Intel boosts the high-end desktop with its Broadwell-E CPUs

Intel’s Broadwell-E high-end desktop CPUs are probably one of tech’s worst-kept secrets by this point. The wraps are officially coming off the company’s biggest, baddest fifth-generation Core CPUs today, so we can finally tell you all about them and the changes within. Compared to Haswell before it, Broadwell-E’s underlying microarchitecture delivers about a 5.5% higher instruction throughput per clock cycle thanks to a combination of microarchitectural tweaks and new instructions. These new “Extreme” chips are fabricated on Intel’s 14-nm tri-gate process, the “tick” to Haswell’s 22-nm “tock.”

The top-end Broadwell-E chip, the Core i7-6950X, adds two more cores, four more threads, and 5MB more last-level cache to the complement of the previous king of the hill, the Core i7-5960Xβ€”all in the same 140W thermal envelope. Those changes mean that builders can now put ten cores, 20 threads, and 25MB of cache in one LGA 2011-v3 socket. As we’ve come to expect from Intel’s highest-end desktop parts, all Broadwell-E CPUs are unlocked for easy overclocking, too.

If those were the only changes in these chips, we’d have a fairly simple story to tell today. The refinements don’t end there, though. Intel is introducing a new technology called Turbo Boost Max 3.0 with Broadwell-E. This tech is a further evolution of the Turbo Boost dynamic overclocking feature introduced with Nehalem Core i7s back in 2008 and its more aggressive second iteration in 2011’s Sandy Bridge CPUs.

Turbo Boost Max 3.0 is meant to mitigate the single-threaded performance deficit that the Extreme chips often exhibit in comparison to Intel’s higher-clocked quad-core chips like the Core i7-6700K. Turbo Boost Max 3.0 (or TBM 3.0 for short) works by finding the core with the most potential frequency overhead, pushing its clock speed to the limits, and prioritizing workloads to run on that core.

To use Turbo Boost Max 3.0, Broadwell-E owners will need to install a driver from Intel that enables the feature. This driver also comes with a basic utility that lets users control some parameters about how TBM 3.0 should behave. By default, the TBM 3.0 driver will prioritize the application with focus, but users can also create a list of applications that should always be run on the chip’s fastest core.

We’re still working up a full Broadwell-E review as of this writing, but we ran some tests in Cinebench to get a rough idea of how much Turbo Boost Max 3.0 can help speed up single-threaded tasks on the Core i7-6950X. Switch between the single- and max-threads results from Cinebench using the buttons below the graphs. (No, there’s not a Core i7-6700K represented in these results, but that chip is only about 3% faster than a Core i7-4790K in single-threaded Cinebenching.)


Long story short, TBM 3.0 does appear to seriously, well, boost the performance of tasks that rely on the strength of a single thread. We don’t usually see Intel’s Extreme CPUs track the company’s higher-end quad-core chips so closely in single-threaded tasks. As one might expect from a chip with two extra cores and four extra threads on the Core i7-5960X, the Core i7-6950X establishes a whole new league of performance when all of its cylinders are firing, too. Good grief.

We also fired up AIDA64 to get a sense of just how much TBM 3.0 is pushing clocks on the best-performing core on our chip. We found that with or without TBM 3.0 enabled, one of the cores on our chip was hitting 4GHz peak Turbo speeds with the Cinebench 1T test running. The difference that TBM 3.0 makes appears to be one of duration. With the feature active, the core sustained that clock speed for the majority of the test, while turning it off caused the core to spend only a little bit of time at its peak speeds. Interesting.

You may have already guessed as much, but the X99 platform will continue to underpin these next-generation Extreme CPUs. That means we still get four channels of DDR4 memory, scads of SATA ports, and enough PCIe lanes to run even the craziest multi-GPU setups that builders can dream upβ€”so long as they pay up for the right CPU, anyway. Existing X99 motherboards should work with Broadwell-E chips with nothing more than a firmware update. Intel is bumping officially-supported DDR4 speeds up to 2400 MT/s with Broadwell-E, but that’s about the only major platform change worth noting. You can read more about this platform in our Haswell-E review.

Get that credit card ready

To get the two extra cores and four extra threads that the top-end Core i7-6950X provides over the Core i7-5960X, buyers will have to pony up a whopping $1723. Intel has never shied away from charging a premium for its most-extreme Extreme CPUs, but those chips have consistently sold for about $1000 even as their core and thread counts have grown over the years. Broadwell-E breaks that tradition with a top-end chip that costs as much by itself as an entire high-end PC from our latest System Guide. Let that sink in for a moment.

With this new price structure, the Broadwell-E lineup beneath the Core i7-6950X looks remarkably similar to Haswell-E before it. For $1089, buyers get the same eight cores and 16 threads in the Core i7-6900K that they got with the Core i7-5960X, so that level of CPU resources isn’t getting any more accessible with these new chips. Turbo Boost Max 3.0 looks like it could be a nice improvement, to be sure, but the value proposition of a $1000 CPU was already muddy. Broadwell-E doesn’t do much to make that prospect any more appealing.

In the $400-$600 range where we usually begin feeling comfortable recommending Extreme CPUs, builders get two options. At $434, the Core i7-6800K gives buyers 28 PCIe lanes, a 3.4GHz base clock, and a 3.6GHz Turbo speed. We’re no happier with Intel’s decision to hobble this CPU’s PCIe connectivity than we were when the Core i7-5820K first showed its face, but Intel seems to have settled on tweaking this particular knob in exchange for giving buyers a chip at this price point. Builders who want to take full advantage of all 40 lanes of PCIe connectivity on the X99 platform will need to cough up about $200 more for the $617 Core i7-6850K. That chip gives builders a 200MHz boost in base and Turbo clocks over the Core i7-6800K for their trouble, too.

Conclusionsβ€”for now

All things considered, Broadwell-E chips appear to continue the slow-but-steady pace of improvement we’ve come to expect from new CPU generations these days. For single-threaded tasks, Turbo Boost Max 3.0 appears to cleverly make up some of the ground that newer chips have gained through architectural improvements by letting these CPUs clock their best-performing core to the sky when it’s needed. That extra single-threaded oomph takes a little bit of the edge off spending tons of money on a chip that’s not built on Intel’s most recent architectural generation. If you were already considering building a Haswell-E system, there’s no reason to favor those chips over the Broadwell-E CPUs launching today.

It’s disappointing that Intel is breaking the cycle of delivering more chip for the same money as it has with past generations of Extreme CPUs, though. Given that AMD hasn’t updated its highest-end desktop CPUs in nearly four years, we’re guessing that Intel is establishing a new pricing tier for its beefiest Broadwell-E chip mostly because it can.

