Intel Skylake i7 6700K CPU Review

Add Your Comments

Just barely a month ago, Intel’s “Tick” release of Broadwell desktop processors hit the scene. The dust has barely settled from that release, and a “Tock” release is upon us. As is usually the case, an Intel “Tock” release generally means a new microarchitecture. This can typically include improved energy efficiency, better performance, and improvements in video transcoding and encryption/decryption. As is the case with the i7 6700K we’ll be looking at today (and most “Tock” releases for that matter), a CPU socket and motherboard chipset change comes along with the release of Skylake processors. Yes, a new socket 1151 motherboard will have to be purchased if you’re considering a move to this platform. Let’s go for a ride and see what this highly anticipated platform has to offer!

Specifications and Features

A detailed specifications list won’t be available until the Intel Developer Forum (IDF) on August 18th. This initial announcement is just for the two enthusiast K series processors (i7 6700K and it i5 6600K). The entire 6th generation launch will have to wait until IDF, but Intel decided to give the enthusiast community top billing this time around ahead of the official launch.

intel_i76700k (13)

Short of having the detailed specifications, we were provided a press deck where we can see many specifications and new features the Skylake processors will bring to the table. Obviously, this initial announcement is aimed at overclockers and gaming enthusiasts with the two unlocked Skylake K series processors making their debut. Below is a list of key features for the i7 6700K we are reviewing today.

Intel® Core™ i7-6700K Processor Key Features

  • Intel® Turbo Boost Technology 2.0: Dynamically increases the processor frequency up to 4.2 GHz when applications demand more performance. Speed when you need it, energy efficiency when you don’t.
  • Intel® Hyper-Threading Technology: Allows each processor core to work on two tasks at the same time for up to 8 total independent tasks (threads) providing parallel processing capability for better multi-tasking with threaded applications.
  • Intel® Smart Cache: 8MB of shared cached allows faster access to your data by enabling dynamic and efficient allocation of the cache to match the needs of each core significantly reducing latency to frequently used data and improving performance.
  • CPU Overclocking Enabled (with Intel® Z170 chipset): Fully unlocked core multiplier, power, base clock and memory ratios enable ultimate flexibility for overclocking.
  • Graphics Overclocking Enabled (with Intel® Z170 chipset): Unlocked graphics multiplier allows for overclocking to boost the graphics clock speed.
  • Integrated Memory Controller: Supports 2 channels of DDR4-2133 and DDR3L-1600 memory with 2 DIMMs per channel. Support for memory based on the Intel® Extreme Memory Profile (Intel® XMP) specification.
  • PCI Express* 3.0 Interface: Supports up to 8 GT/s for fast access to peripheral devices and networking with up to 16 lanes configurable as 1×16, 2×8, or 1×8 and 2×4 depending on the motherboard design.
  • Chipset/Motherboard Compatibility: Compatible with all Intel® 100 Series chipsets with the latest BIOS and drivers.
  • Intel® HD Graphics 530: Integrated 3D performance with support for Microsoft DirectX 12 and Ultra HD 4K resolution display for immersive mainstream gaming. For Microsoft DirectX 12 testing, the performance-tuned Intel graphics driver is expected to be available for download on Intel Download Center by the 6th generation Intel® Core™ processor family product introduction. The Intel® HD Graphics 530 dynamic graphics frequency ranges up to 1150MHz.
    • Vibrant Media: Enhanced, built-in visual features deliver a seamless visual PC experience for rich Ultra HD 4K entertainment and HD gaming.
    • Intel® Quick Sync Video Technology: Media processing for incredibly fast conversion of video files for portable media players or online sharing including support for HEVC (H.265) encode/decode to support Ultra HD 4K.

Moving on to the slides Intel provided, it’s easy to see they are committed to the overclocking enthusiast and gamer. With an estimated 1.2 billion PC gamers worldwide, it makes a lot of business sense to develop products that cater to them. Being #1 in software sales, volume, and growth demands attention. PC gaming isn’t going anywhere any time soon with numbers like these.

intel_i76700k (10)
intel_i76700k (11)
intel_i76700k (12)

The next slides give us a few more details on the Skylake K series processors and the accompanying Z170 chipset. Intel claims impressive performance gains when compared to older PCs. Anywhere from a 30% performance increase over a three-year-old PC to a 10% increase over a PC that’s just a year old (read i7 4790K). Being the first mainstream desktop platform to support DDR4 memory, Intel took it a step further and now includes a finer level of memory tuning, which means an abundance of available memory ratios. Instead of the typical 200/266 MHz increments we are used to seeing with DDR4, we’ll now have ratio spacing of 100/133 MHz.

