Intel Kaby Lake Core i3-7350K versus non-K Showdown

CPU by leeghoofd @ 2017-02-26

Time for an update to the Kaby Lake launch article! Even though the multiplier-unlocked K-skew models are fairly priced, still a lot of PC users/builders opt-in for the cheaper variants. Mostly to cut down on the total system price or to maintain a better price/performance balance between the different components. Myself being labeled as a hardcore tweaker/overclocker has a hard time to grasp the previous. The quest for extra performance is never ending, being it for daily usage or for extreme benchmarking. Yet again is all this power required for daily usage? Do gamers need all this processor power or is a cheaper processor also a worthwhile option? Today we will test the Non-K skew Kaby Lake processors series and also compare them to the freshly released Core i3-7350K with the trustworthy MadShrimps test suite.

Intel Kaby Lake Core i3-7350K and non-K Models

 

Before we start wading through the different benchmarks let us list the new Core i3-7350K and the other Non-K skew Kaby Lake models. From the pretty basic Core i3-7100 to the full-blown Core i7-7700, Intel has got you covered with these non-K skew series.

 

 

None of the i3 models is equipped with the Intel Turbo feature, thus they have to rely on the listed core clocks. One can notice that Intel has done some compensation by allowing some beefy clocks on all these Core i3 processors. The quad core Core i5s of course don't feature a Hyper threading function: not a big issue as real cores are faster and most day to day applications have plenty of calculating power with the available four cores, though the i5 range is clocked significantly lower versus the i3 ones, yet again they can rely on the Turbo feature to achieve higher clocks.

Let us take a closer look at the two flagship K-skew models versus their non-K counterparts, as this will be an interesting comparison throughout this article. It still puzzles me when people invest in a K skew model and they just neglect the fact of overclocking it. Is the slightly higher frequency and therefore higher price justified over the non-K models? Is more always better?

 

Far higher out of the box clocks for the two K skew models versus their non K counterparts. Therefore the TDP is also more elevated at 91W versus 65W for the Non-K ones.

 

 

 

TEST SETUP

Our testsystem remains the same as the one of the initial Kaby Lake article. Thanks to Gigabyte, Tones.be and Seasonic for the provision of the samples!

  • Intel Core i7-7700K, 7700, 7600K, 7600, 7500, 7400, 7350K, 7320 & 7100 CPUs
  • Gigabyte Aorus Z270X Gaming7 (Bios F4Q)
  • Cooled by Be Quiet Silent Loop 240.
  • 16GB G.Skill TridentZ 3600C17 Dual channel memory.
  • MSI Lightning R290X graphics card (2)
  • 2TB Western Digital Caviar Green HDD.
  • Seasonic Snow Silent 760W power supply.
  • Streacom BTC Open Benchtable.

 

 

Benchmark Suite:

 

2D

  •  SuperPI 32M (Single threaded application)
  •  Wprime
  •  Maxon Cinebench R10 & 11.5
  •  Handbrake with 7.8GB Sample file encoding test
  •  Fritz Chess Benchmark

 

3D

  • Futuremark Firestrike Extreme and Ultra
  • Allbenchmark Catzilla 1080 & 1440P
  • Eidos Tombraider
  • 2K Games Bioshock Infinite
  • HWBOT Unigine Heaven
  • Stardock Ashes of Singularity
  • Rockstar Games GTA V

SuperPi 32M and Wprime 32/1024 Test

Super PI is a single threaded benchmark that calculates pi to a specific number of digits. It uses the Gauss-Legendre algorithm and is a Windows port of a program used by Yasumasa Kanada in 1995 to compute pi to 232 digits.For SuperPI 32M, in additional to CPU frequency, memory frequency, memory timings affect the total time as there is heavy memory access throughout the computation.


SuperPi 32M

 

 

 

When using single threaded applications we can clearly see the difference in clock speeds between the i3 and the lower clocked i5 processors, even though these sorts of applications are becoming more and more seldom. However SuperPi  still gives a good indication of the instructions per clock improvements a processor generation can offer.

 

 

Wprime 32/1024

Prime is a leading multi-threaded 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 sqrting, 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.