If you need all the cores and threads the Core i7-6950X offers, our preliminary results suggest it’s a winner. If you don’t, though, the $1729 you’ll shell out for that chip builds a heck of a PC with a Core i7-6700K inside. The top-end Skylake chip’s four cores and eight threads, running at 4.0GHz base and 4.2GHz boost speeds, will probably offer more than enough performance for most people at a price that’s not as stratospheric as the top of Taipei 101. Still, the i7-6950X appears to set a new high-water mark for what Intel’s Extreme desktop processors can do, and that’s always exciting to see at any price.

Comments closed
    • richardjhonson
    • 3 years ago
    • sophisticles
    • 3 years ago

    And still no review.

    • DeadOfKnight
    • 3 years ago

    Wait for Zen. It may not be a direct competitor to this thing, but it should at least push the price down to $1k where the flagship extreme edition used to always sit. Or maybe you’ll want Zen instead. Who knows what the picture will be, but AMD should force Intel to be competitive with CPU prices again assuming Zen is all that it is hyped up to be.

    • sweatshopking
    • 4 years ago

    I’VE APPLIED, BUT THEY NEVER HIRED ME, GUIZE. PROBABLY BECAUSE MY KEYBOARDS. I’D BANG OUT LIKE 40 REVIEWS A WEEK.

      • yogibbear
      • 4 years ago

      Nah they spoke to your reference, who told them that you spent too much time greasing the capslock key. πŸ˜‰

      • the
      • 4 years ago

      YOU SHOULD BE RESPONSIBLE FOR ALL KEYBOARD REVEIWS SSK.

        • Neutronbeam
        • 4 years ago

        NO, HE SHOULD JUST DO REVIEWS OF THE CAPS KEY ON KEYBOARDS. JUST THE CAPS KEY.

      • Meadows
      • 4 years ago

      40 reviews of what?

        • Chrispy_
        • 4 years ago

        OH THIS AND THAT. MOSTLY KEYBOARDS BUT ALSO SOMETIMES SOFTWARE THAT MODS EVERYTHING YOU DO SO THAT IT LOOKS LIKE YOU HAVE THE CAPSLOCK TURNED ON ALL THE TIME

          • Andrew Lauritzen
          • 3 years ago

          I have a theory that he actually holds down shift the entire time he is typing :O

            • Meadows
            • 3 years ago

            Such stamina!

    • ronch
    • 4 years ago

    I know it’s an unfair comparison, but if you need tons of threads and you’re on a tight budget, the FX-8350 is less than 1/10 the price of this thing. Yes I know it’s probably not even half as fast as this thing but again, 1/10 the price. Boards are also far cheaper.

    Expensive CPUs like this make me appreciate my FX-8350 even more. Yes, it’s an old CPU on an old platform and few apps run 8 threads but for those who want lots of threads, hey, it’s just $160 over at the Egg. $160. Good grief that’s cheap.

    Disclaimer : not trying to sell AMD CPUs. I just appreciate the low pricing of what is still a very capable CPU.

    • ronch
    • 4 years ago

    Do you really need this? Because if you’re spending $1,700 on the CPU alone, you’d better really need it. If you’re just one of those RKOI brats wanting to show off what you game with, save some of your parent’s hard-earned money and get a good gaming PC for $1,700. That’s enough to get a very nice i7-6700K with a very decent graphics card and SSD. Oh and that also includes a good 22″ display.

    Then you can donate the money you saved to poor folks like us.

      • Krogoth
      • 4 years ago

      Prosumers who do real work with their hardware can easily justify the cost of the chips and some cases write off some the of cost as “tax credit/business expense”.

        • ronch
        • 4 years ago

        Yep. I suppose many folks really need it. I’m referring to those kids who just buy the most expensive stuff without really knowing what it’s for.

        • NoOne ButMe
        • 4 years ago

        *coughXeoncough*

        If you need throughout Xeon will be better priced. E5-2683 v4 will offer 12% more throughput (base) and cost only $100 more.

        • TheMonkeyKing
        • 3 years ago

        But at the end of the day, you still have to look at the money being spent. Expensing the purchase allows for it to be backed into the cost of a project, a one time deal and more than likely, an upfront cost. (Reimbursement comes later) If you decide to capitalize the cost, then you don’t write down the whole cost but do so over time. This helps your paycheck but again, you see it the following years (for each tax year you write down the cost).

        So if you are big company, you have to justify the expenditure over other technology and if you are a mom and pop shop, do you have that kinda cash laying around? It’s an ultra niche market either way.

    • f0d
    • 4 years ago

    i just realized the 25MB of cache on the 6950X is bigger than the size of my first HDD which was a 20MB MFM

      • tipoo
      • 4 years ago

      You could run Windows 3.1 entirely out of the cache of that thing with some room to spare for applications!

    • albundy
    • 4 years ago

    no idea why it doubleposted.

    • albundy
    • 4 years ago

    in other news, AMD boosts its high-end desktop with releases from intel’s high-end desktop line.

    • dikowexeyu
    • 4 years ago

    So, Intel asks 1700$ for what should be entry level processor today, and then rants about people no more buying PCs.

    WOW, 5.5% higher instruction throughput per clock cycle! I will pay 1700$, 80 Gb of DDR4 (you need 4 GB per thread) and a new mother.

    Next news: Intel reduces his workforce to cut his budget…

    • sophisticles
    • 4 years ago

    For $1700 bucks I can pick up an off lease dual socket server with dual 6 core xeons + HT for 24 total threads that would be a better multithreaded workstation.

    Think about this cpu for a second, $1700 for the top of the line Broadwell-E, about $200 for a motherboard (maybe more, much more), at least $100 for ram, you’re at 2 grand without tax and we haven’t talked about the case, power supply, video card or ssd.

    Here’s a test I would like to see TR do, take the top of the line Broadwell-E, run some x264 and x265 benchmarks, then take the cheapest skylake i3 and run avc and hevc encoding benchmarks using it’s built in quick sync encoder using the latest ffmpeg build that supports said feature or Staxrip and then compare the encoding times of using a pure software solution on the most expensive desktop cpu available against the hardware based solution of the cheapest dual core available.

    Then for laughs remember that Skylake also supports a hybrid vp9 encoding engine (part HW, part SW) and that Kaby Lake will feature full hardware vp9 encoding, and you realize how stupid one has to be to buy one of these cpu’s.

    BTW, what is it with TR being behind everyone with reviews, no GTX1080 review, no review of these cpu’s, just previews.

    Is someone sick, lazy, or do they have a full time job and this site is just a way to earn a few extra bucks?

      • w76
      • 4 years ago

      Software vs hardware encoding isn’t at all comparable in terms of picture quality at a given bit rate. Apples and oranges.

