Intel no longer advertises or displays multi-core Turbo frequency information

[Greg S]

The two most common specs examined when looking at CPUs are clock speed and core count. With improvements in CPU architectures and increasing core counts, varying clock speeds on a per-core basis have provided significant benefits to performance while keeping thermals and power consumption under control.

Up until now, Intel has provided clock speeds for single-core and all-core maximum frequencies. Turbo Boost 2.0 allows each CPU to dynamically change per-core frequencies on any number of cores depending on load and operating conditions.

Moving forward, Intel has decided to stop disclosing information on its Turbo Boost Technology feature beyond single-core maximum frequencies.

Looking at the newest Coffee Lake CPUs on Intel's ARK page, all Turbo frequencies have been taken down besides the single-core maximum frequency.

Here's Intel's stance as provided to ExtremeTech:

[W]e’re no longer disclosing this level of detail as its proprietary to Intel. Intel only specifies processor frequencies for base and single-core Turbo in our processor marketing and technical collateral, such as ARK, and not the multi-core Turbo frequencies. We’re aligning communications to be consistent. All Turbo frequencies are opportunistic given their dependency on system configuration and workloads.

Official response aside, there is some speculation as to what the 'true' reasoning is in not publishing information that would be helpful to know if shopping for a new CPU.

One theory is that Intel wants to avoid any legal hassle for silicon that may not achieve an advertised Turbo frequency on all cores.

It is also possible that Intel has decided to stop publishing this data in order to make it more difficult for the competition to sell their CPUs. Intel currently has chips that can typically achieve higher single core frequencies compared to AMD's Ryzen line-up.

Whatever the reasoning, Intel is not going to be successful in hiding its numbers for long. Individual core frequencies can be easily measured using benchmarking tools.

See our Core i3 and Core i7 Coffee Lake reviews for more information on the latest generation of Intel's CPUs.

Permalink to story.

https://www.techspot.com/news/71334-intel-no-longer-advertises-or-displays-multi-core.html

 

[BSim500]

Here you go:-

i7-8700K - 3.7GHz Base. 4.7 (1C) / 4.6 (2C) / 4.4 (4C) / 4.3 (6C)
i7-8700 - 3.2GHz Base. 4.6 (1C) / 4.5 (2C) / 4.3 (4C) / 4.3 (6C)
i5-8600K - 3.6GHz Base. 4.3 (1C) / 4.2 (2C) / 4.2 (4C) / 4.1 (6C)
i5-8400 - 2.8GHz Base. 4.0 (1C) / 3.9 (2C) / 3.9 (4C) / 3.8 (6C)
i3-8350K - 4.0GHz Base. No Turbo but unlocked
i3-8100 - 3.6GHz Fixed

 

[EEatGDL]

Here you go:-

i7-8700K - 3.7GHz Base. 4.7 (1T) / 4.6 (2T) / 4.4 (4T) / 4.3 (6T)
i7-8700 - 3.2GHz Base. 4.6 (1T) / 4.5 (2T) / 4.3 (4T) / 4.3 (6T)
i5-8600K - 3.6GHz Base. 4.3 (1T) / 4.2 (2T) / 4.2 (4T) / 4.1 (6T)
i5-8400 - 2.8GHz Base. 4.0 (1T) / 3.9 (2T) / 3.9 (4T) / 3.8 (6T)
i3-8350K - 4.0GHz Base. No Turbo but unlocked
i3-8100 - 3.6GHz Fixed

You meant "C" in all the T's.

 

[BSim500]

You meant "C" in all the T's.

Well I thought "Thread" instead of "Core", but I'll change it if it makes it easier to read.

 

[GreenNova343]

You would think they would want to show those speeds, since they're technically higher than the full-core speeds available on the Ryzen & Threadripper CPUs from AMD (even when OC'd).

...

Oh wait, that's right...