Another departure from past CPU architectures is the elimination of BCLK straps. The new reference clock architecture has untied the PCI-E and DMI bus from the BCLK. In a nutshell, this means any changes you make to the BCLK won’t have an effect on the PCI-E or DMI bus speeds allowing the BCLK to operate independently of the CPU core, cache (ring), memory, and BCLK generator. Also worth noting is the integrated voltage regulator has been removed from the CPU for vCore and other major voltages. Those duties will now fall back on motherboard manufacturers as they were a couple generations ago. This change should also help overclocking as on-die heat generation should be less.

Other new technologies found on the Z170 chipset include PCI-E 3.0 storage support, and more PCI-E 3.0 and USB 3.0 high speed I/O lanes (40% more compared to the i7 4790K). The chipset also provides support for CPU based PCI-E 3.0, 1 x16, 2 x8, or 1 x8 and 2 x4 to be precise. New to the Skylake CPUs is the Intel HD Graphics 530, which is unlocked and can be overclocked as well. We’ll save testing the iGPU post-official launch, but we’d like to compare the HD Graphics 530 to Intel’s Iris Pro Graphics found on the Broadwell desktop CPUs when we get to that article… stay tuned!

intel_i76700k (17)
intel_i76700k (14)
intel_i76700k (15)
intel_i76700k (16)
intel_i76700k (18)

From the information on the next slides, all the big names in the industry are ready to support this new platform with a plethora of pre-built PCs or components for the DIY’ers out there.

intel_i76700k (19)
intel_i76700k (20)

We typically wait until the end of a review to reveal the pricing of a particular item. However, in this case, we have a slide to show all the pricing details on these two Skylake K series CPUs. This slide also shows additional specifications on the i7 6700K, such as the 8 MB of cache and the 4.0 Ghz base and 4.2 GHz turbo frequencies. The TDP is 91 Watts and the processors feature four cores and eight threads. It’s not mentioned in the below slide, but the Skylake processors are built around the 14nm manufacturing process, which should definitely improve power consumption. If you look at the memory support column, you’ll see that Skylake processors will support both DDR3 and DDR4 memory. It’s unlikely you’ll see many Z170 chipset motherboards offering DDR3 support, but some manufacturers may offer DDR3 on their lower-end 100 series chipset motherboards. The i5 6600K is trimmed down a little on the cache and frequency and does not offer Hyper-threading. The i7 6700K has a recommended customer price of $350, and the i5 6600K sits at $243. Given Intel’s pricing history on new CPU releases, we feel those prices are right in line.

intel_i76700k (21)
intel_i76700k (22)

Product Tour

Our Intel CPU review samples always seem to arrive in a little black box with a couple anti-static foam pads inside. Nothing fancy, but it gets the job done. Inside, we get our first look at the i7 6700K, which looks a lot like what we’re used to seeing from Intel. The big difference of course, would be the move to socket 1151 and the contact area at the bottom of the CPU to support it. Rumor has it, the K series processors will not include a cooler in the retail box; but that shouldn’t be an issue. Why? Because socket 1151 uses the same hole pattern as previous socket 115X platforms, which means you can use your existing cooler. The people who purchase K series CPUs most likely do so because of overclocking potential, and they’ll probably prefer to use a better cooler than Intel packages with their processors. This helps to hold cost down, which is a wise decision considering most K series users will never use the stock cooler.

Review Sample Packaging

Review Sample Packaging

CPU Top

CPU Top

Bottom Contact Side

Bottom Contact Side

Top Side Angle

Top Side Angle

Bottom Side Angle

Bottom Side Angle

Benchmarks

For comparison samples, we’ll use two previous socket 1150 processors in the Haswell i7 4790K and Broadwell i7 5775C. We’ll also throw in results from the i7 5960X, which will obviously dominate the multi-threaded CPU based testing. We know most of you are interested in a clock for clock comparison with the i7 4790K, so we’ll run our benchmarks with both of those CPUs at the same speed and check that out too. It’s probably an unfair fight given the I7 6700K is using DDR4 memory, but it’s worth a look. Lastly, we’ll compare gaming performance against all four of the CPUs.

