Clevo + Coffee Lake: Status?

I think it is still too early to get excited about anything at this point. Excitement about 50% more cores and threads in a laptop is understandable, but at this point it is really not a big deal for a desktop. With both AMD and Intel HEDT CPUs pushing double or more than 6C/12T it makes the thought of having "only" a hexacore CPU in a desktop now seem pretty ordinary and boring... like comparing a 6700K to a 6950X.

And, in terms of going to a hexacore in a laptop, it's going to be fun for benching on AC if the firmware isn't severely botched up. If the P570WM handled 4930K and 4960X with its abortion heat sink using AC cooling, the new P870 will probably do just fine on AC. We have no idea what kind of volcanic mess it is going to be without AC cooling or how buggered up the stock firmware is going to be. The P870DM-G, P870DM3 and P870KM1 are products I would never have an interest in owning without @Prema firmware. Let me qualify that by going a step further by saying there is no laptop of any brand or model that I want to own without @Prema firmware to keep things in perspective. But, there are a whole bunch of unknowns about how well Clevo is (or is not) going to execute on it. The idea of putting a hexacore CPU in a P7 chassis seems absolutely ludicrous considering it is thermonuclear with a 6700K or 7700K running undervolted with stock clocks and throttles like there is no tomorrow. Unless they do some really insane magic with the thermal management of that one it's going to be a nightmare. Same for the 16L13. There is no way what they have under the hood right now is going to manage an 8700K unless you run it on AC cooling 24/7. And, MSI castrated its power handling already with some kind of hardware limitation. It runs flawlessly and overclocks like a dream until you go a little past stock TDP, then everything goes to hell in a handbasket. If MSI doesn't remove that demonic feature, it will likely not hold overclocks at all because MSI has decided to play the role of TDP Nazis.

 

I'm sure if you do your math... And know the IPC, core counts vs. Kaby/Skylake. You will find correct scores. My math work very well But thats me

 

There is Zero IPC gain between 7700k and 8700k:

So, then what is that CB score at stock settings and stock boost that you have calculated in your head?

You do realize that so far the variance of 1200cb - 1400cb blows under that score, rather than over it due to failure to hold Turbo Boost?

 

Failure to hold Turbo... You mean 8700K can't hold stock Turbo boost. What if you OC? Will it fail in this task as well?

 

It would draw more power, run hotter, maybe not OC as high on poor silicon, I think you know the results of bad silicon, right?

The "Maximum" Turbo boost for the 8th generation as quoted from Intel isn't guaranteed and it's not declared as before, you need to infer it yourself, as per the AdoredTV video details.

The poor silicon gives less than the "Maximum" boost.

“Intel will no longer provide this information”

"We are only including processor frequencies for base and single-core Turbo in our materials going forward - the reasoning is that turbo frequencies are opportunistic given their dependency on system configuration and workloads"
https://www.anandtech.com/show/11859/the-anandtech-coffee-lake-review-8700k-and-8400-initial-numbers

"Nonetheless, we were able to obtain the per-core turbo ratios for each of the new processors for our motherboard. Given Intel's statement above, it seems to suggest that each motherboard might have different values for these, with no Intel guidelines given."

So, you can't count on the Max Turbo at all...it will vary by CPU and motherboard.

 

A bad silicon(8700K) will most likely do 4.3GHz on all 6 cores? Btw. Changes in performance with different memory speeds-timings on Coffee.


What would be written on the web if a lot of Intel's 8700K can't do more than 4.3-4.5GHz ? Would Intel sell a lot of them?

 

theres nothing to miss, i watched adoredtv's video i know what he talked about. but you misunderstood what i said which is what you quoted.

to simplify, 7700k gives score A, 8700k gives score B, if B > A result in a % increase over ~50% then B score is incorrect, that idea is false because score A could also be incorrect. this is basically what i was trying to tell u.

 

His premise of 8700K CB of 45% greater due to 2c/4t increase vs 7700k, but 0% IPC increase difference between 7700K and 8700K is correct.

