Productivity benchmarks - Core i7 8700K vs Core i5 8600K vs Ryzen 7 2700X vs Ryzen 1700 vs Xeon E5-2680 V4

[Artas1984]

Greetings once again from me folks, newcomers and those who know me for a long time already! Over the last 6 months i've been assembling and selling various computers, thus having tested many processors in the course! I always wanted to do another massive CPU battle since THIS, and finally, this will be my biggest CPU benchmark in 6 years. Also this is the first time i present benchmarks in Techpowerup since last year's 4K video card benchmark thread.

This CPU benchmark is dedicated to folks who are looking a for a budget friendly workstation build - for those who can't afford new AMD X399 or Intel X299 systems with Ryzen Threadripper or Core i9 processors, but are checking out Z370 and X470 platforms with flagship Core i7 8700K and Ryzen 7 2700X processors. Not only this will be a massive direct battle between these two chips, but also included are the other alternative chips, like Ryzen 7 1700 and Core i5 8600K, which cost less and can provide enough performance for your WS rig. Without a doubt, older Core i5, Core i7 and Xeon processors have been included in the benchmark to provide comparison and perhaps competition to the modern chips.

PROCESSORS

All overclockable processors have been overclocked to their maximum capabilities! The clock speed of each CPU is the maximum all cores achieve.

Core i7 8700K, 5 GHz and 4.8 GHz for AVX tasks
Core i5 8600K, 5 GHz and 4.8 GHz for AVX tasks
Core i7 6700, 3.7 GHz
Core i7 5775C, 4.3 GHz
Core i5 4570, 3.4 GHz
Ryzen 7 2700X, 4.2 GHz
Ryzen 7 1700, 3.8 GHz
Xeon E5-2680 V4 ES, 2.3 GHz {14 cores/28 threads}
Xeon E5-2630 V4 ES, 2.3 GHz {10 cores/20 threads}
Xeon E5-2620 V4 ES, 2.3 GHz {8 cores/16 threads}

OTHER PARTS

RAM: 4X8 DDR3 1600 MHz CL8 and 2X16 GB DDR4 2133 MHz CL14.
HDD: WD SE WD2000F9YZ 2 TB SATA3 7200 RPM 64 MB
GPU: COLORFUL GEFORCE GTX 1080 TOP V2 8 GB
WINDOWS 7 X64 SP1
NVIDIA FORCEWERE 398.36

Here is screenshot of the 4 more heavily overclocked processors.

At first i was assembling workstations with Broadwell-EP Xeon ES processors, since they are being sold real cheap at ebay. Even the 14 core/28 thread CPU can be found for 400 - 500 $, while 6 months ago it was going on sale for 700 $. The 18 core/36 thread Broadwell-EP Xeon was being sold for 1000 $, so it was a too expensive shot. That got me thinking:

- ''what if it ain't worth buying those Xeons, what if Core i7 8700K or Ryzen 7 2700X are actually faster, what if i am making a mistake by assembling workstation computers with Xeon chips''?

Ye, you get the idea where this benchmark came from now...

These processors have been tested in real time programs with custom presets as well as in standard synthetic benchmarks. Video and photo conversion, 3D and animation rendering, and supportive tasks for video card rendering will be the focus of this benchmark, as i searched that there are no benchmarks in the internet comparing all the processors in one stock. No gaming benchmarks.

LETS START!

CPU-Z MULTI THREAD

CINEBENCH R15

V RAY

CORONA

-----------------------------------------------------------------------------------------------------------

NEXT BENCH-MARKED ARE PROGRAMS WITH REAL PRODUCT [FILE] VALUE UPON COMPLETION

HANDBRAKE CUSTOM VIDEO FILE CONVERSION 1

HANDBRAKE CUSTOM VIDEO FILE CONVERSION 2

HANDBRAKE CUSTOM VIDEO FILE CONVERSION 3

BLENDER CPU RENDERING BMW

BLENDER CPU RENDERING CLASSROOM

BLENDER CPU RENDERING BARCELONA PAVILION

BLENDER CPU RENDERING SPLASH

BLENDER CPU RENDERING MAIN BENCHMARK

IRFANVIEW PHOTO ALBUM CONVERSION

IRFANVIEW PHOTO ALBUM CONVERSION

IRFANVIEW PHOTO ALBUM CONVERSION

IRFANVIEW PHOTO ALBUM CONVERSION

------------------------------------------------------------------------------------------------------------

NEXT IS A TASK FOR THE PROCESSORS TO SUPPORT VIDEO CARD IN RENDERING. THIS TEST SHOWS WHICH PROCESSORS MIGHT BE A BOTTLENECK IN CERTAIN SCENARIOS.

