Howdy, Stranger!
It looks like you're new here. If you want to get involved, click one of these buttons!
Quick Links
Sandy vs Kaby
http://www.hardocp.com/article/2017/01/13/kaby_lake_7700k_vs_sandy_bridge_2600k_ipc_review/1
Interesting article. They take a 2600K and a 7700K and clock them both to 4.5 to see what the difference in architectures bring in non-GPU limited scenarios. There are some other minor differences (DDR3 vs DDR4), but, as the test results show, they don't appear to be very relevant to the results.
the tl;dr
The Kaby architecture presents about a 20% increase over Sandy. That's 20%, over the course of 6 full generations of chips. In some benchmarks it was a bit less, some a bit more, but overall around 20%.
Interesting article. They take a 2600K and a 7700K and clock them both to 4.5 to see what the difference in architectures bring in non-GPU limited scenarios. There are some other minor differences (DDR3 vs DDR4), but, as the test results show, they don't appear to be very relevant to the results.
the tl;dr
The Kaby architecture presents about a 20% increase over Sandy. That's 20%, over the course of 6 full generations of chips. In some benchmarks it was a bit less, some a bit more, but overall around 20%.
Comments
That being said, they are pretty open about their testing, and of pretty much all review sites, I trust HardOCP's methodology and results the most. I don't think their final results are the result of any intentional fabrication, but I won't discount the possibility that there's an error in there somewhere.
Then again, maybe that 10-15% improvement clock to clock isn't really what we have been seeing. 10-15% from generation over generation maybe, but we've also seen clock adjustments there, or a change in RAM, or whatever so it's not a 10-15% IPC improvement (which is mostly what these tests in this article show), but does result in a net improvement.
I think they do plan to do a max overclock vs max overclock test eventually, he mentions in the article they are running some other comparisons. The trouble with that is that not every chip will hit a specific overclock, so how do you find a "fair" comparison there?
http://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/10
and Sandy OCed like a beast, even 5 GHz wasnt anything unsual lol
the only question is: why are you plain out lying lol
over 6 generations Intel increased performance by whooping 25%.
But we seen things slow and speed up depending on when new tech comes out before. We will certainly see a huge leap again within 10 years.
I expect Ryzen to have nearly identical IPC to Intel due to AMD and Intel's relationship. Since they develop on the same platform they must come to a standard for their architectures or they will begin to diverge too much. Both CPU architectures will mostly utilize the silicon the same. Something tells me their cross-licensing agreement also includes coming to a standard way of doing things. Like how x64 became the standard instead of Intel's 64-bit architecture. In that vein I think the general architecture in Ryzen is very similar to the Core architectures as it has proven to be better. They will begin diverging in other ways.
I think at the end of the day, Intel wants AMD as it's own company instead of going out of business. If AMD does fall it would mean a company like Qualcomm or Samsung would get the x86 license by buying up AMD in bankruptcy so Intel does not retain a monopoly. They would offer too much competition by comparison and may eat up a lot of their market share.
I self identify as a monkey.
Same holds true also for AMD: The speed difference between Bulldozer family processors first released in 2011 and the processors sold today is really small.
Here's hoping 2017 will change it and we'll see some good processor development, because if things continue as they are it's only a matter of time before ARM processors will be serious competitors in desktop computers.
I think part of the reason is they've gotten to the point where they can handle everything you throw at it awhile ago. What's the point of a super CPU if you don't need it on a personal computer for literally anything.
I have a 6700k and I definitely do not use all of its potential, and I'm playing online games which is the most demanding thing most people will use a CPU for.
And to get to 5GHz on 7700k youll need beefeiest water cooling you can find and it will still run 90c lol
AVERAGE of > 60000 tested CPUs in now "rare exception" <- prime example of kind of bullshit you post.
You have no clue whatsoever on the matter. Its exactly this king of missinformation and plain out lies that most people are exposed to so they cant get real information but your kind of BS.
I dont make claims that have no connection with reality or lies i deal only with FACTs. But we got a glimpse of your PC slizz, so...yeah lol
To leave CPUs for a moment and talk about GPUs, it's basically trivial to come up with a benchmark in which the Radeon RX 480 will beat a Pascal-based GeForce GTX Titan X outright. It's likewise easy to come up with a benchmark where the RX 480 will lose to a GeForce GTX 1050. This isn't just weird synthetics, either; an RX 480 beating a Titan X outright happens in at least one class of algorithms that real people care about for their own sake--and that makes sense to run on a GPU.
If you want to compare a Radeon RX 480 to a GeForce GTX 1060, finding situations where either doubles the performance of the other is pretty easy to do. But that doesn't mean that such results are typical. If you're asking how much faster Kaby Lake is than Sandy Bridge, the critical question is, faster at what? Because there is no such thing as a "typical" or "average" workload.
But more fundamentally, if one processor architecture is very heavily derivative of the previous, we shouldn't expect huge performance gains from it. Twenty years ago, that wasn't the case, as you could have a new processor of basically the same architecture as the previous but clocked a lot higher. That died around 2004 as physics put up a clock speed wall around 4 GHz.
