In this second part we will analyze in detail the performance of the Intel i7-3960X with Sandy Bridge-E architecture. Several tests will be run with various software, specific tests on the memory, CPU frequency, 3D and consumption scaling and overclocking. Very interesting are the comparison tests the Intel i7-980X.
In our previous article, which you can read at this address, we have analyzed the architecture and main features of this new processor. In this article, after a brief mention of its characteristics, we will analyze only the performance. Scaling tests will also be performed comparing with the Intel i7-2600K and the AMD FX-8150 Bulldozer CPUs. Finally there will be tests on consumption of the various platforms and overclocking tests.
The architectural innovations are not many. Sandy Bridge-E LGA 2011 platform is an evolution of the previous 1155 LGA platform. We find therefore a greater number of cores, bigger L3 cache (up to 15 MB), a greater number of memory channels (4), a larger number of PCI Express links (40), and some additional features such as a more flexible overclocking, supported memory frequency of 2400 MHz and the PCI Express 3.0.All taking advantage of the proven 32nm production process and a greater area available in the die, also considered the absence (of course) of the integrated graphics in this series of CPUs.
The LGA 2011 systems are dedicated to enthusiast users, the most hardcore gamer, and all professionals who want uncompromised performance, especially in price.
System configuration and testing methodologies
• 3DMark Vantage
• AIDA64
DATA COMPRESSION AND MULTIMEDIA
• 7zip
• Winrar
• Cinebench 11.5 and 10
• X264 HD Benchmark
• Hand Brake
• Fritz Chess
• Blender
• POV-Ray
• Euler 3D
• SiSoft Sandra
Other specific tests for CPU frequency, memory frequency, 3D scaling with SLI GTX 580, and finally consumption will follow. There will be a page dedicated to testing and overclocking, however, using a x79 Asus Sabertooth.
CPU tests
3DMark Vantage: CPU test
In 3D Mark Vantage, performance primacy is of the proposed new Intel Sandy Bridge-E. This new processor sharply separates from the excellent Sandy Bridge architecture that is currently seen the i7-2700K as a top range model. The separation is also against the previous Intel architecture with 6 core, the i7-980X, which has lower performance compared to this new proposal.
At a clock to clock comparison performed at 4 GHz, is best seen as the improvement compared to the i7-980X CPU is evident.
AIDA64
AIDA64 is a software product from FinalWire that monitors the system by providing detailed information on hardware components. The software includes a utility inside the bench capable of testing memory and cache inside the processor.
In these tests with AIDA64, we note the sharp increase in terms of bandwidth compared to i7-980X CPU. Although the overall bandwidth of system memory is not the absolute top, you can not help but notice the high bandwidth of the L1 cache, which places it in the first place in overall standings.
The difference with the old 6 core Intel architecture is evident in this comparison, where the improvements are mainly in the L1 and L2 cache. The findings regarding the memory bandwidth leave room for some doubt on the actual benefits of quad channel even in a benchmark focused on memory as well as that of AIDA.
7Zip 9.20
This well-known archive software contains within it a tool that can analyze system performance, reporting a value expressed in MIPS (million istruction per second). The test includes compression, decompression, and overall value.
In this compression and decompression test performed with 7-Zip the two Intel CPUs with 6 physical cores are able to get the absolute primacy of the performance to other AMD and Intel CPUs. The two models i7-3960X and i7-980X, have very similar performance, indicating that this software still does not exploit the full potential of the new architecture.
In the clock to clock test it becomes clear that 7-Zip is not to exploit the benefits of the new architecture, the same thing happened with the CPU i7-2600K which is seen ahead of the old architecture with the Lynnfield CPU i7-875K.
WinRAR 4.01
In this other compression and decompression test performed this time with WinRar, you notice that not making any changes to the Windows registry, the performance is always superior to other cpu both AMD and Intel. It can be seen also that in these conditions the gap with the CPU-8150 FX is not very high. On the contrary, once a disabled core parking of Windows, you can see how the performance gain is of absolute reference. The increase is also clearly evident in comparison of Intel i7-980X CPU that in these conditions is defeated even by Sandy Bridge CPU architecture but on socket 1155.
In this comparison, as opposed to 7-Zip, you can clearly see that the computing power of the new architecture is fully exploited. The tests with Core Parking active show that the i7-980X processor is also exceeded by the i7-2600K CPU with Sandy Bridge architecture.
