4GHz CPU Battle: AMD 2nd-Gen Ryzen vs. Intel 8th-Gen Core

For people talking about gaming results, IT still doesn't matter the GPU is still the bottleneck, there is zero reason of testing CPUs at 1080p gaming and comparing them, the only absolute advent of this reversing is #if next gen GPUs are so ridiculously strong they blast and manage to oversaturate the 3.0 bus, which with Compute is plausible pertaining to Ray-tracing, even then keep in mind 4 - Titan V's on HEDT did the Ray-tracing on the real time render, so the question is, On a mainstream platform At what point will the CPU be weaker(Bottleneck wise) than the GPU?, and affordable for normal mortals?

The 1180 will apparently be a 13Tflop card, the 1080Ti has about 10.6Tflops. Without SLI being a real thing I seriously doubt in the next 5 years we will be at a point where we will see a CPU bottleneck, let alone not even accounting for further CPU releases increasing their strength. The most gains I think we will se on the GPU side will be NVidia's optimizations for DX12 with the new architecture that will give a significant bump in titles making the 5-15 frame difference on the CPU side even more negligible.

The only issue AMD has is Intel's clock advantage on the node process, a 1.0Ghz increase, AMD has a far superior interconnect between Cores and Nodes with the InfinityFabric Mesh. As all things dictate we will have to wait and see how this turns out and changes with the Industry.
 
It's great that AMD is putting up a fight. These new second-gen Ryzen CPUs are in the ballpark, and that is good for consumers.

When core counts go up and developers start to take advantage, AMD might even have an edge. But that'll be a different generation of CPUs. We'll have to see if Intel will address their interconnect issues in future processors.
 
Intel have the more rounded architecture and capability to scale to much higher clocks. The 8700K at least being able to do 20 percent faster clock speeds when pushed hard in overclocking, of course that doesn't translate to a similar win everywhere but it does give it the edge.

However this is talking at the bleeding edge and top end of the consumer market only. AMD have a design that is extremely close overall in IPC to Intel's latest and greatest. For the majority of people going out the buy anything below the highest end parts they do have a viable alternative.
 
For people talking about gaming results, IT still doesn't matter the GPU is still the bottleneck, there is zero reason of testing CPUs at 1080p gaming and comparing them, the only absolute advent of this reversing is #if next gen GPUs are so ridiculously strong they blast and manage to oversaturate the 3.0 bus, which with Compute is plausible pertaining to Ray-tracing, even then keep in mind 4 - Titan V's on HEDT did the Ray-tracing on the real time render, so the question is, On a mainstream platform At what point will the CPU be weaker(Bottleneck wise) than the GPU?, and affordable for normal mortals?

The 1180 will apparently be a 13Tflop card, the 1080Ti has about 10.6Tflops. Without SLI being a real thing I seriously doubt in the next 5 years we will be at a point where we will see a CPU bottleneck, let alone not even accounting for further CPU releases increasing their strength. The most gains I think we will se on the GPU side will be NVidia's optimizations for DX12 with the new architecture that will give a significant bump in titles making the 5-15 frame difference on the CPU side even more negligible.

The only issue AMD has is Intel's clock advantage on the node process, a 1.0Ghz increase, AMD has a far superior interconnect between Cores and Nodes with the InfinityFabric Mesh. As all things dictate we will have to wait and see how this turns out and changes with the Industry.

Finally! Someone who gets the secret sauce of Infinity Fabric :cool: The clock advantage will be further narrowed in the next gen, where in time they feel they can reach 5Ghz and much less power consumption, greater densities.

Intel are probably 2 years behind. So when I speak of the true window, its until 2020, lets say. AMD have had a big burst of speed. The graphics thing is completely true - and going forward there there's interconnects being worked out for them too.

This is what I mean about a cost effective way of making these things. Intel are behind on the interconnect, they pulled the existing one from upper-part Kaby Lake, but the truth is that anything coffee lake is designed for older, higher clocked monolithic type parts. You can't make an 8700k 8 core, for instance. Nothing about it 'works'.

The thing about interconnects like this, is that you can't clock as high. So the more cores, the lower the clock ATM. AMD aren't just in front with the cool Infinity Fabric name, there's limits to what can be done with them; and thats being worked out.

Things won't change until you can buy an intel version at the shop.