SUPPORTIVE TASK FOR VIDEO CARD RENDERING 3DS MAX

SUPPORTIVE TASK FOR VIDEO CARD RENDERING MAYA

SUPPORTIVE TASK FOR VIDEO CARD RENDERING CREO

SUPPORTIVE TASK FOR VIDEO CARD RENDERING MEDICAL

SUPPORTIVE TASK FOR VIDEO CARD RENDERING CATIA

SUPPORTIVE TASK FOR VIDEO CARD RENDERING SOLIDWORKS

SUPPORTIVE TASK FOR VIDEO CARD RENDERING SHOWCASE

-----------------------------------------------------------------------------------------------------------

PERSONAL NOTES ABOUT MY VIDEO BENCHMARK (2X16 GB DDR4 ECC 2400 MHz CL15-15-15-36-300)

---> MKV, x265, 8 bit, 3840X2160, 25 FPS, constant, slower, auto, 64 Mbp/s, 1 pass.

Xeon E5-2683 V4 2.3 GHz 16 C/32 T: 24:32
Xeon E5-2663 V3 3.1 GHz 10 C/10 T: 26:53

 

[Solaris17]

while both Core i7 8700K and Core i5 8600K reached 4.8 GHz with predetermined Asrock Z370 Killer SLI settings, both of them would throttle in some of the more demanding programs to 4.3 GHz no mater what i tried... I left Core i5 8600K all cores overclocked at 4.3 GHz, which is it's default turbo boost anyway, while i overclocked Core i7 8700K to 4.8 GHz despite knowing that in some programs it would go down. In the upcoming benchmark i wrote where that happens. Unfortunately this means that this benchmark will favor AMD chips already, since they are at their maximum overclock, while Intel chips are not!

Politely, while I appreciate you taking the time and having the enthusiasm to make this write up, you must know if you are taking reviews seriously, that by that paragraph alone all these results are invalidated.

 

[Toothless]

Gotta run them all at stock for valid results.

 

[Artas1984]

Politely, while I appreciate you taking the time and having the enthusiasm to make this write up, you must know if you are taking reviews seriously, that by that paragraph alone all these results are invalidated.

I understand perfectly. Biased reviews, even if unintentionally made, can not be taken at face value. While i won't be able to replicate the results with stock processors, i can still manage to fix the problem with Core i7 8700K and Core i5 8600K. I would gladly replace the motherboard in order to present these chips at their fullest, but that requires time. If i will have the option to edit this thread for a long time, like a month, i am willing to fix this. So far i won't be uploading this content to my youtube channel.

As for running all processors at stock, sure - this would be required for a professional presentation, but that's secondary compared to fixing the damage - fairness of the test itself. Also, just testing one processor takes a full day, never intended to bench these processors at stock speeds anyway, because it would consume too much time and give little value as people do not run their Core i7 8700K at 3.7 GHz anyway... However, point taken and agreed, which means there is no need to talk about it further.

For now, take what it is.

 

[lexluthermiester]

Politely, while I appreciate you taking the time and having the enthusiasm to make this write up, you must know if you are taking reviews seriously, that by that paragraph alone all these results are invalidated.

Why? They are valid within the context of the testing methodology.

Gotta run them all at stock for valid results.

Again, why? OP is correct, most people buying a "K" series CPU are OCing it. Stock speed testing would be of little use to show potential gains.

 

[Frick]

Would you consider doing a run with the CPUs at the same clockspeed? Memory speeds will be different of course, but it would still be interesting.

 

[phill]

Appreciate the time it's taken and the effort you've put into it.. Impressive work! Kind of reminds me of my benchmarking days!

I find running things at stock first and then overclocking it, you tend to appreciate the differences more so, but that's just personal preference Been testing my 5960X with WCG (World Community Grid) and been having fun with power usage and all that sort..

Anyways, thanks for the interesting read

 

[Artas1984]

Asrock Z370 Taichi will be coming soon. Will remake the results for 8 generation CPU this week!

I'd love to do stock CPU results myself, but if i had to present the results between overclocked and stock configs, i think the great majority here would agree, that the overclocked results are more relevant. As i said before, remaking the benchmarks with stock Core i7 5775C, Core i5 8600K, Core i7 8700K, Ryzen 7 1700 and Ryzen 7 2700X would take me 5 additional days. Perhaps someone is willing to pay me for that??? This is not just a comparison between two processors in a few random programs, it's as if you had little idea folks how long does it take to do this stuff... Those who understand should respect my decision to overclock the max out of everything straight away.

 

[AltCapwn]

Damn I didn't know Ryzen 2700X was that cool. Even single thread perf is surprising!

 

[Totally]

Damn I didn't know Ryzen 2700X was that cool. Even single thread perf is surprising!

I was thinking about upgrading to it from a 1700X but seeing that there is almost no diff from the 1700, I'm going to hold off.

