complete SB.

Specifications Details:

CPU AMD FX-6350 shape is different from the other models of AMD, on the front you can see the model name of the processor and also the countries where it was made and assemble.

Processor model name is FD6350FRW6KHK has the notation:

F – Family AMD FX

D – Segment Workstations

6350 – Model Number

FR – TDP 125 W

W – CPU Socket Socket Am3 +

6 – Number of Active Cores

K – Amount of Cache Memory L2 – 2 Mb per Dual-Core Module and 8 MB Shared Cache L3

HK – CPU Core Stepping OR-C0

Model AMD FX-6350 Labeling FD6350FRW6KHK CPU Socket Socket AM3 + Base Clock Frequency 3900 – MHz Maximum Clock Frequency With AMD Turbo Core 3.0 4200 – MHz Factor 21 (unlocked) Basic System Bus Frequency 200 – MHz Cache Memory of First Level L1 6 x 16 (Data Memory) – KB

3 x 64 (Memory Instructions) – KB

Cache Memory of Second Level L2 2048 x 3 – KB Cache In The Third Level L3 8 – MB Microarchitecture Piledriver Codename Vishera Number Of Cores / Threads 6/6 Support Document MMX (+), SSE, SSE2, SSE3, SSSE3, SSE4A, SSE4.1, SSE4.2, x86-64, AMD-V, AES, AVX, XOP, FMA3, FMA4 Supply Voltage 0.888 ~ 1.404 – V Maximum Rated Power (TDP) 125 – W Tehprotsess 32 – NM Technology Support AMD64 (AMD 64-bit)

Virtualization (AMD-V)

Balanced Smart Cache

Wide Floating Point Accelerator

Turbo Core 3.0

PowerNow!

Built-In Memory Controller Max Memory 32 – GB Types Of Memory DDR3 (1866 MHz frequency up) Number Of Memory Channels 2

As we have said earlier, this model has unlocked multiplier, which means there is an opportunities for overclocking to test the full performance of the processor.

With AMD Turbo Core 3.0 enable, AMD FX-6350 clock speed was above from 3900 MHz with a multiplier more than X 19, while the core voltage was 1.356 V. In the different tasks the maximum frequency was reach up to 4200 MHz, and core voltage was up to 1.404 V.

In the idle state, multiplier was dropped up to X 7, and clock speed was 1400 MHz and voltagewas dropped to 0.8 V.

Cache memory in AMD FX-6350 is distributed as follows.

First level L1:

16 KB per core is allocated to the data cache with 4 channels of associativity and 64KB for each of the three modules for dual instruction with 2 channels associativity.

Second level L2:

2 MB per dual-core module with 16 channels of associativity.

Third level L3:

8 MB shared L3 cache with 64-E channels associativity.

Memory controller running in dual channel mode and support frequency is 1866 MHz DDR3.

Testing Performance:

As for the testing the new model, we have used both motherboard form AMD and Intel of the same brand – ASUS, with the total memory of 8-GB – 2x 4GB. Other details of the system are listed below:

AMD ASUS F1A75-V PRO (AMD A75, SFM1, DDR3, ATX) GIGABYTE GA-F2A75-D3H (AMD A75, sFM2, DDR3, ATX) AMD ASUS SABERTOOTH 990FX (AMD 990FX, sAM3 +, DDR3, ATX) Intel ASUS P8Z77-V PRO / THUNDERBOLT (Intel Z77, sLGA1155, DDR3, ATX) ASUS P9X79 PRO (Intel X79, sLGA2011, DDR3, ATX) Intel ASRock Z87M OC Formula (Intel Z87, sLGA1150, DDR3, mATX) Coolers Scythe Mugen 3 Supported both AMD and Intel Motherboards Memory 2 x 4GB – GeIL Frost White EVO POTENZA Dual-Channel 2666MHz Graphic Card AMD Radeon R9 280X 2 GB GDDR5 Hard Disk Western Digital Caviar Blue WD10EALX, 1TB Power Supply COOLER MASTER Silent Pro M700 RS-700-AMBA-D3 700W ATX12V V2.3 80 PLUS BRONZE

First of all we have tested the performance of the AMD Turbo Core 3.0 – ON and OFF, the difference was just 1 to 2 %, so it’s doesn’t make any major difference to keep it ON or OFF.