Cinebench 10
Software product from Maxon that allow, through the development of three-dimensional images and content, to test the CPU performance. This release allow to test the CPU using a single core, or all the cores inside the processor.
In this renders test done with the version 10 of Cinebench, we see how the new Intel CPU is the most outstanding performance ever. The distance from the top range of AMD cpu is truly remarkable, rather lower that towards Intel i7-980X CPU.
The 3960X computational power in this test is enhanced in both the single core rendering, both in multithreaded rendering. It's interesting to note the high single core performance improvement of Sandy Bridge architecture in the test with respect to Westmere architecture.
Cinebench 11.5
Software product from Maxon that allow, through the development of three-dimensional images and content, to test the CPU performance. This release allow to test the CPU using a single core, or all the cores inside the processor.
Very similar situation seen in the previous case with the version 10 of Cinebench. But this time the software is able to exploit the new architecture really well, and makes the best scores ever recorded in both single core with a score of no less than 1.58, both in multi-threaded with a score of 10.54. The distance from the other CPU is net. In this test the new Intel i7-3960X CPU has no rivals.
The new Intel has an extraordinary computing power per core, very comparable to the model i7-2600K. Higher than the previous generation Intel for more than 20%. Is still evident that the AMD Bulldozer is lacking from this point of view, with a single core computing power of 60% less.
X264 Benchmark HD 3.0
Software to measure CPU performance using x264 video encoding.
Good score in this test, where if the score in 1st Run may not surprise, that in the 2nd Run certainly does mark a staggering leap lead over the other CPU in a unique way. The gap is very high even against the previous Intel i7-980X.
In the 1st Pass test and in the 2nd Pass test you can clearly see as the difference in performance compared to the CPU i7-980X is very strong, we are at about a +17%. The Sandy Bridge architecture is a winner especially in the 1st Pass Test where the distance to other architectures is very pronounced.
Handbrake 0.9.5
Multi-threaded video encoding software with which we will transform a file (a movie) to MP4; the process includes x264 video encoding, FAAC audio encoding and muxing into final MP4 container. The time it takes for the CPU to perform this task will be taken into consideration.
In the conversion of the movie that we used for this test, the best performance in this clock to clock comparison is entitled to 6-core Intel CPUs. The time scored by the new architecture Sandy Bridge-E is slightly better than the i7-980X CPU.
Fritz Chess
Fritz Chess is the strongest chess engine and the evolution of what he faced the world champion Vladimir Kramnik in Bahrain in October 2002. Millions of chess fans watched the games live on the Internet. Fritz was able to equalize the match (4:4), and Kramnik showed all his respect for the performance of Fritz when he said: "DeepFritz is stronger than deepblue!" The new engine is based precisely on that Fritz8 version of "Bahrain" and was further improved in positional play.
Good score in this benchmark, which does nothing but confirm the general situation, even if the CPU i7-980X does not record a score entirely away. Substantially distanced all other CPUs instead, where the gap is truly remarkable.
In the test of chess, it is clear that the best performance are obtained from the Intel i7 6-core architecture. The new Sandy Bridge-E can still score something more in terms of CPU compared to the 980X, that before the arrival of this new processor had registered the best score.
Blender 2.60
Blender is an open source program modeling, rigging, animation, compositing and rendering of three-dimensional images. It also has features for UV mapping, simulations of fluids, coatings, particles, and other non-linear simulations, creating applications / games 3D. It is available for various operating systems: Microsoft Windows, Mac OS X, Linux, FreeBSD, along with unofficial port for BeOS, SkyOS, AmigaOS, MorphOS and Pocket PC. Blender has a robust feature set comparable in features and complexity, other popular 3D modeling programs.
The computing power of individual core is not surprising. Among the cpu in the standings in our tests, the Intel win the best position sharply outpacing the rival AMD. A sign that the IPC of these cpu is very high relative to the competition, including with regard to the new Bulldozer architecture. It is noted that in the single core the gap against the i7-980X cpu is very large.
Very special this graph, which does nothing but confirm what has already been seen for some tests in the single core. Substantially all of the Sandy Bridge architecture, including of course the i7-3960X CPU, get the same time in this test. The difference with other architectures is very strong.
