I7-7700 vs I7-6700 --> Skylake vs Kaby Lake

How much additional performance and should I get for the upgrade and how much is the likely cost difference?

I hear the I7-7700 is much cooler running? It's hard to find a price comparison and a few places show Kaby Lake as cheaper!

 

The difference is exactly 0% between the 2 architectures - it's all just clockspeed.

 

'Exactly 0% difference' is not exactly true.

See:
http://www.cpubenchmark.net/compare.php?cmp[]=2906&cmp[]=2586

With a ~12.5% boost to single core (responsiveness) performance and a ~10% boost to multithreaded performance - this isn't potential performance that should be left on the table - especially if the better/newer option is the more powerful one.

Not to mention the further benefits Kaby Lake should offer with the right M/B and future Optane SSD's.

For that kind of 'raw' performance increase indicated above, I would be willing to spend up to an additional $250 or so per platform. In reality though; there should be much less difference (as you've seen with the Kaby Lake option actually being cheaper, Morpheus1).

When a new platform is available and you're buying a new system today? The old platform isn't just old: it is ancient and crumbling already...

 

thanks

 

When you buy the new platform aka Kaby today. It's already ancient within short time. Mainstream 6 core i7 coming faster than you think Why should I run out buying 4 core i7 platform today? Not me!!

 

Because in everyday use, without special tasks less (but faster) cores better than slower more cores.

 

Actually - the opposite is true - with a lot of smaller tasks running parallel to your main demanding application, more cores/threads adds a LOT to the overall experience.

 

The key is/was if you need to buy 'today'. (Whenever 'today' is...).

Buying the old platform is a foolish decision when the new (and available) platform will offer so much more over the course of expected ownership.

Waiting for the 'next...' is not within the scope of what I responded to; what I responded to was a need to buy today scenario - yeah; an assumption on my part based on the OP...

 

No way, think just for Iphones; two brutally fast cores, gives much snappier experience than any android devices with 4 and 8 cores...

​

 

Up to 4 cores/threads with all else being equal? I'd agree.

But with greater than 4 cores/threads usually running at (much) slower clock rates for all cores actually doing work?; it is a slower system overall for any workstation type of workload (i.e. almost any single user workflow you can think of...). (Server workloads thrive on more cores - but they run continuously and usually without user intervention once started).

Moving to a dual core processor was a huge jump by itself (a decade ago...) - today anything more than 4C/8T is usually too much $$$$$$ for the small increase in productivity for most workloads/workflows.

Greater than 4C/8T will become the 'norm' soon enough. But today that price performance ratio hasn't been hit by a long shot in 'workstation' workloads for most.

 

Since when are desktops and laptops running custom ARM-based CPUs? What's that? They're not? Oh, okay, why the comparison then?

Not really? Ryzen 7 1700 is twice as fast as a 7700K in productivity and only about 15% slower in games BECAUSE the 7700K has a 1GHz advantage in a scenario where only 6 or 8 threads are being utilized meaning that the extra multi-threaded capabilities on the R7 are not used in the game BUT are used to drive background tasks such as streaming, skype, discord, anything else you've left open like photoshop, game launchers, Anti-virus software, YT video, music and whatever. There's a reason why people are currently buying the R5 1600 over any i5 - cause 4 cores pegged at 90% usage in a game means that you can't really run a lot of background applications whereas having 6 cores and 12 threads means you have a lot of resources at your disposal with which to make sure that both your game/render/whatever is running fine while also being able to use the system for other things. Aka: Multitasking.

 

Because we doing the same tasks on computers; closing, opening tabs, apps, programs all the time. Here /core performance matters. Gaming? Still four fast cores the ideal. but if you do lots of video encoding, I can understand your point...

 

Not really? ARM is fundamentally different compared to X86 - not to mention the Apple A-chips are even more different when compared to the Snapdragons from Qualcomm - really, this comparison is very odd and does not translate to desktop/laptop x86 systems at all. As for the 4 fast cores - then why is the 7700K consistently ahead of the 7600K by upwards of 20%? Cause games are currently tuned to use 6-8 threads/cores and thus a simple quad-core is pretty much low-end/obsolete.

 

3.8GHz vs. 4.2GHz... As I sad faster cores means faster performance, so that is why. Hope they rather going to hit 5GHz on quad core, rather than 3.5 on 6 Cores...