Let’s first compare with the ADM processor, other tested model was AMD FX-4130, (AMD Zambezi Generation) – it was far behind form our tested model AMD FX-6350, the performance difference was about 25-30%, it was due to the difference in Core 4 as compare to 6 and also twice size of cache in the third level of L3. And as compare to the second tested model AMD FX-8350, the performance of the our model was low up to 25 – 30%, and it was due to the difference of Core 8 as compare to 6, but the price of this processor is much higher.

And as compared to the Blue competitors (INTEL) – the average performance of Intel Core i3-4130 was low up to 7 – 10 % and against the Intel Core i5-3330, the performance was higher up to 7 – 10 %.

According to results AMD FX-6350 has good processing power, especially in problems related to encryption / decryption, as well as mathematical calculations.

Overclocking:

This model has unlocked multiplier so that we can easily achieve the overclocking on it. So increasing the voltage up to 1.160 V and multiplier is fixed at X 23, we were manage to get the dynamic clock frequency more than 4700 MHz and the temperature does not exceed to 60 °C while using the bench cooling system instead of normal cooler.

Following results was obtain with the overclocking.

Normal Mode Overclocking Difference Futuremark PCMark 7 PCMark Score 3382 3595 6.30% Computation Suite 7474 7865 5.23% Futuremark 3DMark11 Score 7330 7150 -2.46% Physics 5848 6476 10.74% Futuremark 3DMark Vantage CPU Score 16123 19075 18.31% WinRAR 4.20 5620 5758 2.4% Fritz Chess Benchmark 4.2 10103 11941 18.19% Batman Arkham City DirectX 11 54 55 1.85% Rezident Evil 5 Benchmark DirectX 10 Smoothing x8 106.2 119.9 12.90% F1 2012 DirectX 11 74 79 6.76% Average: 14%

The average difference in the performance was 14 %, which is quite impressive and almost equal to the  AMD FX-8350. But as you know this model has High TPD, so in overclocking mode power consumption of the entire system was 199 watts at maximum load and further increase up to 281 watts at maximum load, so for the overclocking you need more powerful cooler.

Final Thoughts:

The performance of the AMD FX-6350 is very good in all the tasks as compare to Intel Core i3-4130 whichhas almost the same price. AMD FX-6350 has an integrated memory controller that support up to 1866 MHz also compatible with low voltage memory modules 1.35 and 1.2 V.

As for the overclocking clock frequency was increased from 4200 MHz to 4716 MHz. But one thing you should remember computer based on this processor has high power consumption, so care about the organization proper airflow through the case fans, but also to buy a quality power supply appropriate power.

That’s why APU Kaveri is becoming a real competitor for a CPU + GPU. However, so far it is only on entry-level configurations. New APU systems will make the difference soon.

For now AMD did not release other models of the APU AMD kaveri, till now there are only two models AMD A10-7850K and AMD A10-7700K. We already have reviewed the second model AMD A10-7700K, but now we have the opportunity to evaluate the potential of AMD A10-7850K.

Let’s start the review with the packing and cooling system of the processor.

Packaging and Cooling System:

The box of the new APU AMD A10-7850K is little bit small with black and red color, having ADM A-Series Black Edition logo, which mean this model has an unlocked multiplier, and hence a certain overclocking potential.

On the side some details are listed such as:

Frequency of the processor in turbo mode (4.0 GHz)

L2 Cache size (4 MB)

Number of cores (12: 4 8 CPU + GPU)

Series of integrated graphics core (AMD Radeon R7 Graphics)

Standard Cooler Supplied.

And at the last briefly describes capabilities of AMD hUMA and AMD GCN.

As for the content which you will find in the box with the processor chip are:

User Guide

Processor Cooler

Logo Sticker Series APU AMD

Now if look at cooler of the processor, its TDP level – 95 W, same as the previous model AMD A10-7700K, with the low radiator, which has been blown, low-profile 70-mm fan. Heat-sink is made of aluminum core with four sections of the plates. The base is initially coated with a layer of thermal paste.

The performance of the cooling system is quite good, in idle mode temperature was about 25 – 27 °C, and at maximum load it was about 45 to 47 °C. Fan speed was 2500 rpm/min and noise level quite low, maximum fan speed is 3500 rev/min.

Specification Details:

Looking at the processor chip APU AMD A10-7850K is same the other models of the AMD Kaveri, front side has the model No. and the name of the countries where he chip was made (Germany) and assembly of the processor in Malaysia.

