I wouldn't wait. Regardless of process shrink, each generation is 15-30% faster on Intel's selected and skewed benchmark and 3-10% faster in each product slot on most legitimate unbiased benchmarks. If you wait, there is always something better coming.
NV GPU's are worth waiting for I guess - they're very predictable in their increase so you can work out ~roughly how much performance you want at a given price/power combination and what year you're likely to be able to get it.....
Nah. Pascal was a huge jump in performance, but it had a new process and a significantly updated architecture. In Intel terms, it was both tick & tock. Volta won't have a significantly new process, and it appears it will have less architectural changes too.
In other words, it'll be less like Maxwell -> Pascal and more like Kepler -> Maxwell.
Nvidia 6 series, 7 series were pretty much the same except for 780 ti/titan, and 9 series was only a 20-40% jump in performance where pascal was a >50% jump on several cards.
Since GV100 is the largest die of all time, with quite some margin over the second largest one, and since it is actually still at a ~15nm process (compared to TSMC's 16nm), a fully enabled die should be the greatest power hog of all time. That, of course, will be addressed by plenty of "dynamic on/off switching" (dark silicon), and by disabling the tensor cores and some CUDA (shader) blocks in the consumer versions. The fully enabled GV100 for professional cards should still be burning power like crazy though, and most likely require a liquid cooler. Unless TSMC did wonder with their 12FFN process.
I am not sure if Nvidia plans to design smaller Volta GPUs for consumer cards. Perhaps that would be too expensive and not worth the bother. Or would it be more expensive to waste such large dies on consumer cards? In any case GV100 at that process node is at the maximum possible die size. It cannot get any bigger. Which is why Nvidia is thinking about taking the AMD CPU route and using 2+ dies on the same package. They will be forced to do that if by the time Volta's successor is to be developed (at the moment it should be at the preliminary design stage) a 7-10 nm process is not viable.
I am still using Sandy Bridge-E from like 5-6 years ago.
I just don't see a point in upgrading right now. She can clock to 4.8ghz with the push of a button, 5ghz if I push the volts. Which makes it very competitive with Ryzen's 6-core chips anyway.
With my 1866Mhz DDR3 modules, I have more bandwidth than Dual-Channel DDR4 3200Mhz.
Still runs every game and app I throw at it flawlessly.
Has PCI-3.0, USB 3.0. Only thing that is a let down is that it only has Sata 6Gbit rather than 16Gbit. And no USB 3.1, USB C. - But you can live without that, not worth the rush to upgrade.
Maybe in a year or two I can finally get a 10-12 core processor for a good price and my 3930K can finally be left to rest.
It's insane how long this rig has lasted me and it's still got a ton of life left in it.
Aye, I'm on an older i5 set up and with all the hype you'd think I should be crying out for upgrading but simply put there's a reason new PC sales are stagnating and it's because there's just no need to upgrade. I'm on DDR3 with high performance modules like you and see no reason to shell out for DDR4 - I'm going to notice zero appreciable difference. As you say, missing out on the very latest USB standards and my older GTX 780 is now really the equivalent of a fairly mid range card (they say an RX 480 has about 125% of its performance) but that's a drop in upgrade I'll do when the time is right and certainly doesn't warrant a whole new system.
For gaming I kind of blame the consoles for holding things back. With their hardware lifecycles lasting so long in comparison to PC GPUs it means devs just don't really seem to be creating stuff that really pushes the latest hardware as they're developing cross platform and the PC market is usually the smaller one, so why put extra effort into it? So we end up with cards with insane performance on the latest games with things like the 1080 and older cards choochin' along just fine for much longer. It's probably a similar story with CPUs - they can't make games which require masses of CPU horsepower to run well because that would compromise performance on consoles.
Stevo's post actually suggests that even if they did push hardware requirements in games, hardware performance has not progressed sufficiently to meet that demand (at least in CPUs). I suppose GPU's are still moving at a reasonable pace (though not as fast as they once were) and developers could make better use of the resources available. However, the fact that you can still draw a valid comparison between a third generation core processor and an 8th generation core processor at all should tell you how little progress CPUs have made on the performance front. Now, if you want to talk about power efficiency and cost (manufacturing, not necessarily retail), that's a different story.
Yeah fair enough. I didn't read it like that but on a re-read I can see how you'd come to that conclusion. I was referring more to GPUs and being held back, the CPU comment was a poorly thought through afterthought and you're probably right. As for GPUs, looking at the FPS you get out of the latest games with a 1080 these days it's just silly. I remember when we were using the Crysis games as a benchmark and GPUs were struggling to get playable frame rates at high settings. I'm sure that with the lack of any real competition, Intel have been simply putting as little R&D as possible in to ensure something useful comes out, whilst adding in the occasional extra "feature" like the next version of HDCP and changing the socket so as to screw over their customers. I'm not a fan of any particular brand and I certainly won't advocate for one product over another based on anything other than objective criteria.... but I really don't like the way Intel does business. I don't know if AMD are genuinely any better but to me they have a better company image. Maybe it's the whole "liking the under dog" thing.
Interested to hear what you think about manufacturing costs. They're making out that it's getting harder and harder with yields getting lower and lower with each process shrink yet you seem to imply that this isn't the case and CPUs are getting cheaper to make?
You are correct, manufacturing has gotten harder. This was particularly apparent with the Broadwell line and may show up again with Cannon Lake. However, die sizes have also gotten smaller. Outside the ridiculously priced low volume Broadwell-E i7-6950X, prices have largely remained stagnant between generations. Combine this with the fact that Intel's margins have remained lofty and it suggests that the overall cost to manufacture has remained somewhat stable despite all the troubles. Also, keep in mind that yields improve over the life of a process. Now that Intel is sticking with the same process node 3 and 4 generations, it would seem that costs should be dropping for the same sized chunk of silicon using the same or extremely similar manufacturing methods. I acknowledge that things are rarely as simple as they seem, but complexity is usually at odds with reducing cost, so I'm probably painting a pretty conservative picture here.
As far as 14+ and 14++ go, they strike me as marketing terms needed to counter the other foundries' so called smaller processes. Intel has always made improvements to their process over time, they just have more time to make additional improvements this time. Just like every other time, I expect yields will continue to improve despite (or because of) the changes.
@StevoLincolnite: "It's a Sandy Bridge-E chip. It's actually 2nd Gen Core. :P"
I was thinking 2nd Gen, but my hands were typing 3rd Gen. In any case, it just makes it that much more impressive. Kudos to you for getting the right system at the right time.
I upgraded from i5-2500K to i5-6400 (non OCed, cpu died) and have to say the performance is the same. These days it's more about features you can get (access to) than raw performance... old board had 2x6Gbps + 4x3Gbps sata, new one has several PCIe M.2s, all SATA ports are 6Gbps etc. Regardless, if the old one didn't die, I'd still be using it.
@StevoLincolnite: "With my 1866Mhz DDR3 modules, I have more bandwidth than Dual-Channel DDR4 3200Mhz."
Also, at lower latency IIRC.
@StevoLincolnite: "Only thing that is a let down is that it only has Sata 6Gbit rather than 16Gbit. And no USB 3.1, USB C. - But you can live without that, not worth the rush to upgrade."
Not sure what you'd plug into a SATA 16Gbit (SATA Express). I haven't seen any drives use it. All the M.2 SATA drives I've seen use 6Gbit controllers even if the interface should be capable of supporting faster. I'd just grab a PCIe to M.2 adapter card and throw a PCIe M.2 drive on it. You should have more than enough lanes on your platform to support this and the interface will be faster than any SATA standard.