 

 

In the multi-threaded Wprime application we can clearly see that the real cores of the i5 processors make mincemeat of the dual core albeit with Hyper threading Core i3 range. Comparing the i5-7600 and i7-7700 versus their K brethren we see the slightly higher core clocks and even higher Turbo of the latter offer a distinctive advantage.

Maxon Cinebench R10 and R11.5


CINEBENCH is a real-world cross platform test suite that evaluates your computer's performance capabilities. CINEBENCH is based on MAXON's award-winning animation software Cinema 4D, which is used extensively by studios and production houses worldwide for 3D content creation. MAXON software has been used in blockbuster movies such as Iron Man 3, Oblivion, Life of Pi or Prometheus and many more. CINEBENCH is the perfect tool to compare CPU and graphics performance across various systems and platforms (Windows and OS X). And best of all: It's completely free!

 

Cinebench R10

 

 

With Maxon's Cinebench R10 we cross check again Single and Multi-threaded processor performance. Single threaded wise the core I3 73xx even give the i5 7600K a run for its money, though once we ran the multi-threaded test the lack of cores becomes obvious.

 

 

Cinebench R11.5

 

 

With the Cinebench Release 11.5 OpenGL test we can see how the processor can support the MSI Radeon R290X Lightning card. All processors put down pretty solid Frames Per Second. Anyway the score scales with clockspeed and available cores, thus even a lower clocked i5 can still match the faster clocked Core i3's with ease. The CPU test is again a clear victory for the Core i5/7s.

 

 

Deep Fritz 12 and Handbrake Test



Deep Fritz 12

We still include Deep Fritz 12, a chess program with integrated benchmark. Sadly no new updates have been made so the benchmark is using a maximum of 8 cores. Nevertheless still perfectly fine to compare some modern mainstream processors with one another.

 

 

Comparing the performance between these CPUs, we constantly spot the same: The Core I3s are no match for their i5 counterparts, no matter the clock speed. The K versus non K models reign due to higher core speeds and idem ditto Turbo clocks.

 

 

Handbrake:

HandBrake is a tool for converting video from nearly any format to a selection of modern, widely supported codecs. We use a sample 4K video file which you can grab here. We convert into a MP4 format and monitor the average FPS during the Fast 1080P encode. No alterations are made to improve performance anywhere. These are out of the box runs on the Handbrake 1.02 version.

 

 

Nothing new under the horizon as the same behavior is noticed time after time, if you want to encode go for Quad more cores and just avoid Core i3's in general.

Unigine Heaven HWBOT Edition

 

Time for some 3D, starting off with Unigine's Heaven benchmark. Unigine benchmarks can be effectively used to determine the stability of a GPU under extremely stressful conditions, as well as check the cooling system's potential under maximum heat output. The Heaven Benchmark with its current version 4.0 is a GPU-intensive benchmark that hammers graphics cards to the limits. This powerful tool can be effectively used to determine the stability of a GPU under extremely stressful conditions, as well as check the cooling system's potential under maximum heat output.

 

Heaven single R290X

 

 

Running Unigine Heaven Extreme benchmark on a single GPU usually leads to one conclusion: the GPU is the limiting factor, therefore all the scores are close to one another. Here the amount of cores hardly matters, though once we install a second MSI lightning GPU, the differences become more apparent.

 

Heaven R290X Crossfire

 

 

With a second GPU installed, the Core i5s pull away again from the cheaper Core i3 models. The Heaven benchmark is also not optimized to use all available processor cores; in fact a case that we experience with many modern games that still are optimized for a quad core processor.

Futuremark Firestrike Extreme

 

3DMark Fire Strike Extreme is an enhanced version of Fire Strike designed for high-end multi-GPU systems (SLI / Crossfire) and future hardware generations. In addition to raising the rendering resolution, additional visual quality improvements increase the rendering load to ensure accurate performance measurements for truly extreme hardware setups.

 

Firestrike Extreme R290X

 

 

Futuremark's Firestrike Extreme comprises both GPU and CPU tests. With one GPU there's only mild scaling in the total score with added CPU power or adding more cores. Nevertheless the Quad and higher specced cores have the edge in both the PhysX and Combined subtests. GPU subtest wise mild scaling too with higher specced processors.