        • travbrad
        • 4 years ago

        Yeah quick sync encoding is fine for streaming and youtube where you are going to lose some quality anyway but if you are trying to encode stuff in the highest quality possible (and most efficient bitrate/filesize) CPU encoding is still the way to go.

        • sophisticles
        • 4 years ago

        According to the tests I have done with my GTX960, NVENC performs comparably (quality wise) to the “very fast” preset of x264 and x265, I have tested a Haswell based Pentium and it’s avc encoder requires about 20% more bit rate for the same quality as x264 with the “medium” preset. Reports I have seen around the net say Skylake’s quick sync offers comparable quality to x264’s “medium” preset, which is the default setting for x264 and x265, so when you factor in acquisition cost of an i3 based system vs this high end monstrosity, power consumption and encoding speed differential Intel’s quick sync is the much better choice than a pure software approach.

          • w76
          • 4 years ago

          Depending entirely on usage case, though. This is targeted likely at higher end prosumers, such as people that put actual work in to their YouTube channels, in our video-editing scenario. I’d be surprised if people that make a living (either directly or by using it as marketing something else) just run with whatever QuickSync kicks out. Don’t get me wrong, Intel is absolutely taking advantage of its market position and extracting all the money it can, but QuickSync and 10 core CPU’s just address very different needs.

          Edit: Also for my own purpose, without equivalent or better quality than software, when I archive from disks or rips or high-quality captures, I will go for the software option. Why permanently lower the quality or increase the size of my archive? Just another example.

          • Takeshi7
          • 4 years ago

          Not to mention my GTX 960 can encode lossless h.264 video at 1080p 60fps in real time. That’s a feat I’ve never seen a CPU do.

    • Plazmodeus
    • 4 years ago

    This looks like I will finally have a worthy upgrade from my O/C’d Sandybridge i7 ‘budget workstation’. The i7-6800k will give me everything I need for a buff media creation+gaming box, particularly if it O/C’s well enough.

    It hasn’t made much sense for me to move to another quad core i7, particularly when my O/C keeps my 5 year old CPU within sight of anything newer. But a Six core CPU+DDR4+Possible O/C makes a $400 CPU upgrade a no brainer. Very exciting news!

    • samer1970
    • 4 years ago

    Hi everyone ,

    Hello ,

    we all know that games dont use more than 8 threads today …

    so to take advantage of an 8 cores or 10 cores CPU in Gaming you should Disable HT (Hyperthreading) and run the gaming test again to compare it against the 4 cores i7 6700K .

    and test it with SLI as well to reach the i7 6700k bottleneck !

    let me put it more simple ,

    The i7 6700K has 4 cores and can oc to 4.4 ghz easy . this CPU will give us 8 Virtual cores comparable to 2.2 GHZ clock for each virtual core .

    However the 8 Coresi7 6900K , With the HT Turned OFF , will give us 8 cores @ 4.4 ghz EACH !

    Thats double the speed of the 4 cores i7 ! if the game uses 8 threads .

    EVEN if we dont OC the 8 cores , it would be 3.2GHZ VS 2.2 GHZ !!!

    if you ask why Disable HT ? simple because the game will never use 16 Virtual cores !!! and the advantage is LOST .

    Please run the test again for games with HT turned off in the 8 cores and 10 cores cpu .

    and to stress the CPU more , TEST SLI as well , we want the i7 6700K to bottleneck !

    THANKS

    oh and Intel Should release i5 Broadwel-E CPU , 8 cores without HT , CHEAPER and BETTER for GAMERS

      • caconym
      • 4 years ago

      I am but a humble computer animator, but I’m pretty sure hyperthreading doesn’t work like that!

      A single-threaded task will run at the same speed regardless of whether HT is turned on or not. HT doesn’t split the cores in half.

      edit: I’m aware there can be a small penalty for HT with some loads, but nothing like a 50% reduction.

        • the
        • 4 years ago

        You must have missed the Pentium 4 days as Hyperthreading then would result in decreased performance, both in terms of throughput and single threaded performance. (Part of this was the single threaded nature of software back then too.) Even the Nehalem generation had a few scenarios where Hyperthreading resulted in poorer single threaded performance though throughput was generally positive.

        It really wasn’t until the Ivy Bridge generation that Hyperthreading could be left on and not really worry about it.

          • caconym
          • 4 years ago

          I did in fact miss out on P4, because the floating-point dominance of the early Athlons kept me in the AMD camp up until the Core 2 Duo. Even the studios I worked at back then were all AMD, so I don’t think I used a hyperthreaded CPU until like 2007.

          • Bensam123
          • 4 years ago

          Still not true about HTing, do some googling… It doesn’t always work properly and often times there are games that perform worse with it on… Take this for instance. [url<]http://www.techpowerup.com/forums/threads/gaming-benchmarks-core-i7-6700k-hyperthreading-test.219417/[/url<] Those are just FPS numbers too and not frame time benchmarks. Personal experience is HTing causes stuttering and I've turned it off in every CPU since I first noticed it happening (9xx series). Usually leaving it on for a bit when I get a new one, then fiddling and ending up leaving it off. There are more games then you think that are affected by it, it's just generally not covered by major tech websites, much like core parking.

            • Krogoth
            • 4 years ago

            It is more of an issue with thread scheduling with certain games and OS (Windows still sub-par at thread scheduling compared to other OS platforms).

            SMT does have a cost to it. Silly intel shills are going into overdrive.

          • Klimax
          • 4 years ago

          IIRC That was Sandy Bridge were you could keep HT on always.

          And funnily enough, there are games which are for some reason crashing with HT. (Prototype)

          • I.S.T.
          • 4 years ago

          There are still occasional workloads that HT can mess up. Dolphin the gamecube emulator loses about 10% performance due to how HT effects the L1 cache.

      • the
      • 4 years ago

      The real benefit of HT is that background OS tasks and mundane applications that aren’t computationally intensive can still run in parallel with a heavy application.

      Only the Itanium and the PPE based console chips line really divided up clock speeds are you are describing. Intel’s Hyper threading works by having extra computational resources that a single thread simply cannot use, so the core runs two threads to maximize execution utilization. This can mean a decrease in single threaded performance but overall throughput it is generally positive as more raw work is being performed. Both threads would operate at the full 4.4 Ghz but one thread could monopolize the AVX unit while the other thread is integer heavily and they wouldn’t contend for resources.

        • Krogoth
        • 4 years ago

        That’s assuming that OS and applications are coded properly with scheduling. That’s not quite the case with mainstream stuff under a Windows environment.

      • ronch
      • 4 years ago

      Many folks think they totally understand Hyperthreading.

      • coolflame57
      • 4 years ago

      If you are only gaming, WHY THE FREAKING HELL do you need a 10 core/20 thread F**KING MONSTER of a CPU???????
      Games don’t utilize extra threads well anyway. Stick with the 6700k at max for gaming.