They do have USB 3.1 adapter cards. Again, I see no reason you board wouldn't have enough PCIe lanes to support this should it become desirable in the future. No need to live without it if you want it.
@StevoLincolnite: "It's insane how long this rig has lasted me and it's still got a ton of life left in it."
Yeah, maybe more than you've considered. You could probably ride that system until something breaks without it feeling shamefully outdated (assuming periodic video card upgrades).
I was just listing off checklist features that I don't have... And you are right, USB 3.1 cards are available. And I do have plenty of PCI-E Lanes. Back with Sandy Bridge-E, Intel didn't castrate the amount of PCI-E lanes on the hex chips.
At 4.8Ghz-5Ghz... I highly doubt Ryzen 7 would beat my 3930K. I can overclock my CPU by over 50% remember. ;) Or at-least the difference would be small enough not to warrant an upgrade.
You do have a point on the power consumption side of things, but is it worth buying a new CPU, Motherboard and Ram combo? With the Price of Ram at the moment...
And my DDR3 Ram in quad-channel still ends up faster.
Better off just waiting until something is available at the right price or until the system breaks.
^This. Why wait for a 2% performance increase, especially when given Intel's track record, the platform won't support future chips anyway? I'm still rocking a Sandy and Haswell, no plan to upgrade either for a few years.
If anything is making me reevaluate my decision to keep my 4670k for at least a couple of years more it is the Ryzen lineup and the shakedown it is causing right now. New Intel CPU might be interesting again soon.
Waiting for a laptop can still makes sense given the power savings from generation to generation. It might not be huge from the CPU alone, but overall systems are still usually decreasing it.
If a big launch is just around the corner (a couple months perhaps) then waiting *might* make sense. But when you're talking about a launch year(s) down the road waiting only makes sense in a couple of scenarios. One would be if you don't have/can't spare the money to upgrade, and thus are waiting for non-technical reasons. Another case would be if what you're using is already good enough and thus you're not truly "waiting" at all. Or some combination of factors like this - you'd LIKE more horsepower/lower power consumption, but you don't really need it and/or can't justify it.
Pfft, my main PC at home was a Phenom 2 rig that lasted me damn near 10 years. Granted I don't game much, skyrim being the most demanding thing I ran on it. I wish I'd made it to Ryzen release but alas the mobo died on me and I switched to Intel at the end of 2016. I'm the same with phones, I usually get 4 or 5 years of usage out of them, went form Nokia N95 to Galaxy Nexus to Nexus 6 which I'm still on.
The only reason to wait is that RAM prices might decline later in 2018. Although I read a recent support saying that the NAND shortage will extend into 2019 potentially. Right now is a particularly bad time to build a new PC because of the double whammy of inflated RAM and SSD prices + inflated GPU prices due to the mining craze. Hopefully things will settle down sooner rather than later.
I hope so. SSD pricing has been flat (which is still absurd given density increases), but RAM costs have actually gone up quite a bit and GPUs obviously are through the roof. Just waiting for Eth to go Proof of Stake. Sigh.
What a prophecy! it's obvious that analogizing Intel and AMD in terms of their cap is not a true debate (which has been brought up in other comments)! in the other hand, siding with Intel, as the giant of the semiconductor industry, by mentioning its potential capacities/abilities seems like blindly following the stronger! however, the hurry-scurry announcements of Intel - at least - show what Intel could've produced but didn't. and what it means?
Kinda what I was thinking. Why would Intel tip its hand that far down the road if it wasn't an attempt to tell investors that they "still got it." AMD should honestly never be able to compete with Intel in the first place. Intel has a 14x higher market cap. The amount of resources at their disposal should have them so far ahead of anything AMD can produce. Intel has stumbled on process advancements, which has been their strongest advantage.
More like they didnt really NEED to invest hard again. They could milk the cow. Just like they love to force complete motherboard changes almost every iteration.
I would say it's more likely they used their resources to diversify outside of their core strength. How much was spent trying to make Atom a viable competitor to ARM?
Who says Intel isn't that far ahead? Honestly there IceLake design and 10nm likely have been tape in ready for a while now. If you have no competition why compete with yourself? Just keep your new stuff in your pocket until you have to pull ahead in the race again. At least AMD is forcing Intel to move forward again. I build myself a Ryzen R7 system so no bias just reality.
The ideal time to upgrade always changes. - 2018: Coffee Lake (perhaps Ice Lake, we'll see) - 2018: nvme 1.3 drives (let's see if they matter) - 2018: nVidia Volta gaming cards - 2019: PCIe 4.0 (probably not available in 2018, but we'll see) - ????: Prices of DRAM come back to earth (?)
Of course, process nodes, perf/$, perf/W and many other factors go into this.
I was really waiting for ThreadRipper and it didn't disappoint. Yet, I want to see the Intel September releases.
The only thing holding me back is PCIe 3.0, which is on it's last legs. My last system is from 2009 (using PCIe 2.0) and is doing fine otherwise (6-core 4.2Ghz), and I do feel the difference comparred to PCIe 3.0 systems I use.
Also, the DRAM prices currently are insane. Hoping the production/markets stabilize in 2018.
Perhaps 10nm Cannon Lake has been canceled, and moving straight to 10+ instead. Previous reporting has said that an early 10nm process may barely outclass the fully optimized 14++ process.
Otherwise the generation naming just doesn't make any sense.
Well yeah. But ain't Cannon Lake just plain 10nm die shrink of 14nm Coffee Lake(heck coffee lake is just optimized sky lake). Like Broadwell was to Haswell. Maybe it's exactly like Broadwell: short living low tdp processors with iris pro igpus. Great for htpc users, but mainly meh for anything else.
It is possible that intel might end up using both coffee lake and cannon lake chips in their 8th gen, 8xxx lineup; perhaps CFL for desktop parts and CNL for low-power mobile parts.
You can certainly sense the urgency from Intel's camp of late. But to be honest, there is no point looking 2 generations ahead since thing may change over the next couple of years. Bring out Cannon Lake, then we talk about the next gen.
The issue is what you're interested in. If your interest is ONLY in "what CPU can I buy in the near future" or "how can I support my corporate tribe", then what you say is correct. But look at the big picture beyond those. That picture shows an Intel that is breaking down before our eyes.
It started with Broadwell which shipped late and lame (in spite of Intel's protestations beforehand that yields were the most awesome ever). Lying about how well their process is doing seems to be the new normal (witness Optane, still MIA from the market where it actually matters, and being sold into markets where flash is a much better choice).
Meanwhile architectural innovation seems to have ground to a halt with Skylake (it actually stopped with Sandy Bridge, since then they've been treading water, but now they're not even trying --- hence once Lake after another). Now we have this process-based split across Kaby, Coffee, Cannon, and Ice that looks like nothing so much as mad scrambling. It's not that having separate designs and processes for server vs desktop vs laptop is a bad idea; but these do not look anything like separate DESIGNS, more like just separate "WTF can we do to get SOMETHING, ANYTHING that works?". There's no evidence here of design in terms of things like giving Cannon a differently balanced microarchitecture that's a better fit for its target space as compared to say Coffee.
So yeah, it's interesting from the point of view of "How does Intel's monopoly end?" Ten years ago was just before iPhone, just before Nehalem. A lot can change in ten years. So the question is what changes over the next ten years? What all this naming nonsense suggests to me is that Intel is even less prepared for the future than they were for mobile. At the time I suggested that Intel's mobile attempts were a pointless waste of time, based not on dick-measuring how one Intel core stacked up against a competitor ARM core, but based on the fact that Intel's STRATEGY made no sense, it revealed a corporation that was either (perhaps both) deeply incompetent or deeply deluded. I see the same fundamental strategic failures now at the mainstream level. Sure things look fine right now, but we're not talking now, we're talking about 10 yrs from now and how well Intel is poised to cope with the future.