Back side corresponds to CPU socket Socket FM2+. APU AMD Kaveri is not compatible with socket Socket FM2.

Model AMD A10-7850K Labeling AD785KXBI44JA CPU Socket Socket FM2 + Clock Frequency (Normal/Turbo In) 3700 / 4000 – MHz Factor (Normal / Turbo In) 37/40 Base Frequency 100 – MHz Cache Memory Of The First Level L1 2 x 96 (Memory Instructions) – KB

4 x 16 (Data Memory) – KB

Cache Memory Of The Second Level L2 2 x 2048 – KB Microarchitecture AMD Steamroller + AMD GCN Codename AMD Kaveri Number Of Cores (CPU + GPU) 12 (8 + 4) Support Document MMX (+), SSE, SSE2, SSE3, SSSE3, SSE4A, SSE4.1, SSE4.2, x86-64, AMD-V, AES, AVX, XOP, FMA3, FMA4 Maximum Rated Power (TDP) 95 – W Critical Temperature 72.4 – °C Tehprotsess 28 – NM Technology Support Dual Graphics

TrueAudio

Mantle

Turbo Core 3.0

PowerNow!

Eyefinity

Fluid Motion Video

Built-In Memory Controller Types Of Memory DDR3 Maximum Frequency 2133 – MHz Number Of Memory Channels 2 Integrated Graphics Core AMD Radeon R7 Graphics Stream Processors 512 Modules Rasterization 8 Texture Units 32 Clock Speed GPU 720 – MHz Support Technologies DirectX 11.2

OpenGL 4.3

DirectCompute 11

OpenCL 1.2

Shader Model 5.0

At normal usage the speed of the of AMD A10-7850K is about 3700 – MHz, when AMD Turbo Core 3.0 was off and was multiplier x 37 and core voltage was 1.4 V (almost).

After enabling the AMD Turbo Core 3.0, the performance speed was boost up to 4000 – MHz and multiplier was x 40 while the voltage is not changed, it remains 1.4 V (almost).

And in the idle state frequency drops to 1700 MHz, while the multiplier was x 17, and voltage is 0.7 V, which mean energy saving mechanism in APU AMD A10-7850K is improved as compared to the previous family of AMD Richland.

AMD A10-7850K Cache distributed in the same manner as in the AMD A10-7770K:

First level L1:

For each of the 4 cores there are 16 KB for data with 4 channels and associativity for each 2-core module – to 96 KB for instructions with 3 channels associativity

Second level L2:

For each 2-core module assigned to 2 MB with 16 channels associativity;

Memory running in dual channel mode and support with frequency up to 2133 MHz DDR3.

And for the GPU in AMD A10-7850K AMD used Radeon R7 Graphics. No exact name turns some confusion as to APU AMD A10-7700K also found AMD Radeon R7 Graphics, but with other characteristics.

AMD has increased the performance of the features of the GPU in its new APU Kaveri models, such as iGPU advanced microarchitecture AMD GCN, which us the main reason of the popularity of APU AMD.  As compared to AMD A10-6800K (APU Richland), AMD Kaveri received additional 128 unified shader pipelines. The only parameter that AMD Radeon R7 Graphics inferior AMD Radeon HD 8670D (iGPU model AMD A10-6800K) is the clock frequency – 720 MHz versus 844 MHz.

According to the AMD press, while the AMD A10-7850K have a better GUP, but it cannot go ahead in all games than the AMD A10-6800K. Also introduction of shared system memory (AMD hUMA) and heterogeneous queue (AMD hQ), in graphics processor core AMD A10-7850K, which is 8, can work closely in conjunction with 4 cores and perform better part of their tasks. AMD A10-7850K supports Dual Graphics, with which you can combine the power of integrated graphics and discrete graphics card. As a final, AMD recommends using adapter AMD Radeon R7 250 DDR3 (Oland XT).

