What's new
  • Please do not post any links until you have 3 posts as they will automatically be rejected to prevent SPAM. Many words are also blocked due to being used in SPAM Messages. Thanks!

Intel Broadwell-E i7-6950X & i7-6900K Review

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,841
Location
Montreal
Intel’s high end desktop platform has been stagnant since Haswell-E launched in August of 2014 and while nearly 2 years may seem like an eternity for industry watchers, the new Broadwell-E processors are seeking to make up for lost time. From a company whose next moves are typically telegraphed well ahead of time, it is sometimes hard to generate any type of excitement around these launches but this time there’s certainly a few elements that will cause even the most jaded of folks to sit up and take notice. It all starts with the first 10-core, 20-thread chip available to desktop users: the positively insane i7-6950X.

Ah Broadwell-E….where do I start describing what Intel’s transitional microarchitecture has been through to get to this point? Due to a set of unforeseen and likely pretty darn serious mishaps its initial desktop lineup never got past an infantile state before being replaced with the Skylake platform. Haswell’s erstwhile 14nm LGA1150 desktop replacement has soldiered on in relative obscurity with a pair of SKU’s: the i7-5775C and i5-5675C. To this day they represent Intel’s proving ground for high level Iris Pro graphics rolled into a compact and pretty efficient desktop package. Their stillborn capabilities also offer a tantalizing look into what Team Blue could accomplish if they were sincerely interested in torpedoing AMD’s APU lineup.

BROADWELL-E-10.jpg

But enough about the past, you are all here to read about Broadwell-E, a platform which began life as a server-facing product and has now made the inevitable transition to the desktop HEDT space. Indeed the Broadwell architecture does represent a step backwards from the optimizations built into current-generation Skylake processors but as with all of Intel’s ultra high end initiatives, they are seeking to accomplish their goals by a very simple brute force approach.

Some will naturally wonder why Intel is even bothering with a lineup refresh for a product stack that’s priced at unattainable levels for typical users. The LGA1150 Skylake platform is efficient, comes equipped with countless features and is in general quite affordable. Think of it as the finely honed blade versus Broadwell-E’s howitzer. What these new processors do represent however is a recommitment to enthusiasts, gamers, professionals and content creators who want a no-compromise approach to performance but can’t justify a stratospherically expensive Xeon-based system.

BROADWELL-E-58.jpg

When I said expensive, I meant it. While the rest of Intel’s Boradwell-E lineup closely resembles their Haswell-E predecessors in terms of feeds and speeds, the top-flight CPU this time around is a 10 core, 20 thread monster called the i7-6950X. Not only does this processor include a pretty respectable 3.5GHz Boost speed but it also packs in 25MB of shared L3 cache, support for DDR4-2400 quad channel memory and a TDP of 140W. The price for all this multi-threaded goodness is a cool $1723USD, highlighting why this end of the market needs much more competition than it currently has. That sets a new high water mark for desktop enthusiast processors, even outstripping the i7 975 Extreme Edition by a whopping 50%.

Move a bit further down-market from the 6950X’s lofty position as the thousand pound gorilla of Intel’s lineup there’s the relatively pedestrian 8 core, 16 thread i7 6900K which is supposed to take over from the 5960X. It does so by retailing for the same $1089 price while offering a very small 200MHz frequency bump.

While many people will likely enjoy ogling at the performance offered by Intel’s two flagships, most will likely gravitate towards the $617 i7-6850K rather than the $434 i7-6800K. When taking a look at their initial specifications only 200MHz separates these six core processors from one another. However, looking past raw processing horsepower shows us the only differentiating factor is what actually draws many enthusiasts towards this platform in the first place: PCI-E lanes. Like with Haswell-E, Intel’s lowest-tier i7-6800K Broadwell-E CPU gets whittled down to just 28 Gen3 lanes, down from the 40 which grace its siblings.