POV-Ray 3.7
POV-Ray is a ray tracing program available for a variety of platforms. It was originally based on DKBTrace, a program developed by David Kirk Buck and Aaron A. Collins. He was also heavily influenced by the ray tracing program Polyray with the consent of its author Alexander Enzmann. In newer versions the rendering engine has been updated to allow the calculation of the deep global illumination, caustics, and elements to create particle clouds, fire, steam. Now makes use of multithreading, then takes advantage of multicore processors on the computer, or multiple processors.
The time registered in this program let leap the new Intel CPU at the top of our list. Sharply distancing the i7-980X CPU that before the release of Sandy Bridge-E won first place in the standings.
The first place goes to the new Intel i7-3960X, which separates all the others, although we note the i7-980X CPU can be around.
Euler 3D
It is a software that can handle real numbers, complex numbers, intervals, matrices and arrays. Track 2D/3D diagrams and includes a modern programming language. All versions of Euler are freeware with GNU. EULER is very similar to Matlab, but the author points out that this is not a clone. Euler 3D is used for analysis of fluid dynamics, with a freely available integrated benchmarking mode that uses all available cores and threads of the processor.
The generated score is absolute reference point for a desktop CPU. The distance from the other CPU is nothing short of embarrassing. If the Intel i7-980X CPU could still maintain a certain detachment from the Sandy Bridge, with this new Intel CPU has increased considerably this posting going strengthen it significantly.
In this test, we note as the new Sandy Bridge-E architecture is able to fairly distance the old Gulftown architecture on socket 1366.
Sisoft Sandra: Data Encryption
SiSoftware Sandra (System Analyser, Diagnostic and Reporting Assistant) is a diagnostic tool that can provide detailed information about your computer hardware and software configuration. A year ago, SiSoftware Sandra 2010 released offering full support to Windows 7, to 18 months after the launch of Windows 7, we have witnessed a real coronation of the PC to the media hub. A few months ago was released a benchmark dedicated to the Blu-Ray, now the authors have focused on delivering a new benchmark: the Media Transcoding based on Windows 7 Media Foundation. To all was given a new benchmark (GPGPU Cryptography) which allows a direct comparison between the performance recorded by the CPU (using the set of encryption available) and those of GPGPU.
Test that highlights the new architectures, where the presence of hardware encryption for data encryption, using AES and SHA instructions 256, is a winner. As you can see, the scores of Intel CPUs are undoubtedly the best ever compared to rival AMD. This new CPU manages to detach all the other sharply. In particular, the AES 256 is undoubtedly the score of record, by making a leap of nearly double the previous architecture, Gulftown.
In AES 256 cryptographic encryption computing power we have almost twice the i7-980X CPU power, with has similar levels of the i7-2600K processor. The new instructions are used extremely well in the new architecture.
Memory scaling
• 1333 MHz CL. 7-7-7-20 1T
• 1600 MHz CL. 8-8-8-22 1T
• 1866 MHz CL. 9-9-9-24 1T
The reference software we primarily used was AIDA64, where we analyzed the bandwidth of read, write, copy and access times to memory. The second software is WinRar 4.01 (stable version), which typically can achieve good performance increase with faster memory. The last software we used is Cinebench R11.5, where we checked the score on the multi-core.
You can see below the graphs for AIDA 64, where the bandwidth and access time were analyzed:
The second reference software is the known compression/decompression program WinRAR, where in effect for all architectures have shown increased performance rising the frequency of system memory. You can see below the graph we have achieved:
The latest software is Cinebench in the 11.5 version by Maxon that uses the Cinema 4D rendering engine. The benchmark, freely downloadable from the producer, enables us to make a comparison, in this case on multicore, on the various CPUs tried. We can observe the results that gave us such software:
With Cinebench is clearly seen that the memories do not have any effect on the final result, because the platforms have no substantive differences between using 1333 MHz memory and use memory frequency of 1866 MHz.
CPU frequency scaling
Another test that we conducted on this processor has up to verify the scalability of the operational frequency of the CPU from a base frequency of 3 GHz up to a frequency of 5 GHz in steps of 500 MHz As software we used the excellent compression/decompression 7-Zip program and the popular Cinebench R 11.5 rendering program that uses the Cinema 4D rendering engine. The results were very interesting and not entirely discounted. Below you can see the graphs created for the two software used:
Looking at the first graph, 7zip, you can see that all CPUs have an excellent performance as the frequency increases. The new CPUs manufactured by Intel has more than linear scalability, indicating that the internal architecture is ahead of the others. Distance is very high when we are at 5 GHz.
