Ryzen 7 2700X vs. Core i7-8700K: 35 Game Benchmark

[Alpha gammer]

So basically the overclocked intel by around 15-20% beats the AMD by juz that...more or less equal to intel...the AMD with more cores and productivity any hardcore gamer /editor knows the difference in perf,
I got a i5 8600k it don't oc that good, but for value I might really consider AMD's offereings next cycle and save some $$$ heh premium cost's for z370 and i5k, I wish I went back too AMD!!!

 

[amghwk]

I have been using Intel CPUs ever since the Core series came out.

But for fairness sake, shouldn't these benchmarks be based upon the same CPU speed? Or am I missing something?

 

[Strawman]

You are looking at it in the wrong way. At low resolutions the Intel runs significantly faster than the AMD. As the resolution (therefore load) the Intel CPU consistently drops frame rates. This indicates the Intel CPU has an inability to maintain the percentage of lead as the resolution increase due to less headroom. In other words, the title becomes CPU bound with Intel as the resolution increase. The inclusion of CPU load with this tests like this is absolutely necessary to make a proper assessment.

You are absolutely 100% wrong. As the resolution increases the GPU is the limit, therefore both CPUs perform the same. Even a pentium G would have the same performance at 4k. It has nothing to do with more load being put on the CPUs, it's the exact opposite.

 

[Zoggo2]

What I noticed not just here but in a couple of CPU benchmarks is that AMD tends to catch up on Intel the more you increase resolution/quality. It's almost as if it's not working at it's max if not pushed to the limits.

If you read carefully you would know at higher resolutions GPU becomes bottleneck.

 

[Capaill]

That's my point. 72c is an ideal ambient temperature

LOL, I hope you meant 72F. Unless you're a dragon. In which case I'll quietly slip away now.

 

[Capaill]

OK so we are testing high end systems ..... Hands up who is going to run a 720p or even 1080p monitor on their new high end build.
Reality is you are looking at 1440p , ultrawides and maybe 4k if you are spending that much on computer hardware .
Fabulous the intel can make more fps at low resolution but is it even useful? Its like having a 5 legged dog and arguing its 25% better .

You would probably be running 1080p if you have a 240Hz monitor for FPS gaming. That refresh rate is very hard to do on higher resolutions.
That's the joy of computers, we all have our own unique needs - high res gaming, lower res and high refresh rate gaming, VR, streaming, video encoding, budget, fanboyism or a dislike of monopolies, product compatibility, future upgrade plans or various mixtures of all. Reviews like this (and the ensuing comments) might help us decide which way to go. Neither choice is definitively right because our input conditions vary so widely. The bottom line is, if you only want to game right now on a PC, based on games currently available right now, and no other factors matter to you, then you should probably go for the Intel chip. That's the recommendation provided. You decide if it's right for you.

 

[GreenNova343]

I have been using Intel CPUs ever since the Core series came out.

But for fairness sake, shouldn't these benchmarks be based upon the same CPU speed? Or am I missing something?

Not quite. These benchmarks are somewhat useful, in the sense that we see that a) the Intel CPU can be OC'd to a higher frequency, but also b) despite being OC'd ~20% higher, the i7-8700K only averages ~12-13% more performance. If I understand what IPC is supposed to mean, that would imply that when OC'd the i7-8700K has less IPC than the R7 2700X.

But I'll agree that the tests are incomplete. What we really need are tests at stock speed...because both the i7-8700K & R7 2700X's stock speeds (I.e. no Turbo Boost/Turbo/XFR) are identical (3.7GHz per core). I'm sure Techspot's testers can somehow adjust the settings so that both CPUs would be locked in at those speeds (giving us a bottom-level baseline), so that we could get a baseline comparison at equal frequencies.

 

[zazapensive]

Glad as heck your testing is thorough and reliable.
I'd never get the spare time to do this.
thank you.

 

[Strawman]

but also b) despite being OC'd ~20% higher, the i7-8700K only averages ~12-13% more performance. If I understand what IPC is supposed to mean, that would imply that when OC'd the i7-8700K has less IPC than the R7 2700X.
s.

Completely and utterly wrong. The 2700x has 2 extra cores / 4 extra threads, which do close the performance difference in certain games (AC origins fro example). Also, teh 12-13% difference may be the result of a gpu bottleneck. IPC means instructions per clock, the actual frequency of the clock makes no difference.

 

[GreenNova343]

but also b) despite being OC'd ~20% higher, the i7-8700K only averages ~12-13% more performance. If I understand what IPC is supposed to mean, that would imply that when OC'd the i7-8700K has less IPC than the R7 2700X.
s.