Testing Performance:

As for the testing the new model, we have used both motherboard form AMD and Intel of the same brand – ASUS, with the total memory of 8-GB – 2x 4GB. Other details of the system are listed below:

AMD ASUS F1A75-V PRO (AMD A75, SFM1, DDR3, ATX) GIGABYTE GA-F2A75-D3H (AMD A75, sFM2, DDR3, ATX) AMD ASUS SABERTOOTH 990FX (AMD 990FX, sAM3 +, DDR3, ATX) Intel ASUS P8Z77-V PRO / THUNDERBOLT (Intel Z77, sLGA1155, DDR3, ATX) ASUS P9X79 PRO (Intel X79, sLGA2011, DDR3, ATX) Intel ASRock Z87M OC Formula (Intel Z87, sLGA1150, DDR3, mATX) Coolers Scythe Mugen 3 – Supported both AMD and Intel Motherboards Memory 2x 4GB DDR3-2400 TwinMOS TwiSTER 9DHCGN4B-HAWP Graphic Card AMD Radeon r9 280X 2GB GDDR5 Hard DISK Western Digital Caviar Blue WD10EALX – 1TB Power Supply Seasonic X-660, 660W, Active PFC, 80 PLUS Gold

AMD Turbo Core 3.0 does not have the major effect on the performance of AMD A10-7850K, which means turning it ON or OFF doesn’t matter at all. Also the processor in the turbo voltage remains at the same level. Also you have nothing to lose, as the load on the cooling system and the components of the power converter is practically at the same level.

Test: AMD A10-7850K (Kaveri) Against A10-6800K (Richland)

Let compare the performance with the last APU series, AMD A10-6800K, it was ahead with the average of 2 to 3 %. It was because the clock speeds of AMD A10-7850K (4000 MHz) are lower than AMD A10-6800K (4400 MHz) – 400 MHz. And compare to AMD A10-7700K, the advantage is on the side of AMD A10-7850K with the average of 3 to 4%. Both have the same number of cores, faster is something that has higher frequency.

If talk about the Blue competitors, Intel – the performance of the 4-core AMD A10-7850K is quite good in the entire task against the Intel Core i3-4130, with an average of 16 to 19 %.

The performance of AMD A10-7850K in terms iGPU, against the AMD A10-7700K, AMD A10-6800K, Intel Core i3-4130 are already lagging far behind with 14%, 26% and 51% respectively.

Overclocking:

As we have told you earlier, this model has an unlocked multiplier, so it is easy to overclock without any kind of trouble.

With the overclocking we have manage to increase the frequency up to 4300 MHz, with the base rate has been fixed at around 100 MHz, and had to raise the voltage to 1.415 V. In this mode, the processor correctly passed the stress test program LinX 0.6.4, while the multiplier was increased up to x 43.

The temperature of the system does not exceed from the 50 ° C. These results are 16.2% better compared with the normal frequency (3700 MHz), and 7.5% compare to turbo frequency (4000 MHz). These results can be called average, because the same AMD A10-7700K with a nominal 3400 MHz to 4400 managed to accelerate MHz.

System performance in some tests are list below:

Normal Mode Overclocking Difference Futuremark PCMark 7 PCMark Score 3241 3401 4.94% Computation Suite 6992 7363 5.31% Futuremark 3DMark11 Score 6568 6765 3.00% Physics 4295 4663 8.57% Futuremark 3DMark Vantage CPU Score 10966 12465 13.67% TrueCrypt 7.1a (Serpent-Twofish-AES, MB / s) Encryption 151 170 12.58% Decryption 157 178 13.38% Batman Arkham City DirectX 11 (fps) 53 53.5 0.94% Rezident Evil 5 Benchmark DirectX 10 Smoothing x8 (fps) 91.7 96.9 5.67% Average 10.58%

As you can see the performance difference in the games and tests, which was increased up to almost 10.6%.

Increasing transition 300 MHz CPU turbo has no impact on performance. While overclocked allowed to increase its speed by more than 10%. This suggests that the technology AMD Turbo Core 3.0 does not work the best possible way.

When the GPU was overclocked, AMD Radeon R7 its frequency increased from 720 MHz to 960 MHz – 33% compared to the Normal mode, the voltage on the video kernel did not change.

Normal Mode Overclocked Differance Futuremark 3DMark11 Score 2512 2699 7.44% Graphics score 2340 2542 8.63% Physics score 4259 4214 -1.06% Combined score 2369 2510 5.95% Futuremark 3DMark Vantage Score 8691 8993 3.47% GPU score 8165 8510 4.23% CPU Score 10772 10838 0.61% Warhammer 40,000: Dawn of War II – Retribution DirectX 10 60.5 61.7 1.98% Rezident Evil 5 Benchmark DirectX 10 65.5 68.2 4.12% DiRT: Showdown DirectX 11 66.3 71.4 7.69% Average 12.36%

Performance of the GPU is increased up to 12 %.