Ironically, back when Ivy Bridge-E rolled out the most popular processor in its lineup was the $310 i7-4820K simply because it incorporated many of its generation’s benefits without cuts to its expansion capabilities. Now we’re being told to not only pay $100 more but also stomach constrained tertiary specs. So is this heartless up selling or crafty positioning by Intel? That’s for you to decide. On paper at least the 6800K provides a perfect bridge product between the Skylake-based 6700K and the upper echelons of Intel’s current lineup. The loss of a dozen PCI-E lanes is certainly something to take into account before choosing it but with NVIDIA now focusing on double SLI over triple and quad card setups, this chip might look really, really tempting.

To be perfectly transparent, these prices have actually changed since our original briefing with Intel a few weeks ago and press weren’t made specifically made aware of the update. Back then all of the information we had pointed towards the i7-6950X, i7-6900K, i7-6850K and i7-6800K retailing for $1569, $999, $587 and $412 respectively. That’s since changed and they’re all more expensive. There’s some irony hidden in this somewhere but right now I’m failing to see it.

All of these processors are completely compatible with current-generation X99 motherboards provided a simple BIOS update is done. However, the motherboard vendors have been busy behind the scenes and they’ll be in the process or rolling out “new” products at launch.

BROADWELL-E-6.png

The Broadwell-E architecture is a relatively straightforward thing when compared to the outgoing Haswell-E and it personifies what Intel’s Tick / Tock process is all about. Whereas Haswell and Skylake are the two new processor architectures that bookend it, back when it was first launch Broadwell represented the first use of Intel’s 14nm Tri-Gate transistor manufacturing process. Many figured that transition is what led to Broadwell’s late launch and ultimate failure to penetrate the desktop space.

Naturally, there is a bit more to Broadwell than a simple manufacturing process shrink and corresponding frequency uplift. There are also a few minor IPC boosts brought about by more efficient pathway communication and the addition of support for Intel’s new ADX instruction set. Other than the few extra cores jammed into the die courtesy of those 14nm transistors, don’t expect any drastic performance uplifts versus Haswell-E.

Perhaps the most important differentiating factor between Haswell-E and these new Broadwell processors is the implementation of Intel’s Turbo Boost Max 3.0. Like previous Turbo algorithms it is supposed to maximize performance by balancing thermals, power and frequency output. The best example of this is how Boost frequencies increase as the number of cores being utilized decreases. Intel’s third iteration takes this to the next level by adding more granularity to the equation by adding a Windows-based software component, technically allowing Broadwell-E to hit its maximum frequencies (and above) more often. As a matter of fact, the i7-6950X and i7-6900 could hit up to 4GHz in certain single-threaded applications under the right circumstances. More about this on the next page.

Other than the addition of their top-of-the-line i7-6950X, the Broadwell-E platform seems to be a very minor step forward for Intel’s enthusiast lineup. Given the competitiveness of Skylake and what seems to be a lack of any additional features to the 2011-v3 X99 motherboards for these new processors (more about this on the next page), will there be enough to alight interest in the jaded hearts of enthusiasts? Intel certainly thinks so.
 
Last edited:

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,841
Location
Montreal
Understanding Turbo Boost Max 3.0 & Intel’s “Best Core”

Understanding Turbo Boost Max 3.0 & Intel’s “Best Core”


Intel’s Turbo Boost algorithms used to be a very much transparent thing which was typically controlled through the motherboard BIOS. However, the term “control” is a very loose one since the function could be turned on and off but not much else. Broadwell-E’s Turbo Boost Max 3.0 on the other hand places additional modifiers within the hands of end users through a simple but effective software application.

In addition, during the production process, each Broadwell-E CPU is vetted through a process which determines which of its multiple cores is able to run the fastest and most efficient. This “Best Core” will be clearly marked in most motherboards’ BIOS with an asterisk or similar marker.

So what does that mean for performance? Quite a bit actually with Intel promising an “up to” 15% increase in throughput within certain applications.

BROADWELL-E-3.png

Past the obvious call-outs in the BIOS, the Turbo Boost Max 3.0 application itself is the heart of all things “Best Core” and actually needs to be installed to insure the proper functionality of this feature. Within it, there’s an enable / disable function, a setting which indicates that a foreground application will have priority and a box which allows you to add or remove programs to be prioritized.