Here are the components used in the benchmark systems for each CPU, with a big shout out to ASUS and G.Skill for providing the Z170-Deluxe motherboard and Ripjaws V Memory used in today’s review system. Thanks guys!

i7 67000K i7 5775C / i7 4790K i7 5960X
Motherboard ASUS Z170-Deluxe ASUS Maximus VII Formula ASUS X99 Deluxe
Memory G.Skill Ripjaws V DDR4-3600 MHz G.Skill TridentX 2X8 GB 2400 MHz G.Skill Ripjaws4 4X4 GB 3000 MHz
HDD ADATA SP610 512 GB Samsung 840 EVO 500 GB Samsung 840 Pro 256 GB
Power Supply Corsair HX1050 Corsair HX1050 Corsair HX1050
Video Card EVGA GTX 780 Ti Classified EVGA GTX 780 Ti Classified EVGA GTX 780 Ti Classified
Cooling EK-Supreme LTX Water Block
360 mm Radiator
MCP35X Pump
EK-Supreme LTX Water Block
360 mm Radiator
MCP35X Pump
EK-Supreme LTX Water Block
360 mm Radiator
MCP35X Pump
OS Windows 7 X64 Windows 7 X64 Windows 7 X64

The graphs below are based off a percentage value with the i7 6700K being the basis, and therefore always 100%. The higher the percentage the better, except where timed benchmarks are used. Below each graph is the raw data used to make it.

We’ll use our standard suite of benchmarks for testing purposes, with all the comparison samples running at their turbo speed. That means the i7 5960X @ 3.5 GHz, i7 5775C @ 3.7 GHz, i7 4790K @ 4.4 GHz, and today’s i7 6700K @ 4.2 GHz.

First on the docket is AIDA’s suite of CPU, FPU, and memory tests. The CPU tests show the i7 6700K dominating the Broadwell i7 5775C across the board and swapping blows with the i7 4790K. The i7 6700K did show a noticeable advantage over the i7 4790K in the Hash and Photoworxx results, which test CPU performance using the SHA1 hashing algorithm and digital photo processing, respectively.

Looking at the FPU tests, we see the i7 6700K again dominating the i7 5775C and beating the i7 4790K in three of the four tests here. Particularly impressive here is the VP8 test where the i7 6700K topped the entire field by a good margin. VP8 is a video compression test, so we definitely see impressive gains for Skylake there. The 32-bit and 64-bit floating-point performance tests (Julia and Mandel) favored the i7 6700K by a nice margin over the i7 5775C and i7 4790K, but the 80-bit SinJulia test favored the i7 4790K.

AIDA64’s suite of memory tests show impressive numbers for the i7 6700K when compared to the i7 5775c and i7 4790K, which isn’t surprising given the move to DDR4. The i7 6700K even managed to beat out the i7 5960X in the write and latency tests.

AIDA64 CPU Test Results

AIDA64 CPU Test Results

AIDA64 CPU Benchmarks – Raw Data
CPU Hash AES Zlib Ph Worxx Queen
i7 6700K 4827 19127 377.9 28381 51230
i7 5775C 3918 16365 326.8 23655 45880
i7 4790K 4504 19979 377.8 22440 56246
i7 5960X 7167 31802 596.3 30727 78823

AIDA64 FPU Test Results

AIDA64 FPU Test Results

AIDA64 FPU Benchmarks – Raw Data
CPU SinJulia Mandel Julia VP8
i7 6700K 5063 19376 35765 7576
i7 5775C 4625 13184 24513 5873
i7 4790K 5607 18677 34882 7118
i7 5960X 8928 29730 55518 6414

AIDA64 Memory Test Results

AIDA64 Memory Test Results

AIDA64 Memory Benchmarks – Raw Data
CPU Latency Copy Write Read
i7 6700K 42.3 46117 52774 48601
i7 5775C 53.1 42403 37376 37589
i7 4790K 42.5 34859 37663 36095
i7 5960X 64.0 53595 46900 58303

CPU rendering and file compression performance are tested using Cinebench and the 7zip benchmark. Obviously, the i7 5960X will dominate these tests, but the i7 6700K swept the i7 5775C and i7 4790K in all four of these tests. The advantage over the i7 4790K is impressive when you consider the i7 6700K is being tested at 200 MHz below the i7 4790K’s core speed.