Seriously? I comment that some people's primary interest in this news is "how can I support my corporate tribe" (not that there's anything wrong with that, simply pointing out that that that's not the ONLY lens of analysis) and your response is to talk about "crapple". I kinda rest my case.
My understanding is that several years back, Intel decided to adopt a focus on hiring & promoting for reasons other than being the best at what you do. For a company that relies on continued superior engineering to maintain its advantage, that doesn't seem like a good thing - at least in terms of the quality of the output product.
And do you have a source for this beyond Fox News and James Damore?
If we're throwing out wild theories, my wild theory is that the problems all started when America First sentiment mean Intel could no longer hire the best possible engineers from the entire world.
It ends when arm support on windows is passable . Microsoft is working on ARM for windows with some kinda of emulation in hardware by the ARM parts to do x86/win32 apps. https://www.xda-developers.com/microsoft-qualcomm-...
Once the performance is good enough and More ARM vendors come on board, it will be the end of Intel , they can't compete on price with arm.
Intel's going to have to either advance the state of the art (something it certainly did not do with Kaby Lake) or it's going to have to learn to live without a 60% profit margin.
You can't price gouge and stand still at the same time.
Well first of all - Intel can not have a monopoly when there is all those ARM based tablets and phones out there - Intel try to get into phone market - but obviously ARM has a monopoly there.
As for Windows on ARM, well it did not work with Windows RT. And no technically sane person would expect it work with emulation on ARM. I don't think Microsoft is having good success with Windows on ARM. If they did they would have used ARM chip on HoloLens.
Intel is fully aware of ARM monopoly on the phone and some ways tablet markets. That is why they invest millions of dollars in research like in this article with Ice Lake. Fanless computers are where is going and expect more powerful version.
Maybe we will see coffee lake and cannon lake focus on different markets or coexist in roughly the same generation like we are seeing with Skylake and Kaby.
I think, based even just on the slides above, it's clear to me why Cannon lake is on 10nm but coffee lake is on 14+. 10nm saves on power and density, but 14++ looks to have much higher performance. I know which I want for my desktop and which I want for mobile, personally.
Of course with 10nm chips - maybe not Cannon Lake - higher performance will likely exceed 14nm chips like Coffee Lake - but Intel's appears to concentrated on Mobile than Desktop chips. This logic because 10nm means also lower power.
Same. My 3570K is doing okay, but I know its days are numbered. I'm willing to wait another year for a substantial (more than 10%) performance bump, but not for another incremental step forward.
I admit i don't know much about manufacturing these, but i do know marketing and this announcement seems to show they already are making them to some degree. The only reason a company announces something they normally don't is 1. To one up the competitor to show its still in the game. 2. It already has the product but is waiting to see the response of buyers (buyers don't want more if they have extended stock to sell).
In the CPU world a lot of times they can build them... but they have issues building them in mass without a lot of bad chips. It can take a long time to get the manufacturing reliability up to a point where it is profitable. They have been making 8nm chips in small batches for a few years now... but mass market parts wont be available for a long looooong time.
I know you corrected but the other point is Ice Lake was tape in ready not tape out which just basically means the design is ready and then they do the tape-out to a physical part.
"Why is the laptop market splitting between 14++ and 10, and why is the desktop market not going to 10nm but straight to 10+?"
It looks like the answer is right there on the slide. It shows 14++ as being higher performance than 10, and at least equivalent to 10+. On the laptop side, the lower power consumption is probably worth the tradeoff in some cases, but not so much on the desktop side.
Its true, judging by their own scale we have to wait for 10++ to get a real performance advantage just because how well 14++ works. Intels fully optimized 14nm process is quite the beast.
You might well still be right, but that's performance per transistor and the smaller process will have more transistors. We'll see how the balance works out but I think it's safe to say cannon lake isn't going to offer much unless they bump core count again. Seems the times to buy are going to be either Kaby Lake or wait for Ice Lake.
I'm hoping with ice lake intel would finally stop their messy cores/threads practice and go for a straight 4C i3, 6C i5 and 8C i7, all with SMT enabled configuration; but since this is intel, probably not.
That would be a nice line up, 2c\4t celerons (where they belong), 4c pentiums, 4c\8t i3's, 6c\12t i5, 8c\16t i7's. That would definitely be a nice competitive line up, especially since its 2 years down the road. Whether we will see that or not is anyone's guess.
I just built a 6700k based system last year, and I'm a bit sad to see a Hexa core CFL coming at that price point within few days, though it's good for competition.
Thinking whether to upgrade to CFL or not, but after seeing this it feels like waiting out for a year might be much better and wiser, for this Ice Lake thing based on 10 nm +.
Should I upgrade now or wait? I am a moderately demanding gamer of titles like RoTR, GTA V, FC Primal/ FC4, Watch Dogs, BF1, AC Syndicate etc. I have a GTX 970 which is just okay in 17, obviously fails to keep up at 60 frames at near max even, so I have to stick with high settings.
I would wait. While the 6700k was released more than 2 years ago, the 8700k is only 10% to 15% faster clock speeds for 1 to 4 core tasks. You'd barely notice the difference in those.
Only once you get to 5-core and 6-core tasks would the upgrade be worth it. And even then you'd only be ~25% to ~50% faster in those few tasks. Wait out the video card shortage and get a new video card instead.
Intels 10nm process does not equal TSMC 10nm, unfortunately marketing was needed to catch up with Intel. So, TSMC 10nm is probably similar to Intel 14nm (each company decides on their own definition).
This reminds me of the frequency wars - now we having multi-core wars and you are right, this is a lot of marketing, There are rumors also that 7nm coming out - and that is probably closer to 10nm or even 12nm on Intel's side.
One thing it appears is that smaller the die gets, the harder it appears to create the process.
For me personally, I not sure of need for more than 4 cores - but that will change over time as all technologies with chips - when the first 32 bit chips came out - I never thought I need 64 bit. But just as 8 bit - went to 16 bit and then 32 bit - 64 bit is logical. I don't think we need 128 bit now - but who knows in decade from now - things might be change.
I also believe Intel's process allows Intel to put more cores on single die - instead combine multiple dies together to achieved this goal.
Hopefully Atom-like chips with 2W or 4W TDP will make a comeback thanks to the process node power savings. Apollo Lake can't be stuffed into tablets because TDP is too high. I'm happy with my Cherry Trail Windows tablet but a newer design with a faster GPU supporting hi-DPI screens would be great.
My guess is that possibly the Atom name would be retried but Intel will move it in to line similar to what they did with Core-M and Y series chips - People tend to put down Core-M and I am typing this on m3-6y30 based Intel Compute Stick - it actually quite fast compare to say my older original Microsoft Surface Pro
But you are right these chips would be hard to stuff into small tablet - but the price needs to be there to compete with ARM. Intel appears to have a lot of research - most of it is in embedded markets - like Quark processor which has even less power than the Atom.
It should be Sandy Bridge, Ivy Bridge, Haswell then Broadwell, the post have the "well" generation wrong way round.
Since we are still on the lake architecture, does that mean all we get is an 10nm Skylake? Cannonlake generation was suppose to be a new uArc. And we are straight to Ice Lake?