 

going from 3.8 to 4.2 is barely 10% - that does not account for the 20% disparity between an i5 and i7 - why? cause the reason is not clockspeed - even with both at 5GHz, the i7 is still 10-15% ahead - it's cause of HT.

Honestly, if Intel remain on 4-core 125W 5GHz chips, they'll lose a LOT of market share. They're already down several % due to Ryzen cause people are sick and tired of quad-cores - we've had mainstream quad-cores from them for over 10 years. It's time to move on.

 

Yes HT is a little extra. I did not notice one is i5 the other i7. But now you start confusing yourself. HT does not translate exactly to cores (far not), that is why I7 can have such a fast cores!! The slower 6 or 8 Cores without HT or with HT in everyday use won't be faster than I7 Quad! What you pushing is only marketing by AMD, because without extra Cores they could not be faster than Intel and still only in situations where more than four Cores can be utilized, which quite rare for an average Joe...
I finished here, you can turn further for yourself and you buy what you prefer, but do not confuse others...

 

You're missing the key part - you're buying overpriced quad-cores with little no innovation added to them in the last 5 years. Intel's 4-core 8-thread CPU costs 350$. AMD's counter costs 150$ - why the hell would you pay 200% more for 20% more performance? That's ludicrous - I'm not repeating marketing, I'm stating my opinion and it is that 4-core CPUs are on their way out, whether Intel likes it or not.

 

Additionally - the question was whether there is any architectural CPU performance to be had from the 7700K over the 6700K and the answer is no, when you have both at 4.5GHz - the perform EXACTLY the same. The difference comes from the fact that Kaby Lake, once delidded, reaches 5GHz whereas Skylake tops out at 4.7GHz. That is the ONLY difference.

If you don't believe me - try Tom's, they've tested it.

 

Now for mainstream. We are moving fast to more cores.

Forget it. Only in your dreams

Exactly. Mainstream 4 c-8t Core i7 is soon history if you go for performance. On the scratch!! More cores and forget about +5GHz chips from Intel. Will never happen. Why should they? A lot people already laugh of the coming 112w 4C for HEDT platform. Only a cheapo. A step inside this platform if you go for upgrade later.

 

Where do you get your information?

See:
http://www.pcworld.com/article/3176...ming-pc-or-why-you-should-never-preorder.html

Ryzen 1700 barely equals a five year old gaming system...

 

Mainly from Hardware Unboxed, Tech City, Level1Techs (Wendell), OC3D (Tiny Tom Logan) and other similar sources. I've never actually heard of PCworld but if they're claiming that Ryzen is performing worse than Haswell, they are making stuff up quite hard. Ryzen IPC is about 5% lower than Skylake/Kaby Lake - the real difference in single-core performance comes from the fact that people benchmark a 5GHz 7700K vs a 4GHz 1700 - that's not how IPC is measured. IPC is measured by comparing at the same clockspeed to see which chip can do more instructions per clock.

Anyhow, clock for clock, Ryzen is about Haswell-Broadwell levels. And considering that the 1500X, which performs similarly to an i7 4790 non-K, costs 170$, I'd say it's damn good value.

 

Has there been any change in gaming performance with Ryzen? This article is soon two and a half months old. Perhaps not completely up to date.

This is also an interesting article regarding gaming performance gamersnexus.net As I see the reviewer from PC World use the older Core i5-3570K. Sometimes less threads will hamper the performance.

 

Myself? I don't care for overclocking nor for IPC scores.

Real world results are always better than any made up and non-realistic (non-realistic at least me) scenarios online rags like to make up so that whatever they're reviewing; it shows 'well'.

You've never heard of PCWorld? Oh my!

Do you have any links which show a 2x advantage to Ryzen 1700 vs. i7-7700K?

See:
http://www.cpubenchmark.net/compare.php?cmp[]=2970&cmp[]=2874

The single core 'score' above (~48% better with Intel) is much more useful to workstation type (i.e. single user) workloads than the slightly better (~13% better for AMD) multicore performance Ryzen has.

 

Yeah; I see (now) that it is an early article.

I also agree to have an i7 QC with 16GB RAM or more for browsing the internet...

But what AMD delivered (initially) is what I was expecting - I was actually surprised that I found it in a quick search...

