Clevo + Coffee Lake: Status?

45w 6 core Intel BGA will struggle being comparable. Junk is junk as you know. But I'm quite sure a good binned oc'd Coffee desktop version should kill a fully locked down 8 core R1700 in Asusbook. A unlocked 8 core Ryzen R1700(X) in a Clevo is another matter.

 

Well, I don't think "kill" is the right word, maybe "playfully nibble at it's heel's" might be more apt

 

i'd agree papusan. 8 core locked ryzen is a no go for me, even if it defaults to higher clock and thats really the kill joy isn't it. to have the chip so highly factory clocked leaving so little left to overclock because of silicon limit.

a 6c intel cpu at 4.8-5ghz would be able to compete with Ryzen 8c at 4ghz, if ryzen can't be clocked then it'll be around the same while ST performance ryzen being way behind due to frequency difference.

look at this WTFFFF, look at 8700k (i still think its 4.7ghz not 4.5.. 4.5 for score 218 ST is too good to be true). i'd take this with TWO grain of salt because 1. it is from CPU monkey ffs fml. 2. CPU isn't out yet and that score is too good to be true, knowing how intel slowly milk us, its hard to believe an optimization 14nm++ arc will give that much boost. kaby to skylake had no improvement but also due to same design/cache etc, where as 8700k has a lot more cache so that could be the reason. @Mr. Fox @Johnksss@iBUYPOWER @Meaker@Sager @TANWare @ajc9988

 

As Intel only can offer 6 cores for mainstream, maybe they had to do something drastic due AMD and Ryzen who have 8 core chips. See also the CBR-15 multi threads score.
More like [email protected] http://hwbot.org/hardware/processor/core_i7_7800x/ See scores with oc'd 4.6GHz.

This mean minimum 4.5/4.6GHz on all 6 cores if those results is true. Or heavily ipc increase on top of high clocks for Coffee. Which I don't think is possible.

Edit. Only high clocks + high ipc increase can hinder AMD from taking a huge bite of the mainstream class!!

If Intel barely can only offer R 1600x performance they will loose the battle due AMD already have 8 cores chips in their arms for the important mainstream!!

 

still waiting on new CL laptops

 

Great points, but I'll throw some sand in the analysis. X and Kaby score around 190 on that test. The slides leaked show 11% improvement in single threading. If you truly get that, that is about 20 points, which gives you about 210 points on CB15.

But, they also say you only get 51% MT. You have 50% more cores. This means it shows MT and HT scale poorly on Intel versus AMD's SMT, which we already knew (need a couple hundred MHz to go beyond AMD SMT at 4GHz, but Intel can OC hundreds of MHz beyond that point, so still gets like 10% more performance on MT, not discussing software optimizations at the moment). But, considering in real life you only got 4-10%, usually towed the lower end and dependent on program, it isn't unreasonable for them to get gains in CB while not seeing that translation to other uses.

I do think they will cannibalize the 6-core skylake-x here and it will outperform, but no quad channel for those that need it and the same PCIe count as Kaby. I agree with the speeds from @Papusan, roughly, based on the high heat of Kaby and Skylake-X. I think it will beat AMDs 1600, but I don't think the value will be there to fully overcome AMDs price/performance ratio. But, we don't know enough on 14nm++ yet to give certainty, other than they presented it as the same size jump as skylake to Kaby. So, just have to give it time.

Also, i3 is getting quad cores with no HT. We are seeing the death of the dual core. This means programmers will need to change their choosing to accommodate multi threading for higher core count SOON. We will have quad, hexa, and octa-core CPUs filling out the mainstream on both sides. This is getting good.

I also want to note that the six core coffee has been planned for years (2014 or 2015 it was known to have 6 cores). The reaction is jumping around the release and potential price points. But heat will determine Clevo's implementation and potential OC. If they actually stick to the power envelope on core clocks, you could look at a six core all core OC at 4.5 on a laptop (without AC). But, we'll have to wait to see (could be higher, but the heat on skylake-x makes me want to be conservative).