Cinebench and 7Zip Results

Cinebench and 7zip Test Results

Cinebench and 7zip Benchmarks – Raw Data
CPU 7zip
CB R15 CB R11.5
CB R10
i7 6700K 27620 938 10.29 36271
i7 5775C 23517 774 8.39 29556
i7 4790K 27304 905 9.67 34395
i7 5960X 42473 1410 15.26 43884

Video conversion is tested using the x264 and PoV Ray benchmark tools. The x264 Pass 1 test has the i7 6700K beating all the comparison samples, even the i7 5960X. Pass 1 is just the read only portion of the test, but Pass 2 is where the actual encoding is performed and where we see the i7 5960X flex its muscles. The Pass 2 test did have the i7 6700K beating out the i7 5775C and i7 4790K pretty easily though.

x264 and PoV Ray Results

x264 and PoV Ray Test Results

x264 and PoV Ray Benchmarks – Raw Data
CPU PoV Ray
x264 Pass 2
x264 Pass 1
i7 6700K 1997.64 58.77 223.84
i7 5775C 1560.85 46.57 187.13
i7 4790K 1831.70 54.99 211.53
i7 5960X 2845.74 83.42 204.95

Our 2D benchmarks include Intel XTU, wPrime, and SuperPi. The Intel XTU benchmark loves core counts, and as such, the i7 5960X leads the way. The i7 6700K handily beat out the i7 5775C and i7 4790K here. The wPrime benchmark again favors the i7 5960X across the board, but the i7 6700K easily beat out the i7 5775C and i7 4790K in this test run. SuperPi 1M is a single threaded benchmark that likes core speed, and that’s pretty evident by the results below. SuperPi 32M again likes core speed, but also responds well to memory speed and timings. The i7 6700K beat out the i7 5960X and I7 5775C here, but fell a tad behind the i7 4790K.

XTU, wPrime, SuperPi Test Results

XTU, wPrime, SuperPi Test Results

Intel XTU, SuperPi, and wPrime Benchmarks – Raw Data
CPU Intel XTU wPrime 1024M wPrime 32M SuperPi 32M SuperPi 1M
i7 6700K 1344 150.835 4.911 447.564 8.643
i7 5775C 996 183.221 5.899 520.370 10.374
i7 4790K 1118 164.473 5.333 429.282 8.159
i7 5960X 1742 103.647 3.525 536.894 10.359

Our gaming benchmarks use the settings seen below. The synthetic tests are run at default settings, but all the game benchmarks are pretty much maxed out. All the CPUs were tested using the same EVGA GTX 780 Ti graphics card.

Synthetic Tests

  • 3DMark Vantage – DirectX 10 benchmark running at 1280X1024 – Performance preset.
  • 3DMark 11 – DirectX 11 benchmark running at 1280X720 – Performance preset.
  • 3DMark Fire Strike – DirectX 11 benchmark running 1920X1080 – Standard test (not extreme).
  • Unigine Heaven (HWBot version) – DX11 Benchmark – Extreme setting.

Game Tests

  • Batman: Arkham Origins – 1920X1080, 8x MSAA, PhysX off, V-Sync off, The rest set to on or DX11 enhanced.
  • Battlefield 4 – 1920X1080, Ultra Preset, V-Sync off.
  • Bioshock Infinite – 1920X1080, Ultra DX11 preset, DOF on.
  • Final Fantasy XIV: A Realm Reborn – 1920X1080, Maximum preset.
  • Grid 2 – 1920X1080, 8x MSAA, Intel specific options off, Everything else set to highest available option.
  • Metro Last Light – 1920X1080, DX11 preset, SSAA on, Tessellation very high, PhysX off.

The four synthetic tests have the i7 6700K leading the pack in two of the four tests. It fell a bit behind the i7 5960X in 3DMark 11 and Fire Strike, but easily handled the i7 5775C and i7 4790K in all four tests.