AMD are not after world domination, they are only looking to capture around 25-30% of the market at least with the Ryzen's, the more the better. With the IPC race almost over, not just because of reaching the limits, but because of the future work loads that hinge upon and build off the existing software that utilizes IPC (and hence moves to supporting multi-threads), the focus shifts to interconnects and multi cores and bringing those to market and of course on cost/efficiency. Naturally Intel with an in-house fab will have some kind of advantage there - and intel do a lot of different things for different segments. They make billions more than AMD.

The trouble for Intel's fab atm is their competitors have caught up. Do people think its cheap to run a fab? Sure, Intel gets all the profit, but they also get all the costs. Thats why its 'dangerous' as some articles said of AMD's success with Ryzen, because you take out any of Intel's profits and those margins get chipped away. Not that they can't handle that as a business, its just not ideal and they will take a hit.

The whole point of discussing IPC in that light, which the article helps with, is in terms of wanting to buy one. I am merely stating that IPC is great, I lover IPC, but going forward IPC is a lesser consideration than today for a large swathe of reasons.

Did the 3-12% IPC deficit depending on work load (but 5-7% with good memory [and there's other improvements coming]) stop AMD from selling 40% of the entire volume of chips at certain outlets? NO. It did not.

I am merely saying about that, that going forward IPC is so close, and will practically mean nothing soon, that its not a consideration. I loved this article though.

Whats more important going forward is latency and efficiency. The IPC thing atm is close and the clocks are coming.

A 5-15 frame difference in a game is meaningless going forward. The article is spot on for today, and I fully appreciate it, but there's no forseeable situation where things get worse for a Ryzen CPU in those situations, and said situations far exceed whats required anyway.

People buying a Ryzen today, and I think the article said so, but you can simply expect the IPC deficit to close by the time bios's are updates, AGESA is updated, games, applications, windows, etc.

But the point is, all that aside you're not making a mistake buying a Ryzen right now. Unless of course you own a recent enough CPU. The same is true for an 8700k.


Intel have the more rounded architecture and capability to scale to much higher clocks. The 8700K at least being able to do 20 percent faster clock speeds when pushed hard in overclocking, of course that doesn't translate to a similar win everywhere but it does give it the edge.

However this is talking at the bleeding edge and top end of the consumer market only. AMD have a design that is extremely close overall in IPC to Intel's latest and greatest. For the majority of people going out the buy anything below the highest end parts they do have a viable alternative.

Good point! Glad someone said. Fact: the 8700k is a fairly low selling chip by volume. And so is the 1080 TI. Maybe 1 or 2% of people have both. The amount who have either, maybe 5%. Even the 1080 is fairly rare worldwide. In some country's given higher free money to spend its higher/concentrated, but world wide its still extremely low.

More people are likely to be buying lesser intel CPU's and nvidia cards to play games on. Or they're buying the Ryzen. Because IPC is just not a concern. And it will be less so.

My only concern with the article, which does not remove its enjoyment of aptness from me, is that the 8700k is a dead arch, and its not representative of what most, if anyone reading/watching videos about them will experience.

Like the first quoted fellow said, there is no situation where the Ryzen sucks, and its cheaper. IF intel could compete on price, monolithic core and all, don't you think they would have done that. I mean, you even get a free cooler with the Ryzen. The better case is for the I 3's in the medium term for gaming as they're 4 core now, and see where this window of AMD advantage I'm talking about ends or closes, and what the situation is then.

But I think you will find that interconnects and efficiency on core scaling will have run away with the industry by then. And only things like compiling will need high IPC, etc, which will be close from both companies, perhaps swapping spots once in a while
 
Last edited:
Sigh... we keep seeing these again and again, year after year... that AMD still trying to catch up with Intel, but could never really trounce it once and for all and claim clear superiority.

But, if AMD can slash at least 30% off it's current price of the CPUs, I think many will have second thoughts buying an Intel.

So True. Speak power to truth good man!

But you got to be care on these forums here. The AMD fanboys will come hating on you, motivated by their blind hatred for Intel.
 
So maybe stop predicting the future, the same thing was being told about Nvidia in the past, how small AMD chips are the way to go and Nvidia is set to be unsuccessful with its big monolithic ones. How is the situation on the market?

