Coffeelake thread, benchmarks, reviews, input, everything.

[LTC8K6]

And as far as I know, no reviewer has told us if MCE will allow the non-K chips to run at the turbo multiplier, as in the past.

 

[TheLycan]

It's not about the choice of 8600k in a vacuum: your were also considering spending more money to go from 3200 C16 to 3600 C17. You would gain more by staying with 3200 C16 and spending the extra money towards the 8700.

AtenRa made the calculation by also factoring in the included heatsink for 8700, so keep that in mind as well.

You are partially right, but:
- who stays with stock cooler? I plan to buy an aftermarket cooler anyway, for silence. So no extra money spendt here
- 3200 vs 3600, that was my original question. If i get extra performance it will be for both cpus, so i dont understand why you say that is needed only for 8600k

 

[coercitiv]

- who stays with stock cooler? I plan to buy an aftermarket cooler anyway, for silence. So no extra money spendt here

Some stay with the stock cooler, personally I always go with aftermarket coolers.

- 3200 vs 3600, that was my original question. If i get extra performance it will be for both cpus, so i dont understand why you say that is needed only for 8600k

It's not needed only for 8600k, I did not say that. Every dollar you spend on the CPU/RAM combo yields a certain performance level, we call that performance / dollar. As long as the 8600K will not be overclocked close to it's maximum potential it's very likely that 8700 + 3200 C16 RAM will offer better perf. / dollar than 8600K @ 4.5Ghz + 3600 C17 RAM.

Anyway, considering the performance delta between these two options you can simply choose whichever you like the most and enjoy your new system. Here on the forums we tend to put too much emphasis on efficient spending.

 

[Dufus]

IMHO MCE should not even be a consideration on unlocked chips, it just doesn't mean anything. My old 2500k would only run 4.6GHz on all cores due to thermals but with 1 or 2 cores would reach higher, up to 5GHz. How would forcing 4.6GHz be an enhancement. I also don't get why some manufacturers would tie it to XMP unless to stop it's use on CPU's that don't support XMP.

Now locked chips made to run all cores at the top turbo, that I would consider an enhancement. For the record forcing top turbo on all cores when it was opportunistic was actually done as far back as 2009.

 

[Arachnotronic]

IMHO MCE should not even be a consideration on unlocked chips, it just doesn't mean anything. My old 2500k would only run 4.6GHz on all cores due to thermals but with 1 or 2 cores would reach higher, up to 5GHz. How would forcing 4.6GHz be an enhancement. I also don't get why some manufacturers would tie it to XMP unless to stop it's use on CPU's that don't support XMP.

Now locked chips made to run all cores at the top turbo, that I would consider an enhancement. For the record forcing top turbo on all cores when it was opportunistic was actually done as far back in 2009.

MCE is just a way for motherboard makers to try to win benchmarks. Once one motherboard company implemented it, every motherboard company had to follow, lest they be left behind.

It's not a bad feature as long as it's turned off by default and users can consciously turn it on. But having it on automatically is bad.

 

[Dufus]

Depends on the perspective maybe, i.e. good for business.

 

[TheLycan]

Some stay with the stock cooler, personally I always go with aftermarket coolers.

It's not needed only for 8600k, I did not say that. Every dollar you spend on the CPU/RAM combo yields a certain performance level, we call that performance / dollar. As long as the 8600K will not be overclocked close to it's maximum potential it's very likely that 8700 + 3200 C16 RAM will offer better perf. / dollar than 8600K @ 4.5Ghz + 3600 C17 RAM.

Anyway, considering the performance delta between these two options you can simply choose whichever you like the most and enjoy your new system. Here on the forums we tend to put too much emphasis on efficient spending.

Agree. I for sure am looking at perf/$. So a 35$ cooler i get anyway for silence and it will help with overclocking. Z370 board is bought anyway, because i dont have anything else. So it all comes to paying 20% more for cpu for overall +2% performance, when i get overclocking for free. It could make sense to get 8700 if one will not overclock, or if you could decrease with a b390 board. But, with overclocking, best gaming cpu is 8600k, or maybe 8400 without overclocking.

Anyway, would it make sense paying 30$ more to get 3600 ram, instead of 3000?

