Intel Core X Series Launched: 18 Cores and Maddening Differentiation

Discussion in 'STH Main Site Posts' started by Patrick Kennedy, May 30, 2017.

  1. #1
    eva2000 and William like this.
  2. Rain

    Rain Active Member

    Joined:
    May 13, 2013
    Messages:
    226
    Likes Received:
    71
    I wish the eight core model had 44 PCIe lanes; it seems way too artificially gimped. Other than that, everything else kind of makes sense.

    While I know Intel is still in a good position, I'm a bit surprised they showed nearly all their cards before AMD Threadripper's pricing was made available. This is going to be interesting.

    It kind of saddens me that my i7-3930K, happily sitting at 4.5GHz, has more PCIe 3.0 lanes* than the new i9-7820X with it's higher launch price... *(Sandy Bridge-E i7 CPUs support PCIe3 speeds, but were not formally PCIe3 certified like the Xeon parts were)
     
    #2
  3. msvirtualguy

    msvirtualguy Active Member

    Joined:
    Jan 23, 2013
    Messages:
    457
    Likes Received:
    233
    I gotta say, requiring me to spend $1k for more lanes than 28 has me MORE interested in what AMD has up it's sleeve with Threadripper.

    I think that was a really dumb move on Intel's part seeing as how NVMe is becoming mainstream and people are going to want multiple NVMe/M.2-U2 etc...especially as x299 matures.
     
    #3
  4. MiniKnight

    MiniKnight Well-Known Member

    Joined:
    Mar 30, 2012
    Messages:
    2,941
    Likes Received:
    857
    That's stupid. Suckier next year is when you need to closely look at specs to see which one I'll buy used. I'm skipping this gen to protest.
     
    #4
  5. Nanotech

    Nanotech Active Member

    Joined:
    Aug 1, 2016
    Messages:
    595
    Likes Received:
    99
    Not sure if anyone saw this but LGA2066 processors are using thermal paste instead of being soldered to the IHS:

    Intel's Skylake-X and Kaby Lake-X CPUs will not be soldered | CPU & Mainboard | OC3D News

    For HEDT that's absolutely unacceptable especially at the prices Intel is charging. This is a visible break from their previous HEDT which had all used solder for the IHS. Why this is important is due to the following:

    "The problems that arise from using non-soldered CPUs occur because the TIM used by Intel will not have the same thermal conductivity as a directly soldered CPU, which will cause Intel's CPUs to run hotter, especially when overclocked."

    So not only when it is overclocked but likely running at stock speeds consumers will likely end up delidding the IHS just to run at acceptable thermals.
     
    #5
    eva2000 likes this.
  6. msvirtualguy

    msvirtualguy Active Member

    Joined:
    Jan 23, 2013
    Messages:
    457
    Likes Received:
    233
    Wow...just wow...Intel feeling pressure from AMD and rushing this?
     
    #6
  7. Nanotech

    Nanotech Active Member

    Joined:
    Aug 1, 2016
    Messages:
    595
    Likes Received:
    99
    I would definitely believe so. Intel was originally just going to launch up to 12 core variants but recently changed it to include 18 core variants. Now those 14,16,and 18 core variants won't be released till sometime in 2018. We also know that their pricing the 18 core at $2,000 USD. For $2,000 and having it with thermal interface material instead of solder on the IHS is absolutely unacceptable. It's not just that part but their entire lineup. As a reference here are temperatures of processors under load with AVX:

    https://lh3.googleusercontent.com/-...ig79gR_k7YszcTD_kmchNgCJoC/w630-h776/0001.PNG

    The 7700K is a TIM processor. I would expect the new Skylake-X to hit similar temperatures under stock and probably much higher when overclocked. Note the temperatures of the soldered HEDT parts from the previous flagship X99 chipset (Broadwell-E).
     
    #7
  8. Biren78

    Biren78 Active Member

    Joined:
    Jan 16, 2013
    Messages:
    550
    Likes Received:
    94
    Your article implies that if I had a dual E5 V3 or V4 workstation I lose top clocks but I'm paying maybe half of what the 18C is and I can get similar net performance. I'd also get more PCIe more SATA more RAM using ECC RDIMMs.
     
    #8
  9. i386

    i386 Well-Known Member

    Joined:
    Mar 18, 2016
    Messages:
    1,665
    Likes Received:
    400
    I think they are okayish considering these are consumer cpus. (For comparison Ryzen 7 has 24 pcie lanes and ThreadRipper 64 pcie lanes)
     
    #9
  10. eva2000

    eva2000 Active Member

    Joined:
    Apr 15, 2013
    Messages:
    241
    Likes Received:
    48
    wow thanks for heads up !
     