In addition to all of that the Turbo Boost Max 3.0 program will also allow you to set core affinity on a by-core basis per application so, for example, a user could dedicate four cores for a specific game while the other six will be reserved for something like video rendering. This can essentially override the “Foreground App Has Priority” setting for applications that are running in the background but nonetheless using processor cycles. For those of you who think this sounds familiar, Microsoft has promised their Windows 10 Redstone update will seamlessly allow for the assignment of applications to specific cores. Intel’s application meanwhile brings this functionality to previous versions of Windows as well.

At its most basic, the driver installed alongside the software prioritizes workloads so they run on the fastest CPU cores. That’s where the Core list to the right side comes into focus. It is ordered from fastest (in this situation Core 9) to slowest whereby any adjacent core numbers may be equally fast as the one above and below it.

With the software / driver handoff doing its thing a single threaded application would run on the “Best Core” or Core 9 in this case while a dual threaded program would on Cores 9+4, triple thread on 9+4+8 and so on. It is meant to be highly dynamic and essentially boost Broadwell-E performance metrics in both single and multi threaded applications. As you’ll see in the benchmarks, it looks to be working in many situations.

BROADWELL-E-4.png

Moving past the basic settings, there’s even more options under the Advanced menu. Here the Evaluation Interval (in 100ms segments) can be changed with the default value being 1 second. This is the speed at which the application scans for demanding workloads before engaging specific cores. The Utilization Threshold meanwhile is specified by Intel as “Demanding work is work whose percent processor utilization over the period specified by Evaluation Interval exceeds a threshold specified by this advanced setting (Intel default = 90%)”

For the other two areas, Intel provides a pretty good explanation as well: When the box to the left of the Affinitize All Demanding Work advanced setting is checked (Intel default), the software takes actions to run the demanding work of all applications on processor cores listed in the Core List in an ordered fashion with the following priority: 1) Foreground Applications (if Foreground App Has Priority control is checked), 2) Applications in the Turbo Boost Max Applications list, 3) Other applications.

When the box is un-checked, the software takes actions to run the demanding work of Foreground Applications (if Foreground App Has Priority control is checked), and applications in the Turbo Boost Max Applications list only.

Affinity Controls - button selection is one of:
a. Affinitize To All Cores
b. Affinitize Only To Cores With Diversity

Affinity Controls specify whether the Intel® Turbo Boost Max Technology 3.0 Application / Driver places demanding work in priority order on all cores in the core list or whether it only places demanding work on higher performance (diverse) cores, which results in OS scheduler policy placing work on the residual cores. The Intel default setting is Affinitize To All Cores.
 

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,841
Location
Montreal
The X99 Chipset; More Lives than Disco?

The X99 Chipset; More Lives than Disco?


The new Broadwell-E processors obviously represent a large step towards enhancing the efficiency and performance of Intel’s enthusiast platform. Unfortunately, the X99 platform itself has remained relatively untouched despite several interfaces like U.2, M.2, Thunderbolt and USB 3.1 requiring an ever increasing amount of bandwidth. That doesn’t bode well for a series of motherboards that promises –in general at least- to cost more than their Z170 siblings.

Now before you start running at Intel with pitchforks and torches, remember this is actually not without precedent. X79 continued from Sandy Bridge-E onto Ivy Bridge-E without any noteworthy changes since its Platform Controller Hub included (at least in Intel’s mind) enough PCI-E lanes, SATA ports and other I/O elements for that generation’s needs. Enthusiasts had other ideas since, when compared to Intel’s less expensive Z97 its capabilities in the realms of native USB 3.0 support and Thunderbolt compatibility were missing in action.

The situation is actually the same now with Z170 and the X99 / Wellsburg PCH; Intel’s mainstream platform has included enhanced storage compatibility through the use of native PCI-E 3.0 lanes whereas X99 muddles along with PCI-E 2.0. As a result, motherboard vendors are being forced to come up with unique and sometimes performance-sacrificing methods to include the interfaces their high end clients expect.