The 7700k scores were invariant - stable across all the reviewers results within the measurement error/variance, while the 8700k results are over a wide range of variance - too large to be explained by test variance.

Notice the relatively straight line graph for the 7700k results, vs. the wide range of the 8700k results. Even taking off the top 4 scores due to "accidental" 4 core Turbo boost of 4.7ghz, you still have a variance of 1200cb - 1400cb, that's where the "bad" silicon comes in to play, at the lower scores the CPU's can't hold Turbo boost Maximum - or never make it at all.

 

this also gets into a grey area and how one would identify as "IPC" increase. since cache/IMC and rest of uncore are part of CPU, increasing those would result general* IPC increase, where as default CORE architecture remains unchanged.

reason i say general IPC increase is because if a software that doesn't use the faster IMC or the increased cache, then you see no changes in IPC.

in our example better to explain as, 8700k 12MB cache which has 4MB or L3 cache than 7700k and a faster IMC, however if a software store its needed files in less than 8MB of L3 cache, we will never have the need for that last 4MB of cache, thus increasing in cache doesn't increase the speed. once 8MB of cache is exceeded, then we start to see benefit, because L3 cache is much faster than memory, which if info is drawn from memory, the faster IMC in 8700k will also benefit towards IPC. this is just comparing two things aside from the core itself, there are obviously other parts of CPU that may contribute to IPC but most remains unchanged from 7700k.

 

thats why i said in the review scenario 8700k were most likely wrong, go back a few post man..

 

The test results to find IPC differences speak for themselves


The 7700K score is within a couple of points of result compared to the 8700k when comparing IPC at 4.5ghz.

There is no IPC gain. There may be other losses or gains, but IPC is zilch difference between Kabylake 4 core and Kabylake 6 core.

 

See. 4.5GHz in CBR-15 from NBC.net



[email protected]


Of course you can get different scores!! Different tasks in the background and different memory will make different scores. How difficult must it be?

 

a test done by people doing brute force is no different than Edison, which is why there are individuals who truely understand science and concept are called geniuses, people like Einstein and Tesla.

those tests are simply general test and i dont expect them to fully understand cpu architecture or what resources software choose to use but even after explaining if you still dont get it then i can say no more.

 

But the same reviewers that got the wide variance of 8700k results got well grouped results for the 7700k CB test, that's the point - the 8700k tests are all at once badly done (top 4 scores), and have a wide variation when properly at stock speeds - assumed, there still might be bugaboo's to be found in BIOS configurations.

One score isn't a valid comparison in this case, the graph of all the reviewers on 7700k and 8700k CB results is more telling. And a variance of 1200cb - 1400cb is too wide to count on individual scores. Your CPU may not do as well as the top score, and more likely fall into the bottom performing Max Turbo category.

 

I don't think we need to continue if you won't look at the graphs and see the issues related to the results.

 

By this collection of CB results from a wide range of reviewers, taking off the top 4 scores for being non-stock, the remaining scores say that for the large part - 45% -, you won't get advertised Max Turbo, so I guess it's a good thing Intel doesn't guarantee it.

So, yeah, Intel won't sell a lot of them when people discover they aren't performing.

At least this "production" run has these problems. Maybe the next production runs in a few months will be better?

 

the idea is not about the graph and not about reviewers testing a "wide variety" of 8700k. all 8700k are designed the same way, difference would be silicon so if clocked the same, they yield no difference.

the term in IPC, once again, should technically be including the entire CPU which includes extra cache and faster IMC. if a software using the extra cache/IMC then you will see it being much faster than 7700k at same frequency per core, just how it is.

a person riding a consumer motorcycle at max speed would not mean if the consumer bike is replaced with a race bike. hardware is different, software needs to take advantage of that difference. in this case the software didn't take advantage of that difference thus no IPC gain. if it did, we will see IPC increase.

a nice example would be samsung SSD and its TLC cache to boost computing experience. hence me and @tilleroftheearth dont really like sammy ssd that much when it comes to steady state or consistent workloads. MLC ssd ive no problem with tho.