It's because people would be disappointed & less than impressed when they see how the full-core Turbo speeds are almost identical to the Kaby Lake full-core Turbo speeds -- especially if the same # of cores are active:
-- i7-7700K: 4.2GHz base; 4.5 (1C) / 4.4 (2C) / 4.4 (3C) / 4.4 (4C)
-- i7-7700: 3.6GHz base; 4.2 (1C) / 4.1 (2C) / 4.1 (3C) / 4.0 (4C)
-- i5-7600K: 3.8Ghz base; 4.2 (1C) / 4.1 (2C) / 4.0 (3C) / 3.9 (4C) (estimated; couldn't find the exact Turbo frequencies for the i5-7600K, but the range between its base & max Turbo of 400MHz is identical to that of the Skylake-S i5-6600K, which has stepping of 100/200/300/400MHz for 4/3/2/1 cores)
-- i5-7400: 3.0GHz base; 3.5 (1C) / 3.4 (2C) / 3.3 (3C) / 3.2 (4C) (also estimated; I couldn't find an exact match for this particular range in the Skylake or Haswell chips, but I'm assuming it's going to be similar to the unlocked 7600K. I'm not sure this was the best locked comparison, however, as the i5-7500 might be better; with a base of 3.4 GHz & 3.8 for 1C, I think it would probably end up being 3.7 @ 2C, 3.6 @ 3C, & 3.5 @ 1C)
-- i3-7350K: 4.2Ghz base. No Turbo but unlocked (unlike the i3-8350K, it still has HyperThreading, so it can at least still pretend to be a quad-core CPU, although technically it's a 2C/4T vs. the 8350K's 4C/4C)
-- i3-7100: 3.9GHz base. No Turbo (technically a 2C/4T CPU vs. the 4C/4T 8100; would be interesting to see how they compare once the non-Z chipset boards come out)

Of course, when the Kaby Lake Core i3 chips came out, the problem that Intel had was that their primary competition wasn't from AMD; Ryzen wasn't out yet, the A-series APUs just didn't have much punch, & the FX chips were well past any prime they had. Their competition were the Kaby Lake Pentiums. $64 for a G4560 got you a 3.5GHz, 2C/4T locked CPU -- half the price of the 7100, 1/3rd the price of the 7350K -- & was able to compete well with the pre-Ryzen CPUs. And to be honest, if you're still looking to build a true "budget" gaming PC -- or especially if you just want a HTPC for streaming video to your TV -- I'm not sure you even need Coffee Lake.

 

[EEatGDL]

You meant "C" in all the T's.

Well I thought "Thread" instead of "Core", but I'll change it if it makes it easier to read.

The thing of using T instead of C is that with T you're implying that you have to thinker with HT (enabled/disabled), but it works based on cores enabled by BIOS regardless of hyper-threading being enabled or not. Unfortunately, from what I've observed on Xeon [where this is a lot more noticeable], you have to disable cores; sending cores to sleep (C-states different from C0) doesn't have the desired effect in turbo frequencies.

 

[Bao Nguyen]

You would think they would want to show those speeds, since they're technically higher than the full-core speeds available on the Ryzen & Threadripper CPUs from AMD (even when OC'd).

...

Oh wait, that's right...

It's because people would be disappointed & less than impressed when they see how the full-core Turbo speeds are almost identical to the Kaby Lake full-core Turbo speeds -- especially if the same # of cores are active:
-- i7-7700K: 4.2GHz base; 4.5 (1C) / 4.4 (2C) / 4.4 (3C) / 4.4 (4C)
-- i7-7700: 3.6GHz base; 4.2 (1C) / 4.1 (2C) / 4.1 (3C) / 4.0 (4C)
-- i5-7600K: 3.8Ghz base; 4.2 (1C) / 4.1 (2C) / 4.0 (3C) / 3.9 (4C) (estimated; couldn't find the exact Turbo frequencies for the i5-7600K, but the range between its base & max Turbo of 400MHz is identical to that of the Skylake-S i5-6600K, which has stepping of 100/200/300/400MHz for 4/3/2/1 cores)
-- i5-7400: 3.0GHz base; 3.5 (1C) / 3.4 (2C) / 3.3 (3C) / 3.2 (4C) (also estimated; I couldn't find an exact match for this particular range in the Skylake or Haswell chips, but I'm assuming it's going to be similar to the unlocked 7600K. I'm not sure this was the best locked comparison, however, as the i5-7500 might be better; with a base of 3.4 GHz & 3.8 for 1C, I think it would probably end up being 3.7 @ 2C, 3.6 @ 3C, & 3.5 @ 1C)
-- i3-7350K: 4.2Ghz base. No Turbo but unlocked (unlike the i3-8350K, it still has HyperThreading, so it can at least still pretend to be a quad-core CPU, although technically it's a 2C/4T vs. the 8350K's 4C/4C)
-- i3-7100: 3.9GHz base. No Turbo (technically a 2C/4T CPU vs. the 4C/4T 8100; would be interesting to see how they compare once the non-Z chipset boards come out)