"Sandy Bridge and Ivy bridge are both getting pretty old by now, and upgrading to Skylake will yeld a 30-45% (45% if your a sandy user) increase in performance. You will also get the latest technologies from Skylake, namely power efficiency, a variety of USB ports, high speed LAN, and probably the biggest upgrade will be high speed storage. Together, all these features makes it a worthy upgrade to Skylake."
http://www.tomshardware.co.uk/faq/id-2752766/skylake-upgrade.html
So Skylake 45% faster than Sandy - and Kaby is faster than Skylake.
As many people have said however whilst extra cpu speed is good a lot of the advancements are in other stuff such as mentioned in the summary.
Consequently Qualcomm has been a competitor for some time - whose market cap is sometimes more than Intel's (depending on respective share prices); ARM as well - now part of Softbank - having been bought in mod-2016 for £23.4 billion.
AMD, by comparison, is tiny. Lower market cap, fewer employees, lower sales etc.
AMD and RYZEN will be good for consumers and a way to push Intel into actually making better CPU's/
There were a few reasons he didn't give power numbers:
A ) His tools for measuring CPU power draw on the motherboard didn't work with the Z67 motherboard he was using for the Sandy Bridge. He could have done at-the-wall power, but...
B ) This was more to look at IPC gains in a desktop environment, where power draw really isn't that critical.
C ) Your looking at overclocks of some sort on both chips, and that's going to skew power draw away from stock considerably. Without a good number of chips to be able to draw an average from, your looking at the silicon lottery again.
Its same kind of BS you were exposed to so far.
but you dodnt even read it correctly:
"For haswell users, it's a no in general. The key advantage of Skylake is feature set, not performance. So if your looking for a good boost in speed then don't switch to Skylake, rather switch to Haswell-E which is a massive boost in performance to Haswell and Skylake.
Another good reason to not upgrade is overclocking. Your i5 K sku or i7 K sku overclocked at 4.5Ghz or beyond is actually a little bit faster than the 6600K or 6700K at stock speeds. While you could upgrade and overclock, would a 10% increase in performance be worth $500 of parts?
Now, if your the select few that uses lots of usb C and USB 3.1 devices, plus you need high speed storage from twin M.2 slots then upgrading makes perfect sense."
the only thing he said is the platforn, NOT speed. And since hardly anyone actually uses new stuff it completely pointless lol
Uneducated people mix OCed and stock. And thats why i said that buying locked Intels CPus is waste of money.
For instance: if you bought unlocked Sandy Bridge, youre fine even today, as someone buying latest and greates from Intel gets ~20% improvement. If you bought 2500k/2600k you have faster CPU than ANY Kaby Lake i5/i77xxx chip except -k Kaby Lake becuase OC frequency will easily nullify ANY IPC improvemments Kaby Lake has over Sandy Bridge lol
OTOH. if you bought locked chip, especially lowest clocked SKU youre screwed as newer chips have higher frequencies (and especially Sandy Bridge had low stock frequencies)
Sandy Bridge is still best generation Intel have done, all after that has been mish mash "it uses a little less power (useless on a desktop) with 3% IPC improvement". Well, Kaby Lake didnt get even that. its straight Skylake rebrand with 200-300 MHz higher stock clocks and 100-200 MHZ better OC.
Wont even comment on the "genius" that thinks that over 60000 chips isnt good sample to see what these chips can do lol
And I read it perfectly. The OPs post is about Sandy - not Haswell. The OPs article was also about Kaby. So - because I did read the OPs post I pointed out that the summary was a comparison with Skylake. Why do you talk about Haswell though - can't read? Newslash: if you own Skylake there is even less reason to upgrade to Kaby.
As for feature sets sure mileage will vary. In the opinion of what you consider a random crap site called Tomshardware though the combination of speed & features was sufficient to merit an upgrade. And it tested Sandy to Skylake overall at 45%.
This article kinda just kicks itself in the rear because the test subjects are being altered instead of using factory specifications.
I touched on this several times to not trust big numbers on websites advertising new hardware.You get it home and it almost always performs worse.
Nobody buys a PC to simply run benchmarks lol,you need to test a machine on the application you plan on using it for as NOTHING else matters.
Never forget 3 mile Island and never trust a government official or company spokesman.
The point was not to see how much faster a 7700K is than a 2600K, it was to see how much Kaby Lake improved over Sandy Bridge. If those two statements mean the same thing to you, then you are not the intended audience for the article.
theres a good reason why Intel locks its chips adn why it charges retarded tax for "privilege" of OCing, because as everyone can see, if you have OCable -k Sandy Bridge, 6 generations later you have no reason whatsoever to upgrade performance wise as most youll get is ~25% but more like ~20%, if you dont - youre screwed lol
And i think i have mentioned quite a few times now how non -k chips from Intel are very bad buy. Thats also why i3 was very bad buy and why now Intel charges 190$ for unlocked i3 (whch is more than i5) and here are results of 5GHz 190$ i3 OCed to 5GHz (lottery to even achieve) against stock 180$ i5 6500
so after you buy expencive Z270 mobo (so you can OC), expencive cooler, are lucky to even reach 5 GHz, pay more for the chip, then that OCed chip uses more power than i7, its STILL slower than 3,2 GHz 6500 lol
Now, if you could OC cheapest 110$ i3 on cheaper motherboard.....that would kill any i5 sales except unlocked i5s lol etc etc but Intel says no, you have to buy more expencive chip with more expencive mobo/cooling and still end up with something a bit slower, which kills whole point of OCing i3