After all the years they barelly matched what we have had here since Sandy Bridge. I will tell you a secret but Intel will introduce completely new architecture too on 10nm.

Let's wait how it turns out because I've read the same predictios in the past, Phenom, then Phenom II, then Phenom II X6, then Bulldozer. Some people never change.

Not a secret any more, now that you told them! I agree, but might take a while before they abandon the "Core" design. 10th gen Ice Lake is called 10th gen because Nehalem was first gen, and the basic architecture is still the same, with little improvements in every new gen. Ice Lake will be the first with embedded hardware mitigations for Spectre and Meltdown, but will have the same basic microarchitecture. So maybe in 2020?
 
Great article, IMO.

I have to laugh, though, at statements like "a whopping 11% slower". I thought only marketing departments used such terminology.

Realistically, is anyone going to notice this while gaming? It would seem that such differences are only going to reveal themselves in benchmarks, but an 11% difference when your frame rates are over 100 in the first place seems like an insignificant amount especially when it takes 300 to 400ms to blink and, in theory, anyway, these frame rates are well beyond what is considered minimum frame rates for smooth game play.

I do get all the talk about overclocking and the fact that Intel has the lead at this point, and I also see that there is greater power consumption on the part of AMD - which the author notes that this is, in part, because of the overclock to run the AMD parts at 4G.

I am not upgrading at this node as I am on an E5-1650v2 which still meets my needs, but if I did upgrade, I would go with an AMD CPU because gaming is not the most important thing for me.

But despite being trounced by a whopping 11% in gaming, the most recent earnings report for AMD is excellent - https://finance.yahoo.com/news/amd-surging-stellar-earnings-report-202300559.html

And this is where it counts. AMD excels in MT and, at least as I see it, this is where they needed to be because it really counts in the lucrative enterprise market - which just might finance future improvements in gaming.

Gaming improvements will come in time; my hope is that AMD does not get IntelPricingItis when those improvements arrive.
 
What do we have here 32/64-bit CPU again.. When it's 64/128-bit let me know. Or better yet 128/256-bit. Still x86/x64 baseline again. GHZ speed seems quick enough anything over 3.7GHz is good eough, my I say anymore. The review is interesting don't get me wrong, but the final point is Intel is the leader of the CPU wars it will always be AMD made up theirs based on Intel model. Might added some beefy L1 or L2 or higher caches and some cool buzz words like APU, still functions and cores might stand out test and marks excites up Tech Guru..I say whatever brand you want get it. Lets all say CPU.. Relax..
 
RYZEN is good considering the price and multithread ratio, AMD said 3% more IPC and we have that, Intel always says IPC while combining IPC and clock improvements, maybe they up 1% IPC and 10% clock but says "we up 11% IPC" lies, the surprise is that AMD is offering better SMT performance per clock, lower Power draw per Core at same clocks (even @14nm) and better price point, AMD is a little giant, David and Goliath Story

Good review, greetings
 
This turned out more of an benchmark of the cache than the IPC itself. The results vary a lot more than I expected. Nonetheless this should put the "speculations" to rest and give us proper results to work with.

No and no, if he test Geekbench is a cache test but this is a IPC test, AMD improved cache a lot, from 17 to 11 cicles (lower is better) on lv2 cache, and other Caches was improved on RYZEN+ but RYZEN's ARK is full optimized, they just could get 3% more IPC with all changes, they need redesign again, 7nm will provide a redesign

Greetings
 
I dont know why anyone would say Intel is still king of gaming CPU's. The language CPU parlance these days is over the top. Do people really pine over CPU's like that?

The eye is tricked at even 24 frames a second. You'll be hard pressed with the advancement in GPU's to appreciate the difference with a ryzen or intel cpu going forward - heck, even today. Techlevel1 did a side by side blind comparison.

The tech reporting industry is out of touch with the majority of consumers when every test about new CPU's focus' (but not exclusively albeit) on the 1080TI and 8700k of which only a couple of % of people both own. Even then, those results only apply to certain titles, at certain lower resolutions, and only for certain detail levels.

When company's provide samples to reviewers, the nature of them is in question. Thankfully this article was balanced, but fell ill to certain biases, like "a whopping 11% slower"...

Sure... on the benchmark. In life, zero.