    #10
  11. Nanotech

    Nanotech Active Member

    Joined:
    Aug 1, 2016
    Messages:
    595
    Likes Received:
    99
    More information on it as well:

    Intel’s new Skylake-X chips still use low-quality TIM | KitGuru

     
    #11
    eva2000 likes this.
  12. Jeggs101

    Jeggs101 Well-Known Member

    Joined:
    Dec 29, 2010
    Messages:
    1,466
    Likes Received:
    216
    It's a timid release. 1x 4GHz 18C is better than 2x 2GHz 18C E5 for workstations. If you've got OCing to 5GHz why get E5?
     
    #12
  13. ATS

    ATS Member

    Joined:
    Mar 9, 2015
    Messages:
    96
    Likes Received:
    32
    Yeah that's complete BS. It isn't low-quality TIM. In fact, it is one of the best most stable TIMs you can get on the market. That TIM can be thermally cycled close to infinite times with basically no pump out and has essentially no time based degradation. Also anyone that thinks they switched from solder to TIM to save pennies is literally insane, they likely switched for reliability reasons.
     
    #13
  14. Klee

    Klee Well-Known Member

    Joined:
    Jun 2, 2016
    Messages:
    1,213
    Likes Received:
    365
    If AMD had nothing coming down the pipe everyone would be super pumped up for this but seeing some of the info on what AMD has coming out with its real hard to be real excited about this.
     
    #14
  15. KioskAdmin

    KioskAdmin Active Member

    Joined:
    Jan 20, 2015
    Messages:
    156
    Likes Received:
    32
    If AMD didn't have something coming, we wouldn't have these chips.

    But f- Intel for doing this garbage with PCIe and RAM. 4 channel memory. 44 PCIe. Done.
     
    #15
    eva2000 likes this.
  16. Klee

    Klee Well-Known Member

    Joined:
    Jun 2, 2016
    Messages:
    1,213
    Likes Received:
    365

    Just Intel being Intel.
     
    #16
  17. Nanotech

    Nanotech Active Member

    Joined:
    Aug 1, 2016
    Messages:
    595
    Likes Received:
    99
    No contrary to your claim it's not "complete BS".

    1. Dow Corning isn't even close to the best TIM you can get on the market (especially since it's the same TIM used for their other TIM based processors such as the 6700K, 4770K, 7700K etc...)
    2. TIM should only be used on small dies (this is a large die hence your statement about thermal cycles is not applicable)
    3. Xeon E3's are soldered and work fine with various thermal environment and cycling.
    4. Previous HEDT were always soldered.
    5. The other issue is the gap between the die and the tim/ihs.
    6. Refer to processors that use TIM and see what temperatures they run at stock/overclocked versus soldered.

    They didn't switch for reliability reasons they switched to save costs. Refer to this for thermal cycling on smaller and medium dies:

    http://overclocking.guide/wp-content/uploads/2015/11/Rjc_cycling-300x282.png

    The Truth about CPU Soldering - Overclocking.Guide

    I am going to quote again from this article:

    If you think soldered dies are inferior and have issues with degradation then you need to refer to mainstream Sandy Bridge processors which are still working to this day. The fact that you are defending a practice of putting TIM on processors up to $2000 on a HEDT platform is why I decided to post this rebuttal. You can't claim someone is "insane" because they view Intel's practices of putting TIM as acceptable on HEDT to save pennies.

    The reality is that X299 is lackluster and a rushed response to Threadripper while offering less features than Threadrippper and being priced more than Threadripper would potentially be. That's not even talking about that you need the 10 core processor just to get the full 44 lanes as the lower lineups all have 28 lanes.
     
    #17
    Last edited: Jun 1, 2017
  18. Stephan

    Stephan IT Professional

    Joined:
    Apr 21, 2017
    Messages:
    98
    Likes Received:
    33
    How come nobody has commented yet that the new C422 chipset apparently has a high enough TDP that it requires a small (noisy) fan? E.g. on Gigabyte GA-C422-WS board. No such thing on C236 or X99, I consider this a step backward.
     
    #18
  19. ATS

    ATS Member

    Joined:
    Mar 9, 2015
    Messages:
    96
    Likes Received:
    32
    1) DC TIMs are the best you can get on the market for what matters: reliability. Their TIMs have virtually zero pump out, have free flowing fill capability, and are basically unaffected by either thermal cycling or time degradation.
    2) TIM can be used on all size dies. Solder still has issues even with large dies and still suffer from thermal cycling.
    3)Xeon E3 are using the same TIM as their various Core iX cousins
    4) Previous gasoline was leaded
    5) gap is always an issue irrespective of solder or TIM
    6) Intel cares about reliability above all else in thermal solutions. Temperatures are within operational range and overclocking is buyer beware as always.

    The cost differential between TIM and solder is noise. They aren't going to switch to save literally pennies per part.

    Cost is a non-issue for solder vs TIM. If they are going to TIM, it is for a viable engineering reason.