 

[notb]

Why? They are valid within the context of the testing methodology.

There is no valid context. These tests are not even seriously work-oriented. WinRAR? Seriously? :-D

Again, why? OP is correct, most people buying a "K" series CPU are OCing it. Stock speed testing would be of little use to show potential gains.

Because workstations are not overclocked.

 

[R0H1T]

Gotta run them all at stock for valid results.

What do you mean by stock? Does Intel come with stock cooler, with unlocked chips? Are the reviews valid where LC is used? Why are there no reviews at fixed clocks?

There is no valid context. These tests are not even seriously work-oriented. WinRAR? Seriously? :-D

Because workstations are not overclocked.

Says who? You would be right about most companies, but not individuals who do work from home or employ OCed systems for doing part of their job from home.

 

[John Naylor]

I have no interest in seeing benchmarks at stock settings. The only interest I have is seeing tests at the speed I will be running them at. Every workstation we ever used or built was overclocked. The most obvious instance is with video cards ... when AMD and nVidia were competitive, both coukld be overclocked fairly well, but when AMD was no longer able to keep up, they started to very aggressively OC their cards in the box it. So when looking at the 1060 / 480 choice for example, one could have based by choice on the "stock" advantage that AMD hadfor example ... or I could venture into the "real world" nad base the comparison on how they will actually perform once a user takes the machine home from out workbench.

Over 90% of CAD workstations are Intel I7s w/ GTX GFX cards, because despite perception, 2D and 3D CAD is faster on these platforms where workstation class CPUs and Quadro's excel on rendering. And in 25 years of building CAD boxes, I have never built nor seen one that was not overlocked. Don't build a lot of video editing workstations but each one we were asked to overclock it "bawlz to the wall". Yeah if ya buying 25 workstations from Dell, they are not going to ship them to you overclocked, but klet's face it ... that customer is not reading enthusiast web forums.

 

[lexluthermiester]

There is no valid context. These tests are not even seriously work-oriented. WinRAR? Seriously? :-D

If you do not understand the test and why they're being used, that is only your problem.

I have no interest in seeing benchmarks at stock settings.

Exactly. Most reviews do testing at stock speeds.

 

[DeathtoGnomes]

you dont see Overclocking vs Stock in most reviews because it takes more time and energy and reviewers usually have a deadline.

 

[Vayra86]

Nice to see a wide range of processors here, and it gives a good ballpark impression especially of OC performance (minus the coffee lake throttling issue). Thanks for the effort. Glad to see you've avoided using processors 'as if they are others' such as the 8600k > 8700k without HT idea.

Looking forward to the Taichi results especially with a RAM kit that suits CFL; 2133 DDR4 really isn't recommended here so it will have massively impacted your current results.

 

[Artas1984]

ATTENTION!!!

THIS WHOLE BENCHMARK HAS BEEN REMADE WITH CORE I7 8700K AND CORE I5 8600K!!!

Both of them have been retested at 5 GHz overclock and 4.8 GHz overclock for AVX tasks.

PLEASE LOOK AT THE RESULTS AGAIN!!! WINRAR TESTS HAVEN BEEN REMOVED DUE TO INCONSISTENCY.

 

[Vayra86]

5.0 Ghz! Nice!

 

[notb]

Says who? You would be right about most companies, but not individuals who do work from home or employ OCed systems for doing part of their job from home.

Exactly: in a corporate environment - says your IT department.

Individuals may OC as much as they want.
But expecting a workstation owner to overclock is like expecting a van driver to tune the engine.

If you're a typical working photographer, video editor, scientist et cetera working on your private PC - chances are you're spending time doing your job, not overclocking. More importantly: you most likely don't have the knowledge needed.
Most PC enthusiasts who spend time reading CPU reviews, overclocking and benchmarking are not using their PCs for work.
PC enthusiasts may do amateur renders, they may crunch for WCG, they may do all other stuff that takes advantage of powerful PCs... but it isn't working, right?

From a practical standpoint, gains from OC are way too small to be worth the fuss and the risk of PC breaking down.

 

[DeathtoGnomes]

some of the new benchmarks are missing 5ghz sample. It doesnt surprise me the 2700X stomps the rest in the CPU-Z and Cinebench benchmarks while the ones that intel topped in were margin of error single digit percentage gains, regardless of margin still beat the 2700X.

When it comes down to pricing vs performance, they still cant beat a 2700X.

 

[Artas1984]

some of the new benchmarks are missing 5ghz sample.

Their ain't missing anything. With 5 GHz oc, 4,8 GHz was the offset mode (-2) for tasks with AVX CPU instruction sets. It was impossible for programs like Blender and Handbrake to maintain CPU at 5 GHz. Even at 4.8 GHz the workload and temps were prime95 stress like.

