Who's buying a 6 core Coffee Lake CPU? (Poll Inside)

[Bouowmx]

It depends on 3 things. How high clocks 35W TDP CPUs will have, price of them, and whether it would be possible to lower the TDP to 35W of Core i5 8400.

Can you guys give me any link about possibility of lowering TDP for any Intel CPU? It would be very informative for me.

In addition to changed Turbo Boost power limits in BIOS, they can also be changed in Windows using Intel Extreme Tuning Utility in section Advanced Tuning.

Furthermore, while not directly controlling power, Throttle Stop Speed Shift Energy-Performance Preference (EPP) parameter, range 0-255, can be changed to control how much load is needed to trigger frequency increase, and how high frequency will get under load, affecting power. Higher values for more load and extreme values to reduce maximum frequency.

Or, disable Turbo Boost altogether.

 

[RichUK]

Maybe.

 

[coercitiv]

Furthermore, while not directly controlling power, Throttle Stop Speed Shift Energy-Performance Preference (EPP) parameter, range 0-255, can be changed to control how much load is needed to trigger frequency increase, and how high frequency will get under load, affecting power. Higher values for more load and extreme values to reduce maximum frequency.

ThrottleStop can control power. I can confirm it works on Z170 + 6600K combo.

 

[Glo.]

Here's a screenshot from a Z270 motherboard review.

In the second group of settings you can see those in charge of configuring short term power usage, long term power usage, as well as the max current limit. Together with the multiplier settings this makes just about any configuration possible, frequency is dynamically adjusted based on multiplier settings and power limits.

Thank you my friend, very informative.

Does it work rock solid, tho? Is this "power gate" type of switch, that will make your CPU never exceed this target?

 

[bigboxes]

I think you will find that their "love" is a result of their hate.

Nope. Only an Intel fanboi would see it that way. I'm all for competition. Bring on the best performance/price!

 

[coercitiv]

Does it work rock solid, tho? Is this "power gate" type of switch, that will make your CPU never exceed this target?

It's part of the chip power management, so I would consider it as solid as the 15W TDP setting for low power mobile parts. The only thing I would be worried about would be BIOS support, if you truly intend to do this make sure the motherboard of your choice has these options exposed, otherwise they may be locked in or available only through software.

 

[Borealis7]

seeing as i bought a 7700K at launch (Jan '17) it's a solid No for me. the i5-8*00K is going to roughly match my 7700K in price and have more cores and slightly less clockspeed, and only 9 months later. luckily i got a sweet deal on that Kabylake, so no buyers remorse here.

 

[FIVR]

Who would pay more for a 6 core that uses 2x the power vs an 8 core with more I/O and better upgrade possibilities?


This 6 core is a desperation move.

 

[AtenRa]

Hmmm I find that comparison a bit odd as the 1600X is a mid range chip and is more comparable to an i5 8600K (if that, I expect the MT performance to be similar between the two but ST peformance would be significantly faster on the 8600K)

The 8700K is on a different tier of performance altogether compared to a 1600X, the only thing they have in common is that they are both 6C/12T chips but the 8700K has a ~20% advantage in both IPC and clockspeed ceiling (4GHz vs 5GHz)

Where this ~20% IPC comes from ???

 

[frozentundra123456]

Where this ~20% IPC comes from ???

Where doe 8700k using twice the power of 8 core ryzen come from? Actually, that is an even more outrageous claim than a 20% ipc difference.

 

[R0H1T]

It was just an estimate on my part, and hey its only one benchmark, but from the other Coffee Lake thread:



8700K is 34.6% faster at Geekbench 4 single thread compared to a 1800X.
Granted it does have higher single core turbo frequencies of 4.7GHz vs 4.0GHz (I think?) on the 1800X but still... that's a significant difference in IPC no matter how you spin it.

Considering your outrageous claims of 20% IPC advantage & 5GHz overclock (potential) it's not that hard to guesstimate ~ http://www.anandtech.com/bench/product/1851?vs=1826

Unless you want to backtrack & say that clock for clock IPC is 20% not in favor of CFL, since you obviously didn't factor in turbo boost?

 

[AtenRa]

Where doe 8700k using twice the power of 8 core ryzen come from? Actually, that is an even more outrageous claim than a 20% ipc difference.

Agreed, he may just translated 65W vs 140W of TDPs though but still this is not true.

 

[crashtech]

People use the term IPC wrong all the time. Even I play a bit fast and loose with it occasionally. CFL won't have a 20% IPC advantage, it will be in single digits or so, but it will certainly have much greater clockspeed potential, which will increase per-core throughput to levels far beyond what today's Ryzens can achieve. Typed on a Ryzen 1700X system, btw, in case anyone feels like pasting me with the fanboy label.

 

[mikk]

Considering your outrageous claims of 20% IPC advantage & 5GHz overclock (potential) it's not that hard to guesstimate ~ http://www.anandtech.com/bench/product/1851?vs=1826

20% is quite accurate because it is definitely the case in games and even for applications it is not far off if we discount the better SMT scaling from Ryzen and keep in mind Coffeelake will further extend its IPC lead due to the much bigger L3 cache and slightly higher memory support on 6C Coffeelake. Also there is no performant AVX/AVX2 support for Ryzen, therefore Intels IPC lead for x265 is much higher than 20%. There is nothing outrageous in this 20% claim.

 

[R0H1T]

20% is quite accurate because it is definitely the case in games and even for applications it is not far off if we discount the better SMT scaling from Ryzen and keep in mind Coffeelake will further extend its IPC lead due to the much bigger L3 cache and slightly higher memory support on 6C Coffeelake. Also there is no performant AVX/AVX2 support for Ryzen, therefore Intels IPC lead for x265 is much higher than 20%. There is nothing outrageous in this 20% claim.