Completely and utterly wrong. The 2700x has 2 extra cores / 4 extra threads, which do close the performance difference in certain games (AC origins fro example). Also, teh 12-13% difference may be the result of a gpu bottleneck. IPC means instructions per clock, the actual frequency of the clock makes no difference.

A "GPU bottleneck", on a GTX 1080TI at 720p? Really? You're going to claim that?

And for probably 90% of the games they used in the benchmarks, once you get past 4 cores/threads (6-8 cores/threads tops), the number of cores/threads had little to no effect on performance -- as Techspot found in their initial testing (https://www.techspot.com/review/1497-intel-core-i7-8700k/page3.html), which showed little/no difference between the 4C/8T i7-7700K, 6C/12T i7-8700K, & 4C/4T i5-7600K. Instead of a 50% improvement due to the 50% increase in cores/threads (or even perhaps a more reasonable 20-25% increase),

 

[Evernessince]

LOL, I hope you meant 72F. Unless you're a dragon. In which case I'll quietly slip away now.

Yep, typo.

 

[MaikuTech]

Seems like those benchmark shows Intel is faster due to higher clock speeds. Imagine if they can make Ryzen Turbo boost to 5.0-gigahertz. Probably would have perform better.

Heh I like to see a quick retest with that 8700k at 4.2ghz against that Raisin 2700x 4.2ghz.
I have a feeling those results won't be so powerfull and it will show amd beating that core i7.
Let me stop before one of the editors/moderators blows a gasket and tosses me outside. >_>

 

[Evernessince]

Heh I like to see a quick retest with that 8700k at 4.2ghz against that Raisin 2700x 4.2ghz.
I have a feeling those results won't be so powerfull and it will show amd beating that core i7.
Let me stop before one of the editors/moderators blows a gasket and tosses me outside. >_>

They did a video like that already

At the same frequency the 8700K wins by an average of 3% (guesstimate) in games and loosing by a good chunk in anything multi-threaded. That's core for core as well so 6 core vs 6 core. AMD's multi-threading SMT is a better implementation than Intel HT.

If anything Intel should really be worrying about it's datacenter customers as this along with the superior core numbers and competitive IPC gives then a big advantage in that arena.

My guess is Intel's true next gen CPU will be one created by Jim Keller in about 3 years. That's the reason he was brought on I think.

 

[Strawman]

A "GPU bottleneck", on a GTX 1080TI at 720p? Really? You're going to claim that?

Yeap. Plenty of games that a 1080ti is the bottleneck at 720p. AC origins for example is one of them. BF1 has a 200 fps cap. Battlefront 2 has a huge GPU bottleneck. Shall I go on ?

And for probably 90% of the games they used in the benchmarks, once you get past 4 cores/threads (6-8 cores/threads tops), the number of cores/threads had little to no effect on performance -- as Techspot found in their initial testing

I never said otherwise. Actually, that's exactly what I said. I said the 2 extra cores benefit only on certain games, like ACo and Ashes of the singularity. And bf1. And watchdogs 2.

Also, you are linking me to benchmarks that you can't even understand what they mean. In your link there is no difference between 8700k and 7700k because there is a HUGE gpu bottleneck...

 

[Strawman]

Heh I like to see a quick retest with that 8700k at 4.2ghz against that Raisin 2700x 4.2ghz.
I have a feeling those results won't be so powerfull and it will show amd beating that core i7.
Let me stop before one of the editors/moderators blows a gasket and tosses me outside. >_>

And besides academic inquiry...what's the point?

 

[Chase Fenton]

FYI: Shopping shortcut mistake. Says links for i5-8700K and Ryzen 5 2700X but its actually i7-8700K and 7 2700X. Anyone else notice?

 

[Abraka]

There's something fishy with your calculations. In the Overall comparison @720p your first table shows that Dirt4 is only 1% faster on AMD, but the very next table shows 17 fps difference in favor of AMD. How can 17 fps be equal to 1% ??????

It can only happen if that game had 1700 fps average frame rate. Did it? Nope. It had around 170 fps, so this is actually 10% difference, not 1%.

If you're that "good" in basic arithmetic then this entire article is BS.

 

[Kitsone]

Restricting the 2700X to 4.2GHz per core is lowering game performance. That OC will only help fully threaded apps. To get better game performance you have to OC using BCLK and/or Precision Boost Overdrive (PBO). That type of OC will get overall better performance in mixed thread workloads.
TLD: You're doing it wrong with the 2700X, OC only the BCLK and/or PBO- not all-core.