Fortunately the market has not bought into such dire max 3-7% with better RAM "poor IPC" suggestions from the media, and gone out and bought a lot of Ryzens to boost the stock price. I do not think a single one of them will be lamenting the "whopping" IPC deficit.

Because its bollocks.


This reputation for AMD has been way overplayed and has not been true since the bulldozer days, or 2006 or so.

AMD is just another corporation with an active markerting department trying to get people to give them there money.

However if as it has been written:


Then AMD will come close to earning that back that goodwill.

This, too, is bollocks. AMD is already far, far ahead in terms of price/performance/value for the vast majority of people. Goodwill? You just got through telling us that they're a corporation like anyone else.

Holding them to a double standard?

No, what we need is for intel to make a non-monolithic CPU design, cut out bits and pieces and glue them together and add a fan.

Its called value. They have really pushed this IPC nonsense far and wide. And we're talking a gap of about 5% on average, which is bugger all, and by the time you hit 4-6 cores its turned into a 15% advantage.

What we as consumers also need is the tech-press to stop buddying up to the company's to secure parts so they can keep the ad money/patreon (YT) rolling in, and tell it true more often instead of just slipping in caveats now and then and drowning it out... its just awareness.

I want Intel to do well, too. I actually own an 8700k, a Haswell, and had an FX8350. They all made it into my parents hands at some point.
 
Last edited:
Sigh... we keep seeing these again and again, year after year... that AMD still trying to catch up with Intel, but could never really trounce it once and for all and claim clear superiority.

But, if AMD can slash at least 30% off it's current price of the CPUs, I think many will have second thoughts buying an Intel.

So True. Speak power to truth good man!

But you got to be care on these forums here. The AMD fanboys will come hating on you, motivated by their blind hatred for Intel.
Considering that the 1800 went for $500 at launch, and the 2700 is priced at $329 at launch, that is a drop of 34.2% off the top CPU price.

Even so, these CPUs will sell.
 
I'm just not sure IPC matters anymore in the way the article suggests. I fully appreciate the article but it just doesn't matter much now. Someone gaming (where the biggest 'loss' is in IPC/value, which is not much, to be sure) should be buying a core i3 or i5. And if they want the extra/equivalent performance of ryzen as compared to the 8700k in multi threads, sure, get the Ryzen for tor the free cooler with the option to update without a board to Zen 2; or make a line for intel and the board and reach the end but know sure, you still have a good CPU. However, the gpu matters more for games.

It should be literally the first paragraph and reiterated multiple times in multiple ways throughout a review. Especially those Youtubers.

[QUOTE="
From another source "I mean the general rumour was Intel deciding to ditch Core and start an all new architecture early last year. They knew Zen was within spitting distance of Kabylake on a significantly worse process and they also knew that come 2019 AMD would have the first proper iteration of Zen on an on par process with their 2019 chips. In other words AMD had a new architecture with likely a lot of low hanging fruit to focus on while Intel is struggling to get 5% gains in IPC on a older architecture. They knew that AMD was going to at best be competitive and at worst be ahead(best/worst in terms of from Intel's pov) and so knew it was time to start work on a new ground up architecture.

It's a year later, this suggests to me that they've gone in circles for a year and not found the lead people they needed. They've lost people over the years and gotten used to just iterating what they have and buying new tech rather than innovating themselves. I would say this means Keller to direct a ground up architecture ready to compete with AMD who also have a new ground up architecture coming 3-4 years from now for both companies.

Remember that when Zen was properly announced it was said AMD are both have a plan in place for 3 major generations of zen with 2 major updates, some minor process/stepping tweaks as with Zen+ and they are working on a ground up architecture to replace it already for after that.

AND


I think you'll find that Intel will be struggling from 2019 till they get a new architecture out.

AMD has two major iterations of Zen planned but the biggest update is they are going from Zen 1 vs Kabylake on a significantly worse process to Zen 2 vs Intel 10nm with a similar and maybe even slightly better process in 2019. AMD is competitive both in performance and power despite a 18-19nm process vs a very mature Intel 14nm. Glofo's 7nm is a real, true competitive 10nm. AMD will gain more power saving, more die space reduction and much more clock speed at 10nm than Intel will.