HWBot Heaven Results

HWBot Heaven Results

3DMark Fire Strike

3DMark Fire Strike

3DMark Vantage Results

3DMark Vantage Results

3DMark 11 Results

3DMark 11 Results

Other than Batman: Arkham Origin where the i7 6700K and i7 4790K finished in a dead heat, the i7 6700K led the way in all our game benchmarks. The i7 6700K’s margin of victory was very slim in most cases, but it did perform better overall compared to the other CPU samples. You’ll notice Crysis 3 is missing from the test results as it currently will not operate on the Skylake platform.

Batman: Arkham Origin Results

Batman: Arkham Origin Results

Battlefield 4 Results

Battlefield 4 Results

Bioshock Infinite Results

Bioshock Infinite Results

Final Fantasy XIV: ARR Results

Final Fantasy XIV: ARR Results

Grid 2 Results

Grid 2 Results

Metro: Last Light Results

Metro: Last Light Results

Overclocking

We were able to stabilize the CPU at 4.8 GHz, while still keeping the memory set to 3600 MHz. The CPU voltage was set to 1.42 in the BIOS and CPU-Z reported 1.44 once in Windows. A few minor adjustments were also needed for LLC, VCCIO, and SA voltages to get things stable at this speed. I ran all the benchmarks we normally perform when doing a motherboard review at this overclock, which is the same overclock the i7 4790K was usually tested at. So, as promised, we have clock for clock comparison between the i7 6700K and i7 4790K.

As you look through the pictures below, the i7 6700K at 4.8 GHz is on the left and the i7 4790K at 4.8 GHz is on the right. The i7 4790K results were pulled from the ASUS Maximus VII Formula review. Obviously, the i7 6700K has an advantage because it’s using much faster DDR4-3600 MHz memory, but the results are interesting to see nonetheless.

Beginning with the three Cinebench tests, we see the i7 6700K performing quite a bit better than the i7 4790K. Percentage wise, CB10 is 9% better, CB 11.5 is 10% better, and CB 15 is 8.5% better.

CB R10 - i7 6700K @ 4.8 GHz

CB R10 – i7 6700K @ 4.8 GHz

asus_m7f-172

CB10 – i7 4790K @ 4.8 GHz

CB R11.5 - i7 6700K @ 4.8 GHz

CB R11.5 – i7 6700K @ 4.8 GHz

CB R11.5 I7 4790K @ 4.8 GHz

CB R11.5 – i7 4790K @ 4.8 GHz

CB r15 - i7 6700K @ 4.8 GHz

CB r15 – i7 6700K @ 4.8 GHz

CB R15 - i7 4790K @ 4.8 GHz

CB R15 – i7 4790K @ 4.8 GHz

The x264 tests also show an advantage for the i7 6700K by a little over 10% on both the Pass 1 and Pass 2 tests.

x264 - i7 6700K @ 4.8 GHz

x264 – i7 6700K @ 4.8 GHz

x264 - i7 4790K @ 4.8 GHz

x264 – i7 4790K @ 4.8 GHz

PoV Ray continues the trend with the i7 6700K performing just short of 12% better than the i7 4790K.

PoV Ray - i7 6700K @ 4.8 GHz

PoV Ray – i7 6700K @ 4.8 GHz

PoV Ray - i7 4790K @ 4.8 GHz

PoV Ray – i7 4790K @ 4.8 GHz

7zip compression testing again shows a substantial advantage for the i7 6700K. Here we see just over a 12% difference.

7zip - i7 6700K @ 4.8 GHz

7zip – i7 6700K @ 4.8 GHz

7zip - i7 4790K @ 4.8 GHz

7zip – i7 4790K @ 4.8 GHz

The first 2D benchmark is wPrime. Here we see just over 12% better performance for the i7 6700K.

wPrime - i7 6700K @ 4.8 GHz

wPrime – i7 6700K @ 4.8 GHz

wPrime - i7 4790K @ 4.8 GHz

wPrime – i7 4790K @ 4.8 GHz

SuperPi finally breaks the trend a little, and we mean very little. The 1M test showed a slight advantage for the i7 4790K of just over 1%. The i7 6700K quickly turned the tide and was less than 1% faster than the i7 4790K in the 32M test. Both of these tests are well within the margin of error, so it’s pretty much a draw here. Keep in mind that since SuperPi 32M is highly memory speed/timing dependent, these results could be a lot different with tighter timings.