 

As Atom and tiller said, plus was not mentioned, all those extra cores will create more heat, lessen battery time. Hope Core count war isn't coming, intel will stick to Quad for laptops, like Apple with dual for iPhones. Was really good example for everyday use, fuk the cores!

 

Ha, ha exactly, an Intel core i5 quad beats amd ryzen 8 cores in gaming by good margin sucking half the power. That's it guys.

 

You do realize that the R7 1700 has a 65W TDP at stock vs the 80W TDP of i5s? The reason is Samsung's LPP design on which Ryzen is based.

Check any productivity benchmark such as Cinebench or Video rendering or hell, even streaming using CPU encoding. Ryzen performs much better than you're making it out to be. Enjoy

 

Already announced 6 core Intel mobile A new Turd will be born. 45W and 6c/12t

Yeah, only 65W

 

I said stock - and true enough, at stock, the 8-core CPU uses as much power as Intel's 4-core CPU meaning that Ryzen's efficiency is much higher thanks to the LPP platform.

 

You do realize though that the Ryzen chips don't come close to achieving their design TPD ratings?

So, I see you can't give me any links for 2x the performance advantage for Ryzen 1700 vs. i7-7700K. Cool...

Btw; that video is useless; my time is worth a lot more than the hour + it would require to digest that video in context.

Productivity isn't Cinebench for me; nor is any form of video editing... if we take out the rendering benchmarks (which very few people, including me don't depend on...) - what increase in productivity does Ryzen offer at any cheaper overall (platform) cost?

 

You fail to see a lot of things. That power reading is taken from the WHOLE system POST PSU inefficiency - take 10W for the drives, 10W for the fans of the case and coolers, 5-10W for the dGPU idles (Ryzen has no iGPU like Intel), 15W for peripherals and you have about 70-75W - That's what it draws and outputs about 65-70W as heat. It's spot on but again, you refuse to listen to anything that does not support your point of view. Why should I even bother anymore with you? You refuse to review any evidence to contradicts your assumption that Intel is somehow perfect. Keep on buying overpriced quad-cores. Doesn't matter to me. Doesn't matter to anyone. Though at least don't lie and mislead others. There's a reason Ryzen has been sold out in quite a few retailers for weeks. There's also a reason people buy it over a 7700K or 6800K - cause it's flat out better for anything involving more than 4 cores. And for games, it's about 10-20% slower. But does that really matter? When the game is running above 120fps, does it matter if it's at 130 or 120? In addition, getting the 7700K to perform better requires you to use a top end GPU like a 1080 Ti at 1080p. if you run any game at 1440p or 4K - the results are identical. If you test with a 1070 - results are identical. Sad truth is, Intel's current products are already outdated in their fundamental design. Their products are simply not going to last nearly as long in the grand scheme of things. In addition, the price tags are, frankly, slightly insulting. 350$ for a quad-core in 2017 should not be the case as proven by the 170$ 1500X which performs similarly to last gen's 350$ quad-core i7 (4790)

EDIT: Until you go and actually read up on the subject from the people over at OC3D and L1C, I will refrain from further replying as this discussion will otherwise not amount to anything.

 

LOL! A good one. Did anyone else read the entire article?

@tilleroftheearth, I tend to agree along the lines of your other posts when it comes to the CPU. I don't game, so FPS, Firestrike, or whatever gaming benchmark one points to doesn't matter. It is more about productivity, the platform, and what work can be done. Not some fluffy FPS measurements or other benchmarks. Posting an article that runs some FPS comparison between 3 different games doesn't do much to prove a point regarding productivity AMD Ryzen.

Also, this line from the article was the best... "The Core i5-3570K was long ago overclocked from its stock 3.5GHz to 4.2GHz to put more pep in its step. Because gamers with similar rigs are likely to have done the same, I left that in place." But there's no mention regarding overclocking the Ryzen based machine.

 

I fail at nothing I put my mind to. And this topic is something that interests me a lot.

What I gather from you is that you can't provide links for the posts you type... I have provided third party evidence for my views, at least.

Go ahead and ignore me; and ignore the links I provide too.

But I'm not lying or misleading others. Not by a long shot.

You on the other hand have bought the AMD marketing bs hook, line and sinker.