Intel can't afford to shelve anything being made. They brought him in to make something asap precisely because they know AMD are going to be competitive if not faster than Intel across the next 2-3 years. AMD is already working on the next full architecture after Zen concurrently. They already have something new planned for 2021-22.

Core architecture has been through so many iterations there is only so much they can improve IPC per iteration with the same base chip. AMD is likely to have larger IPC jumps for the main Zen 2 Zen 3 chips.

If AMD pull out something better than updated Zen with their next architecture Core is going to be way behind. Intel brought him in to make a top notch chip and make something competitive that their entire line up will be replaced with as soon as possible, but I'd put that at 2021-22 also.
[/QUOTE]

Lines up with what I am saying. They were already on a worse node but with a solid arch, designed by a guy other than Keller, and within spitting distance of the iCore's. This year the move up to a node thats as good or can be better with some time. Thats the end of 2018 and they're bringing out something thats modular with infinity fabric BUT higher clocks and tweaks, with plans to get more low hanging fruit, followed by another redesign shortly after....

I think you will find Intel will struggle beyond 2019. I said "struggle" not fail miserably. I actually pegged it previously here for between 2-4 years from now. IPC is not something people will dither over. Not only is it good for both sides right now without fear of missing out, even on lower end chips (especially for gamers), going forward it simply won't matter - and doesn't to the majority of people anyway - and because there will either be nothing separating them, or because Intel is at the end of the arch-progress, AMD could even be a bit ahead.

And it won't matter because the CPU's, Ryzen in particular will see a larger jump in over the next few generations; remembering there is an entirely new arch coming after Zen 3.

Yes, Intel is in trouble on this, and the secret sauce is the Infinity Fabric. AMD will be ramping up clocks with Zen 2. I mean Intel just have not bothered with a new arch. They've even lost customers like Apple because of this and in the server space its dicey. Well, not exclusively there's other factors, but they are down a peg right now. And the way software is being written, except for things like compiling etc, means the entire IPC argument is out the window for the majority of users.

Articles like this interest me greatly, however it may be time to start for those people who game, by stating what the end result of a CPU pushing a GPU is - meaning, don't worry too much about the CPU in gaming, make sure you have a really good GPU.
 
Last edited:
Sigh... we keep seeing these again and again, year after year... that AMD still trying to catch up with Intel, but could never really trounce it once and for all and claim clear superiority.

But, if AMD can slash at least 30% off it's current price of the CPUs, I think many will have second thoughts buying an Intel.

So True. Speak power to truth good man!

But you got to be care on these forums here. The AMD fanboys will come hating on you, motivated by their blind hatred for Intel.
Considering that the 1800 went for $500 at launch, and the 2700 is priced at $329 at launch, that is a drop of 34.2% off the top CPU price.

Even so, these CPUs will sell.

They sold really well yes. And if we think these ones sold well then wait until Zen 2. Today the share price jumped, and if they can leverage the next few cycles, by the time we get to the next redesign after Zen 3, its boom time. Some server things will need to go right, and GPU's, but hey.

With most of the Zen 2 team having moved onto the GPU's now, seeking, ironing out, and testing, this can really only mean that AMD is confident the Zen 2 can beat intel hands down.

Improved latencies, improved clock speed, improved internals, improved IPC, low hanging fruit sorted, don't be surprised by the refresh if they're around the high 4 GhZ's.

Zen 2 is set to be a cracker of a chip, and Intel won't be able to answer it directly for a couple of years at the earliest. Then AMD have a new socket after Zen3 and new redesign, possibly going up against Intels redesign.

mind you, Intel wants to hit 5.3 with their next bunch of chips at some point. The pressure is on them. Or people will complain apparently (10nm compares with the 7nm of amd). But do expect high 4's with Zen 2 and 3. No one knows if intel can do that; but its expected the zen 2 will

A high 4's Zen 2 is looking like a monster plus innately faster; and its said they may get over 5 on this new node, maybe, at some point. (for intel and the more mature things they do, I don't mean 5.3 golden sample, I mean out the box, ready to turbo boost - guaranteed; they hope for this to be a certainty, owing to cost to produce and expectation/old arch, etc; more legs for their production side; but by now it simply costs so much to implement these on opportunity-cost measures, so it must pay off; or they enter sinking costs.)
 