Totally specular situation with Cinebench test that shows the same results seen with 7zip. The new Intel CPU has a much steeper than linear performance increase, and the advantage increases as we go up in frequency.
3D scaling
Looking at the 3D Mark Vantage we can see how the scaling of Intel is less marked than the AMD, especially when switching from 4 GHz to 5 GHz, in this regard we can be seen the slope of the segment generated by the graph.
Turning to the 3D Mark 2011, you can see little marked inclination of the segment of Intel CPUs. You can also see a good inclination of the segment generated by AMD, where the frequency of 5 GHz is also going over the Intel i7-2600K.
Noting however the fps generated by Unigine, which reflects a more real gaming. Going from the default frequency to that of the 5 GHz the increase in fps is not as pronounced, indicating that already in default, Intel can provide high computing power. As you can see the fps generated are also higher than the AMD.
Consumption Scaling
Turning off the turbo for all CPUs, we note that for the FX-8150 consumption are increased (quite abnormal behavior), while for Intel CPU consumption go down slightly or remain the same as in the case of 2600K. The reason for this increase on the FX-8150 may be related to the TDP management system, built-in with turbo that could automatically turn off some tricks to limit consumption of Bulldozer architecture.
Bringing the frequency to 4 GHz, which we recall has not requested an increase of operating voltage for any CPU, we can observe that even in this case the consumption recorded by the AMD FX-8150 are very close to Intel i7-3960x with a minimum difference. In this case, the i7-2600K CPU shows a consumption of less than the other two with a gap of about 50W.
Further increasing the frequency and pushing up to 5 GHz, overvolt is required for all CPUs. For the FX-8150 the vcore was equal to 1.52v; for the i7-2600K CPU voltage of 1.6V and finally for the i7-3960X CPU voltage of 1.48v. Again we see how the two top range CPU of the manufacturers are marking their consumption very similar and this time the i7-2600K, despite the high vcore, does mark a radical departure from the other with a difference of more than 200W.
We conclude that the efficiency of Intel, even with the model i7-3960X confirms what has already been seen in performance. We have a good efficiency in relation to its excellent performance. The overclocking instead shows a very high consumption, which entails the adoption of adequate cooling, but also a very strong power supply. As we have pointed out in reviews of x79 motherboards, the little space available for the power supply section of the CPU requires a high efficiency and an adequate cooling for the mosfets.
Overclocking [Mini Guide]
In this section we try to provide a guide to overclock these processors, as well as understand and analyze the impact that the three components involved have: Bclk, HCl, CPU multiplier and RAM multiplier on the final result. The main difference between the socket 1155 desktop-class CPU is related to the presence of the final "K" letter or not. With the new-born Sandy Bridge-E bound to the socket 2011 Intel introduced initially in the first phase three CPU classes, precisely: I7-3960X, i7-3930K and the i7-3820 as shown in the table.
The freedom of action in terms of overclocking at least with the first two specimens is total and Intel has really heard the prayers of those in the past that had difficulty with multipliers, with BUS frequency and RAM to achieve balance and at the same time, the maximum frequency and thus the more stable computing power of all components involved. The i7-3820 even if it shares the architecture of the past is not an end Extreme CPUs, so it does not have the multiplier unlocked upwards, moreover, has only 4 cores (8 Logic). This CPU can be overclocked only by changing the frequency of buses. After this first introduction, we go deeper.
The factors at play that determine the final frequency are primarily entrusted to a clock generator that unlike the Z68 and P67 platforms is external to the chip and is entrusted to a buffered CK505 chip. This chip generates the base frequency of BCLK. Then based on the Bclk frequency, the remaining frequencies are generated through the Host Clock Multiplier. It acts as multiplication ratio, where the default is equal to 1.00. The other available multipliers are 1.25x, 1.66x and in some motherboards 2.5x. These last two multipliers are rather difficult to apply with the current generation of Sandy Bridge-E CPU. Looking at the new champion i7-3960X, the default frequency is 3300 MHz where the impact of the turbo, if adequate cooling is given (very sensitive from this point of view since it has internal sensors), brings the frequency up to a maximum of 3900 MHz. The table below simplifies the discussion.