 

the biggest issue with AMD at hand right now is 1. lower clock. 2. iGPU. coffeelake if priced correctly just as previous gen i7 like 7700k then with it's high frequency peformance with 6cores and a decent iGPU it'll blow AMD's 1700x or 1800x away because no graphic unit, well only if OEM sees intel's new i5 as a potential chip in mobile platform which i have no doubt seeing it coming. AMD needs to bring their APU with ryzen and vega ASAP.

majority of enthuiast cares but not for most, however mainstream people might listen to their enthuaist friend blabering on and on about intel. i'd recommand my friend or anyone to go buy AMD, but if i know they are super into ST performance and know what they are getting themselves into (yearly upgrade, chipset limitation, last gen 14nm) then i'd agree them getting intel.

also i still think ryzen's 8c won't scale as well as intel 6c, well 8c in general don't scale as well as 6c. since we are not talking about software optimization or TSX being used here then even infinity fabric can't run away from scaling.

my only concern left is that, new coffeelake laptop will have similar issue with p870km1 which you can't use the iGPU.. even if you remove discrete GPU still can't use them, thats BS

 

[QUOTE="ole!!!, post: 10586494, member: 454055]also i still think ryzen's 8c won't scale as well as intel 6c, well 8c in general don't scale as well as 6c. since we are not talking about software optimization or TSX being used here then even infinity fabric can't run away from scaling.[/QUOTE]

W0t???

That's the first time I'm seeing someone saying AMD's architecture scales worse than Intel's monolithic architecture.

AMD's Zeppelin chip is extremely versatile while Epyc and Threadripper have proven that the whole stack works extremely well with fast RAM, an improvement hardly seen in with Intel Core and Xeon. Infinity Fabric is also fast and effective.

If what AMD promises is true, they're effectively one-upping Intel's architecture two years early, meaning we may well see Zen 2 and AMD's 7 nm arch next year completely leapfrog Intel's 10 nm and the pathetic attempt at milking one node for four years, i.e. Covfefe Lake. Cannon Lake was supposed to be out this year, and it was a major node following Skylake. Intel has been delayed by lousy innovation, anti-competitive practices, etc. Instead we have Ice Lake.

Intel has 2 things to do: innovate, and innovate cheaply. Because even if Intel has the better product in terms of raw performance, if anyone values his/her money, they'd go and get an AMD system.

This argument not for notebooks, because we haven't seen AMD's Raven Ridge APUs yet.

 

im talking about cores scaling and sharing resources. imho 8cores and lower prob wont be affected as much as more than 8 cores.

 

@ajc9988 heres tommy's ipc for 7700k and 7600k at stock.

7700k at 4.5ghz, 7600k at 4.2ghz if we do just match, per 100mhz we get 4.377 & 4.333 CB15 ST score respectively. the difference is only about ~0.0444 per 100mhz and we can safely assume that difference came from the extra 2MB L3 cache on 7700k over 7600k. 8700k has 4MB more than 7700k so double of 0.0444 per 100mhz we get about 0.088 ish, times by 45 assuming 8700k is clocked at 4.5ghz we get 3.96 extra score from just extra 4MB of L3 cache. add 3.96 to 197 we get about 200 to 201.

now theres many issue with this math as its pure 100% scaling and thats assuming this is really how number of MB cache relates directly to CB ST performance. this will be my guess since i have nothing else to go about other than 7700k and 7600k.

earlier chart i posted shows about 200 and 218 for 7700k and 8700k, toms shows 197 for 7700k thats a difference of 3 points basically 1.53% for 7700k, if we lower 218 by 1.53% ish we get about 214-215. now going from 200-201 to 214-215 is a decent jump, though definitely not 11% but the extra L3 cache already factored in and theres still a noticeable difference.

INTEL DEFINITELY HELD BACK AND MILKED US

 

Intel is upgrading KBL chipset to Z370 just to support 6 core CFL LGAs now.
If they could have done that on current Z270, they would have.
This change in roadmap was caused forced by AMDs move...they got used to milk the cow, but this time they had little choice...