 

[PeterScott]

Adored says he does not believe 8400 can sustain 3.8Ghz under 6 core loads, at least not the retail chips and not on cheaper 300 series boards (enforced 65W TDP and lower max current than Z boards). Problem is there's a very clear difference between being able to sustain these clocks in heavy duty loads like CineBench / Blender / Prime etc and keeping them up in usual consumer loads such as games, browsers, archiving programs etc.

No, that isn't the problem. It's the baseline assumption that 8400 can't sustain 3.8GHz x 6 cores, and spreading misinformation on it.

The main loads most people want all core turbo is for heavy duty continuous loads like rendering, and so far in CineBench/Blender/x264 testing/reviews, the 8400 has sustained it's 3.8GHz all core turbo speed.

Saying those are cherry picked chips and retail ones will fail is nothing but FUD, based on ZERO evidence.

That is the problem.

 

[Bouowmx]

From a gaming perspective (focusing on lowest latency memory), going from 2x8 GB of 3200 MT/s C14 (8.75 ns) to 3600 MT/s C15 (8.33 ns) is $40. Considering total cost of build, I do not think $40 is a big deal, but 3600 MT/s is a bigger challenge for motherboard compatibility. In the case of incompatibility, set the modules back down to 3200 MT/s C14, in which you would be better off not paying extra in the first place.

 

[deathBOB]

I don't see how this could be Intel's or the motherboard manufacturer's fault in any way - simply looking at the core clocks would reveal MCE was on right?

 

[PeterScott]

From a gaming perspective (focusing on lowest latency memory), going from 2x8 GB of 3200 MT/s C14 (8.75 ns) to 3600 MT/s C15 (8.33 ns) is $40. Considering total cost of build, I do not think $40 is a big deal, but 3600 MT/s is a bigger challenge for motherboard compatibility. In the case of incompatibility, set the modules back down to 3200 MT/s C14, in which you would be better off not paying extra in the first place.

Can you alter memory timings in Non Overclocking boards? IIRC for non OC boards/chips you can't raise the memory speed beyond Spec (2666 for CL 6 core), but can you tweak it to run with less latency?

 

[TheLycan]

From a gaming perspective (focusing on lowest latency memory), going from 2x8 GB of 3200 MT/s C14 (8.75 ns) to 3600 MT/s C15 (8.33 ns) is $40. Considering total cost of build, I do not think $40 is a big deal, but 3600 MT/s is a bigger challenge for motherboard compatibility. In the case of incompatibility, set the modules back down to 3200 MT/s C14, in which you would be better off not paying extra in the first place.

I am checking 3000 cl16 vs 3600 cl17. 30$ difference

 

[Bouowmx]

Can you alter memory timings in Non Overclocking boards? IIRC for non OC boards/chips you can't raise the memory speed beyond Spec (2666 for CL 6 core), but can you tweak it to run with less latency?

Correct, no memory overclock, but timing tuning allowed.

I am checking 3000 cl16 vs 3600 cl17. 30$ difference

Considering performance/price, if total cost of build is high ($1500 or up), then I think $30 extra is worth. And published very recently: https://www.techpowerup.com/reviews...Memory_Performance_Benchmark_Analysis/10.html

 

[TheF34RChannel]

I am checking 3000 cl16 vs 3600 cl17. 30$ difference

IMO with 3000 you're missing out, I'd recommend 3200 as a minimum which also is the current sweet spot. I myself am coming from 3000 and opted for 3333 CL16 (due to no availability) (and was told here the set isn't so good because of the latency*).

FlareX was made for Ryzen so it does not have XMP compatible profiles because AMD doesn't use XMP. Not sure if it'll work, someone else here may be of better assistance.


*After reading the TPU memory review I am no longer worried about my set being worse than the ideal 3200 CL14: let's face it, for my needs [gaming] the difference will be approx. 1% = I don't care I don't do benchmarks where every little bit counts, so I'm good with my set.

 

[TheF34RChannel]

And published very recently: https://www.techpowerup.com/reviews...Memory_Performance_Benchmark_Analysis/10.html

Nice find, thanks for the link! Perhaps the most useful bit, a piece of the conclusion, for those who may not have the time to read through the review:

"If you have a bit more money to invest, the next [after 2666] good option is 3200 MHz memory, with CL14 or CL15. 3200 MHz CL16 is roughly equal to 3000 MHz CL14 in speed, so consider that option too. Coffee Lake also sees good gains from improved timings, which often makes it more sensible to buy lower latency memory than to go for the highest clock speed you can find."