Of course, when the Kaby Lake Core i3 chips came out, the problem that Intel had was that their primary competition wasn't from AMD; Ryzen wasn't out yet, the A-series APUs just didn't have much punch, & the FX chips were well past any prime they had. Their competition were the Kaby Lake Pentiums. $64 for a G4560 got you a 3.5GHz, 2C/4T locked CPU -- half the price of the 7100, 1/3rd the price of the 7350K -- & was able to compete well with the pre-Ryzen CPUs. And to be honest, if you're still looking to build a true "budget" gaming PC -- or especially if you just want a HTPC for streaming video to your TV -- I'm not sure you even need Coffee Lake.

Actually the i3-8350K is particularly interesting, as it's essentially a i5-7600K rebadge with extra 2MB cache for almost half the price. Seems like we have a new king in budge gaming rigs.

 

[enemys]

You would think they would want to show those speeds, since they're technically higher than the full-core speeds available on the Ryzen & Threadripper CPUs from AMD (even when OC'd).

...

Oh wait, that's right...

It's because people would be disappointed & less than impressed when they see how the full-core Turbo speeds are almost identical to the Kaby Lake full-core Turbo speeds -- especially if the same # of cores are active:
-- i7-7700K: 4.2GHz base; 4.5 (1C) / 4.4 (2C) / 4.4 (3C) / 4.4 (4C)
-- i7-7700: 3.6GHz base; 4.2 (1C) / 4.1 (2C) / 4.1 (3C) / 4.0 (4C)
-- i5-7600K: 3.8Ghz base; 4.2 (1C) / 4.1 (2C) / 4.0 (3C) / 3.9 (4C) (estimated; couldn't find the exact Turbo frequencies for the i5-7600K, but the range between its base & max Turbo of 400MHz is identical to that of the Skylake-S i5-6600K, which has stepping of 100/200/300/400MHz for 4/3/2/1 cores)
-- i5-7400: 3.0GHz base; 3.5 (1C) / 3.4 (2C) / 3.3 (3C) / 3.2 (4C) (also estimated; I couldn't find an exact match for this particular range in the Skylake or Haswell chips, but I'm assuming it's going to be similar to the unlocked 7600K. I'm not sure this was the best locked comparison, however, as the i5-7500 might be better; with a base of 3.4 GHz & 3.8 for 1C, I think it would probably end up being 3.7 @ 2C, 3.6 @ 3C, & 3.5 @ 1C)
-- i3-7350K: 4.2Ghz base. No Turbo but unlocked (unlike the i3-8350K, it still has HyperThreading, so it can at least still pretend to be a quad-core CPU, although technically it's a 2C/4T vs. the 8350K's 4C/4C)
-- i3-7100: 3.9GHz base. No Turbo (technically a 2C/4T CPU vs. the 4C/4T 8100; would be interesting to see how they compare once the non-Z chipset boards come out)

Of course, when the Kaby Lake Core i3 chips came out, the problem that Intel had was that their primary competition wasn't from AMD; Ryzen wasn't out yet, the A-series APUs just didn't have much punch, & the FX chips were well past any prime they had. Their competition were the Kaby Lake Pentiums. $64 for a G4560 got you a 3.5GHz, 2C/4T locked CPU -- half the price of the 7100, 1/3rd the price of the 7350K -- & was able to compete well with the pre-Ryzen CPUs. And to be honest, if you're still looking to build a true "budget" gaming PC -- or especially if you just want a HTPC for streaming video to your TV -- I'm not sure you even need Coffee Lake.

Actually the i3-8350K is particularly interesting, as it's essentially a i5-7600K rebadge with extra 2MB cache for almost half the price. Seems like we have a new king in budge gaming rigs.

Budget gaming rigs rarely go for a $120+ mobo, and without it K-series CPU is nonsense. And when a locked 6-core i5 can be had for $10 more than an unlocked 4-core i3, the 8350K starts looking even worse. It'll be 8100 or 8400, depending on budget, but only after cheaper chipsets appear in shops next year.