With previous Asrock Z370 Killer SLI that offset was 4,3 GHz, and no manual oc was possile anyway, that's why i had to change the MB.

 

[notb]

some of the new benchmarks are missing 5ghz sample. It doesnt surprise me the 2700X stomps the rest in the CPU-Z and Cinebench benchmarks while the ones that intel topped in were margin of error single digit percentage gains, regardless of margin still beat the 2700X.

Margin of error single digit %?

Ugly truth on an ugly histogram below (variable = 2700X result / 8700K result).
2 synthetic tests (CPU-Z and R15) are in orange and show very well how realistic these tests are.
1.54 outlier is the "Blender classroom main scene" - I don't know why it prefers Ryzen so much (it looks suspicious to say the least).
Other than above 3, all Ryzen winners are < 10%.

When it comes down to pricing vs performance, they still cant beat a 2700X.

Well... this is even more subjective than the raw performance results.
2700X is $35 cheaper on Amazon US - a 10% difference on the CPU itself and around 2% difference on a $1500 setup.

 

[Artas1984]

1.54 outlier is the "Blender classroom main scene" - I don't know why it prefers Ryzen so much (it looks suspicious to say the least).

You can not draw a performance summary here, because the tests were done based on different categories - points, time and fps. Also you are you being rude by stating ''it looks suspicious to say the least''. Your statement that workstation computers are not meant for overclocking is the biggest BS piece of crap i have heard in this forum in 9 years. Just leave this thread if you don't like it. There are plenty of folks here who disagree with you. So many video, photo and graphic designers, i have in mind, are exactly looking for over-clockable workstations, some are even looking at old Nehalem Xeon computers for productivity, all they care is how well such computers might overclock and how much productivity performance they can gain instead of buying predetermined HP or Dell overpriced buckets.

 

[DeathtoGnomes]

Margin of error single digit %?

Ugly truth on an ugly histogram below (variable = 2700X result / 8700K result).
2 synthetic tests (CPU-Z and R15) are in orange and show very well how realistic these tests are.
1.54 outlier is the "Blender classroom main scene" - I don't know why it prefers Ryzen so much (it looks suspicious to say the least).
Other than above 3, all Ryzen winners are < 10%.
View attachment 105654

Well... this is even more subjective than the raw performance results.
2700X is $35 cheaper on Amazon US - a 10% difference on the CPU itself and around 2% difference on a $1500 setup.

You're right that is an ugly chart.
We know why it prefers Ryzen, its better in those instances. Spinning results to match your desired outcome is no surprise here. As for the price, its still cheaper is it not?

 

[notb]

You can not draw a performance summary here, because the tests were done based on different categories - points, time and fps.

The results are real positive numbers and I can surely compare them and calculate ratios. I don't understand your problem.

Also you are you being rude by stating ''it looks suspicious to say the least''.

I didn't mean a deliberate action. Get a chill pill.

Assuming Cinebench is close to a best case scenario for Zen, how exactly is this result possible?
Maybe something went wrong? How many times did you run each test?

Your statement that workstation computers are not meant for overclocking is the biggest BS piece of crap i have heard in this forum in 9 years.

You should try other forums/communities. "Go out" once in a while.

So many video, photo and graphic designers, i have in mind, are exactly looking for over-clockable workstations

Tiny niche.
If you really want to learn something about the user groups mentioned above, forget computer enthusiast forums.
Find one focused on photo and video (e.g. dpreview.com).

But even these places are biased.
1) every online forum (especially those about electronics) will have an overrepresentation of computer geeks.
2) if you're spending time on a forum, you're not making money as a photographer - so pros aren't the most active members (simple as that).

First thing to notice on dpreview forum (https://www.dpreview.com/forums): there is a similar activity in PC and Mac discussions.
Not many Mac discussions on TPU, right?

Even when you, some are even looking at old Nehalem Xeon computers for productivity, all they care is how well such computers might overclock and how much productivity performance they can gain instead of buying predetermined HP or Dell overpriced buckets.

If you don't understand why people buy "HP or Dell overpriced buckets" than this discussion is clearly pointless. You're talking about budget-efficient computers with a lot of oomph. I'm talking about computers used for professional work. We'll never meet.

You're right that is an ugly chart.

Standard business aesthetics. We're talking about workstations after all, right?

We know why it prefers Ryzen, its better in those instances. Spinning results to match your desired outcome is no surprise here.

But why is the difference so large?
It's way above core count advantage, which means that Zen cores are suddenly way faster Intel's. It's as if i7 was running on base clocks all the time.
Have you seen other benchmarks that give 8C/16T Ryzen over 50% advantage over 6C/12T i7?

As for the price, its still cheaper is it not?

Sure. $35 less. And it also includes a cooler which is fairly OK - at least for people that don't OC (like me).