    X299 is not a response to Threadripper. It was always planned and is shipping basically on time that roadmaps have had it at for quite a while. Threadripper (and quite honestly the 12-18c i9s) are solutions in search of a problem. There is little point in low clock speeds and high core count outside of server workloads. We can talk more about Threadripper when it is more than some nebulous concept with zero concrete facts.

    As far as PCIe lanes, meh. They pretty much go unused in 99.9% of the cases. The number of machines that are used with multiple GPUs outside of servers are minimal. SLI and XF are terrible solutions that don't provide any benefit the majority of the time. Most use cases for multiple GPUs outside of server have bandwidth to spare at 1 lane per.
     
    #19
  20. Nanotech

    Nanotech Active Member

    Joined:
    Aug 1, 2016
    Messages:
    595
    Likes Received:
    99
    I've yet to see Dow Corning any better than MX-2,MX-4 or any of the best TIM brands available on the market for consumers and it's definitely inferior to liquid thermal pastes. There's a reason why enthusiasts and consumers buy those TIM's and not Dow Corning. As for your reliability on Dow Corning you should look at the thermal paste shelf life of Dow Corning compared to something such as MX-2 or MX-4. It's a big difference and your statement is incorrect and misleading (factually). If you don't believe me then take for example TIM on the DBX-B and compare with the MX-2/MX-4. Dow Corning is at best maybe a few years (under perfect conditions) whereas MX-2/4 is 8 years+. I know because I've researched their TIM's before and seen the slides and information available.

    Solder is a non-issue. Just because you can use it on larger die sizes doesn't mean you should. Read above as to what the effects are.

    Xeon E3 are soldered and not using TIM depending on the TDP as there are different versions.

    How is that even relevant?

    Gap becomes an issue when you use TIM. Read above and refer to the above link. Gap when soldered wasn't an issue.

    Intel doesn't care about reliability. This is planned obsolescence. Intel is saving money by using cheap bulk TIM from Dow Corning which is known to degrade within a few years. Reliable and well known/proven TIM brands can last many years compared to Dow Corning. HEDT platforms have always been popular with overclockers and now that it will overclock even worse than Broadwell-E where do you think enthusiasts have gone? I know many who have gone with X99 upgrades or Ryzen or elsewhere because of Intel's TIM choice on X299. You can try and justify it but Intel's userbase is looking to X399 or X99 and not X299.

    They are saving money by switching to TIM. Read above. Note how Intel has yet to respond to why they chose to go to TIM instead of solder especially on HEDT. This isn't a mainstream socket or chipset we are discussing but their flagship.


    Not when HEDT used solder and mainstream parts used solder before and have been doing so for a while (especially on HEDT for many years). There is an obvious cost saving for Intel even on HEDT which is not a viable engineering reason. It comes down to cost and not engineering (read above about small dies and tim versus medium/large dies with tim and solder).


    When did I say that X299 was a response to Threadripper? X299's increased core options of 14c/16c/18c are response to Threadripper but X299 has been present on the roadmap. However the prices likely changed after Threadripper and so did the lineup as well. As for your comment on Threadripper there is more than enough information present on it. X299 is a rushed platform with severe limitations compared to AMD's HEDT. The very fact that they are using TIM on HEDT parts up to $2,000 should tell you all you need to know when you choose to call Threadripper a concept with zero concrete facts. If that were the case (which it isn't) then it would all just be rumor.

    PCIe lanes are one of the options that differentiate a HEDT from a mainstream platform along with cores and additional memory channels. Many enthusiasts who go HEDT specifically look at lanes because they use multiple devices and want to get more lanes. For example multiple graphics cards (not just for gaming but other usage scenarios), networking cards, raid cards, sound cards, PCIe based SSD's, M.2, U.2 etc... Saying that it goes unused in 99.9% is incorrect and misleading. Also SLI and Crossfire aren't terrible solutions. While they don't provide 100% scaling in most scenarios they provide a noticeable improvement and are still popular solutions (not necessarily for 3-way or 4-way but 2-way is).

    This discussion isn't going anywhere. You are just defending Intel blatantly when X299 offers little in terms of substantial improvements over X99. Especially since it's an enthusiast chipset and platform but cuts corners where possible. I've already provided facts and information that is correct and verified.
     
    #20
    Last edited: Jun 1, 2017
Similar Threads: Intel Core
Forum Title Date
STH Main Site Posts Intel Xeon E-2288G Review 8 Cores and 5GHz in the Entry Segment Oct 31, 2019
STH Main Site Posts Intel Core i3-9100F Benchmarks and Review for Servers Oct 25, 2019
STH Main Site Posts Intel Xeon W-3275 Review A 28 Core Workstation Halo Product Oct 14, 2019
STH Main Site Posts New Intel Core i9 Platform Improvements and Dramatically Lower Costs Oct 1, 2019
STH Main Site Posts Intel Ice Lake Cores Released 10nm Generation is Here Aug 1, 2019

Share This Page