Last edited:
Have you tried optimizing the ram subtimings? They are usually awfully bad on Ryzen and that's why they are behind in gaming. I've seen a 2700x tear me a new one in AC origins with optimized subtimings on his 3466 c 14 ram.
 
Love these comparison's, nice work Steve. Think you could throw in some more architectures and do 3 or 4ghz for those? Would be interesting to see what intel and amd did from like 2004-2006 until today at say 3ghz.
 
Yeah, Intel is doomed.

While they have already 10nm chips, AMD is selling Ryzen+ as 12nm and in fact, it is just on 14nm+ and they don't even compare those two processes but to 16nm finfet of TSMC...

Intel 10nm is in development hell as it is being delayed again. Ryzen 2 will probably arrive before Icelake at this point.
 
The IPC looks worse on AMD for gaming only when compared to CFL. When compared to SKL-X, they are neck and neck.

AMD still holds the value edge. Being able to get a 12 thread cpu for $200 that can overclock to 4.3 ghz all core on a $100 board is very respectable.

Intel really has nothing that gets this kind of all around performance for the same price.
 
Intel's superiority comes from the fact that you can overclock most of their chips like a mofo. With good cooling one can easily push 4.7 to 4.8 GHz on an 8700K and thus positively smoke the Ryzen 2600X which is stuck below 4.3 GHz. AMD desperately needs Ryzen to clock higher, much higher than it can right now if they have any hope of catching up to Intel because right now Intel is the performance king in gaming because of its ability to clock high.
No you can't OC most of their CPUs only K and X.

This test was a little skewed by the 2600X only clocked 4.15GHz @ 1.42V so many reviews achieved 4.2 1.40V (on 470 MB) or less. Plus you can increase the power limit and XFR2 can clock cores to 4.4+ no all core OC needed. The major problem concerning gaming is using a <5% GPU instead of one that matches the CPU choice dollar wise. The engineering types want to make sure the tests are not GPU bound so they use a high priced GPU which obscures the fact that if you use a main stream card there will be little or no difference in gaming performance. Using a GPU like 1080Ti makes those scatter charts purporting to show value by FPS vs cost misleading at best because the frame rates are artificially high because a $1100-1500 card was used.
Don't get me started on the nonsense average percent of percent faster or slower used by Hardware Unboxed erroneously instead of average FPS or range of FPS faster or slower. Especially when its only 5-10 FPS for most games. That gets worse when 4K @ Ultra preset is used to compare GPUs where the 1080ti barely manages 46 FPS most get 32-36 FPS AVG when percent is used it cannot be compared with 1080p@ Ultra results of 146 FPS and most get 132-136 FPS AVG the real difference is 10-14 FPS in each case.
 
Last edited:
It is nice to see that AMD did improve IPC with this new revision. At the same frequency, with the same number of cores (read 2600X and 8700K) on productivity apps they are on par which is VERY good, given where they were with Excavator. In games, yes, there is a difference, but it is negligible in my opinion. The only single thing to do for AMD is to either try to up the IPC more or increase frequency like 8700K. In any case, I don't see any reason not to buy AMD now...
 
...
AMD still holds the value edge. Being able to get a 12 thread cpu for $200 that can overclock to 4.3 ghz all core on a $100 board is very respectable.

Intel really has nothing that gets this kind of all around performance for the same price.

Yet let the evidence show that such a generic claim is clearly easily refuted see:

http://www.microcenter.com/product/486089/Core_i5-8600K_Coffee_Lake_36_GHz_LGA_1151_Boxed_Processor
http://www.microcenter.com/product/505112/Z370M_DS3H_LGA_1151_mATX_Intel_Motherboard

$220 for the CPU + $105 for mobo - $30 for the combo discount. Easily gets you a CPU for less than $200 and mobo for less than $100, and is more than par for 2600x. I'm sure the owners of the 8600K can tell you that getting to 4.3Ghz actually means underclocking.
 
THANK YOU for this article!

While AMD is still in the back seat compared to Intel, they are doing very well considering the market share (money) they get compared to Intel. I am an overclocker. If I get more value for the money and performance/overclock than Intel, then I may consider AMD. I haven't bought AMD since the old Athlon days. I would like to support them, but I am looking at it from a customer point of view. Now being the worst time in history to build a new rig, it will be quite some time before I even look at the market.
 
Back