Meanwhile I am eagerly waiting for AMD with Zen2 on 7nm next year.
No Zen2 on 7nm next year. The process is very late and the cpu integration too. Apparently AMD want EUV at 7nm, quad patterning is too complex for actual AMD resources. Go to EETimes, there is a nice article on this.....AMD exec claims: "AMD is among chip designers getting an early taste of 7nm process technologies"
The bad thing about intel is that all their existing desing all the way to tiger lake (which was supposed to be a die shrink of Ice Lake) are based on the nehalem/sb arch, they were never intended to be scalable multicore chips with good production/yield ratio.
Zen is the contratry, is a design for the true mass multicore era for both supercomputing, servers all the way to consumers, all in just 1 design. Zen really screwed intel plans for the next 5 years.
10cores plus for consumers on intel side has proven to be a patch work for the panic mode.
Everybody has their opinions but if Intel wanted combine more than one of it larger core cpus together to make a massive cpu. they could do it. I think Zen is more of patch, by patching multiple cores in one package especially when it uses the PCIe bus internally to connect them.
It funny, most people don't see what really going on here - Intel's major threat is not AMD but ARM in the mobile industry - That is why Cannon Lake is for low power fanless computers and not huge desktops that most people don't use except a decade ago.
Besides the entire process switch between 14nm and 10nm, I more curious on architecture changes of both Cannon Lake and Ice Lake.
My thought which could be wrong - is that Cannon Lake is similar architecture of Coffee Lake, but using the 10nm process along with it advantages - but it sounds like it might be only limited to low power laptops
Icy lake will be architecture improvements and also intended for full line up of chips.
My hope is to actually have a Quad or Six core ultra low power notebook in fan less box one day - but that is probably Ice Lake.
One small note - just adding more cores - is not the only architecture change - new technology to make the applications run faster and better is also. I pretty sure that 10nm chips have AVX-512 for example - but there could be other changes that Intel has not disclosed.
"My thought which could be wrong - is that Cannon Lake is similar architecture of Coffee Lake, but using the 10nm process along with it advantages"
Why are you not sure? It has always been the plan from day one, years ago, for cannon lake to simply be a an improved skylake on 10 nm but because intel hit issues with that node, they had to release kabylake and now coffee lake on improved 14 nm nodes.
Ice lake is a bit of a question mark though. Intel always used a completely new name for their new architectures, sandy bridge, haswell, skylake, but this time ice lake's name still belongs to the Lake family.
This makes me think that it's still based on the same skylake architecture. I could be wrong though. Maybe intel likes Lake names and plan to use them even for new architectures.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
106 Comments
Back to Article
TEAMSWITCHER - Tuesday, August 15, 2017 - link
This is going to be a great upgrade point... New MacBook Pro... New Windows Gaming Desktop... Both is 2018. Waiting is the hardest part.goatfajitas - Tuesday, August 15, 2017 - link
I wouldn't wait. Regardless of process shrink, each generation is 15-30% faster on Intel's selected and skewed benchmark and 3-10% faster in each product slot on most legitimate unbiased benchmarks. If you wait, there is always something better coming.Qwertilot - Tuesday, August 15, 2017 - link
NV GPU's are worth waiting for I guess - they're very predictable in their increase so you can work out ~roughly how much performance you want at a given price/power combination and what year you're likely to be able to get it.....bryanlarsen - Tuesday, August 15, 2017 - link
Nah. Pascal was a huge jump in performance, but it had a new process and a significantly updated architecture. In Intel terms, it was both tick & tock. Volta won't have a significantly new process, and it appears it will have less architectural changes too.In other words, it'll be less like Maxwell -> Pascal and more like Kepler -> Maxwell.
firerod1 - Tuesday, August 15, 2017 - link
Nvidia 6 series, 7 series were pretty much the same except for 780 ti/titan, and 9 series was only a 20-40% jump in performance where pascal was a >50% jump on several cards.niva - Wednesday, August 16, 2017 - link
Only 20-40% jump in performance. Spoiled brats.Santoval - Tuesday, August 15, 2017 - link
Since GV100 is the largest die of all time, with quite some margin over the second largest one, and since it is actually still at a ~15nm process (compared to TSMC's 16nm), a fully enabled die should be the greatest power hog of all time. That, of course, will be addressed by plenty of "dynamic on/off switching" (dark silicon), and by disabling the tensor cores and some CUDA (shader) blocks in the consumer versions. The fully enabled GV100 for professional cards should still be burning power like crazy though, and most likely require a liquid cooler. Unless TSMC did wonder with their 12FFN process.I am not sure if Nvidia plans to design smaller Volta GPUs for consumer cards. Perhaps that would be too expensive and not worth the bother. Or would it be more expensive to waste such large dies on consumer cards? In any case GV100 at that process node is at the maximum possible die size. It cannot get any bigger. Which is why Nvidia is thinking about taking the AMD CPU route and using 2+ dies on the same package. They will be forced to do that if by the time Volta's successor is to be developed (at the moment it should be at the preliminary design stage) a 7-10 nm process is not viable.
Santoval - Tuesday, August 15, 2017 - link
edit : "Unless TSMC did wonderS with their 12FFN process."StevoLincolnite - Wednesday, August 16, 2017 - link
I am still using Sandy Bridge-E from like 5-6 years ago.I just don't see a point in upgrading right now.
She can clock to 4.8ghz with the push of a button, 5ghz if I push the volts.
Which makes it very competitive with Ryzen's 6-core chips anyway.
With my 1866Mhz DDR3 modules, I have more bandwidth than Dual-Channel DDR4 3200Mhz.
Still runs every game and app I throw at it flawlessly.
Has PCI-3.0, USB 3.0.
Only thing that is a let down is that it only has Sata 6Gbit rather than 16Gbit.
And no USB 3.1, USB C. - But you can live without that, not worth the rush to upgrade.
Maybe in a year or two I can finally get a 10-12 core processor for a good price and my 3930K can finally be left to rest.
It's insane how long this rig has lasted me and it's still got a ton of life left in it.
philehidiot - Wednesday, August 16, 2017 - link
Aye, I'm on an older i5 set up and with all the hype you'd think I should be crying out for upgrading but simply put there's a reason new PC sales are stagnating and it's because there's just no need to upgrade. I'm on DDR3 with high performance modules like you and see no reason to shell out for DDR4 - I'm going to notice zero appreciable difference. As you say, missing out on the very latest USB standards and my older GTX 780 is now really the equivalent of a fairly mid range card (they say an RX 480 has about 125% of its performance) but that's a drop in upgrade I'll do when the time is right and certainly doesn't warrant a whole new system.For gaming I kind of blame the consoles for holding things back. With their hardware lifecycles lasting so long in comparison to PC GPUs it means devs just don't really seem to be creating stuff that really pushes the latest hardware as they're developing cross platform and the PC market is usually the smaller one, so why put extra effort into it? So we end up with cards with insane performance on the latest games with things like the 1080 and older cards choochin' along just fine for much longer. It's probably a similar story with CPUs - they can't make games which require masses of CPU horsepower to run well because that would compromise performance on consoles.