As mentioned a moment ago, the final frequency is a function of the CPU Multiplier, of HCL and the frequency of BCLK. This framework provides the first rudiments of a freedom of action, for example, avoiding stress on the classic work with motherboards that are out of specific bus frequencies. Another concept that is important to understand is the impact that both the BCLK and the HCL have on the memory. Comes in handy in this sense, the multiplication/speed ratio that allows extreme flexibility in the modules under overclocking.
Exclusively for the overclocking tests we used a motherboard made by Asus, it is the Sabertooth x79 model, as we would have limited by the Intel board in the results obtained. The rest of the test system has remained unchanged.
Overclocking [Achievements]
Before performing the overclocking test we disabled the inner features in the various CPUs such as EIST, SpeedStep in order to relieve the CPU from voltage fluctuations due to the strong amperometric absorption due to the high frequency and the resulting strong workload. Thanks to the solid power system DIGI+ accompanying the Sabertooth x79, it can also act via the software interface Asus DIGI+ adjusting the level of Load-Line Calibration, very important to make the supply voltage as stable as possible to the CPU in presence of strong absorption.
We chose Lynx software to bring the level of CPU usage to 100% and test the so-called Rock-Solid, namely the stability of the achieved level of overclocking. Lynx is based on the Linpack libraries with which computes linear equations by returning a value "GFlops" (billions of floating point operations per second). Overclocking means deal with the temperature reached by the processor that is the leading i7-3960X Esacore CPU introduced by Intel in the consumer market. These are the results taking into account that we have entrusted to the CPU-Z tool for the detection of voltages provided.
In general, returning to what was said in the previous guide, we noted a particular difficulty climbing the bus, even using the host clock multiplier. The easiest way to overclock is therefore to use the CPU multiplier, quite similar to what happens in the LGA 1155 CPU. In some cases you should apply the multiplier to 1.25x bus, but reaching a situation of stability of all the cores will be much more complicated. The maximum achievable frequencies are usually slightly lower than those of the LGA 1155 CPU. This can easily be explained by the virtually double number of transistors in the die, which multiply the probability of incurring in small manufacturing defects and therefore less stable CPU frequency increases. The barrier of 6GHz for Sandy Bridge CPU is not therefore derived from impossibility to go up by bus (a problem identified on the LGA 1155 and passed on with LGA 2011 through the HCM), but it is evidently a true architectural and production process limit. These CPUs, in fact, unlike the AMD counterpart, poorly tolerate extremely low temperatures and is usually sufficient to dissipate the heat produced. The benefits of extreme cooling solutions are therefore limited in terms of overclocking margin.
Conclusions
Performance | |
Efficiency | |
Overclock | |
Price | |
Overall |
Intel can still improve this CPU, the CPU Sandy Bridge-E integrates 8 cores already inside them, although two have been disabled. The position of advantage over the competition allowed it to not press the accelerator too much, giving up to many features of the x79 chipset to avoid potential errors such as happened with the bug of the P67.
Tests show without a shadow of doubt that this new processor is the current benchmark in desktop CPUs, although in some cases the gap with the architecture Gulftown is not so marked, but it is mostly caused by the failure to exploit this new architecture by the softwares. Very impressive computing power per core shared by all Sandy Bridge CPU architecture.
Analyzing the frequency scaling tests, the advantage of this SB-E platform is very marked. The results show a behavior more than linear with increasing frequency. Less obvious improvements derived from the use of RAM memory more or less driven. The scaling for the 3D stands at levels very similar to another Sandy Bridge CPU architecture, the i7-2600K.
Consumption measured by our instrumentation make this new processor in direct competition with the Bulldozer CPU AMD FX-8150: especially in overclocking, consumption of the two platforms are very similar, but of course for the Intel platform the performance advantages are very obvious and of leadership levels.
Even for the overclocking results we have obtained quite satisfactory results, although not markedly different, despite a greater flexibility in managing the clock, compared to Intel Sandy Bridge on socket 1155.
The price is undoubtedly high, we talk of a price in the order of € 900 including VAT, but Intel has always accustomed to high prices of its Extreme Edition CPU where those who want to buy it does not surely care at the price, but demands maximum performance as a reference. This CPU does accommodate these Enthusiast users. Will be very interesting to test the i7-3930K CPU which has a lower purchase price and the specifications are very similar. Stay tuned for these new tests!
Translated by Marco Comerci