 

[total]

Well, taking into account that their thermal compound has been sucking since Sandy Bridge, their CPUs may easily overheat when having more cores than they have now, I mean, an 8-core CPU may require a top-tier cooler to even reach those multicore turbo boost frequencies, and Intel can easily make future Turbo Boost depend on current temperatures, I.e. the lower temperature their CPU has - the higher frequency it will have. And it will be pretty hard to find out, because all "top" reviewers use top-tier cooling systems and obviously they will not notice anything suspicious. Moreover, regular users most likely will not notice it too, because who would spend much time on testing those frequencies? Regular users only play games, and having a lower frequency by let's say 300 or even 400 MHz will not influence framerate much. Personally I don't like how Intel treats their customers, it should not be like that.

 

[senketsu]

INTEL ARK now becomes INTEL DARK, checking for new data....nah, did you really think there was gonna be any

 

[Theinsanegamer]

Well, taking into account that their thermal compound has been sucking since Sandy Bridge, their CPUs may easily overheat when having more cores than they have now, I mean, an 8-core CPU may require a top-tier cooler to even reach those multicore turbo boost frequencies, and Intel can easily make future Turbo Boost depend on current temperatures, I.e. the lower temperature their CPU has - the higher frequency it will have.And it will be pretty hard to find out, because all "top" reviewers use top-tier cooling systems and obviously they will not notice anything suspicious. Moreover, regular users most likely will not notice it too, because who would spend much time on testing those frequencies? Regular users only play games, and having a lower frequency by let's say 300 or even 400 MHz will not influence framerate much. Personally I don't like how Intel treats their customers, it should not be like that.

Intels turboboost already does that, and has been doing that since Ivy Bridge I believe.

Doesnt really matter for me, my next machine will be ryzen based, and my next laptop will be raven ridge based. Still extremely suspicious that intel does this right when competition begins heating up. They must be awfully embarrassed at how poorly these new coffee lake chips compare to haswell..

 

[total]

my next machine will be ryzen based

Ryzen has a different problem - how to make non Samsung chips-based RAM work @ 3200 MHz and higher. Seriously, I tried to find out how much would a Razen-based PC cost, and a kit that 100% works with Ryzen is only 3000 MHz CL16 and it costs the same as 3200 MHz CL16 kits, but all those 3200 MHz kits are hynix-based and I found at least 1-2 forum posts that they are not fully compatible with Ryzen, meaning they do not work as 3200 MHz kits, only as 2933 or 3060 MHz.
So if somebody wants to use 3200 MHz RAM or higher, it's much easier to build a PC based on Coffee Lake.

 

[Athlonite]

The real story here is will Intel now start selling CPU's that have below average turbo speeds where as those core/turbo speeds were a guaranteed speed of

i7-8700K - 3.7GHz Base. 4.7 (1C) / 4.6 (2C) / 4.4 (4C) / 4.3 (6C)

may now became

i7-8700K - 3.7GHz Base. 4.5 (1C) / 4.4 (2C) / 4.3 (4C) / 4.2 (6C) or lower

And they'll legally be able to do it because they don't advertise the multicore turbo speeds so you can't say hey that's not right anymore... look forward to CPU's that once couldn't cut the mustard to now be sold as mustard cutters

 

[Athlonite]

Ryzen has a different problem - how to make non Samsung chips-based RAM work @ 3200 MHz and higher. Seriously, I tried to find out how much would a Razen-based PC cost, and a kit that 100% works with Ryzen is only 3000 MHz CL16 and it costs the same as 3200 MHz CL16 kits, but all those 3200 MHz kits are hynix-based and I found at least 1-2 forum posts that they are not fully compatible with Ryzen, meaning they do not work as 3200 MHz kits, only as 2933 or 3060 MHz.
So if somebody wants to use 3200 MHz RAM or higher, it's much easier to build a PC based on Coffee Lake.