 

[KCfromNC]

Correct, no memory overclock, but timing tuning allowed.


Considering performance/price, if total cost of build is high ($1500 or up), then I think $30 extra is worth. And published very recently: https://www.techpowerup.com/reviews...Memory_Performance_Benchmark_Analysis/10.html

For the most part this looks like an unintentional test of the noise inherent in the benchmarks they picked.

 

[Abwx]

Saying those are cherry picked chips and retail ones will fail is nothing but FUD, based on ZERO evidence.

So far here are the very chips that Hardware.fr was delivered by whom you know :

http://www.hardware.fr/articles/970-2/core-i7-8700k-core-i5-8600k-core-i5-8400-core-i3-8350k.html

Not sure that these are all off the shelves CPUs...

 

[PingSpike]

For the most part this looks like an unintentional test of the noise inherent in the benchmarks they picked.

Yeah, they should show minimum framerates or something because the gaming benchmarks look the same as the sdram cl2 versus cl3 benchmarks from 15-20 years ago. "Fast RAM makes a big difference!*"

*Assuming 3% counts as big.

 

[PingSpike]

So far here are the very chips that Hardware.fr was delivered by whom you know :

http://www.hardware.fr/articles/970-2/core-i7-8700k-core-i5-8600k-core-i5-8400-core-i3-8350k.html

Not sure that these are all off the shelves CPUs...

Sure they are, the hottest new product from your gaming PC is the new Intel Confidential N/A. Its going to be a sleeper hit.

 

[dullard]

For the most part this looks like an unintentional test of the noise inherent in the benchmarks they picked.

Yeah, they should show minimum framerates or something because the gaming benchmarks look the same as the sdram cl2 versus cl3 benchmarks from 15-20 years ago. "Fast RAM makes a big difference!*"

*Assuming 3% counts as big.

Did you both read the full 11 pages, or just look at the few gaming graphs?

The results are pretty much what you would expect. Below the top supported memory frequency (2666), performance plunges. It plunged way more than 3% in several benchmarks (such as 2133 was 10+% slower than 2666 in WinRAR Compress and CFD). Above that top supported 2666 frequency, the performance gains are measurable but quite small. Latency matters far more than high memory frequency in that region (above 2666).

 

[psolord]

I am confused why the broader community suddenly decided to have conversation about Multi-core enhancement, now.., when it could have had it 5 years ago. Manufactured drama.

Well AMD was toast back then. So a little faster or a little slower didn't matter.

However now there's competition after a long time and even small +/-5% deltas matter. Especially for cpus which see meaningful performance gains every few years.

 

[Markfw]

Sure they are, the hottest new product from your gaming PC is the new Intel Confidential N/A. Its going to be a sleeper hit.

I would question any benchmark made with a NON-retail stamp (ES or Intel confidential) as any benchmark could be flawed for good or bad because its not identical to the retail product. I know since I still have an 8400 dual-core ES

 

[DrMrLordX]

You are going to flip out a few of the newbies by bringing up one of these:

https://www.newegg.com/Product/Product.aspx?Item=N82E16819115037

"Wait a 8400 is a hexcore, not a dualcore, waaaaat"

 

[inf64]

8700K is an impressive chip. Still, there are few things that stand out with this launch. We have zero availability, odd bios settings for MCE depending on the motherboard vendor and zero upgrade ability of this platform. It is Kabylake with 2 more cores that comes with almost no added power cost, that is awesome. When will this be available for purchase at recommended prices though?

 

[PeterScott]

So far here are the very chips that Hardware.fr was delivered by whom you know :

http://www.hardware.fr/articles/970-2/core-i7-8700k-core-i5-8600k-core-i5-8400-core-i3-8350k.html

Not sure that these are all off the shelves CPUs...

Rather that just posting a link, perhaps you could tell us, what it is supposed to prove.

I was talking about i5-8400 FUD, and this doesn't change the view on that at all.

Adored just posted unsubstantiated FUD.