    • techguy
    • 4 years ago

    This is price-gouging at its worst. Intel offers no justification for the 70% markup for their top-end HEDT SKU, I hope no one buys this thing. I also hope AMD brings some competition to the high end before year’s end, otherwise this will become the new normal.

      • Ninjitsu
      • 4 years ago

      Well the 8-core is still priced the same…so technically there’s no markup on the equivalent thing.

      There IS a price increase for the lower two SKUs, though.

        • techguy
        • 4 years ago

        The 6900k is priced higher than the 5960x which has a listed price of $999-1059. Granted, it’s not much of a price increase at only $1089. My point is the top-end SKU is now 70% more expensive (the 5960x was last generation’s top SKU) and that is what I am lamenting.

        • the
        • 4 years ago

        But how much faster is the new 8 core vs the old 8 core?

        • w76
        • 4 years ago

        [quote<]Well the 8-core is still priced the same...so technically there's no markup on the equivalent thing.[/quote<] Not entirely true; depending on what all they're doing or adding, going from 22nm with Haswell to 14nm here with Broadwell my guess is that, counter to the trend established decades ago, it's possible you might be paying more per transistor, while their cost (in terms of area on a wafer consumed per core) should have fallen substantially.

    • anotherengineer
    • 4 years ago

    It’s a good thing cpu pricing isn’t dictated by multi-tread cinebench.

    Because if an fx-8350 is $160 for 642 points, then the 6950x at 1790 points would be worth about
    (1790/642)*$160= $446

      • chuckula
      • 4 years ago

      Good thing we are pretending the FX-9590 never launched, because given a probably inflated score of about 700 or so for the $900 FX-9590, the 6950x at 1790 points would be worth about:

      (1790/700) * $900 = $2301.

        • anotherengineer
        • 4 years ago

        Don’t be a hater.

        Wouldn’t you want a 6950X for $446? I know I would πŸ™‚

        Edit – Not pretending the FX-9590 doesn’t exist. In fact it’s still available from newegg for $230.

        [url<]http://www.newegg.com/Product/Product.aspx?Item=N82E16819113347&cm_re=fx-9590-_-19-113-347-_-Product[/url<] Not sure why you would you put it down as $900?? I guess that was the original launch price way back when??

          • ronch
          • 4 years ago

          AMD launched those silly nothing-but-overclocked-FX-8350s at $900, hoping AMD fanbois with more fanboism and money than rationality would bite. After the dozen or so 9590s were sold at $900 and sales took a steep dive to 0, AMD had to gradually lower the price. They could’ve dropped the price straight down to $230 but fanboism irrationality is not black and white; it’s like a graph where more fanbois would bite as the price decreases. 12 folks would bite at $900, 11 would bite at $800, and about a hundred bit at $230. Me, well, Rob Taylor and Adam Kozak can’t fool me even if they held each other’s hands and kissed. I know that the 8350 is STILL the top FX SKU. And those ‘new, energy efficient’ FX models? I could clock my 8350 at 3.3 and get the exact figures as those EE chips off the wall.

        • ronch
        • 4 years ago

        Too bad Intel just shrugged it off.

        If AMD can’t fool its fans, it can’t fool Intel.

      • Chrispy_
      • 4 years ago

      It’s a good thing the price/performance curve is a [i<]curve[/i<] πŸ˜‰

        • anotherengineer
        • 4 years ago

        It’s a curve for a bit, get almost straight up linear near the end though πŸ˜‰

    • El_MUERkO
    • 4 years ago

    Was thinking of an upgrade from my 4960X but now… … … meh!

      • TEAMSWITCHER
      • 4 years ago

      Even people running 3960X SandyBridge-E systems don’t need to upgrade just yet. I wish I had built one of those a few years back – the longevity of those systems is legendary.

    • brucethemoose
    • 4 years ago

    Have we gotten an official die size yet?

    If the 6950X is as big as a 5960X, with 2 disabled cores, pricing it at $1800 seems rather silly.

      • chuckula
      • 4 years ago

      Yes, posted in the earlier articles. It depends upon the die variant that Intel chose but IIRC the smallest size die in thelineup includes 10 cores. It is about 250 mm^2 or so, which is nearly identical to a Kaveri die for reference.

      They might be using the next lager size up with some cores disabled for the 6950X though.

        • brucethemoose
        • 4 years ago

        $1700 for a high volume chip that size seems ridiculous.

        The smallest Sandy Bridge-E die, for example, was almost 300mm^2 and cost *far* less than $1700.

    • Bauxite
    • 4 years ago

    If you’re going to blow nearly 2 grand, might want to wait and see what version 16xx E5 xeons become available.

    Or add another grand and get 22 cores on fleabay.

      • brucethemoose
      • 4 years ago

      Most people don’t know that the 16xx Xeons are overclockable though. They’d still sell plenty of 6950Xs if a cheaper, unlocked 10C Xeon existed.

    • Bensam123
    • 4 years ago

    Agh… I thought the low end would price match the 5820k at 380… But they decided to throw a extra $50 on top of it. Intel really doesn’t want normal people touching this stuff. That’s a damn shame as video games definitely could use a health dose of CPU processing power increase. Division, H1Z1, and even Overwatch pegging my current processor running at 4.4 we’re starting to see limitations and the next 5-10% increase in performance from Kaby Lake isn’t going to be enough to keep this charade going.

    I was totally going to hop on the six core when it came out from my 4690k, but now I might just wait to see what zen is later this year. Dropping $600 is a lot of money, that doesn’t even count the memory I’ll have to upgrade as well.

    AMD please…

      • BestJinjo
      • 4 years ago

      All of those games you described would run faster on a 4.8Ghz 6700K with fast DDR4 than any Broadwell-E processor. Looking at overclocking, BW-E won’t hit 4.7-4.8Ghz either. It’s a dud for gaming as 4 fastest cores with HT are the best combination for games until nearly every AAA game uses DX12 properly. By then we’ll buy Skylake-E/Icelake-E.

        • Bensam123
        • 4 years ago

        I disagree. When you see your cores all getting pegged, a extra 5-10% performance isn’t going to change that. You can’t cherry pick the silicon lottery either.

        Hyperthreading does crap all and sometimes even reduces performance for gaming. None E i7s are a huge sham and basically steal $100 from your wallet compared to i5 parts.

        For instance: [url<]http://www.techpowerup.com/forums/threads/gaming-benchmarks-core-i7-6700k-hyperthreading-test.219417/[/url<]

          • Andrew Lauritzen
          • 4 years ago

          There are pretty much no reasonable games right now that actually load anywhere near 100% of an Intel quad core desktop. Ashes might be the sole exception depending on settings. Beyond that you might not be analyzing it properly…

          If this were the case you’d already see HSW-E and the 6 core parts showing significant gains, but you don’t. Frankly if hyper-threading is doing nothing then the game isn’t pegging all your HW threads either.