I don't see this 20% advantage you speak of, in games, even without Zen's SMT ~

Though to get a clearer picture we need tests at fixed clocks, without that we're basically guesstimating since turbo speeds will skewer the results from either side. You might have a point if we're going strictly stock vs stock.

 

[R0H1T]

Oh gee seems like my 20% remark really hit a nerve on some people! I clearly stated in my follow up post that it was an estimate and not a definite figure. Keep in mind that Coffee Lake is supposed to have slightly increased IPC over Kaby Lake as well.

With regards to Ryzen gaming IPC (is there such a metric? perhaps gaming performance is more accurate) this article shows that it is only roughly on par with Sandy Bridge... yes... Sandy Bridge.

https://www.techspot.com/review/1474-ryzen-vs-older-budget-cpus/page2.html

Compare the R3 1200 scores to the 2500K. Both 4C/4T chips. At best they are about equal, at worst the R3 is about 5 - 10% slower. Yes I'm aware the 2500K is clocked at 4.4GHz compared to 4GHz on the R3 1200, but keep in mind gaming performance doesn't scale linearly with clockspeed either as the GPU also plays a large role in the average framerate.

The Ryzen are doing fine with a 10% clock speed deficit, the SMT does work wonders though, not always but in many cases.

Also no one's offended, but this "my hyperbole is less hyperbolic than your hyperbole" doesn't work for me!

 

[mikk]

I don't see this 20% advantage you speak of, in games, even without Zen's SMT ~

This is not an IPC test, no wonder.

 

[R0H1T]

'Doing fine' against a Sandy Bridge chip from 2011 is not really that impressive though, is it? For all of Ryzens strengths gaming isn't one of them and we really shouldn't try to pretend otherwise.

Anyway we are getting off track now, back to Coffee Lake discussion shall we? I'd hate for this thread to be derailed because of my 20% figure...

For the price R3 sells for, I doubt you can find better gaming chips.

CFL should do well, the sudden explosion in VFM offerings from Intel will ensure that it does, though the real battle will only start when Ryzen APU debuts.

This is not an IPC test, no wonder.

You claimed it was 20% better even in games, how do you establish IPC for them? Force them all to run on a single core?

 

[aigomorla]

IF you guys are gonna get PMSY with RYZEN COMMENTS get the hell out and STAY OUT inside an INTEL Thread, until we see offical numbers.

Sigh i hate seeing the same people end up derailing the thread with Ryzen this... Ryzen that, when Intel hasn't even lifted the bed sheet off the product.

So if you want to Ryzen this... Ryzen that... GOTO A RYZEN THREAD.

 

[mikk]

You claimed it was 20% better even in games, how do you establish IPC for them? Force them all to run on a single core?

Yes exactly. The problem is you don't understand these tests, your link is a joke.. You are not aware of GPU Bottlenecks or you never heard of it. Some AMD Fanboys could come up with 8K benches and claim they are equally fast. You are doing something like this here. It doesn't work. There are hundreds of tests in the web and it is well known for serious testers that Intel has a big IPC lead in games. 20% is quite accurate, there is an 20% advantage for i5-7500 compared to Ryzen R3 1300x in games. Both are 4/4 with an allcore turbo of 3.6 Ghz.

 

[DooKey]

IF you guys are gonna get PMSY with RYZEN COMMENTS get the hell out and STAY OUT inside an INTEL Thread, until we see offical numbers.

Sigh i hate seeing the same people end up derailing the thread with Ryzen this... Ryzen that, when Intel hasn't even lifted the bed sheet off the product.

So if you want to Ryzen this... Ryzen that... GOTO A RYZEN THREAD.

Thank you.

 

[crashtech]

A certain amount of speculation regarding CFL's expected performance vs its competition is inevitable, and not entirely off-topic. The problem isn't the mention of the competition, it's the emotional baggage people attach to these pieces of silicon.

 

[R0H1T]

Yes exactly. The problem is you don't understand these tests, your link is a joke.. You are not aware of GPU Bottlenecks or you never heard of it. Some AMD Fanboys could come up with 8K benches and claim they are equally fast. You are doing something like this here. It doesn't work. There are hundreds of tests in the web and it is well known for serious testers that Intel has a big IPC lead in games. 20% is quite accurate, there is an 20% advantage for i5-7500 compared to Ryzen R3 1300x in games. Both are 4/4 with an allcore turbo of 3.6 Ghz.

You say IPC then you apply that logic to games, then backtrack & say that test's not IPC & now again 20% more in games?

You do realize that it's you shifting the goalposts right, how about you add a caveat to every statement of yours so that you don't have to resort to Trump like tactics? As for GPU bottleneck, where did you see an 8k bench in there or do you think 1080p is irrelevant to your claim?

 

[Arachnotronic]

IF you guys are gonna get PMSY with RYZEN COMMENTS get the hell out and STAY OUT inside an INTEL Thread, until we see offical numbers.

Sigh i hate seeing the same people end up derailing the thread with Ryzen this... Ryzen that, when Intel hasn't even lifted the bed sheet off the product.

So if you want to Ryzen this... Ryzen that... GOTO A RYZEN THREAD.

Seriously, this was a much needed comment. Thanks, aigo.

 

[PeterScott]

I am surprised that the Poll is so overwhelmingly i7. You save a lot going with an i5, and the performance benefits of hyper-threading are often not that great. I voted i5...