 

Firestrike Extreme R290X Crossfire

 

 

By adding the second GPU, we lift the graphics bottleneck from the benchmark, though this extra graphic power needs serious backup from the processor, hence why we observe good scaling with the quad core processors. The Dual Core i3's are sort of bottlenecking the Crossfire setup in this test.

Futuremark Firestrike Ultra 4K

 

Can your PC handle the world's first 4K gaming benchmark? Fire Strike Ultra's 4K UHD rendering resolution is four times larger than the 1080p resolution used in Fire Strike. A 4K monitor is not required, but your graphics card must have at least 3GB of memory to run this monstrously demanding benchmark.

 

Firestrike Ultra 4K  R290X

 

 

Raising the resolution has a serious impact on the ability to render fast Frames Per Second by your favorite graphics adapter. It is typical the higher the resolution and the higher the chosen detail settings, the less impact your processor will have. Therefore the scores are way closer here than with the Firestrike Extreme benchmark.

 

Firestrike Ultra 4K R290X Crossfire

 

AllBenchmark Catzilla 1080P and 1440P

 

Catzilla is a computer test focused to check your computer graphics capabilities. The test can analyze if your computer is suitable for latest games or watching HD movies in best quality. It can also suggest what hardware modules of your computer should be replaced in order to get best performance.
The benchmark will also check software configuration of your system and will suggest improvements that may cause your computer to work faster and more efficiently.

 

Catzilla 1080P & 1440P  R290X

 

 

Catzilla 1080P & 1440P   R290X Crossfire

 

Eidos Tomb Raider & Ashes of Singularity

 

For our game tests we opted to include only those games that have an integrated benchmark build in.  The Eidos Tomb Raider game is one of the four games we have been using for a few years, still a solid looking game that can go hard on graphics card when all the eye candy is enabled. Testing the 1080P and 2160P resolutions with all settings on the Ultra profile. We didn't run Crossfire tests as our copy did not scale at all with a second R290X GPU installed.

 

 

At 1080P, no matter the processor we got all-round superb FPS in the Tomb Raider benchmark. When running the game at 4K resolutions we bottlenecked our single R290X GPU, making it max out at around 35 frames.

 

 

Ashes of Singularity

 

 

AoS was run in DirectX 11 mode on the Extreme Preset. We only took the Average FPS into account. At 1080P resolution we observe some mild scaling with added cores and speed. At 4K we are running into GPU bottleneck territory again. Maybe time for a faster GPU MadShrimps ?

 

Grand Theft Auto V

 

Grand Theft Auto V for PC offers players the option to explore the massive world of Los Santos and Blaine County in resolutions of up to 4k and beyond, as well as the chance to experience the game running at 60 frames per second.GTA V for PC offers players a huge range of PC-specific customization options, including over 25 separate configurable settings for texture quality, shaders, tessellation, anti-aliasing and more.

 

Grand Theft Auto V  R290X

 

 

Single card action, is without a surprise pretty spot on between most processors, as the game engine is maxing out the GPU capabilities.


Grand Theft Auto V  R290X Crossfire

 

 

When adding the second R290X GPU in the mix we spot the Core i3 as the Achilles heel of the game setup. Even though the Frames Per Second are still good enough for immersive play, a lower clocked quad core has got way more power under the hood.

Overclocking the Core i3-7350K

A few years back Intel thrilled the enthusiast community with the Pentium G3258 Anniversary Edition. A very affordable CPU that had massive overclocking potential; sadly the lack of HT killed the price/performance ratio for daily usage.

With Kaby Lake Intel has released their fastest Dual core to date, sporting a fully unlocked multiplier, the Core i3-7350K, this time this 7350K also has Hyper threading capabilities. Thanks to the unlocked multiplier there is no need for special BIOSes as with Sky Lake Non-K series to go high on frequency. However you will need a Z170 or Z270 motherboard to use the unlocked multiplier properly.