            • Bensam123
            • 4 years ago

            Try the Division. I have a widget open on a second monitor that always keeps track of CPU utilization. How often do you look at your utilization?

            HTing doesn’t work properly because all threads aren’t created equal and Windows isn’t amazing at figuring out what goes where. HTing doesn’t magically double the amount of cores you have. Do some looking around. HTing does relatively nothing for games and most the time even hurts performance.

            • Andrew Lauritzen
            • 4 years ago

            I look at CPU and GPU usage of games as a major portion of my job πŸ™‚ Which is why I mentioned using various tools that just give you ballparks like task manager and so on are not really that great in these days of turbo and P-states and other “reactive” methods. Furthermore the tools do nothing to tell you if useful work is actually being done or if the game is just spinning a thread polling something, which is pretty common. It’s trivial to make task manager or similar tools say 100%. I can do that in a few lines of code πŸ™‚

            > “HTing doesn’t work properly because all threads aren’t created equal”
            Common misconception – there’s not a “main” and “hyper-thread” on a given core. The two threads *are* created “equal”. The lack of 2x boost has to do with the fact that only certain parts of the execution pipeline are duplicated, which is really the entire point in SMT. Again, HT does “relatively nothing” for games precisely because games are generally not CPU bound, and generally poorly multithreaded.

            Again, if your notion here was true then the 5960x should run the division significantly faster than the 6700k, right? I severely doubt that is the case.

            • Bensam123
            • 4 years ago

            So when a CPU sits at 100% constantly with a few dips in between, it’s not at 100%? The widget I’m using is not the same as Windows Task Manager.

            As I mentioned in my first post, it’s OC’d to 4.4ghz and sits at about 4.3 to 4.2 in games depending on what’s going on. I also have a widget to monitor the multiplier.

            I highly, highly doubt that apps are trying to cheat Windows task manager into thinking it’s using more processor then it really is. On top of that you can hear audio stutter sometimes and FPS dip when it gets maxed out.

            Windows still assigns threads to cores and doesn’t always intelligently distribute them to cores that have less strenuous workloads. More so when threads come down to latency and you have two very latency dependent threads happening on the same core.

            I never mentioned ‘significantly faster’ as with my original post, games are *starting* to bump into CPU caps and as a result you see stutter, frame drops, and if the game is coded well enough just a reduced frame rate. It happens even in Overwatch on occasion. Yes that game has hiccups. Hitting a CPU cap isn’t the same as having your GPU maxed out. GPUs tend to deal with 100% workloads much more gracefully then your OS and everything else that’s riding on top of it hitting your CPU cap… especially when it comes to latency sensitive applications, like gaming.

            You need at least 10-20% of breathing space on your CPU otherwise wanky shit starts happening. With Windows 10 this gap has narrowed to about 10% as it’s better dealing with workloads before vomiting all over itself.

            I have a Haswell, IPC is the same as getting a Haswell-E . If I can hit 4.2g with a 5820k I’m in roughly the same boat with single threaded performance. That has six real cores and not four fake ones attached. Once again… a none-E i7 is a extra $100 for basically worse performance in games. That also applies to Skylake.

            I was only talking about gaming performance, I’m sure there are all sorts of scientific simulations and acrobatics you can do with your CPU to make sure HTing does a good job. Most people don’t care about that stuff though and when you’re just using your PC you generally never bump into a CPU cap, regardless of HTing.

            You didn’t even take time to look at the link I posted before writing two responses.

            Don’t worry, not everyone is good at their job… especially if they’re just maintaining the status quo and reiterating what they think is common consensus. πŸ™‚

            • Andrew Lauritzen
            • 4 years ago

            Respectfully, you’re way out of your depth here on almost every point. If you’re not interested in learning, I’m not going to waste my time. FWIW I did even read your link and they are way out of their depths too…

            • DarkUltra
            • 4 years ago

            How about this:

            [i<]The performance with both cards suffers a little with eight cores enabled, and it drops even more when Hyper-Threading is turned on.[/i<] [url<]https://techreport.com/review/29090/fable-legends-directx-12-performance-revealed/4[/url<]

            • Andrew Lauritzen
            • 4 years ago

            The article itself discusses possible reasons for what they saw. Unfortunately it’s impossible to speculate further with the data given as the benchmark isn’t even publicly available and the game has been canceled. Thus it’s hard to know if the game itself was even running a static workload or adapting its own thread pool and tasking based on the machine thread count. I’ll also note that the difference in behavior between the Fury X and 980 Ti even in DX12 where the driver is not allowed to launch threads heavily implies that this is a (user mode) software effect.

            Obviously the latter situation – which is extremely common – confounds OS and hardware scheduling efficiency with the *software* efficiency of the game’s thread pool implementation and scheduling (usually work stealing) as well. Then there are a myriad of secondary effects that running more threads has from power to cache pressure.

            I don’t think anyone is claiming that the OS scheduling is perfect or there are no situations in which SMT will regress performance (whether it be in software or hardware). But it’s not super-common and it’s usually not by much.

            This is all tangential to the original discussion though – games simply *do not* put significant workload onto modern CPUs at regular settings, and they tend to do a terrible job at spreading their work across many cores, as both of the provided links show. Thus for gaming frequency and IPC is still king, hence why the 10 core BDW-E ties (when GPU bound) or loses to the 6700K in pretty much every game benchmark I’ve seen (again Ashes with specific CPU-heavy configs might be the sole exception).

            • Bensam123
            • 4 years ago

            Except they tested a crapton of modern games in the thread I linked you. They don’t count because it’s a person instead of a hardware site testing it?

            And as I stated, some games are starting to do that. Go test OW. Go test Division, especially in the darkzone. I can literally start the game up and go in the dark zone and have all four of my cores pegged.

            You’re unequivocally just wrong in this case. There is no clever wording that can change that. Your data is either out dated or you never tested it in the first place.

            • Bensam123
            • 4 years ago

            I’m not the one who isn’t interested in learning.

            So you looked at the link, saw a compilation of benchmarks showing you’re wrong and because of it they’re ‘out of their depths’. You even stated HTing doesn’t really help in single threaded gaming benchmarks. Why is it far fetched to assume it actually hurts it? Do some testing on your own.

            As far as OW using a lot of CPU resources, you can test that yourself as well as Division if you don’t want to believe the plebs and their ‘depth’. It happens every night. when I stream it. Occasionally it’ll bump against my CPUs hardcap and I’ll get stuttery audio.

            Jesus dude.