BurntMyBacon - Wednesday, August 16, 2017 - link
Stevo's post actually suggests that even if they did push hardware requirements in games, hardware performance has not progressed sufficiently to meet that demand (at least in CPUs). I suppose GPU's are still moving at a reasonable pace (though not as fast as they once were) and developers could make better use of the resources available. However, the fact that you can still draw a valid comparison between a third generation core processor and an 8th generation core processor at all should tell you how little progress CPUs have made on the performance front. Now, if you want to talk about power efficiency and cost (manufacturing, not necessarily retail), that's a different story.philehidiot - Wednesday, August 16, 2017 - link
Yeah fair enough. I didn't read it like that but on a re-read I can see how you'd come to that conclusion. I was referring more to GPUs and being held back, the CPU comment was a poorly thought through afterthought and you're probably right. As for GPUs, looking at the FPS you get out of the latest games with a 1080 these days it's just silly. I remember when we were using the Crysis games as a benchmark and GPUs were struggling to get playable frame rates at high settings. I'm sure that with the lack of any real competition, Intel have been simply putting as little R&D as possible in to ensure something useful comes out, whilst adding in the occasional extra "feature" like the next version of HDCP and changing the socket so as to screw over their customers. I'm not a fan of any particular brand and I certainly won't advocate for one product over another based on anything other than objective criteria.... but I really don't like the way Intel does business. I don't know if AMD are genuinely any better but to me they have a better company image. Maybe it's the whole "liking the under dog" thing.Interested to hear what you think about manufacturing costs. They're making out that it's getting harder and harder with yields getting lower and lower with each process shrink yet you seem to imply that this isn't the case and CPUs are getting cheaper to make?
BurntMyBacon - Wednesday, August 16, 2017 - link
You are correct, manufacturing has gotten harder. This was particularly apparent with the Broadwell line and may show up again with Cannon Lake. However, die sizes have also gotten smaller. Outside the ridiculously priced low volume Broadwell-E i7-6950X, prices have largely remained stagnant between generations. Combine this with the fact that Intel's margins have remained lofty and it suggests that the overall cost to manufacture has remained somewhat stable despite all the troubles. Also, keep in mind that yields improve over the life of a process. Now that Intel is sticking with the same process node 3 and 4 generations, it would seem that costs should be dropping for the same sized chunk of silicon using the same or extremely similar manufacturing methods. I acknowledge that things are rarely as simple as they seem, but complexity is usually at odds with reducing cost, so I'm probably painting a pretty conservative picture here.As far as 14+ and 14++ go, they strike me as marketing terms needed to counter the other foundries' so called smaller processes. Intel has always made improvements to their process over time, they just have more time to make additional improvements this time. Just like every other time, I expect yields will continue to improve despite (or because of) the changes.
tienchien1 - Sunday, August 20, 2017 - link
What's your job?In business, anyone will do as intel. We are not gods.
StevoLincolnite - Thursday, August 17, 2017 - link
It's a Sandy Bridge-E chip.It's actually 2nd Gen Core. :P
Here is a comparison between Sandy Bridge-E and Broadwell-E. (Sadly Anandtechs bench doesn't let me compare against Skylake-X.)
http://www.anandtech.com/bench/product/1727?vs=552
The difference is inconsequential.
Then if we compare Broadwell-E and Skylake-X... Performance still hasn't moved much.
http://www.anandtech.com/bench/product/1727?vs=190...
5-6 years... And if I spent almost $1,000 AUD on a new CPU, Board and Ram I wouldn't even get a 50% improvement.
BurntMyBacon - Thursday, August 17, 2017 - link
@StevoLincolnite: "It's a Sandy Bridge-E chip.It's actually 2nd Gen Core. :P"
I was thinking 2nd Gen, but my hands were typing 3rd Gen. In any case, it just makes it that much more impressive. Kudos to you for getting the right system at the right time.
HollyDOL - Friday, August 18, 2017 - link
I upgraded from i5-2500K to i5-6400 (non OCed, cpu died) and have to say the performance is the same. These days it's more about features you can get (access to) than raw performance... old board had 2x6Gbps + 4x3Gbps sata, new one has several PCIe M.2s, all SATA ports are 6Gbps etc. Regardless, if the old one didn't die, I'd still be using it.BurntMyBacon - Wednesday, August 16, 2017 - link
@StevoLincolnite: "With my 1866Mhz DDR3 modules, I have more bandwidth than Dual-Channel DDR4 3200Mhz."Also, at lower latency IIRC.
@StevoLincolnite: "Only thing that is a let down is that it only has Sata 6Gbit rather than 16Gbit.
And no USB 3.1, USB C. - But you can live without that, not worth the rush to upgrade."
Not sure what you'd plug into a SATA 16Gbit (SATA Express). I haven't seen any drives use it. All the M.2 SATA drives I've seen use 6Gbit controllers even if the interface should be capable of supporting faster. I'd just grab a PCIe to M.2 adapter card and throw a PCIe M.2 drive on it. You should have more than enough lanes on your platform to support this and the interface will be faster than any SATA standard.
They do have USB 3.1 adapter cards. Again, I see no reason you board wouldn't have enough PCIe lanes to support this should it become desirable in the future. No need to live without it if you want it.
@StevoLincolnite: "It's insane how long this rig has lasted me and it's still got a ton of life left in it."
Yeah, maybe more than you've considered. You could probably ride that system until something breaks without it feeling shamefully outdated (assuming periodic video card upgrades).
StevoLincolnite - Thursday, August 17, 2017 - link
I was just listing off checklist features that I don't have... And you are right, USB 3.1 cards are available. And I do have plenty of PCI-E Lanes. Back with Sandy Bridge-E, Intel didn't castrate the amount of PCI-E lanes on the hex chips.Lolimaster - Wednesday, August 16, 2017 - link
Except you can now get the Ryzen 7 1700 with 50%+ more performance while consuming 3 times less... and with upgrades all the way to 2020.StevoLincolnite - Thursday, August 17, 2017 - link
At 4.8Ghz-5Ghz... I highly doubt Ryzen 7 would beat my 3930K. I can overclock my CPU by over 50% remember. ;)Or at-least the difference would be small enough not to warrant an upgrade.
You do have a point on the power consumption side of things, but is it worth buying a new CPU, Motherboard and Ram combo? With the Price of Ram at the moment...
And my DDR3 Ram in quad-channel still ends up faster.
Better off just waiting until something is available at the right price or until the system breaks.