Final Thoughts:

AMD did their best making emphasis on the integrated video core capabilities. Indeed, in the APU important than absolute performance in the test benchmarks, and a good balance between the processor and graphics. So far AMD A10-7850K is the best in all APU released till now. Its two dual-core processor module clocked at 3700 – 4000 MHz, provide good performance in all applications that the average user uses every day.

You can increase the CPU performance with the little optimizations and it is unlocked multiplier, so no problem with the overclocking, one thing you should remember that the overclocking potential of the novelty is not as large as that of the previous generation APU, but still the best. And for the games you can play most of the games with the medium or low graphic settings.

In fact, we just got a lineup expansion decisions under the platform Socket LGA1150. Not so long ago the same as AMD did, releasing AMD APU Richland, which is an optimized version of AMD Trinity APU.

Intel came to the updating processor line with a much larger scale. It has announced more than 40 new models 27 of them for desktop configurations, and 17 for mobile.

New released models are fully compatible with platform and Socket LGA1150 motherboards equipped with Intel chipsets Z9 and Z8 series. In the latter case it is necessary to simply update the BIOS. There is no need to describe in detail about the micro-architecture features of Intel Haswell Refresh, since it is identical to that used in the models of Intel Haswell.

For now we have Intel Core i5-4690T – energy efficiency with reduced TDP, let’s start with its appearance and detailed specifications.

Appearance:

To review we have got only the chip, so some of the content like box of the processor and cooling system are not available so that its description is missing, as we have told before that there is no difference between the Haswell series and Haswell Refreash.

Because of a family of processors Intel Haswell Refresh platform compatible with Socket LGA1150, then the location of contacts on the back side is fully consistent with this connector.

Specifications Details:

Looking at the table with the specification, you can see that Intel Haswell Refresh only slightly optimized versions of Intel Haswell. Indeed, we have in fact a complete copy of the technical characteristics of the technical characteristics of the model Intel Core i5-4670T. The only difference is in the frequency formula.

Model Intel Core i5-4690T Labeling QF4R CPU Socket Socket LGA1150 Clock Frequency (Nominal / Turbo In) Mhz 2500/3500 Factor (Normal/Turbo In) 25/35 Reference Frequency 100 – MHz Cache Memory Of The First Level L1 4 x 32 – KB

4 x 32 (Data Memory) – KB

Cache Memory Of The Second Level L2 4 x 256 -KB Cache In The Third Level L3 6 – MB Microarchitecture Intel Haswell Core Intel Haswell Refresh Number Of Cores / Threads 4/4 Support Document MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, EM64T, VT-x, AES, AVX, AVX2, FMA3 Maximum Rated Power (TDP) 45 – W Tehprotsess 22 – NM Technology Support VT-x, VT-d, TXT, TSX, Turbo Boost 2.0 Built-In Memory Controller Max Memory 32 – GB Types Of Memory DDR3 Maximum Frequency 1600 – MHz Number Of Memory Channels 2 Integrated Graphic Core Intel HD Graphics 4600 Unified Shader Pipelines 20 Modules Rasterization 4 Texture Units 8 Rated Frequency GPU 350 – MHz GPU Frequency Mode Turbo Boost 1200 – MHz Maximum Number Of Supported Displays - Support Instruction And Technology DirectX 11.1

Shader Model 5.0

Intel Quick Sync Video

Intel Clear Video

Intel Core i5-4690T normal operation speed without Intel Turbo Boost 2.0, is equal to 2500 MHz, supporting 100 MHz and multiplier x25. At the time core voltage was 0.933 V.

After enabling the Intel Turbo Boost 2.0, then multiplier increases by 10 points to the value of x35. Processor clock speed will be increased to the level of 3500 MHz, and the voltage – up to 1.201 V. This is a big difference in the frequency and voltage of the power in different modes due to the desire of manufacturers to reduce power consumption when the CPU have multi tasks.

At maximum loads scheme of processor cores as follows:

1 core – factor x35

2 cores – factor x34

3 core – factor x32

4 cores – factor x31

Processor frequency varies between 3100 – 3500 MHz  with supply voltage, fall below 1.201 V), but most of the time he spends at the rate of 3100 – 3200 MHz. All this is done for the sake of reducing energy consumption and reduce heat dissipation.