 

I think you may have gotten the point of the graphs after all

The 8700k silicon isn't performing, with about 45% of the CB scores being under the average of 1425cb - with a range of 1262cb - 1364cb.

We still don't know for sure if the 1425cb is totally stock, as this may all be down to motherboard makers interpretation of Intel's specs for "stock" defaults.

 

i dont need to look at it to understand the graph i mean most reviewers do graph the same way, but it looks like you dont understand what IPC really is tho. but i'll leave it at that.

 

Yeah, I don't like the results either, no IPC improvement demonstrates that this isn't really Coffee Lake architecture it's 6 core Kaby Lake, but we were already told that by Intel, the real Coffee Lake is out past 2H18 at this point, a date range which can change again.

Architecture changes compared to Kaby Lake
Coffee Lake features largely the same CPU core and performance per MHz as Skylake/Kaby Lake. [7] [8]
https://en.wikipedia.org/wiki/Coffee_Lake

 

I did a bench with [email protected] as stock 8700K. For joo bruh Aka tested with all time high BGA clocks.


This is from my older [email protected]


As joo can see... There ain't big difference. Then do your math with 50% more cores and with same IPC. I wonder what the numbers will be? Dan'g I can't do math. Too old You're welcome!!

Edit. Very strange. Almost on par with @Talon's [email protected]. How is this possible? I will try... 957+50% more cores = 1435.5cb. Is this correct math?

 

CFL is a refresh, so of course its same architecture. we should only focus on the changes and differences to see where the real benefits are. doing a review to show IPC increase imo is kinda silly but i guess many do wanna see it, me included

 

Strange. Almost same score for 4.3GHz. How is this possible?

Same score with a few things up.

 

all that matter is if that 870TM1 cpu heatsink comes with 2 fans and if I can get my hand on a 6ghz 5.2ghz which i probably could from caske for wallet rape at $1000 CAD... totally worth if for 2 fans cpu heatsink.

if its 2 fans then run cpu at 5ghz at say 1.3v will still yield to a decent temp without having to do max fan all day.

 

You can. The 4.4 beta is able to do so without any issues. The percentage is only the usage, which is kinda useless to know if you don't know the clock speeds. Like on Mr.fox video you can see how his CPU throttles down to 4.2ghz, if the same is happening to you because you undervolted to much, then the temps are less impressive, but still good.

 

That's okay, the 7700k cpu base clock is 4.2Ghz.
If the cpu throttles past 4Ghz (goes below) then you would have noticed. The game would have stuttered or freeze slightly.

I think this only happens when gaming on BGA...

You can clearly see my 1080 SLI GPUS on 1911Mhz and only hitting 49c max. Now that's impressive.

I checked last night by the way, with my cpu speed, it never goes below 4.2GHz when gaming

 

Not even in the slightest. Even if you clock your CPU to 3ghz it would run fluid.

Nobody is questioning your temps, regardless of the performance of the CPU they are still very good, even tho at a stable 4.5ghz it would be more impressive, comparded to mr.fox furnace of a notebook. That throttles down to 4.2ghz half of the time while maintaining 85-90c temps. You defnitely did something right thats for sure, the GPU temps are also really good even tho u get some SLI sync problems, probably caused by the poor SLI support of pascal cards in general. Games like Metro and witcher would probably make ur notebook hotter, even tho I don't think u'll get past mid 60s.

 

I'm in no rush, waiting for Ice Lake personally, 5ghz 8 cores 16threads and hopefully a nice little 10% bump in IPC. that will be my baby for a solid 5-6 years, so don't want to jump the gun just yet.

 

Same here, hopefully with volta too. I'm sick of intel's chase game...