Lolimaster - Wednesday, August 16, 2017 - link
For $270.Strunf - Wednesday, August 16, 2017 - link
Ryzen 7 1700 + MB + Memory... for $270? I think not.Murloc - Tuesday, August 15, 2017 - link
no point in waiting for newer CPUsSamus - Tuesday, August 15, 2017 - link
^This. Why wait for a 2% performance increase, especially when given Intel's track record, the platform won't support future chips anyway? I'm still rocking a Sandy and Haswell, no plan to upgrade either for a few years.valinor89 - Tuesday, August 15, 2017 - link
If anything is making me reevaluate my decision to keep my 4670k for at least a couple of years more it is the Ryzen lineup and the shakedown it is causing right now. New Intel CPU might be interesting again soon.ingwe - Tuesday, August 15, 2017 - link
Waiting for a laptop can still makes sense given the power savings from generation to generation. It might not be huge from the CPU alone, but overall systems are still usually decreasing it.Cliff34 - Tuesday, August 15, 2017 - link
I agree. I am waiting because I am a laptop user. For desktop users, it doesn't make too much of a difference.Alexvrb - Tuesday, August 15, 2017 - link
If a big launch is just around the corner (a couple months perhaps) then waiting *might* make sense. But when you're talking about a launch year(s) down the road waiting only makes sense in a couple of scenarios. One would be if you don't have/can't spare the money to upgrade, and thus are waiting for non-technical reasons. Another case would be if what you're using is already good enough and thus you're not truly "waiting" at all. Or some combination of factors like this - you'd LIKE more horsepower/lower power consumption, but you don't really need it and/or can't justify it.niva - Wednesday, August 16, 2017 - link
Pfft, my main PC at home was a Phenom 2 rig that lasted me damn near 10 years. Granted I don't game much, skyrim being the most demanding thing I ran on it. I wish I'd made it to Ryzen release but alas the mobo died on me and I switched to Intel at the end of 2016. I'm the same with phones, I usually get 4 or 5 years of usage out of them, went form Nokia N95 to Galaxy Nexus to Nexus 6 which I'm still on.Stochastic - Tuesday, August 15, 2017 - link
The only reason to wait is that RAM prices might decline later in 2018. Although I read a recent support saying that the NAND shortage will extend into 2019 potentially. Right now is a particularly bad time to build a new PC because of the double whammy of inflated RAM and SSD prices + inflated GPU prices due to the mining craze. Hopefully things will settle down sooner rather than later.Stochastic - Tuesday, August 15, 2017 - link
*recent reportAlexvrb - Tuesday, August 15, 2017 - link
I hope so. SSD pricing has been flat (which is still absurd given density increases), but RAM costs have actually gone up quite a bit and GPUs obviously are through the roof. Just waiting for Eth to go Proof of Stake. Sigh.zodiacfml - Tuesday, August 15, 2017 - link
A new process node is worth it for mobile...and server parts.Netmsm - Tuesday, August 15, 2017 - link
It's just an attempt to halt in selling Ryzen.TEAMSWITCHER - Tuesday, August 15, 2017 - link
I won't ever build an AMD machine again. That ship has sailed.Netmsm - Tuesday, August 15, 2017 - link
And that's only your way and your problem too not realistic ones!coder543 - Tuesday, August 15, 2017 - link
That seems like a sound, thought-out argument built upon troves of market analyses.Thanks for sharing with the rest of us.
artk2219 - Tuesday, August 15, 2017 - link
You're not quite living up to your username man :-/ .Netmsm - Tuesday, August 15, 2017 - link
What a prophecy!it's obvious that analogizing Intel and AMD in terms of their cap is not a true debate (which has been brought up in other comments)! in the other hand, siding with Intel, as the giant of the semiconductor industry, by mentioning its potential capacities/abilities seems like blindly following the stronger!
however, the hurry-scurry announcements of Intel - at least - show what Intel could've produced but didn't. and what it means?
MonkeyPaw - Tuesday, August 15, 2017 - link
Kinda what I was thinking. Why would Intel tip its hand that far down the road if it wasn't an attempt to tell investors that they "still got it." AMD should honestly never be able to compete with Intel in the first place. Intel has a 14x higher market cap. The amount of resources at their disposal should have them so far ahead of anything AMD can produce. Intel has stumbled on process advancements, which has been their strongest advantage.tamalero - Tuesday, August 15, 2017 - link
More like they didnt really NEED to invest hard again. They could milk the cow.Just like they love to force complete motherboard changes almost every iteration.
jardows2 - Tuesday, August 15, 2017 - link
I would say it's more likely they used their resources to diversify outside of their core strength. How much was spent trying to make Atom a viable competitor to ARM?FreckledTrout - Tuesday, August 15, 2017 - link
Who says Intel isn't that far ahead? Honestly there IceLake design and 10nm likely have been tape in ready for a while now. If you have no competition why compete with yourself? Just keep your new stuff in your pocket until you have to pull ahead in the race again. At least AMD is forcing Intel to move forward again. I build myself a Ryzen R7 system so no bias just reality.Lolimaster - Wednesday, August 16, 2017 - link
Jim Keller is known as the GOD of cpu's for a reason.Athlon
Athlon XP
Athlon 64/Opteron
Zen
halcyon - Wednesday, August 16, 2017 - link
The ideal time to upgrade always changes.- 2018: Coffee Lake (perhaps Ice Lake, we'll see)
- 2018: nvme 1.3 drives (let's see if they matter)
- 2018: nVidia Volta gaming cards
- 2019: PCIe 4.0 (probably not available in 2018, but we'll see)
- ????: Prices of DRAM come back to earth (?)
Of course, process nodes, perf/$, perf/W and many other factors go into this.
I was really waiting for ThreadRipper and it didn't disappoint. Yet, I want to see the Intel September releases.
The only thing holding me back is PCIe 3.0, which is on it's last legs. My last system is from 2009 (using PCIe 2.0) and is doing fine otherwise (6-core 4.2Ghz), and I do feel the difference comparred to PCIe 3.0 systems I use.
Also, the DRAM prices currently are insane. Hoping the production/markets stabilize in 2018.
nevcairiel - Tuesday, August 15, 2017 - link
Perhaps 10nm Cannon Lake has been canceled, and moving straight to 10+ instead. Previous reporting has said that an early 10nm process may barely outclass the fully optimized 14++ process.Otherwise the generation naming just doesn't make any sense.
nevcairiel - Tuesday, August 15, 2017 - link
On the other hand, a 10th generation CPU is still the successor of the 8th generation, isn't it.jabbadap - Tuesday, August 15, 2017 - link
Well yeah. But ain't Cannon Lake just plain 10nm die shrink of 14nm Coffee Lake(heck coffee lake is just optimized sky lake). Like Broadwell was to Haswell. Maybe it's exactly like Broadwell: short living low tdp processors with iris pro igpus. Great for htpc users, but mainly meh for anything else.nevcairiel - Tuesday, August 15, 2017 - link
Even then Broadwell got its own generation number though.eddman - Tuesday, August 15, 2017 - link
It is possible that intel might end up using both coffee lake and cannon lake chips in their 8th gen, 8xxx lineup; perhaps CFL for desktop parts and CNL for low-power mobile parts.name99 - Tuesday, August 15, 2017 - link
The new model is Tick Tock Sproing Thud. It's funny because it's true.HT to SemiAccurate.
watzupken - Tuesday, August 15, 2017 - link
You can certainly sense the urgency from Intel's camp of late. But to be honest, there is no point looking 2 generations ahead since thing may change over the next couple of years. Bring out Cannon Lake, then we talk about the next gen.name99 - Tuesday, August 15, 2017 - link
The issue is what you're interested in.If your interest is ONLY in "what CPU can I buy in the near future" or "how can I support my corporate tribe", then what you say is correct. But look at the big picture beyond those.
That picture shows an Intel that is breaking down before our eyes.
It started with Broadwell which shipped late and lame (in spite of Intel's protestations beforehand that yields were the most awesome ever). Lying about how well their process is doing seems to be the new normal (witness Optane, still MIA from the market where it actually matters, and being sold into markets where flash is a much better choice).
Meanwhile architectural innovation seems to have ground to a halt with Skylake (it actually stopped with Sandy Bridge, since then they've been treading water, but now they're not even trying --- hence once Lake after another).
Now we have this process-based split across Kaby, Coffee, Cannon, and Ice that looks like nothing so much as mad scrambling. It's not that having separate designs and processes for server vs desktop vs laptop is a bad idea; but these do not look anything like separate DESIGNS, more like just separate "WTF can we do to get SOMETHING, ANYTHING that works?". There's no evidence here of design in terms of things like giving Cannon a differently balanced microarchitecture that's a better fit for its target space as compared to say Coffee.
So yeah, it's interesting from the point of view of "How does Intel's monopoly end?"
Ten years ago was just before iPhone, just before Nehalem. A lot can change in ten years. So the question is what changes over the next ten years? What all this naming nonsense suggests to me is that Intel is even less prepared for the future than they were for mobile. At the time I suggested that Intel's mobile attempts were a pointless waste of time, based not on dick-measuring how one Intel core stacked up against a competitor ARM core, but based on the fact that Intel's STRATEGY made no sense, it revealed a corporation that was either (perhaps both) deeply incompetent or deeply deluded.
I see the same fundamental strategic failures now at the mainstream level. Sure things look fine right now, but we're not talking now, we're talking about 10 yrs from now and how well Intel is poised to cope with the future.
Hurr Durr - Tuesday, August 15, 2017 - link
Oh look, usual crapple shill yapping about corporate tribes of all things.name99 - Tuesday, August 15, 2017 - link
Seriously? I comment that some people's primary interest in this news is "how can I support my corporate tribe" (not that there's anything wrong with that, simply pointing out that that that's not the ONLY lens of analysis) and your response is to talk about "crapple".I kinda rest my case.
twtech - Tuesday, August 15, 2017 - link
My understanding is that several years back, Intel decided to adopt a focus on hiring & promoting for reasons other than being the best at what you do. For a company that relies on continued superior engineering to maintain its advantage, that doesn't seem like a good thing - at least in terms of the quality of the output product.name99 - Tuesday, August 15, 2017 - link
And do you have a source for this beyond Fox News and James Damore?If we're throwing out wild theories, my wild theory is that the problems all started when America First sentiment mean Intel could no longer hire the best possible engineers from the entire world.
Se how that works?
mak1977 - Tuesday, August 15, 2017 - link
How does Intel Monopoly End.It ends when arm support on windows is passable . Microsoft is working on ARM for windows with some kinda of emulation in hardware by the ARM parts to do x86/win32 apps.
https://www.xda-developers.com/microsoft-qualcomm-...
Once the performance is good enough and More ARM vendors come on board, it will be the end of Intel , they can't compete on price with arm.
BillBear - Tuesday, August 15, 2017 - link
Intel's going to have to either advance the state of the art (something it certainly did not do with Kaby Lake) or it's going to have to learn to live without a 60% profit margin.You can't price gouge and stand still at the same time.
HStewart - Wednesday, August 16, 2017 - link
Well first of all - Intel can not have a monopoly when there is all those ARM based tablets and phones out there - Intel try to get into phone market - but obviously ARM has a monopoly there.As for Windows on ARM, well it did not work with Windows RT. And no technically sane person would expect it work with emulation on ARM. I don't think Microsoft is having good success with Windows on ARM. If they did they would have used ARM chip on HoloLens.
Intel is fully aware of ARM monopoly on the phone and some ways tablet markets. That is why they invest millions of dollars in research like in this article with Ice Lake. Fanless computers are where is going and expect more powerful version.
sor - Tuesday, August 15, 2017 - link
Maybe we will see coffee lake and cannon lake focus on different markets or coexist in roughly the same generation like we are seeing with Skylake and Kaby.Drumsticks - Tuesday, August 15, 2017 - link
I think, based even just on the slides above, it's clear to me why Cannon lake is on 10nm but coffee lake is on 14+. 10nm saves on power and density, but 14++ looks to have much higher performance. I know which I want for my desktop and which I want for mobile, personally.guidryp - Tuesday, August 15, 2017 - link
I agree. It makes perfect sense as a transition strategy, and they get to amortize their 14nm process even more.I pretty much expect ~3 years down the road 7nm will be used with Mobile processors first.
ilt24 - Tuesday, August 15, 2017 - link
guidryp, FYI: back in February Intel said that starting with 7nm server chips would be the first to use the new process followed by client chips.HStewart - Wednesday, August 16, 2017 - link
Of course with 10nm chips - maybe not Cannon Lake - higher performance will likely exceed 14nm chips like Coffee Lake - but Intel's appears to concentrated on Mobile than Desktop chips. This logic because 10nm means also lower power.Hurr Durr - Tuesday, August 15, 2017 - link
It looks like intel is flirting with the Osborne effect here on the upcoming 6-core 14nm parts.Old_Fogie_Late_Bloomer - Tuesday, August 15, 2017 - link
I'm definitely keeping my eye on the 6C/12T 14nm chip to finally upgrade from Sandy Bridge.Hurr Durr - Tuesday, August 15, 2017 - link
I`m sort of on the fence considering if I really should leave Ivy this refresh, or the one coming after it.Stochastic - Tuesday, August 15, 2017 - link
Same. My 3570K is doing okay, but I know its days are numbered. I'm willing to wait another year for a substantial (more than 10%) performance bump, but not for another incremental step forward.imaheadcase - Tuesday, August 15, 2017 - link
I admit i don't know much about manufacturing these, but i do know marketing and this announcement seems to show they already are making them to some degree. The only reason a company announces something they normally don't is 1. To one up the competitor to show its still in the game. 2. It already has the product but is waiting to see the response of buyers (buyers don't want more if they have extended stock to sell).CaedenV - Tuesday, August 15, 2017 - link
In the CPU world a lot of times they can build them... but they have issues building them in mass without a lot of bad chips. It can take a long time to get the manufacturing reliability up to a point where it is profitable. They have been making 8nm chips in small batches for a few years now... but mass market parts wont be available for a long looooong time.name99 - Tuesday, August 15, 2017 - link
So basically Intel said NOTHING here that wasn't already known and obvious...Socaltyger - Tuesday, August 15, 2017 - link
Winter is coming.tamalero - Tuesday, August 15, 2017 - link
The graph just shows how much stagnation was there..almost one year for every tick tock and node change. and they were stuck with 14nm for 4years..
zmeul - Tuesday, August 15, 2017 - link
Ian, mate ... Intel officially stated Ice Lake taped out back in Junehttps://twitter.com/intelnews/status/8728447568453...
zmeul - Tuesday, August 15, 2017 - link
Canon Lake, sorryFreckledTrout - Tuesday, August 15, 2017 - link
I know you corrected but the other point is Ice Lake was tape in ready not tape out which just basically means the design is ready and then they do the tape-out to a physical part.twtech - Tuesday, August 15, 2017 - link
"Why is the laptop market splitting between 14++ and 10, and why is the desktop market not going to 10nm but straight to 10+?"It looks like the answer is right there on the slide. It shows 14++ as being higher performance than 10, and at least equivalent to 10+. On the laptop side, the lower power consumption is probably worth the tradeoff in some cases, but not so much on the desktop side.
nevcairiel - Tuesday, August 15, 2017 - link
Its true, judging by their own scale we have to wait for 10++ to get a real performance advantage just because how well 14++ works. Intels fully optimized 14nm process is quite the beast.wallysb01 - Wednesday, August 16, 2017 - link
You might well still be right, but that's performance per transistor and the smaller process will have more transistors. We'll see how the balance works out but I think it's safe to say cannon lake isn't going to offer much unless they bump core count again. Seems the times to buy are going to be either Kaby Lake or wait for Ice Lake.eddman - Tuesday, August 15, 2017 - link
I'm hoping with ice lake intel would finally stop their messy cores/threads practice and go for a straight 4C i3, 6C i5 and 8C i7, all with SMT enabled configuration; but since this is intel, probably not.artk2219 - Tuesday, August 15, 2017 - link
That would be a nice line up, 2c\4t celerons (where they belong), 4c pentiums, 4c\8t i3's, 6c\12t i5, 8c\16t i7's. That would definitely be a nice competitive line up, especially since its 2 years down the road. Whether we will see that or not is anyone's guess.zodiacfml - Tuesday, August 15, 2017 - link
Cannon Lake is sure to be compelling but AMD's integrated GPU in Raven Ridge could be great for gaming on a budget.ithehappy - Tuesday, August 15, 2017 - link
I just built a 6700k based system last year, and I'm a bit sad to see a Hexa core CFL coming at that price point within few days, though it's good for competition.Thinking whether to upgrade to CFL or not, but after seeing this it feels like waiting out for a year might be much better and wiser, for this Ice Lake thing based on 10 nm +.
Should I upgrade now or wait? I am a moderately demanding gamer of titles like RoTR, GTA V, FC Primal/ FC4, Watch Dogs, BF1, AC Syndicate etc. I have a GTX 970 which is just okay in 17, obviously fails to keep up at 60 frames at near max even, so I have to stick with high settings.
dullard - Tuesday, August 15, 2017 - link
I would wait. While the 6700k was released more than 2 years ago, the 8700k is only 10% to 15% faster clock speeds for 1 to 4 core tasks. You'd barely notice the difference in those.Only once you get to 5-core and 6-core tasks would the upgrade be worth it. And even then you'd only be ~25% to ~50% faster in those few tasks. Wait out the video card shortage and get a new video card instead.
jimjamjamie - Wednesday, August 16, 2017 - link
Leave that 6700k alone it's perfectly fine, swapping out the 970 for a 1070/1080 would be money better spent.Bluetooth - Tuesday, August 15, 2017 - link
Any new features coming with these new Lake-X Cores? Like for H256 or vectors, etc. ?fazalmajid - Tuesday, August 15, 2017 - link
So how does Intel's vapor ware 10nm process compare to TSMC's 10nm FinFET as used in shipping products like the new iPad Pros' A10X.Bluetooth - Tuesday, August 15, 2017 - link
Intels 10nm process does not equal TSMC 10nm, unfortunately marketing was needed to catch up with Intel. So, TSMC 10nm is probably similar to Intel 14nm (each company decides on their own definition).HStewart - Wednesday, August 16, 2017 - link
This reminds me of the frequency wars - now we having multi-core wars and you are right, this is a lot of marketing, There are rumors also that 7nm coming out - and that is probably closer to 10nm or even 12nm on Intel's side.One thing it appears is that smaller the die gets, the harder it appears to create the process.
For me personally, I not sure of need for more than 4 cores - but that will change over time as all technologies with chips - when the first 32 bit chips came out - I never thought I need 64 bit. But just as 8 bit - went to 16 bit and then 32 bit - 64 bit is logical. I don't think we need 128 bit now - but who knows in decade from now - things might be change.
I also believe Intel's process allows Intel to put more cores on single die - instead combine multiple dies together to achieved this goal.
Pork@III - Tuesday, August 15, 2017 - link
Ice Lake or Ice Fake :Dserendip - Wednesday, August 16, 2017 - link
Hopefully Atom-like chips with 2W or 4W TDP will make a comeback thanks to the process node power savings. Apollo Lake can't be stuffed into tablets because TDP is too high. I'm happy with my Cherry Trail Windows tablet but a newer design with a faster GPU supporting hi-DPI screens would be great.HStewart - Wednesday, August 16, 2017 - link
My guess is that possibly the Atom name would be retried but Intel will move it in to line similar to what they did with Core-M and Y series chips - People tend to put down Core-M and I am typing this on m3-6y30 based Intel Compute Stick - it actually quite fast compare to say my older original Microsoft Surface ProBut you are right these chips would be hard to stuff into small tablet - but the price needs to be there to compete with ARM. Intel appears to have a lot of research - most of it is in embedded markets - like Quark processor which has even less power than the Atom.
rtho782 - Wednesday, August 16, 2017 - link
I'd love to see Intel fab GPUs for nVidia....iwod - Wednesday, August 16, 2017 - link
It should be Sandy Bridge, Ivy Bridge, Haswell then Broadwell, the post have the "well" generation wrong way round.Since we are still on the lake architecture, does that mean all we get is an 10nm Skylake? Cannonlake generation was suppose to be a new uArc. And we are straight to Ice Lake?
Meanwhile I am eagerly waiting for AMD with Zen2 on 7nm next year.
Gondalf - Wednesday, August 16, 2017 - link
No Zen2 on 7nm next year. The process is very late and the cpu integration too. Apparently AMD want EUV at 7nm, quad patterning is too complex for actual AMD resources. Go to EETimes, there is a nice article on this.....AMD exec claims:"AMD is among chip designers getting an early taste of 7nm process technologies"
They are "tasting"
http://www.eetimes.com/document.asp?doc_id=1332049
Lolimaster - Wednesday, August 16, 2017 - link
Zen2 is coming Q3-Q4 2018.minde - Wednesday, August 16, 2017 - link
mobile processors u 15w from coffeelake or cannonlake will be with 4corehttps://en.wikichip.org/wiki/intel/cores/coffee_la...
Lolimaster - Wednesday, August 16, 2017 - link
The bad thing about intel is that all their existing desing all the way to tiger lake (which was supposed to be a die shrink of Ice Lake) are based on the nehalem/sb arch, they were never intended to be scalable multicore chips with good production/yield ratio.Zen is the contratry, is a design for the true mass multicore era for both supercomputing, servers all the way to consumers, all in just 1 design. Zen really screwed intel plans for the next 5 years.
10cores plus for consumers on intel side has proven to be a patch work for the panic mode.
HStewart - Wednesday, August 16, 2017 - link
Everybody has their opinions but if Intel wanted combine more than one of it larger core cpus together to make a massive cpu. they could do it. I think Zen is more of patch, by patching multiple cores in one package especially when it uses the PCIe bus internally to connect them.It funny, most people don't see what really going on here - Intel's major threat is not AMD but ARM in the mobile industry - That is why Cannon Lake is for low power fanless computers and not huge desktops that most people don't use except a decade ago.
HStewart - Wednesday, August 16, 2017 - link
Besides the entire process switch between 14nm and 10nm, I more curious on architecture changes of both Cannon Lake and Ice Lake.My thought which could be wrong - is that Cannon Lake is similar architecture of Coffee Lake, but using the 10nm process along with it advantages - but it sounds like it might be only limited to low power laptops
Icy lake will be architecture improvements and also intended for full line up of chips.
My hope is to actually have a Quad or Six core ultra low power notebook in fan less box one day - but that is probably Ice Lake.
HStewart - Wednesday, August 16, 2017 - link
One small note - just adding more cores - is not the only architecture change - new technology to make the applications run faster and better is also. I pretty sure that 10nm chips have AVX-512 for example - but there could be other changes that Intel has not disclosed.eddman - Thursday, August 17, 2017 - link
"My thought which could be wrong - is that Cannon Lake is similar architecture of Coffee Lake, but using the 10nm process along with it advantages"Why are you not sure? It has always been the plan from day one, years ago, for cannon lake to simply be a an improved skylake on 10 nm but because intel hit issues with that node, they had to release kabylake and now coffee lake on improved 14 nm nodes.
Ice lake is a bit of a question mark though. Intel always used a completely new name for their new architectures, sandy bridge, haswell, skylake, but this time ice lake's name still belongs to the Lake family.
This makes me think that it's still based on the same skylake architecture. I could be wrong though. Maybe intel likes Lake names and plan to use them even for new architectures.
Tyler_Durden_83 - Thursday, August 17, 2017 - link
On intel official website linked in the article it actually says that ice Lake is the 8th generation, not the one afterIan Cutress - Friday, August 25, 2017 - link
Android Password Breaker hacking tutorials hacking ebooks hacking news hacking tools android technology https://myhacker.net