BROADWELL-E-1.jpg

From a high level view the changes to X99 are limited at best but there is a positive spin to the lack of movement as well: since the LGA 2011-v3 socket and tertiary functions remain the same, these boards are compatible with Haswell-E CPUs. In addition, Broadwell-E processors are simple drop-in upgrades for first gen X99 boards provided a BIOS update is done. Now there are some very minor changes to path circuitry being done on these newer products but there’s nothing on them which would impede their predecessors from operating just as well.

Anyways, onto those “changes”. Both the firmware and BIOS support has been updated to work properly with Broadwell-E CPUs while Intel’s Rapid Storage Technology stack has been enhanced as well. The latter is of particular importance since it allows for more efficient communications through multi-drive NVMe and PCI-E storage connections. For the record, you can apply its changes to your existing X99 motherboard as well once the drivers become available.

Another item being rolled into Broadwell-E is compatibility with DDR4-2400 memory. However this is something of a red herring since Haswell-E processors had no problems working with substantially faster RAM speed bins and neither should these new chips. This is a purely system integrator driven specification and the market Broadwell-E is targeted towards will naturally push their modules far beyond this.

High level graphics card usage is what the Broadwell-E platform lives and dies by. That means things Z170 users can only dream of accomplishing; setups with two high end GPUs feeding through a pair of Gen3 x16 interfaces or triple / quad card rigs. There are some caveats with 28-lan Broadwell-E processors and we’ll get into that below. Also, NVIDIA recently put a dent in Intel’s plans by announcing their focus on dual card configurations but that’s a topic best broached another time.

Despite these minor revisions X99 looks very much like a platform which was launched three years ago, at least on the PCH side. Granted it does incorporate native USB 3.0 ports but those PCI-E 2.0 lanes are completely insufficient for feeding current and future high bandwidth storage solutions. Z170 meanwhile incorporates PCH-based Gen3 lanes which allow a lot more flexibility for motherboard designers. Broadwell-E motherboards will either need to rely upon their CPU-originating PCI-E 3.0 lanes or use expensive bridge chips to multiply the number of lanes they have available.

BROADWELL-E-7.jpg

In most situations the lack of additional PCI-E 3.0 lanes won’t be a hindrance provided you don’t want to run a serious amount of storage hardware. Compatibility also largely depends upon which CPU you choose. On most mid-range (read: $300 or less) X99 motherboards with a 40-lane processor installed either an M.2 or U.2 connector will become unusable when more than two GPUs are used. Higher end boards with those aforementioned bridge chips likely won’t encounter those hiccups. Just remember there’s no certainty to any of these statements since motherboard manufacturers are free to set up their products however they like.

When switching to the i7-6800K, things become a lot less clearly defined on a number of different fronts. Not only will dual discrete GPU systems be running at x16 / x8 speeds instead of x16 / x16 but quad card setups will be impossible without a bridge chip. In addition those 28 lanes leave previous little available Gen3 bandwidth for secondary devices like PCI-E based storage so many motherboard vendors will disable certain ports when the i7-6800K is detected.

Like previous generations, many of the changes between first generation X99 boards and their replacements fall into two areas: cosmetic ones and features that were present on Z170 boards being carried over to X99. For example, ASUS is including upgraded sound subsystems, RGB lighting and other items onto their new X99 Gen2 products while MSI, EVGA and Gigabyte are all doing it as well.

In some ways Z170 is being utilized as a proving ground for technologies which should eventually be rolled into Intel’s promised Skylake-E systems. Unfortunately that means Broadwell-E and X99, solutions supposedly designed for enthusiasts who want cutting edge tech, end up being left out in the cold more often than not. The question remains though: does this incapacitate the processors themselves? That’s up to buyers to decide.
 

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,841
Location
Montreal
Test Setups & Methodology

Test Setups & Methodology


For this review, we have prepared a number of different test setups, representing many of the popular platforms at the moment. As much as possible, the test setups feature identical components, memory timings, drivers, etc. Aside from manually selecting memory frequencies and timings, every option in the BIOS was at its default setting.

BROADWELL-E-59.jpg

For all of the benchmarks, appropriate lengths are taken to ensure an equal comparison through methodical setup, installation, and testing. The following outlines our testing methodology:

A) Windows is installed using a full format.