Otherwise we would have gotten only KBL refresh LGA CPUs this year and the 'proper' CFL next year.

 

Caused by AMDs move. Rather say forced by AMDs move... Maybe it's same, but my English

 

Clevo cuts the iGPU from use in hardware (no leads for it), so all related to the iGPU is moot from the start for this thread. They haven't used it on the high performance laptops since I started buying them with the HM series. It prevents all optimus based B.S.! On desktop, even then, many have dedicated cards now that beat it and use those. But I'll give you the clock point, even though it is clock and IPC that determines performance (Intel currently leads in both, with the most performance seen in single thread because of poor scaling).

In this sense of scaling, are you referring to the performance per core by dividing MT score by number of cores, or are you talking raw end score for MT? The first, you may be right with coffee if they made improvement, but you have to control for IPC and clocks to test that scaling. The second, in raw performance, I'm calling software optimization, not a raw win. So, I need to know which is being referred to.

 

this scaling. http://images.anandtech.com/graphs/graph6290/49955.png

note this is for HPC workload and well optimized software that takes advatnage of many cores, and MOST likely intended for intel's cpu other than IBM/optereon. (because intel was the dominated cpu in server space).

i am assuming that all cpu, architecture will get affected by core scaling, though zen's design and IBM new power9 might be an exception. intel will definitely be affected unless software supports TSX and the chip also has TSX. i donno know if AMD has TSX but if its a consumer software, going by that chart we can semi guess that 6 cores scale better than 8 cores.

@the clevo talk, thats bull crap. i want to use IGPU instead of optimus or w/e its called.. unless it is more power efficient than intel's junk iGPU, anyone know which is more efficient? also aside from efficiency, i'd want iGPU to work because if discrete goes dead then computer won't boot.. iGPU has its own purpose.

 

Optimus simply uses the IGP, the desktop chipset has always been the line where the IGP is never enabled. It does keep the design and BIOS more simple to develop.

Is the added cost worth it for the platform? I'd say it's arguable.

 

Routing display output through the iGPU lowers performance by a few %, increases gpu to display lag, and kills any possibility of SLI or GSync.

It would be nice to have a cooler running unit at idle that gets more battery life but not at the expense of gaming performance and technology, and if what optimus offers is so important, the 'second tier' units (P670) offer it.

I did enjoy the 3+ hr battery life with optimus (and had zero problems with it in 4 years) in my P170EM when I used to move around more.

CL six core will be interesting. If Clevo just drop it in the same thermal solution as the 7700K, I seriously doubt it'll realise the +50% from extra cores due to temps from +50% higher power draw

 

hold on so optimus is still using iGP then? then how come johnkss told me p870km1 cant use iGP does it not have that optimus feature? what if you only get 1 GPU without SLI? what john told me was even without any discrete GPU we can't use iGPU. do you know anything @Meaker@Sager

 

the P870XX models are designed to not use the iGPU because they're marketed as high-end devices and clevo knows that enthusiast gamers don't want optimus so they design the board to bypass it.

 

not cool mang.. leave the option to us it has it's usage even for high end enthuaiast

 

No feature is free, especially on a custom designed project like desktop CPUs in a mobile form factor.

 

I'd only want it to see if windows 10's multi gpu support actually works with an Intel gpu and the dedicated nvidia gpu.

Sent from my SM-N910G using Tapatalk

 

That sounds like an entire pile of headaches...

 

It's not a requirement, just a curiosity.If it doesn't work then no big deal.

 

shouldnt computer mobo default to use iGPU if it doesnt come with onboard vga? so we're looking at them not routing connection from display to iGPU but to dGPU instead but isnt it kinda pointless i mean iGP is on the CPU so it should be easy to do it, i mean everything on the mobo is connected to the cpu it would seem more work to purposely not have it route to iGP.