You may want to take a little read here before spouting off

http://www.guru3d.com/articles-pages/amd-ryzen-7-agesa-1006-performance-update-review,1.html

they test quite a few sets of 3200MHz problems they found were mostly mobo related especially with the Asus Crosshair VI Hero which was quite picky they found using a gigabyte board they could run most sets that would fail on the CHVIH except for several GSkill kits that all failed to run at the rated speeds of 3200MHz

 

[total]

You may want to take a little read here before spouting off

You too
https://www.reddit.com/r/Amd/comments/607ka6/trident_z_rgb_am4_compatibility_from_g_skill/df42w4q/
https://www.reddit.com/r/Amd/comments/62vp2g/clearing_up_any_samsung_bdie_confusion_eg_on/dgsnlzj/
https://www.reddit.com/r/Amd/comments/62vp2g/clearing_up_any_samsung_bdie_confusion_eg_on/dhc8p5n/
https://www.reddit.com/r/Amd/comments/62vp2g/clearing_up_any_samsung_bdie_confusion_eg_on/dj4fk6q/
http://forum.gigabyte.us/post/7711/thread
http://www.tomshardware.co.uk/answers/id-3490084/3200mhz-ryzen.html#r20039681
Plus, there are some comments from Russian-speaking customers https://can.ua/g-skill-aegis-ddr4-3000mhz-8gb-f4-3000c16s-8gisb/p75308/#tab=comments that bought 2 G.SKILL Aegis DDR4 3000MHz 8GB XMP sticks and couldn't make them work, I can translate that for you:
7 Sep 2017 Alexander: I bought 1 memory stick and it has been stable for a week @ 2997 MHz, I ordered another one, hopefully they are from the same batch and will work well together.
7 Sep 2017 Roman: What CPU do you have?
8 Sep 2017 Alexander: Ryzen 5 1600, but today I received the second stick and my PC will boot with them at the 2nd-3rd attempt only when working @ 2933 MHz, the first stick alone works great, the second one doesn't really, I tried XMP profiles, I will try to configure them manually when I have time, if it doesn't help - I will ask for replacement.
8 Sep 2017 Yuri: What CPU and mobo do you have? I bought a 16 GB kit of the similar sticks and it will not work @ 2933 MHz, they work only @ 2133 MHz, while one of them is stable @ 3200 MHz. My mobo - MSI X370 SLI Plus. My CPU - AMD Ryzen 5 1600Х.

 

[Athlonite]

@total he bought two separate sticks of ram what did he expect that they'd just magically work right from the get go then he's an idi@t. If your looking at running dual channel ram then you buy a kit of two Dimms not one at a time also none of those have stated if they've updated their BIOS's to the latest with AGESA 1.0.0.6 now 1.0.0.7

 

[total]

@Athlonite I found proofs that those kits didn't work, while you're basing your assumptions on nothing. AMD released new agesa, so what? Where is a single comment from a user that bought a hynix or micron based memory and it worked well? Find proofs and then talk.

 

[cliffordcooley]

Where is a single comment from a user that bought a hynix or micron based memory and it worked well? Find proofs and then talk.

Why would anyone post that their memory is working as intended? By nature the only comments you will find is from problematic memory. Your request is unreasonable.

 

[total]

@cliffordcooley and again - I base my statements on actual user reports, while you base your statements on an assumption that the new patch should've fixed that. There's info that those kits didn't work, and they don't work until proven otherwise.

 

[cliffordcooley]

@cliffordcooley and again - I base my statements on actual user reports, while you base your statements on an assumption that the new patch should've fixed that.

No I didn't say anything about a patch. I said people without issue don't post comment. Therefor the fact you are requesting may not exist. But that means nothing at the end of the day, because there could be thousands if not millions of people who could prove you wrong.

 

[total]

@cliffordcooley find at least one per each memory kit specified to finally PROVE something
Meanwhile look there http://forum.corsair.com/v3/showthread.php?t=166631 and see that all 3000 MHz kits and higher are supported only by "select" motherboard. Then go let's say there https://www.msi.com/Motherboard/support/X370-XPOWER-GAMING-TITANIUM#support-mem-2 and read: "The memory compatibility depends on the CPU memory controller when the clock rate is over 3000 MHz.".
What I'm saying is that you either need to find exactly what is specified in your motherboard QVL which may not be easy and those kits often cost more than unspecified ones, and kits that are 100% compatible with Ryzen are based on Samsung chips and cost more too, so your attempt to may be save some money by buying "cheap" Ryzen stuff may easily be doomed to fail.
While Intel allows you to buy pretty much any RAM and stick it in your motherboard and just use your PC. That's all.