            • Andrew Lauritzen
            • 3 years ago

            Here’s the thing – at the level of anecdotal discourse you are wanting to discuss things (evidenced by your links), there’s a simple way you can prove that games are bumping up against the actual throughput limitations of four core CPUs: find a game that performs significantly (say 25% or more) better on an 8 core HSW-E/BDW-E than a 6700k.

            Other than Ashes and potentially some other strategy games (unlikely, but possible), I don’t know of any. Frequency/IPC is king for games because any CPU bottlenecks are related to serial execution of 1-2 threads.

            Let’s keep all the rest of the discussion off the table for now until you’ve demonstrated this basic effect you are claiming.

            • Bensam123
            • 3 years ago

            I once again never said significantly faster. I said that they’re bumping up against the CPU cap which leads to stuttering, FPS drops, and just lower all around FPS if things are programmed semi-decently like in The Division, but even in the Division you still end up getting stuttering, inconsistent gameplay, and the audio cutting out.

            Going from say a 4690k to a 5820k (both haswell) operating at the same frequency wont double or even increase FPS by 50%. It will however remove the bottleneck I’m talking about, which is occasionally bumping against 100% utilization and as such you wont get stuttering, FPS drops, and/or audio drop outs. In the case of the Division increased FPS all around.

            So what is ‘actual throughput limitations’? Are we talking about theoretical performance of a CPU? Because that really does not matter. We’re talking about real world performance here. I’m talking about games that exist and CPUs that don’t always perform up to par in them because they’re reaching max utilization.

            It doesn’t matter if it can be coded better or if there is something that will properly utilize HTing better, this is what is happening currently and what needs to be dealt with. Milking performance is not what this is about and never will be for the same reason people crap all over AMDs piledriver even though it had much better multithreaded performance then Intel chips at the time.

            Go play the Division, for the third time or even Overwatch. This happens to me every night maybe once per hour I’ll end up with stuttering and jerking in Overwatch. It lasts maybe a couple seconds. I also play Overwatch for 7+ hours at a time so it’s entirely possible you wont see the same thing happening simply by starting it up and sitting on the menu.

            “Let’s keep all the rest of the discussion off the table for now until you’ve demonstrated this basic effect you are claiming.”

            Go test it yourself. You said it yourself that this is your job. I’m not doing this for you. I already told you what’s happening. You don’t want to believe it, then you aren’t very good at your job.

            My point still stands, if you want to argue it, how about YOU actually presenting some meaningful information instead of posturing and talking in a condescending tone. I’ve offered benchmarks for HTing, I’ve talked about my personal experience I can replicate every night, and other people have offered benchmarks as well.

            • Andrew Lauritzen
            • 3 years ago

            > It will however remove the bottleneck I’m talking about, which is occasionally bumping against 100% utilization and as such you wont get stuttering, FPS drops, and/or audio drop outs. In the case of the Division increased FPS all around.

            I’m trying to give you the benefit of the doubt here, but what you’re saying just doesn’t make any sense. What is this “CPU cap” you’re talking about? What is “100% utilization” if not the game actually making use of every cycle on every core for meaningful work (which you admit isn’t really happening else it would be faster with more cores)? You’re implying that the game normally is using some small fraction of the CPU but all of a sudden spikes up and uses the entire thing for long enough to blow audio buffers (100+ms) and miss frames? That can certainly happen with a single threaded bottleneck and/or poorly affinitized/prioritized threads, but it’s highly unlikely for that to happen with the entire CPU. Either something is wrong with your system (or background tasks) or you’re not interpreting that data you’re getting properly. Again, this notion of “100% utilization” is not well-defined.

            > Go test it yourself. You said it yourself that this is your job. I’m not doing this for you. I already told you what’s happening. You don’t want to believe it, then you aren’t very good at your job.

            I’ve never seen the behavior you describe and I play lots of games. I play Overwatch frequently as well and have never had this sort of stuttering or any audio dropouts despite using VOIP all the time while playing.

            Of course my anecdotes are no better than your anecdotes, but the point is your reasoning is not plausible – and that’s something that my experience does allow me to say with credibility. I’m not disputing that what you are seeing is happening (I’m sure it is – why would you make this stuff up), I’m saying your interpretation of the causes is almost certainly incorrect as this stuff can be fairly subtle and without the right tools and experience it’s easy to misdiagnose.

            • Andrew Lauritzen
            • 3 years ago

            Took a brief GPUView capture of Overwatch last night while playing to see if I could see anything weird going on (despite, as I said, never seeing anything during gameplay). Here’s the results:

            [url<]http://pasteboard.co/1x496zwq.png[/url<] [url<]http://pasteboard.co/1x47jNlK.png[/url<] [url<]http://pasteboard.co/1x4bhlqs.png[/url<] First image: the GPU queue (see "3D queue" under GTX 980 section) is entirely full all the time and the CPU is practically idle on all of its threads (all the black bars in the main part of the image are idle time). The CPU is actually dropping into P-states a lot of the time because it has nothing to do... Second image: Zoomed in on a dozen or so frames. You can see the details on the steady state here... GPU always busy, CPU mostly idle other than small chunks of serial work on 2 threads (game render thread and NVIDIA driver thread). Third image: The game's device context and threads. As expected, it is completely GPU bound. At the end of each frame (where the red hashed stairstep bit hits the bottom) you can see that the application render thread starts to work (the orangy-bar 4th from bottom) it gets unblocked after Present). Shortly after that the NVIDIA user mode driver thread (nvwgf2um.dll) starts to work as well. The game then does a little bit of parallel work on a few threads and another chunk of serial work on the blue thread. The render thread and whatever the blue thread is never overlap, so these are likely effectively serial as well. Based on that, on this config it looks like we could run roughly ~2x faster before getting CPU bound on the single-threaded work. The blue lines are vsync's on a 144Hz display. They don't align with the game work because vsync is disabled in the game for benchmarking purposes, but they give you a reference point for ~7ms of time on the zoomed in shots. It looks like with an infinitely fast CPU I could be hitting close to vsync locked @ 144Hz on this machine. Since the bottleneck is serial in nature here, a higher clocked part like the 6700k would do even better. This is all consistent with what I've been saying, and very typical of most games. I just wanted to walk through it a bit to give you the idea of how this stuff works under the hood and maybe inspire you to do some more in depth analysis when you see issues vs. jumping to conclusions.

            • dodozoid
            • 3 years ago

            People like you are one of the reasons I read TR. There are valuable things to learn not only in articles but in the comments aswell.
            Thanks for insights like this and for your patience…
            (BTW do you realise how much time did you spend arguing on the internet?)