SuperPi 1M - i7 6700K @ 4.8 GHz

SuperPi 1M – i7 6700K @ 4.8 GHz

SuperPi 1M - i7 4790K @ 4.8 GHz

SuperPi 1M – i7 4790K @ 4.8 GHz

SuperPi 32M i7 6700K @ 4.8 GHz

SuperPi 32M i7 6700K @ 4.8 GHz

SuperPi 32M - i7 4790K @ 4..8 GHz

SuperPi 32M – i7 4790K @ 4..8 GHz

Intel XTU shows a huge performance increase of over 17% for the i7 6700K when compared to the i7 4790K.

Intel XTU - i7 6700K @ 4.8 GHz

Intel XTU – i7 6700K @ 4.8 GHz

Intel XTU - i7 4790K @ 4.8 GHz

Intel XTU – i7 4790K @ 4.8 GHz

Memory testing includes a lot of data, so we won’t bore you with all the percentage numbers. Latency numbers are pretty close between the two processors, but read/write/copy performance is substantially better with the DDR4 memory in play.

AIDA64 Cache & memory - i7 6700K @ 4.8 GHz

AIDA64 Cache & memory – i7 6700K @ 4.8 GHz

AIDA64 Cache & memory - i7 4790K @ 4.8 GHz

AIDA64 Cache & memory – i7 4790K @ 4.8 GHz

MaxxMEM - i7 6700K @ 4.8 GHz

MaxxMEM – i7 6700K @ 4.8 GHz

MaxxMEM - i7 6700K @ 4.8 GHz

MaxxMEM – i7 4790K @ 4.8 GHz

With the addition of DDR4 to the Skylake platform and other improvements made to the CPU, we think it’s safe to say a 10% (or more) overall performance increase over Haswell is definitely possible. Using simple math to add all the percentage increases in performance and then divide by the number of tests, we come up with an overall performance increase of 9.5%. That doesn’t account for the memory testing though, which will push that overall value even higher. It’s hard to complain about that!

Temperatures and Power Consumption

Temperature testing doesn’t show a whole lot of difference between the comparison samples below, but that’s no surprise given they all were tested with the same water cooling system in place. Overclocking temperatures will vary greatly depending on your individual cooling scheme, but the water cooling setup on our test bench had no problem keeping the i7 6700K cool. Even using up to 1.42 V to the CPU, the maximum temperature ever recorded was about 84 °C. Below are the results under stock conditions, which show the i7 6700K running slightly cooler at idle, but a hair warmer under load compared to the other CPU samples. Different stress test utilities will yield varying numbers here, but we typically use AIDA64’s built in stress test and average the results from all four cores.

Temperature Comparison @ Stock (Normalized to 25 °C)
i7 6700K i7 4790K i7 5775C
Stock – Idle 27 °C 28 °C 28 °C
Stock – Load 60 °C 59 °C 59 °C

As far as power consumption goes, we see better overall numbers at the wall from the Skylake platform when compared to Haswell. The Broadwell platform tops the power consumption comparison as we would expect. These numbers will vary depending on your particular system configuration, but we definitely see improvements in this area over Haswell.

Power Consumption (Stock)
i7 6700K i7 4790K i7 5775C
Stock – Idle 114 124 110
Stock – Load 190 211 153

Pushing the Limits

We were able to get to the desktop and finish a quick SuperPi 1M run at 5.0 GHz. That’s about all we could accomplish though, as anything multi-threaded instantly locked up. We even tried lowering the memory ratio down to 3000 MHz, but that didn’t help matters. What we learned as far as this particular CPU goes, is 24/7 stability will be somewhere between 4.8 GHz and 4.9 GHz depending on the amount of voltage you’re willing to use on a daily basis.

SuperPi 1M @ 5.0 GHz CPU / 3600 MHz Memory

SuperPi 1M @ 5.0 GHz CPU / 3600 MHz Memory

Conclusion

On a strictly enthusiast level, the advantages the Skylake/Z170 platform has over the Haswell/Z97 platform are tough to turn a blind eye to. Our clock for clock testing tells the story of a 10% or better performance increase when the two were compared. Intel claims a 10% performance increase over Haswell, and they are spot on there.

The i7 6700K makes it clear there is a new king in the mainstream CPU market. If you couple that with a price that’s likely to be just a few dollars more than the i7 4790K, then it’s pretty much a no-brainer as to which platform to go with if a new PC build is on your horizon. If you’re currently sitting on a high-end Haswell platform, an upgrade is an appealing option as well. However, you will have the added expense of having to purchase a motherboard and DDR4 memory.