B) Chipset drivers and accessory hardware drivers (audio, network, GPU) are installed.

C)To ensure consistent results, a few tweaks are applied to Windows 10 and the NVIDIA control panel:
  • UAC – Disabled
  • Indexing – Disabled
  • Superfetch – Disabled
  • System Protection/Restore – Disabled
  • Problem & Error Reporting – Disabled
  • Remote Desktop/Assistance - Disabled
  • Windows Security Center Alerts – Disabled
  • Windows Defender – Disabled
  • Screensaver – Disabled
  • Power Plan – High Performance
  • V-Sync – Off
 

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,841
Location
Montreal
System Benchmarks: AIDA64 / Cinebench / PCMark / WPrime

AIDA64 Extreme Edition


AIDA64 uses a suite of benchmarks to determine general performance and has quickly become one of the de facto standards among end users for component comparisons. While it may include a great many tests, we used it for general CPU testing (CPU ZLib / CPU Hash) and floating point benchmarks (FPU VP8 / FPU SinJulia).


CPU PhotoWorxx Benchmark
This benchmark performs different common tasks used during digital photo processing. It performs a number of modification tasks on a very large RGB image:

This benchmark stresses the SIMD integer arithmetic execution units of the CPU and also the memory subsystem. CPU PhotoWorxx test uses the appropriate x87, MMX, MMX+, 3DNow!, 3DNow!+, SSE, SSE2, SSSE3, SSE4.1, SSE4A, AVX, AVX2, and XOP instruction set extension and it is NUMA, HyperThreading, multi-processor (SMP) and multi-core (CMP) aware.


BROADWELL-E-30.jpg


CPU ZLib Benchmark

This integer benchmark measures combined CPU and memory subsystem performance through the public ZLib compression library. CPU ZLib test uses only the basic x86 instructions but is nonetheless a good indicator of general system performance.

BROADWELL-E-31.jpg


CPU AES Benchmark

This benchmark measures CPU performance using AES (Advanced Encryption Standard) data encryption. In cryptography AES is a symmetric-key encryption standard. AES is used in several compression tools today, like 7z, RAR, WinZip, and also in disk encryption solutions like BitLocker, FileVault (Mac OS X), TrueCrypt. CPU AES test uses the appropriate x86, MMX and SSE4.1 instructions, and it's hardware accelerated on Intel AES-NI instruction set extension capable processors. The test is HyperThreading, multi-processor (SMP) and multi-core (CMP) aware.

BROADWELL-E-32.jpg


CPU Hash Benchmark

This benchmark measures CPU performance using the SHA1 hashing algorithm defined in the Federal Information Processing Standards Publication 180-3. The code behind this benchmark method is written in Assembly. More importantly, it uses MMX, MMX+/SSE, SSE2, SSSE3, AVX instruction sets, allowing for increased performance on supporting processors.

BROADWELL-E-33.jpg


FPU VP8 / SinJulia Benchmarks

AIDA’s FPU VP8 benchmark measures video compression performance using the Google VP8 (WebM) video codec Version 0.9.5 and stresses the floating point unit. The test encodes 1280x720 resolution video frames in 1-pass mode at a bitrate of 8192 kbps with best quality settings. The content of the frames are then generated by the FPU Julia fractal module. The code behind this benchmark method utilizes MMX, SSE2 or SSSE3 instruction set extensions.

Meanwhile, SinJulia measures the extended precision (also known as 80-bit) floating-point performance through the computation of a single frame of a modified "Julia" fractal. The code behind this benchmark method is written in Assembly, and utilizes trigonometric and exponential x87 instructions.


BROADWELL-E-34.jpg


CineBench R15 64-bit


The latest benchmark from MAXON, Cinebench R15 makes use of all your system's processing power to render a photorealistic 3D scene using various different algorithms to stress all available processor cores. The test scene contains approximately 2,000 objects containing more than 300,000 total polygons and uses sharp and blurred reflections, area lights and shadows, procedural shaders, antialiasing, and much more. This particular benchmarking can measure systems with up to 64 processor threads. The result is given in points (pts). The higher the number, the faster your processor.