            • Andrew Lauritzen
            • 3 years ago

            Glad you appreciated it πŸ™‚

            I often struggle with how much time to spend on this stuff especially when it is buried in comments, but the optimistic side of me hopes that it’s worth trying if people are willing to learn.

            So yeah on one hand I spend a bunch of time that could better be used elsewhere arguing on the internet, but on the other hand sometimes folks like you stumble upon it and that makes it worth it. Thanks for commenting to that end.

            • Waco
            • 3 years ago

            I’ll add another thank you. I get so tired of this “magic” level of expertise from those that are only observing random data points with very little depth. That “expertise” eventually turns into incorrect “common knowledge”, and then it’s a pain in the ass to change everyones’ mind. It’s like reading buzzword soup from sales reps that absolutely believe they have the product that will solve all of your problems. You can try to call them on it, but you usually just get yelled at.

            So, thank you!

            • Mikael33
            • 4 years ago

            Do you know who you’re unsuccessfully arguing with?
            [url<]https://www.linkedin.com/in/andrew-lauritzen-2717b84[/url<]

            • Bensam123
            • 4 years ago

            Why would I care who someone is on their linked in? They’re wrong in this case. Saying they aren’t wrong doesn’t change that they are.

            • Captain Meatfart
            • 3 years ago

            Because he’s an Intel engineer who helped design DX12, you ridiculous git. Also, his style of argumentation doesn’t rely on defensive hostility. Also, when you kept prattling on about him needing to provide proof of his points, he did precisely that and explained in detail what was happening during each step of the render chain — all by testing the very game you kept insisting would prove him wrong. Then you promptly vanished.

            Because you are a ridiculous, prattling git.

            • Mikael33
            • 3 years ago

            I nearly forgot about this post, thanks for the hilarious reply, though I shouldn’t expect more from someone who didn’t realize even after several replies that you were in way over your head, the fact you were making bad arguments and doubling down on them against someone who works for Intel is just icing on the cake.

            • DrDominodog51
            • 4 years ago

            [quote<]I can do that in a few lines of code :)[/quote<] But are those lines in NASM assembly? πŸ˜€

            • Krogoth
            • 4 years ago

            There are a number of games out there that are painfully CPU-bound on modern chips but they tend to fall under strategy and sandbox genres not silly shooters.

            The performance issues and hic-ups with some games under SMT actually come from scheduling issues with OS (NT 6.x is still inferior to *nix at this) and applications (It is a PITA to properly multi-thread a program with SMT). SMT only make sense for professional-tier applications or you are doing heavy multi-tasking. It makes little sense for mainstream usage patterns.

            • Chrispy_
            • 4 years ago

            I tend to find that when I stumble on a game that runs badly even on low graphics settings (ie, it’s a CPU bottleneck) that a mad overclock does far more than more cores.

            Games are mutithreaded, yes.
            Multithreaded games still choke because of [i<]one thread[/i<] throwing a tantrum; Speed up that thread with a 30% overclock and you get a 30% performance increase. Going back to the Lauritzen/Bensam argument - I'm happy as a layman to just pretend that "games reserve CPU power on all available threads". I know that's oversimplified but it's a model that works and correlates with my real-world experience. Not only that, but if you launch a CPU-hungry game with specific processor affinity, you'll find that your gaming experience is actually better because the game doesn't hog the whole processor. I shortcut all my games to run on just six threads so that there's always a couple of spare threads to let voice comms, web browsers, and the OS to run unhindered on.

            • Bensam123
            • 4 years ago

            CPU affinity only really affects things if you’re bumping against your CPU cap. IE the processor is getting maxed out and yup it helps specifically with all the system processes running in the background.

            Changing affinity doesn’t make a difference if you have 10-20% of overhead available I’ve mentioned already and will just artificially limit the frame rate in your games and/or make things operate worse regardless.

            If you’re bumping up against your CPU cap you’re going to end up with a rough experience regardless.

      • Voldenuit
      • 4 years ago

      [quote<]Division, H1Z1, and even Overwatch pegging my current processor running at 4.4 [/quote<] Are you running on 144 Hz or 4k? If not, the reason Overwatch could be pegging your FPS is because the default setting caps FPS at +10 over your monitor scan rate. You can unlock that setting and get moar FPS, but it won't do you much good. OW has been a surprisingly good performer on low and mid-range systems. [url<]http://www.techspot.com/review/1180-overwatch-benchmarks/page3.html[/url<]

        • Bensam123
        • 4 years ago

        144hz… it’s uncapped.

        Getting more FPS increases the simulation and as such the fluidity and responsiveness of the game regardless of your monitor refresh rate. There is a reason pros are running CSGO at 300fps.

          • Chrispy_
          • 4 years ago

          LOL, server still ticks at 20fps tho πŸ˜‰

            • Bensam123
            • 4 years ago

            Yup… one of the biggest problems with Overwatch at the moment. I talk about this every night. πŸ™

      • squeeb
      • 4 years ago

      If Overwatch is pegging your CPU, it sounds like you have bigger issues. That game puts absolutely no strain on my ancient FX-6300 build.

        • Bensam123
        • 4 years ago

        You never watch your CPU.

    • tipoo
    • 4 years ago

    [url<]https://www.youtube.com/watch?v=bKx4HzZvvEg[/url<]

      • derFunkenstein
      • 4 years ago

      everthang in the computer need my face on it, is what i’m sayin

    • JosiahBradley
    • 4 years ago

    Once again the performance for dollar metric doesn’t change. Sure you get 2-4 more cores but at a linear price increase. Also looking at preliminary overclocking results you might be better off with less cores at a higher clock anyhow.

    • jihadjoe
    • 4 years ago

    $1723: “We don’t really wanna make this chip, and you don’t really need it, but if you’re crazy enough to buy it…”

    Anyone who actually needs 10 cores will be better served by a proper Xeon with ECC support. The workstation-focused [url=http://ark.intel.com/products/91750/Intel-Xeon-Processor-E5-2687W-v4-30M-Cache-3_00-GHz<]E5-2687W [/url<] is just ~$420 more expensive, and comes with two more cores, ECC and dual-socket support.

    • chuckula
    • 4 years ago

    Did Intel send over samples?

      • derFunkenstein
      • 4 years ago

      well Jeff posted a couple preliminary benchmarks so I would have to guess yes.

        • chuckula
        • 4 years ago

        Good to see that at least Intel is still sending over hardware for review.

      • Neutronbeam
      • 4 years ago

      Is Intel sending over any samples to ME? There, FTFY. πŸ™‚

        • chuckula
        • 4 years ago

        I’ll take two review samples and an order of fries, TO GO.

    • Chrispy_
    • 4 years ago

    Thing is, this chip is only valuable where the 25MB cache is a genuine asset. That’s large dataset analysis and is usually farmed out to server arrays with hundreds or even thousands of cores in the companies I’ve consulted for and worked with.