In idle form its decrease up to x8, and frequency drops to 800 MHz, voltage is 0.688 V. This implies that in terms of energy saving mechanism as nothing has changed compared to the CPU Intel Haswell.

Intel Core i5-4690T cache memory is distributed in the 4-core models from the family of Intel Haswell:

First level L1:

For each of the 4 cores allocated to 32 KB data with 8 channels of associativity and 32 KB for instructions and with 8 channels associativity

Second level L2:

For each core is given by 256 KB with 8 channels associativity

Third level of L3

6MB for all nuclei with 12 channels of associativity

Memory running in dual channel mode and support modules with up to 1600MHz DDR3. Maximum memory supported up to 32-GB.

Graphic core installed in Intel Core i5-4690T is Intel HD Graphics 4600. According GPU-Z 0.7.8 utility, it uses 20 unified shader pipelines, 4 blocks ROP and 8 texture units. Normal frequency is 350 MHz iGPU and dynamic can be increased up to 1200 MHz. The performance of the graphics processor Intel HD Graphics 4600 of family Intel Haswell Refresh remained at the same level as used in the previous model Intel Core i5-4670T.

Testing performance:

As for the testing the new model, we have used motherboard from ASUS, with the total memory of 8-GB – 2x4GB. Other details of the system are listed below:

AMD ASUS F1A75-V PRO (AMD A75, SFM1, DDR3, ATX) GIGABYTE GA-F2A75-D3H (AMD A75, sFM2, DDR3, ATX) AMD ASUS SABERTOOTH 990FX (AMD 990FX, sAM3 +, DDR3, ATX) Intel ASUS P8Z77-V PRO / THUNDERBOLT (Intel Z77, sLGA1155, DDR3, ATX) ASUS P9X79 PRO (Intel X79, sLGA2011, DDR3, ATX) Intel ASRock Z87M OC Formula (Intel Z87, sLGA1150, DDR3, mATX) Coolers Scythe Mugen 3 (LGA 1155/1366, AMD Socket AM3 + / FM1 / FM2) ZALMAN CNPS12X (LGA 2011) Memory 2x 4GB DDR3-2400 TwinMOS TwiSTER 9DHCGN4B-HAWP Video Card AMD Radeon R9 280X 2 GB DDR5 Hard Disk Western Digital Caviar Blue WD10EALX, 1TB Power Supply Seasonic X-660, 660W, Active PFC, 80 PLUS Gold

First of all you need to know that Intel Turbo Boost 2.0 have a major impact on system performance. Deactivating it will reduce speed up to 17%. For demanding applications such difference will definitely affect the speed of their work and will be noticeable.

As compared with the last generation, which have the same energy-efficient – Intel Core i5-3570T, the performance of the tested model Intel Core i5-4690T was better, not very much but up to 10 %.

As compared to the AMD processors, the tested model has very good performance, but in some task its little bit behind than the AMD, like tested model have 30 to 23 % better performance than the AMD A10-7700K.

If compare the performance of integrated graphics cores, the novelty from Intel is already far behind with more than two-fold difference in the results.

And as compared to the 6-core and 8-core model of the AMD FX series, AMD-8350 FX, it is ahead of 18 to 20 %.

Overclocking:

Since all processors have Intel Haswell Refresh multiplier locked, then the special achievements in terms of overclocking should not count. It was possible to increase the speed from 100 to 103.5 MHz, which in turn from 3500 to 3620 MHz (3.4% increase), and during experiment the CPU voltage did not rise, the technology Intel Turbo Boost 2.0 has been disabled, and the CPU multiplier was recorded manually at x35.

An average performance increase was only 2.5%, which was quite low overclocking the processor. In real applications, this increase remain unnoticed and virtually no impact on the speed of the vast majority of tasks.

Final Thoughts:

Intel Haswell Refresh dose not have anything new, slightly optimized version of the CPU generation Intel Haswell. And gain extra 100-200 MHz (depending on model) virtually no impact on the performance of the whole system, what we have seen in Intel Core i5-4690T.

The average difference in performance compared to the model Intel Core i5-3570T of the previous generation (Intel Ivy Bridge) was at 10%.

Going for the new Intel Haswell Refresh, only have one reason if your current CPU is very old, Duel-Core or Core-to-Due, nothing else.

Thinking to build a new computer in a compact package, the Intel Core i5-4690T may already be a more attractive option. This model has a relatively low power consumption and heat dissipation, but it has 4 full core and high performance.