BROADWELL-E-35.jpg



PCMark 8


PCMark 8 is the latest iteration of Futuremark’s system benchmark franchise. It generates an overall score based upon system performance with all components being stressed in one way or another. The result is posted as a generalized score. In this case, we didn’t use the Accelerated benchmark but rather just used the standard Computational results which cut out OpenCL from the equation.

BROADWELL-E-37.jpg



WPrime


wPrime is a leading multithreaded benchmark for x86 processors that tests your processor performance by calculating square roots with a recursive call of Newton's method for estimating functions, with f(x)=x2-k, where k is the number we're squaring, until Sgn(f(x)/f'(x)) does not equal that of the previous iteration, starting with an estimation of k/2. It then uses an iterative calling of the estimation method a set amount of times to increase the accuracy of the results. It then confirms that n(k)2=k to ensure the calculation was correct. It repeats this for all numbers from 1 to the requested maximum. This is a highly multi-threaded workload. Below are the scores for the 1024M benchmark.

BROADWELL-E-38.jpg
 
Last edited:

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,841
Location
Montreal
Productivity Benchmarks: 7-Zip / Blender / Handbrake

7-Zip


At face value, 7-Zip is a simple compression/decompresion tool like popular applications like WinZip and WinRAR but it also has numerous additional functions that can allow encryption, decryption and other options. For this test, we use the standard built-in benchmark which focuses on raw multi-threaded throughput.

BROADWELL-E-40.jpg


Blender


Blender is a free-to-use 3D content creation program that also features an extremely robust rendering back-end. It boasts extremely good multi core scaling and even incorporates a good amount of GPU acceleration for various higher level tasks. In this benchmark we take a custom 1440P 3D image and render it out using the built-in tool. The results you see below list how long it took each processor to complete the test.

BROADWELL-E-41.jpg


Handbrake


Video conversion from one format to another is a stressful task for any processor and speed is paramount. Handbrake is one of the more popular transcoders on the market since it is free, has a long feature list, supports GPU acceleration and has an easy-to-understand interface. In this test we take a 6GB 4K MP4 and convert it to a 1080P MKV file with a H.264 container format. GPU acceleration has been disabled. The results posted indicate how long it took for the conversion to complete.

BROADWELL-E-43.jpg
 
Last edited:

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,841
Location
Montreal
Productivity Benchmarks: POV Ray / WinRAR

POV Ray 3.7


POV Ray is a complex yet simple to use freeware ray tracing program which has the ability to efficiently use multiple CPU cores in order to speed up rendering output. For this test, we use its built-in benchmark feature which renders a high definition scene. The rendering time to completion is logged and then listed below.

BROADWELL-E-44.jpg



WinRAR


WinRAR is one of those free tools that everyone seems to use. Its compression and decompression algorithms are fully multi-core aware which allows for a significant speedup when processing files. In this test we compress a 3GB folder of various files and add a 256-bit encryption key. Once again the number listed is the time to completion.

BROADWELL-E-45.jpg
 

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,841
Location
Montreal
Single Thread Performance

Single Thread Performance


Even though most modern applications have the capability to utilize more than one CPU thread, single threaded performance is still a cornerstone of modern CPU IPC improvements. In this section, we take a number of synthetic applications and run them in single thread mode. The only addition to our normal benchmarks is Dolphin which uses a simple Nintendo GameCube emulation test on a single core.

BROADWELL-E-36.jpg
BROADWELL-E-39.jpg
BROADWELL-E-42.jpg
 

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,841
Location
Montreal
720P Gaming – 3DMark / DOOM / Dying Light

720P Gaming – 3DMark / DOOM / Dying Light


BROADWELL-E-46.jpg

BROADWELL-E-47.jpg

BROADWELL-E-48.jpg
 

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,841
Location
Montreal
720P Gaming – GTA V / Middle Earth / Witcher 3

720P Gaming – GTA V / Middle Earth / Witcher 3


BROADWELL-E-49.jpg

BROADWELL-E-50.jpg

BROADWELL-E-51.jpg
 

Latest posts

Twitter

Top