What does this mean? Click to find out!

What does this mean? Click to find out!

Dino DeCesari (Lvcoyote)

Leave a Reply

Your email address will not be published. Required fields are marked *

Discussion
  1. Great review Dino. It is my understanding that Intel is suing tim under the lid instead of solder. Do you have any info? I'm not asking you to delid :D I'm just curious.
    Thanks guys. As far as the TIM goes, we haven't received a detailed specifications list as of yet. Today was a "Soft" release with just the K series CPUs being announced/released. More detailed information will be released at IDF on August 18th. We should be able to find out more information then.
    I would have to guess with the temps we had that it may be solder...? But really not sure. That is just a guess and here is why I say that...: I was sitting at 4.9Ghz and 1.449v and didn't hit 70C. BUT... the voltage difference (1.3v stock to 1.449v) is not as much as the jump from 1.1v to 1.45v I had to use with 4790K.
    The FIVR is also gone so that saves a few C there as well... so, in the end, we will have to wait like Dino said. :)
    EarthDog
    I would have to guess with the temps we had that it may be solder...? But really not sure. That is just a guess and here is why I say that...: I was sitting at 4.9Ghz and 1.449v and didn't hit 70C. BUT... the voltage difference (1.3v stock to 1.449v) is not as much as the jump from 1.1v to 1.45v I had to use with 4790K.
    The FIVR is also gone so that saves a few C there as well... so, in the end, we will have to wait like Dino said. :)

    This sums up my first thoughts when I was talking with you this morning about the temps you saw.
    Availability update............newegg doesn't have a clue when....
    Newegg chat:
    Joe Shields: I was checking in on your expected availability for the new 6700K CPU released today.
    Agent Marvin X.: I am checking on it for you, Joe. One moment please.
    Customer Joe Shields: 10-4... thanks!
    Agent Marvin X.: Thank you for holding.
    Agent Marvin X.: I am sorry I cannot locate the item. Do you have a link of it? I will further check it for you.
    Customer Joe Shields: It is the new CPU from Intel that was released today.
    Customer Joe Shields: i7 6700K
    Customer Joe Shields: Retail availability is supposed to be today according to Intel.
    Customer Joe Shields: Link to our review on it... http://www.overclockers.com/intel-skylake-i7-6700k-cpu-review/
    Agent Marvin X.: Thank you for that information.
    Agent Marvin X.: I am sorry that we do not have the item on our website currently.
    Customer Joe Shields: Correct. I am asking when you expect it to be available.
    Agent Marvin X.: After double checking, I am sorry that we are not sure when the item can be available on our web site, Joe. Please check it online from time to time.
    Customer Joe Shields: Ok, thank you!

    Microcenter also does not know either...
    Grrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrr!!!
    I really don't agree with using uber expensive DDR4-3600 for the review. It's $400 for a 2 x 4GB set at Newegg. An equivalent set of DDR4-2800 is only $70 which makes an upgrade to Skylake affordable and would make the benchmarks more realistic.
    DaveB
    I really don't agree with using uber expensive DDR4-3600 for the review. It's $400 for a 2 x 4GB set at Newegg. An equivalent set of DDR4-2800 is only $70 which makes an upgrade to Skylake affordable and would make the benchmarks more realistic.

    By that logic we shouldn't be using DDR3-2400 10-12-12 for the 4790K tests.
    It's an overclocking site, we're going to be using fast memory.
    I'm not sure he had a choice...? But I dont imagine most results to be a lot different with using slow DDR4.
    Moving forward, I believe we settled on DDR4 3000 CL15-15-15-35 speed and timings for our motherboard reviews on this platform. I believe that is a sweet spot at the current time between showing what DDR4 can do and pricing.
    On the flip side, I dont see a point of testing DDR4 at DDR3 speeds (2666/2800)? We want to show progress too.
    Great write up as usual Dino. Thanks for spending the time to do it properly.
    I understand the necessity of switching to the faster DDR4, but how much of that 10% increase do you think is due to the faster ram? Maybe 3-4%? Just curious. Is there even a way to test that?
    Blaylock
    Great write up as usual Dino. Thanks for spending the time to do it properly.
    I understand the necessity of switching to the faster DDR4, but how much of that 10% increase do you think is due to the faster ram? Maybe 3-4%? Just curious. Is there even a way to test that?