 

[Evernessince]

Restricting the 2700X to 4.2GHz per core is lowering game performance. That OC will only help fully threaded apps. To get better game performance you have to OC using BCLK and/or Precision Boost Overdrive (PBO). That type of OC will get overall better performance in mixed thread workloads.
TLD: You're doing it wrong with the 2700X, OC only the BCLK and/or PBO- not all-core.

I actually wouldn't mind an article on BLCK overclocking with the 2700X. It seems very popular in the community and hasn't been something we've seen much of.

Apparently you can get up to 4.5 GHz by doing that.

 

[Sausagemeat]

lol and you sound like an intel fan boy.

The purpose of showing the 720 numbers was just to remove the gpu bottleneck for this review. In real life no one builds a system in 2018 to game at 720p. And the reason those amd guys are talking about future proof is they are talking about the socket not the CPU and they are 100% correct. Intel has shown a history of a new socket at every new generation.

AM4 is not really “future-proof” if buying now, that socket is only supported for 2 more years. If you need to upgrade your CPU within 2 years then you’ve bought the wrong CPU in the first place!

 

[HardReset]

AM4 is not really “future-proof” if buying now, that socket is only supported for 2 more years. If you need to upgrade your CPU within 2 years then you’ve bought the wrong CPU in the first place!

I want CPU that has IPC around same that current CPU's have and turbo clock speed at least 5.5 GHz on stock. What CPU should I buy? There is none available right now. What should I do? Wait until CPU like that is released. Very likely AM4 platform will have that kind of CPU in two years but LGA1151 platform won't. So...

 

[deathtrap]

As many games show today,
a stunningly big chunk of the CPU-Limit is solveable by software (engine) modernization.
Partly because of too IPC-demanding and/or bad vectorized code,
and sometimes softwareparts not optimized for or even not recognizing the Zen-µArch.

I hope all gamedevelopment wich started after the Zen-Launch is using newest "visual-studio" and such stuff.

I whould buy the 2700X for sure.

 

[Lionvibez]

AM4 is not really “future-proof” if buying now, that socket is only supported for 2 more years. If you need to upgrade your CPU within 2 years then you’ve bought the wrong CPU in the first place!

There is no such thing as future proof when its comes to computers. However my point still stand Intel changes Sockets far more than AMD.

 

[amghwk]

Not quite. These benchmarks are somewhat useful, in the sense that we see that a) the Intel CPU can be OC'd to a higher frequency, but also b) despite being OC'd ~20% higher, the i7-8700K only averages ~12-13% more performance. If I understand what IPC is supposed to mean, that would imply that when OC'd the i7-8700K has less IPC than the R7 2700X.

But I'll agree that the tests are incomplete. What we really need are tests at stock speed...because both the i7-8700K & R7 2700X's stock speeds (I.e. no Turbo Boost/Turbo/XFR) are identical (3.7GHz per core). I'm sure Techspot's testers can somehow adjust the settings so that both CPUs would be locked in at those speeds (giving us a bottom-level baseline), so that we could get a baseline comparison at equal frequencies.

Yes, I too would have preferred comparing at the stock speeds, when both were advertised at same speed (the number of threads notwithstanding) .

In fact, I would also like to see how these CPUs perform on non-overclockable boards like the H370 chipset for the 8700K for raw performance head to head.

 

[jWilzz]

Incorrect on both counts; from a professional review site as why 720p is important to testing, please read and educate yourself as not to be confused again.

"Of course, nobody buys a PC with a GTX 1080 to game at 720p, but the results are of academic value because a CPU that can't do 144 frames per second at 720p will never reach that mark at higher resolutions either. So these numbers could interest high refresh-rate gaming PC builders with fast 120 Hz and 144 Hz monitors."

I don't need to educate myself. The majority of the market is not gaming on 120hz or 144hz monitors.

Which will require you to have 1080ti class graphics cards to two card in SLI to achieve those numbers.

The amount of people doing that is tiny.

I'm aware of why they are posting the numbers but you are deluding yourself in the relevance of them.

Most people are still gaming on 1080p/1440p 60hz monitors.

That graph you posted is for 4.8 I said 5Ghz which most people have to delid for.

Cryorig H5 ultimate, scythe Mugen, and Fuma all 3 of these cost more than 50CAD I'm in Canada.

all of the Noctua's are over 80CAD.

Wait what?! Require 1080ti or SLI just to run 120 or 144hz?

I have a 1060 3gb with an i5-6600k and run on 144hz no problem. Gta, witcher 3, wow, all perform fantastic.