 

 

Our Core i3-7350K sample was running rock stable at 4800MHz with 1.3Vcore. Not bad, but not a golden CPU as we have seen some 5GHz samples on the Internet. Nevertheless a 600MHz gain versus the stock clocks is something we can only applaud as an enthusiast. However does it make a big difference?

 

2D TESTS:

 

 

With the extra 600MHz unlocked, the little Core i3-7350K packs some serious punch, though in heavy multi-threaded benchmarks it still can't keep up with a real Quad Core i5 processor. These Core i3 CPUs are great for an office PC or are a solid base as a daily surf platform, though for gaming or more multi-threaded applications Core i5 or i7 is the way to go.

 

 

 

The performance difference is clearly distinguishable, though real cores beat HT-enabled ones easily...

 

 

 

Temperature wise there was also a serious increase over the stock frequencies. With Prime95 running we hardly breached 60°C on the stock clocks. With all the CPU cores running at 4800MHz we measured 80°C on simple tower cooler; still in the green zone, though always keep track of the core temperatures when performing any overclocking on air or water.

 

 

 

Overclocking the Core i3-7350K 3D

Achieving a nice performance increase in 2D is one thing, how does the Core i3-7350K at 4800Mhz perform in some of our 3D tests.

No matter how we look at it, there is close to no benefit from the increased clocks on the i3-7350K, this mainly as our Radeon R9 290X GPU is being maxed out by either the benchmarks or the Game engines running at the maximum detail settings.

 

 

 

 

Depending on the test, we achieve only a mild increase or even in worst case scenario close to no increase in gaming performance. We can easily conclude that even overclocked this Core i3 still has no significant advantage over the lower clocked Core i5/i7 processors in most 3D situations.

 

 

Conclusion

Time to wrap it up! By looking at the numerous tests we performed in 2D as in 3D, we can conclude that for a basic day to day system one of these Kaby Lake Core I3 processors are more than powerful enough. Which model to best opt for? No matter the budget I would aim for the cheapest variant, the Core i3-7100. The higher specced i3 models are just a few hundred MHz faster and this will not warrant the prize difference versus the performance gained.

The Core i3-7350K leaves me with a mixed feeling: yes it can be one of the fastest dual core CPUs out there, thanks to the multiplier which has been kindly unlocked. Clock gains over 600MHz are easily achievable with just a few mouse clicks. Even though the performance goes up, it sadly can never weighs up to the ridiculous high price Intel asks for it. Retailing in the Benelux region at the same price level as the Core i5-7400 and only 20-25 Euros cheaper than a far better performing Kaby Lake Core i5-7500. Last but not least you need a motherboard sporting a Z170 or Z270 chipset to unlock the OC potential.

This Core i5-7500 could be the best price performance for a budget game setup. The higher specced i5 versions have a lesser price/performance ratio. The performance difference isn't worth the extra cash. If budget is less of a problem then go straight for the Core i5-7600K. Clock it up to +4700MHz speeds and enjoy superb all-round performance.

For the i7 range, the Core i7-7700K is the only way to go, even if one doesn't have the intention to overclock it; it is the fastest Quad Core Intel has in their lineup. With unleashed power under the hood, ready to be triggered, by you!  Best of all it is only 35 Euros more expensive than the non K Core i7-7700. In this price range, opt for the best!

 

 

Intel's Kaby Lake lineup has got processors for any user out there. Since there was not much competition, prices have skyrocketed over the years. The latter is the Achilles heel for some models, especially for the unlocked Core i3-7350K, which is positioned at the same price range as the Core i5-7400. No matter how you look at it, the i5 series are more all-purpose CPUs and rely on the crunching power of four real cores.

Anyway the upcoming release of AMD's new Ryzen CPUs might shake up the current establishment. Let the prize wars begin (again)!

 

 

 

 

I wish to thank Tones.Be for providing us with the entire Kaby Lake processor line-up. Kuddos guys!

 

And Gigabyte for providing us with the Gigabyte Aorus Gaming 7 motherboard for extensive testing

 

Seasonic Europe for the Snow Silent 750W PSU

 

and Streacom for the Open Benchtable setup.

  翻译: