Intel Announces Sunny Cove Brings 18% IPC Improvement

Discussion in 'HardForum Tech News' started by cageymaru, May 28, 2019.

  1. Snowdog

    Snowdog [H]ardForum Junkie

    Messages:
    9,432
    Joined:
    Apr 22, 2006
    What? Pretty sure AMD had nothing to do with Intel fixing 10nm.

    If you are up to date on Intel's process issues, you would know that they designed their next generation core specifically for 10nm, so with the process delayed, so was the new core, which is why we got the same *Lake cores over and over, while Intel was stuck on 14nm.

    With 10nm finally coming online, we finally get the new core.
     
  2. IdiotInCharge

    IdiotInCharge [H]ardForum Junkie

    Messages:
    11,931
    Joined:
    Jun 13, 2003
    You could do several PhD's on what just these two companies' marketing departments have done.

    That you speak with emotion regarding one and not the other, especially given that one cannot be discussed without the other due to context, exposes not just your bias but your ignorance toward the term 'marketing', and very likely, the phrase 'publicly traded company' too.
     
  3. ZodaEX

    ZodaEX 2[H]4U

    Messages:
    3,726
    Joined:
    Sep 17, 2004
    Wrong. Humans are emotional creatures. Just because we embrace emotion does not mean we are ignorant. It's just being healthy and embracing what we naturally are. Herp, Derp.
     
  4. IdiotInCharge

    IdiotInCharge [H]ardForum Junkie

    Messages:
    11,931
    Joined:
    Jun 13, 2003
    The topic is microcomputer processors. Emotion can factor into reactions for those with interest, but the whole tribalistic asshattery involved displays a lack of basic critical thinking, most especially when speaking about publicly traded companies.
     
    Deleted member 184142 and DooKey like this.
  5. Hagrid

    Hagrid [H]ardForum Junkie

    Messages:
    8,473
    Joined:
    Nov 23, 2006
    So there isn't much of a gain? Are people shutting off their HT? That would make 4/8 just a 4 core.

    Are their new chips all fixed, partially, not at all?
     
  6. MMitch

    MMitch Gawd

    Messages:
    790
    Joined:
    Nov 29, 2016
    People get emotional when faced with kung fu marketing ploy like the disclaimer we got served by Intel, it's normal, can't really blame them. The important part to remember is ALL "plublicly traded companies" make risk assessment decision based mainly on making money for share holders. I think it's fair to call them out but it would be childish to think they're the only one doing it. I would also argue that deflecting those acts when pointed out by saying "All companies do it so move on" is short sighted at best too.

    In the end it's up to us consumer to be informed and keep those companies in line, just assume the worst from every marketing dept, at the end of the day their job is to sell their cr*p at the highest price possible in mass ;)
     
    blackmomba likes this.
  7. ZodaEX

    ZodaEX 2[H]4U

    Messages:
    3,726
    Joined:
    Sep 17, 2004
    Nah, they are critically thinking. They are just to idiotic to do it up to your standards.
     
  8. IdiotInCharge

    IdiotInCharge [H]ardForum Junkie

    Messages:
    11,931
    Joined:
    Jun 13, 2003
    Intel gave one number- 18%- with significant uncertainty, and it's for an ultrabook SoC that will be thermally limited in all currently planned retail releases (those being ultrabooks). I'm betting that Sunny Cove will be quite a bit faster than the SoC's which were already faster / more efficient than what AMD has produced in the segment, i.e., whatever weaknesses they have they'll still be the best product available.

    With respect to the fixes, that's where the uncertainty comes in. Intel gave a number and didn't go into the exacting detail of the platforms tested and included a disclaimer that not all security patches may have been run. Given that more vulnerabilities are being discovered weekly, however remote the possibility of their exploitation and limited their impact to the consumer they are, it's understandable that Intel affixed the disclaimer- and that those that have emotional grudges against Intel (or just like to be snarky) portray the numbers as having been run with no mitigations in place. We also don't know the extent of Intel's hardware fixes in the Ice Lake cores, and those could substantially reduce any performance impact.

    I think that the answer is somewhere in the middle, and since these are ultrabook SoCs only, I'm pretty optimistic about the performance they'll bring to that segment, especially since AMD hasn't really decided to compete.
     
  9. MMitch

    MMitch Gawd

    Messages:
    790
    Joined:
    Nov 29, 2016
    To be fair, if they were blurry because there could be more vulnerabilities in the future, they could simply say all mitigation applied when tested as of date XYZ.
    You know what they say, talk good of it or talk bad of it but please talk about it ;) I'm pretty sure the choice of words and the grudge it created has been "risk evaluated" and is a strategic move. Sorry if my choice of words for my point is not exact, English isn't my first language but I'm sure we get the point.
     
    IdiotInCharge likes this.
  10. Falkentyne

    Falkentyne [H]ard|Gawd

    Messages:
    1,572
    Joined:
    Jul 19, 2000
    There wa no such thing as a Pentium 4 4.73 ghz Extreme edition. That may have been 3.73 ghz. And if you put 1.75v into it it would degrade like mad too.
     
  11. MarkVI

    MarkVI Limp Gawd

    Messages:
    207
    Joined:
    Sep 30, 2014
    I hate this crap where they play with the scale of the graph to make it look like vast improvement. You can make anything look like a huge change when you don't start from zero and zoom in enough...
     
  12. THUMPer

    THUMPer 2[H]4U

    Messages:
    3,052
    Joined:
    May 6, 2008
  13. Snowdog

    Snowdog [H]ardForum Junkie

    Messages:
    9,432
    Joined:
    Apr 22, 2006
    IdiotInCharge and DooKey like this.
  14. wyqtor

    wyqtor Limp Gawd

    Messages:
    406
    Joined:
    Dec 30, 2011
    Will we also see the 18% IPC improvement on the (still)14 nm(+)ⁿ desktop, or will this be limited to the mobile platform only?
     
  15. Snowdog

    Snowdog [H]ardForum Junkie

    Messages:
    9,432
    Joined:
    Apr 22, 2006
    Intel made a big mistake basically tying their next gen CPU core tech to the 10nm process, making it extremely difficult to back port to 14nm.

    It looks like 14nm is stuck with *Lake cores.
     
  16. ChadD

    ChadD [H]ardness Supreme

    Messages:
    4,206
    Joined:
    Feb 8, 2016
    Considering how terrible the clocks seem to be with the icelake mobile parts... it seems 10nm is still very broken.

    The only thing Intel has really done to deal with it is start preparing for a move to 7nm in 2021. I don't think Intel will be doing much of anything with desktop parts until then. Accept perhaps a few more PR stunt SKUs with no real volume shipments.

    AMD is going to be eating Intels lunch for the next couple years and they be mostly spinning.

    Still its Intel... they can afford a few lean years. They can change the conversation a bit with GPU stuff... and super computer wins for their CPU+GPU solutions. They shouldn't have much issue buying time till 2021 and a 7nm chip launch.

    On the big iron server business... they are also really pushing Clear Linux these days. Its possible they can refine that software and use that to help push their product, even if AMDs server chips are technically superior hardware wise. Computing is always about hardware and software. I love AMD but Intel has them beat pretty hard on the software side of things in those markets.
     
    Darth Kyrie likes this.
  17. OrangeKhrush

    OrangeKhrush [H]ard|Gawd

    Messages:
    1,504
    Joined:
    Dec 15, 2016
    The iGPU component I think is going to run hot, maybe the first in a long line of inefficient Raja designs. They should really consider dropping iGPU's for their "gaming orientated" desktop SKU's.
     
  18. Snowdog

    Snowdog [H]ardForum Junkie

    Messages:
    9,432
    Joined:
    Apr 22, 2006
    Intel CPUs have traditionally done dual duty as laptop and desktop parts so the IGP has value.

    But I agree, definitely should have dropped it for 8 core+ designs. That vast majority of 8 core core laptops probably have discrete GPUs, and the 8 core would then be smaller and cheaper to build.

    I wonder if the rumored 14nm 10 core desktop part (if it exists), will still have an IGP, which would be even more senseless.
     
  19. ChadD

    ChadD [H]ardness Supreme

    Messages:
    4,206
    Joined:
    Feb 8, 2016
    Wonder if at the top they where thinking... if we don't integrate them and pretend to be selling a complete PC. We are admitting (worst of all at the high end) that PCs need NV or AMD.

    I imagine non igpu Intel CPUs are much more likely once Intel can ship CPU+dGPU solutions. Until then I have a feeling we will never see Intel ship a chip that REQUIRES parts from those other guys. ;)
     
  20. juanrga

    juanrga Pro-Intel / Anti-AMD Just FYI

    Messages:
    2,550
    Joined:
    Feb 22, 2017
    Sunny Cove uses 10+ and it gets up to 4.1GHz on a 15W SoC. It seems rather good, specially when only some few months ago people with 'sources' was claiming that 10nm was cancelled. 10++ would bring clocks higher than 14nm.

    I have been reading the same since Bulldozer.
     
    ZeroBarrier likes this.
  21. juanrga

    juanrga Pro-Intel / Anti-AMD Just FYI

    Messages:
    2,550
    Joined:
    Feb 22, 2017
    There are lots of changes in Sunny Cove: caches, ROB, wide,...

    Intel announced mobile line. AMD announced a desktop line. There is no competition.

    Zen2 APUs are 4C/8T.

    Same song ever. People said the same after Zen launch: Intel is killed! Reality is that Intel lost ~3% marketshare in PCs and increased sales in datacenter after Naples launch.

    Also Intel 10nm is more dense than the '7nm' node used by AMD for Zen2. I guess Intel 7nm will be more dense than Samsung '3nm'.
     
    ZeroBarrier likes this.
  22. juanrga

    juanrga Pro-Intel / Anti-AMD Just FYI

    Messages:
    2,550
    Joined:
    Feb 22, 2017
    By "gimp tests" do you mean those Computex Rome demos where AMD crippled performance of Xeons by up to 30%? Or do you mean years ago when AMD disabled turbo and quad channel on Broadwell during Zen demos? Or do you mean when AMD selected GPU-bound scenarios to claim same gaming performance than Intel? Or do you mean something else by AMD?
     
    ZeroBarrier likes this.
  23. IdiotInCharge

    IdiotInCharge [H]ardForum Junkie

    Messages:
    11,931
    Joined:
    Jun 13, 2003
    Skylake cores, specifically. Unfortunately their Skylake replacement is named 'Ice Lake'...

    I'm seeing lower clockspeeds, and I'm seeing preliminary benchmarks that put the 15w 10nm parts on par with higher-clocked 15w 14nm parts. Lower clockspeeds with the same performance means longer battery life and longer boost duration, and that means more performance actually accessible for the user.

    No, no, and... no.

    No, there's no reason to assume that the IGP is going to run hot. They're not boosting the Ice Lake IGP significantly over the Skylake IGP, there are only modest increases in the number of IGP CUs on taop.

    No, Raja had nothing to do with this IGP. This part has been ready and waiting for three years.

    No, they shouldn't drop their IGPs from any part. Intel's IGPs are great at a great many things, including but far from limited to outputting video signals to monitors.

    This isn't an impossibility, however, note that 'switchable graphics' can save significant resources (namely battery) when real GPU grunt is not needed. And given that they'd still want those IGPs on desktops, it's hard to imagine them taking it off outright.

    Note that IGPs are likely to be more efficient at minimal workloads regardless of how efficient dGPUs for mobile applications become simply because IGPs are already adjacent to or integrated with the CPU, which must receive power and cooling. Powering up the dGPU for everything results in more power, signal, and cooling infrastructure being active.
     
  24. defaultluser

    defaultluser [H]ardForum Junkie

    Messages:
    12,610
    Joined:
    Jan 14, 2006

    No, they didn't. It was 3.8 GHz, nowhere near the goal of 10ghz publicized at the release (let-alone the 5GHz you think it was).

    https://en.wikipedia.org/wiki/Pentium_4

    Keep in mind, there were SEVERAL fab revisions (released at 180nm, finished at 65nm),but none could get above the 4GHz mark.

    The 4ghz part was planned, but canceled.

    You're getting old, man, and need to get your brain checked :D You're not recalling that the first Intel chip to push towards 5 GHz (overclocked) was Sandy Bridge, half a decade later.
     
  25. MMitch

    MMitch Gawd

    Messages:
    790
    Joined:
    Nov 29, 2016
    How can 15W against 15W means more battery life!? I mean if they both use 15W ... Anyway I get what you meant I suppose.
     
    ChadD likes this.
  26. IdiotInCharge

    IdiotInCharge [H]ardForum Junkie

    Messages:
    11,931
    Joined:
    Jun 13, 2003
    Well, they both have a TDP of 15W, but for the same amount of work Ice Lake in Sunny Cove may actually use less than Sky Lake (currently Coffee Lake-something). Ice Lake may also do more work at the same TDP / power draw / battery life. Having both means that Ice Lake should be more efficient across the board.
     
  27. defaultluser

    defaultluser [H]ardForum Junkie

    Messages:
    12,610
    Joined:
    Jan 14, 2006

    A more efficient processor can get more work done for the same amount of power, saving battery life (faster time to idle).

    It can also allow higher performance when you need it, as you can get higher performance in that same 15w envelope.

    They both have a thermal design limit of 15w, so the cooling doesn't have to change form one generation to the next. A more efficient processor just gets more work done at the same 15w max power the laptop's cooling handles.
     
    Last edited: Jun 5, 2019
  28. Dan_D

    Dan_D [H]ard as it Gets

    Messages:
    54,760
    Joined:
    Feb 9, 2002
    Sorry, I mistyped. I don't know why I did that.
     
    defaultluser likes this.
  29. ChadD

    ChadD [H]ardness Supreme

    Messages:
    4,206
    Joined:
    Feb 8, 2016
    I'll believe the magic better performance with the same power draw at 500+mhz lower clocks when I see it from a reliable testing source.

    If true great... however it sounds to me like best case it can equal an older chip. I have no doubt when intel says 18% better IPC they are talking clock for clock. The MHZ difference we have seen on the 10nm chips so far will account for a much larger deficient then 18%.
     
    Last edited: Jun 5, 2019
  30. IdiotInCharge

    IdiotInCharge [H]ardForum Junkie

    Messages:
    11,931
    Joined:
    Jun 13, 2003
    Yeah, I was seeing it from notebookcheck- which means that the variables were definitely not equalized in either direction. Same laptop family though.
     
  31. MMitch

    MMitch Gawd

    Messages:
    790
    Joined:
    Nov 29, 2016
    Yeah I understood what you meant originally, I was just playing with you ;)
    A fully loaded 15W 10nm vs a loaded 15W 14nm should use the same amount of power. Yeah the total work done would be more but at their max load they would be the same load on the battery hence why I asked how it would be more efficient. Now if you meant day to day use, well we can agree but at the end of the day none will be drawing their max load the entire span of time so the 15W per say is irrelevant in this math. I expect the newer part to be matched with better parts too (better battery, chipset, etc) so that shuffle the variables even more ;)

    I may have mixed TDP and power draw too, slow afternoon.
     
    IdiotInCharge likes this.
  32. IdiotInCharge

    IdiotInCharge [H]ardForum Junkie

    Messages:
    11,931
    Joined:
    Jun 13, 2003
    If you put 'max' in front of power draw and TDP, it works!

    The challenge is that we wind up discussing task efficiency and max performance at the same time. The 10nm parts appear to be superior at both, but they have different implications for comparisons.
     
  33. MMitch

    MMitch Gawd

    Messages:
    790
    Joined:
    Nov 29, 2016
    Sorry I have too, "it just works" ;)
     
    Darth Kyrie and IdiotInCharge like this.
  34. IdiotInCharge

    IdiotInCharge [H]ardForum Junkie

    Messages:
    11,931
    Joined:
    Jun 13, 2003
    Fun part?

    It does.

    That doesn't mean that we can't simultaneously count on say EA to bungle everything they touch :D


    [the challenge is going to be not in getting the hardware to work- it's in using the resources well as opposed to using them in their 'just works' state...]
     
  35. MMitch

    MMitch Gawd

    Messages:
    790
    Joined:
    Nov 29, 2016
    You're right on point. Right now, having m0ar c0res mainstream, developer need to find ways to multi thread. I understand that your vision was wider on this point but I think the focus for desktop is that for now, mobile is another story.. they need to fit a free energy magnet source in there (ahah, I have a friend who wants to buy in in those scam... can't talk him out). I would love to see more OS baked features to assign core for each tasks either manually or automatically. Some sort of AI algorithm over the cloud for compute efficieny (wow my manager would be proud) may help for better core assignment although people running VM,games,server,etc at the same time isn't the norm.

    Wonder if someday we'll see specific RAM channels assigned to specific CPU core(s).

    And thanks for the EA prank, d*mn I hate them for ruining NHL HUT this year LOL. Imagine having to build crates to unlock "cores"...
     
  36. OrangeKhrush

    OrangeKhrush [H]ard|Gawd

    Messages:
    1,504
    Joined:
    Dec 15, 2016

    For the Xeons it was the price competitive part not all Xeons, it is a rather big deal. As for the tests who knows, but it is funny how intel ran to their "own" tests with special parameters to which they disclose in the disclaimer is not real world results and results may be very different just to try salvage the situation.

    GPU limited, in as far as I know PUBG has been an intel favoured game, so much so Intel has their logo's on buildings in the game, not sure it is a GPU limited game.
     
  37. IdiotInCharge

    IdiotInCharge [H]ardForum Junkie

    Messages:
    11,931
    Joined:
    Jun 13, 2003
    I've always just thought it was poorly coded- and CPU mattered less than GPU, in that if you are using an AMD GPU for PUBG, flip a coin for decent performance. The only result from benchmarking that game is when it doesn't not work.
     
  38. juanrga

    juanrga Pro-Intel / Anti-AMD Just FYI

    Messages:
    2,550
    Joined:
    Feb 22, 2017
    Price is irrelevant. Xeons performance was crippled by AMD in its 'demo', not strange that AMD tried to hide that they played dirty and didn't give us configurations of its 'demo'.

    Intel favored game? Do you really mean that Intel is better for gaming?
     
  39. Pieter3dnow

    Pieter3dnow [H]ardness Supreme

    Messages:
    6,789
    Joined:
    Jul 29, 2009
    If AMD hid it you don't know what it was then you come to a conclusion that is not based on facts but rather on speculation ...
     
  40. OrangeKhrush

    OrangeKhrush [H]ard|Gawd

    Messages:
    1,504
    Joined:
    Dec 15, 2016
    Juanrga lacks the logic to think like that. He is a timing fanboy, and it is fun to watch him spin the same damage control over and over.

    What is his proof that AMD gimped the test? They just enabled security checks that Intel has off due to their core integrity problems. Intel's test is more unrealistic so I believe a live demo over Intel running an internal bench.
     
    Uvaman2, Verado, clockdogg and 3 others like this.