 

You are talking a mux since it's the same connector which means BIOS switches, an extra part to the PCB and driver/firmware testing.

 

so when did this start? clevo started non HEDT since 4790k then 5775C then 6700K finally 7700k. did no iGP started since haswell?

 

No desktop based CPU system has ever had it.

 

Intel® Core™ i7 Desktop Processors with Intel® HD Graphics
https://www.intel.com/content/www/us/en/support/processors/000005530.html

 

I think he was referring to no desktop CPU laptop system not that the CPU's themselves don't have it

 

Thanks, I tried to keep myself alert by reading NBR while waiting earlier, it's tough to get enough context when answering by phone, that's why I rarely do it.

 

There are many reasons not to, including heat production. Also, why do that when every unit sells with a dGPU. If there is never a unit sold without a dGPU (aside from bare bones, which means you'll buy one), there isn't a need.

 

Did you not stop and think that not knowing an Intel mainstream desktop CPU has IGP would be a bit odd for me?

 

well thats ashame boys, so to counter the issue with GPU failure only option is to get dual GPU 2nd GPU as backup incase primary fails lmao. unless the bios got something in it that can use 2nd GPU pcie slot as primary to boot without having to swap GPU.

 

Not when reading your wording:

If you said "No desktop CPU based laptop has ever had it", that would have been clear.

I answered what you said.

Besides the link has good coverage, whether it's in a laptop or not now, maybe it should be? With all the complaining about battery life, why not stick a mux switch in there and support igp operation on battery only, definitely don't introduce Optimus.

 

What I am saying is, you should have read that and gone "really, for meaker that's a really odd gap" and perhaps read into the context before picking out a single post. It should have been a flag to read into the conversation further it's usually a good idea to check before trying to correct me. I'm not perfect but I'm more often than not correct

In this case I am a representative of sager posting in the clevo/sager forum as well so very likely talking about our systems which is a filter to keep in mind too.

 

if thats the case... fess up when will 8700k come in a laptop yeah?

 

Yes, inquiring minds are curious. Spill the beans.Please.

 

Yes breaking NDAs always goes down so well with companies

 

So long as new laptops are available in a few months I'll be happy.

Sent from my SM-N910G using Tapatalk

 

so 8700k out early october, means laptop out possibly and hopefully by christmas.

forever wait

its probably worth it, you should try!

 

Ugh?
WTF! P870KM dead in less than half of a year?!!

 

Welcome to the ZM pain (did the same thing).

 

naw, more like 10 months, because we might not see the new clevo with 8700k available for at least another few months i'd say, cpu wont be out till oct so nothing is for sure.

also, theres always new tech and its no surprise chipset/socket change with intel should be a normal thing now. its not like people have to upgrade yearly for the best lol.

 

And yet you guys circlejerk around the LGA socket so much, only for Intel to screw you and your wallets in the rear end

I think implementing MXM in notebooks will be a much better future-proofing move rather than using desktop LGA CPUs and chipsets, unless we see Ryzen properly come to notebooks.

Basically I'd like to see the opposite of the W650KK: BGA HK CPU, with an MXM 3.0A/B slot.

 

Oh you should have seen the rage at suggesting such a thing when we went from mobile socket to BGA, people much prefer purchasing multiple $1000 extreme editions and binning for the best one

 

I don't exactly have the time nor money to spend on tweaking for the maximum—not very worth it, in my opinion. It's like paying $250 for a performance boost from an iGPU to a GTX 1060, and then paying $600 more to get to a 1080 Ti, which is not exactly 4x as powerful as the 1060...

 

I'd like to see an HK system done with a desktop class power delivery system and vapor chamber cooling. The smaller footprint should make a heatsink easier to produce.

 

i can live with an 8XXXHK.

 

Any chance of 2k screens in the next generation of laptops? That's my one real wish. 2k IPS screens.

 

You do know 2K is 1920x1080 (1080P) unless you are referring to the native 2K of 2048x1080... Which won't make sense... Unless I'm Missing something??