 

Why isn't Intel getting more flak for this stunt?
https://www.reddit.com/r/intel/comments/75ib6t/why_isnt_intel_getting_more_flak_for_this_stunt/

"Why isn't Intel getting more flak for this BS paper launch stunt? Have they Officially responded to the lack of product that was shipped to retailers?
Why would you do the release when you had no product available? Really f'ing a lot of people over here who rushed out and bought hardware and are now just sitting on it."

For those of you looking out for 8700k siliconlottery is expecting stock by Oct. 27, if it helps any.
https://www.reddit.com/r/intel/comments/7620cz/for_those_of_you_looking_out_for_8700k/

" ImTheNigerianPrince 2 points 7 hours ago
It was Oct 20 a few days ago, so I wouldn't be surprised if they pushed it back again."

 

Last time I detail this thread. But proof that I put my money where my mouth is.


Sent from my iPhone using Tapatalk

 

You guys need to stop with the CPU clocks don't matter for gaming BS.
Even a 1060 Max Q is nerfed by a 7300HQ @3.1ghz in a number of games.
What's your standard of ' fluid' there?

 

woah you got high hopes there! icelake 10nm+ silicon is below 14nm++ as per intel's slide, so we may expect a 4.7ghz 8 cores instead (more cores = higher voltage = higher temps). 10nm++ is when we might be able to see more chips at 5ghz 8 cores become more of a norm.

double standard, he'll justify everything to general usage. but when picking out the bad benchmark, he'll go for the extremes, all for that 1 bad example to justify his statement.

 

And 7300hq will struggle equal as 6700hq (This with 1070). Same IPC and same max Turbo Boost for single core.

And i5 BGA, Oh'well

 

Yeah, my desk can look similar only way less tidy.

Nice ball mouse by the way

 

Where have yoo been? I'm sure you will find more if you starts to use the search button.

 

Thanks. That’s from the customer I visit on each Friday. I love my trackballs and I so want the new Logitech one that came out but I can’t complain about what I already have.


Sent from my iPhone using Tapatalk

 

Yeah no way I’m going back to BGA for performance. For a surface pro or equivalent sure but nothing that is going to be pushed to the limit.


Sent from my iPhone using Tapatalk

 

Power concumption numbers for stock clocked 7700K vs. 8700K (Wprime 1024M Stress test).

[email protected]


[email protected]

 

2 more cores, 100mhz less with 90mv less, excellent silicon. add an additional fan to that 870TM cpu heatsink and we have a deal.

 

I can see why some people love them for some kinds of work, probably a lot more comfortable than a mouse for extended periods.

 

Yeah I also do game on it as well from time to time (Diablo 3 mostly for the trackball). It is great when I don't have a lot of desk space.

 

We should see samples in a week or two. MP is setup for middle of November. Check our website future section for update. Heatsinks are improved and some are redesigned.

 

Stock delidded cpus?

 

@Eurocom Support , @Donald@HIDevolution
Hi Guys,
I wasn't sure which one to contact so I tagged you both. I'm willing to volunteer my time to test the new Coffee Lake P870TM1 model. You don't have to send me a crazy expensive configuration. Just an 8700K, 1080 SLI, #3 2TB 960 Pro's, a couple 4 TB 850 EVO's, and an unlocked BIOS, the wireless card isn't a big deal, I'm easy, I'll take any one you have. I'm a perfect test subject as I don't really know what I'm doing, so I'm sure you'll get this machine back a half dozen times to fix it and see what happened. I know donating my time is a generous offer but I'll do what it takes to help. You can PM me for my address. I forgot about RAM, 32Gb is fine

 

Are you seriuos!?

 

ABSOLUTELY! My generosity knows no bounds

 

How about you send me that P570WM for testing. I'll drop a 12-core Xeon and the upcoming 1080 MXM gpu in there.

 

That would be a bad a** set up. My six core cranks through my Architectural software better than my 7700K. That 12 core Zeon you had must have been a beast. Will the p570wm handle a 1080?

 

need to see if theres extra fan!!