    Easy way would be to downclock the ram to lower speeds and see how much scaling is involved in the results.
    Blaylock
    Great write up as usual Dino. Thanks for spending the time to do it properly.
    I understand the necessity of switching to the faster DDR4, but how much of that 10% increase do you think is due to the faster ram? Maybe 3-4%? Just curious. Is there even a way to test that?
    The only way to test that would be to run on a board that has DDR3 and DDR4 on it (since the CPU has IMCs for both IIRC) at the same speed and timings to minimize other variables.
    Memory bandwidth really doesn't have a lot of effect on many things. I would guess 1-2% at like speeds. But you (EDIT: WE) can't really run DDR3 on this platform at the moment, you have take the results as they are.
    I also tested IPC in the MSI Gaming M7 review but used DDR4 2666K versus my other reviews at DDR3 2666. The timings were WAY different though. Again, memory bandwidth in a lot of tests, there really isn't much difference.
    Thank ATM for seeing this: http://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/7
    Comparing default DDR4 to a high performance DDR3 memory kit is almost an equal contest. Having the faster frequency helps for large frame video encoding (HandBrake HQ) as well as WinRAR which is normally memory intensive. The only real benchmark loss was FastStone, which was one second in 48 seconds.
    End result, looking at the CPU test scores, is that upgrading to DDR4 doesn’t degrade performance from your high end DRAM kit, and you get the added benefit of future upgrades, faster speeds, lower power consumption due to the lower voltage and higher density modules.

    For discrete graphics card testing, only three differences stand out here. For GRID on the R7 240 the DDR4 set loses by 3.2%, but for the GTX 770 the DDR4 wins on Mordor by 6.4% and on GRID by 2.3%. All other differences are below 2%, mostly on the side of DDR4.
    There are reviews around the web with ~1-3% difference between DDR4-2133 and 3200 ... I wouldn't expect more than 5% difference. DDR3-2400 CL10 will be somewhere in the middle because of lower timings.
    Actually broadwell is faster clock to clock than skylake but it's limited in OC. On the other hand what we see in various reviews doesn't look really good on skylake. Most results are about 4.7GHz ~1.35V so about the same as on 4790K.
    I decided to get 6600K as it's $100 cheaper in local stores and I have 5820K anyway. It should arrive tomorrow. What is worse is that I can't find any motherboard worth to buy and all are weirdly expensive.
    Actually broadwell is faster clock to clock than skylake
    You would be the first person to say that. The couple of reviews I have read say the opposite. Can you link one up that shows what you are saying between broadwell and skylake?
    EDIT: I should clarify.. Skylake is faster (IPC) than Broadwell except in the rare instance where the eDRAM on die is used. :)
    ATMINSIDE
    By that logic we shouldn't be using DDR3-2400 10-12-12 for the 4790K tests.
    It's an overclocking site, we're going to be using fast memory.

    Actually, the fastest DDR3 I could find at Newegg is DDR3-3100. So DDR3-2400 is middle of the road like the DDR4-2800 I suggested for the Skylake. And further, why was slower DDR4-3000 used for the Haswell-E setup? At least the two DDR4 systems could have used the same RAM.
    None of the RAM choices made for this review make any sense at all when you're trying to compare relative performance.
    DaveB
    Actually, the fastest DDR3 I could find at Newegg is DDR3-3100. So DDR3-2400 is middle of the road like the DDR4-2800 I suggested for the Skylake. And further, why was slower DDR4-3000 used for the Haswell-E setup? At least the two DDR4 systems could have used the same RAM.
    None of the RAM choices made for this review make any sense at all when you're trying to compare relative performance.

    You're talking quad channel vs dual channel now.
    When DDR3 came out there wasn't anything running 2400MHz out of the box.
    With DDR4, we're still right out of the gate.
    I see what you are getting at Dave, but, how far down the rabbit hole do you want to go? The DDR4 3600 kit was only dual channel. If he used those in the X99 platform and took the time to retest everything, then someone would likely complain its not quad channel and it doesn't show the X99 platform as it should be used and THAT isn't a valid comparison either. Its tough to please everyone, but I feel he did a great job with the limited time he had to give us all the best information with the tools(parts) we have available.