For example: assuming that the performance was the same (which everyone knows it's not...) saving half the price of an $350 processor on a platform that for me is in the $2K+ range is not a 'victory' in any sense of the word. Especially when in the most important workstation usage scenarios the AMD platform is 20%+ slower...

But; you believe what you will and I'll do the same.

The rest of the people reading this thread can decide for themselves if saving 9% on a one time cost (if they scenario is similar to mine) is worth their time at ~20% each time they use their systems.

Take care.

 

The point though in that article is that an upgrade (especially going from Intel to AMD) isn't just comparing spec's - it is comparing real world setups to whatever the competition offers 'today'.

Agreed that the lack of O/C'ing the Ryzen machine (also) was a misstep for the reviewer - but for myself; I don't O/C anything either... so I'm still cool with that.

If you have to compare Ryzen to a five year old platform (even O/C'd) - then that is pretty telling that anything newer is just not worth upgrading from too.

 

The article was an indication of nothing. A fluff piece. Meaningless.

Since when does a comparison of FPS on three games have anything to do with the performance of a CPU? A run through three games is hardly a large enough sample size, and FPS tells more of the GPU than anything else.

Let's get to the real questions that you and I both want answers to. How much work I can get done? For me what about multiple VMs with multiple compilers churning through code? How fast can it render this image or encode that video? Those are the questions to ask when evaluating this CPU.

 

I took away more than that from that article; AMD is marketing Ryzen for 'cheap gaming' and also 'ultimate computing power'. It is neither.

For my workflows? I have given up on online rags to give me meaningful 'scores' - I simply have to do it myself on my own actual workflows/workloads. The key though is that I would compare the platform I have now to the platform I am evaluating.

Nothing more 'real world' than that.

 

Now you're talking.

 

Isn't Kabylake also more power-efficient than Skylake?

 

Not really - it's the exact same CPU core with a slightly different iGPU and modified power delivery to allow for higher clockspeed.

 

See:
http://www.pcworld.com/article/3111...ings-to-know-about-intels-kaby-lake-cpus.html

With 12% to 19% better performance over Skylake (nominal/real world) at the same TDP rating; it is much more efficient. For the same workload it will finish faster and use less total power.

And if you use 4K streams? The power efficiency is drastically improved.

 

Not really - TDP is NOT related to the REAL WORLD power draw of these CPUs

As TomsHardware have demonstrated, the 7700K is overvolted and pushed to the limit even out of the factory drawing well over 100W even at stock. The rated TDP is absolute BS and not accurate whatsoever on Kaby Lake i7s

 

Performance is also painfully similar.

 

Clock for clock perform 6700K and 7700K exactly same. Kaby lake have better Oc headroom and will use less voltage for same clocks. Hence will use a bit less power. iGPU is no existence in laptops with LGA. Aka No gain. And Optane isn't useful for laptops either for now. But the point is you shouldn't go for older tech, although they will give almost same performance.

 

Agreed; it's not. But Global Wattage from AC source isn't remotely a processor's TDP either.

Metro Last light Redux is no indication of performance. As a matter of fact; no game engine is - at least not for my workloads.

Clock for clock isn't anything to compare either. I don't buy latest tech to run at what my last platform worked at...

 

Comparison is 4.6 vs 4.8 - the average OCs for both chips. Most people aren't experienced techies and they don't know how to delid and tweak the 7700K so as to get 5GHz. Add to that the fact that the 7700K, even at stock under a 240mm AIO, runs quite hot (A friend has one with a Corsair H100i - gets about 75-80*C load temps when rendering for a few hours. Before, his older 3770K would only get 70*C with a small air cooler)

And power draw difference is only based on different CPUs - everything else is the same so I can safely say the 7700K draws more power under load than the 6700K due to the higher clockspeed it's forced to run on so as to be worth the investment over the cheaper 6700K

 

Even in multi-core productivity - you're looking at a 10% difference AT MOST and that is only IF there is a difference in clockspeed.

 

Here i link a related reading for don_svetlio whom i left here after a certain number of hopeless posts. So good that my theory proves to be right also for x86 systems. More slower cores not just making general tasks running slower, but also games!

Benchmarks Find Intel Core i7-7700K Better Than i7-7800X for Gaming​

 

X299 is, generally speaking, been a massive failure so far.