    For a lot of the target market, such as CPU render and CPU analysis through parametric iteration, the software most people are running is designed to run as a distributed workload across multiple machines. There’s an entire industry of competing products to manage distributed workloads and they compete with the in-built distributed renderer/analysis tool of the primary product people are using.

    Given that (pulling a number out of my ass) 90% of the target market for these things are using distributed-workload software, Two “cheap” 6700Ks are matching the multi-threaded performance of this $1723 chip and doubling the single-threaded performance. You could actually buy quadruple the compute power and enjoy the flexibility of a farm for the cost of a single 6950X workstation and that’s even accounting for multiple software licenses and extraneous hardware.

    • ptsant
    • 4 years ago

    I’ll wait for Zen 8-cores. If Zen gives 80-90% performance at a much better price, I’ll buy it. Otherwise, I’m seriously tempted by the 6850K. The $600 is a lot, but you also get 8-slot RAM and I believe single-threaded is not going to improve tremendously. This is a very future-proof processor.

      • tipoo
      • 4 years ago

      Yeah, if it hits Haswell per-core performance and offers 8 of them at a good price, that would be tempting, for the not huge gulf between Skylake and Haswell.

      • BestJinjo
      • 4 years ago

      No point to the 6850K. PCIe 3.0 x8/x8 is at most a 2-3% bottleneck for 1440p/4K gaming. There are more than enough lanes on the 6700K/6800K for 1080 SLI and fastest M.2 SSDs. 6700K overclocked will beat everything in this line up for gaming, assuming you pair it with at least DDR4-3200.

    • Krogoth
    • 4 years ago

    Not shock at the current pricing model for top of the line model. It is a prosumer product for those who don’t need ECC support. There’s probably little overclocking and thermal headroom on these chips.

    i7-6950 just makes i7-6700K look like a good buy for normal desktop users.

    • the
    • 4 years ago

    Whoa, $1723 for 10 cores? Granted it is $1000 cheaper than the Xeon E5-2689 v4 which also packs 10 cores at 3.1 Ghz but comes with ECC memory support and dual socket support. However, if you want 10 cores, the Xeon E5-2640 v4 comes in at $939 with the same ECC and dual socket support but at a default clock of 2.4 Ghz (and a nice 90W TDP). Or if you have $1723 to burn, might as well step up to $1745 for the Xeon E5-2680 v4 and its 14 cores at 2.4 Ghz consuming 120W of power.

    Basically the only reason to pay the $1723 is for the high clock speed with an unlocked multiplier. Otherwise, Xeons are arguably a better value when it comes to core count. I haven’t even explored the option of getting two Xeons with more total cores for under that $1723 price.

      • DreadCthulhu
      • 4 years ago

      There is also the used chips from ebay option, if you need a lot of cores; the slower clock speed Xeons are silly cheap there. You can get a Xeon E5-2658 v3 , 12 cores @ 2.3ghz, for [url=http://www.ebay.com/itm/Intel-Xeon-E5-2658-v3-ES-LGA2011-3-12C-Compatible-with-X99-i7-5820K-5930K-5960X-/151989476551?hash=item2363475cc7:g:xzMAAOSw5dNWs997<]$159, as an example. [/url<] Rather tempting, though silly for my use case.

        • smilingcrow
        • 4 years ago

        ES = Engineering Sample so you need to check the stepping as the early ones may not be supported on consumer single socket X99 boards. Some manufacturers are friendly enough to say which is the earliest stepping they support in the BIOS support page.

    • Klimax
    • 4 years ago

    Was thinking about getting 10 cores. Sorry, price kills that. We’ll see next CPU update.

    • excession
    • 4 years ago

    Jeff, when are we going to see some more hardware reviews on this hardware review site? TR feels like a news aggregator at the moment.

      • homerdog
      • 4 years ago

      They don’t even have the 1080 review up yet =(

      • TwoEars
      • 4 years ago

      Heavy lies the crown.

        • albundy
        • 4 years ago

        Thats why you have to HODOR!

      • PrincipalSkinner
      • 4 years ago

      +1

      • adampk17
      • 4 years ago

      I’m not sure how I managed to do it but I down voted this 3 times when I meant to up vote it 3 times. I’ve had this twinge of fear that TR’s best days are behind it. I sincerely hope not.

      • torquer
      • 4 years ago

      TR has often fallen prey to the oft-delayed hardware review. I love this site and have paid to be a gold subscriber because I believe it is the best tech site out there. Many here agree, but many of those people also think to love something means you can never offer legitimate criticism. There is no legitimate excuse for such a delay on the 1080 review (or anything else with such a long delay). It is of primary interest to the core audience and I might say even more of an interest and draw than trade show news bites.

      It is not good enough to excuse it away or give the Valve-regarding-half-life line of “you’ll see it when you see it.” That drives readers and views to other sites. It solves no problems and creates others. A quality product can and should be able to be produced in a reasonable timeframe and thats not happening. Unfortunately that reality is bolstered by many commentators here who suggest that the only options are a rushed hackjob article or a pristine pulitzer worthy piece 6 months after the product launches. The truth and beauty is somewhere in between.

      Don’t encourage complacency in the things and people you love. You do them and yourself a disservice. I’ve let my gold membership lapse in protest.

        • Chrispy_
        • 4 years ago

        Got to admit I used to be a gold subscriber but I didn’t feel it was worth it this year as the frequency and number of actual TR in-house articles has fallen significantly.

        TR articles still seem to be some of the best reads on the web when they eventually arrive but I do now go elsewhere for the latest news and come back to TR later to get a juicer second opinion on it.

        Not only that, but increased delays also increase expectations of the review in question so whilst I’m sure TR’s article will touch areas that other reviewers haven’t yet, it’s only a matter of time before the things an incoming TR article would cover start to appear on forums like Reddit anyway. That takes the “late but worth waiting for” and turns it into “just late”.

        I’m not criticizing TR writers, but it’s obvious to me that revenue for all tech sites around the web is dwindling and the last thing in the world that will bolster that revenue with new readers and subscribers is overdue reviews and a reduction of original content πŸ™

          • torquer
          • 4 years ago

          Agreed, sadly.

          • NeelyCam
          • 4 years ago

          [quote<]Got to admit I used to be a gold subscriber but I didn't feel it was worth it this year as the frequency and number of actual TR in-house articles has fallen significantly.[/quote<] Same here

          • Arbiter Odie
          • 4 years ago

          TR is trying to do Computex coverage. TR is understaffed. The review process for a GPU is time consuming. The product of those data is slow reviews. TR needs to get more writers hired, and that takes time.

          Think of it like a friend who’s got cancer. You’ve gotta keep supporting them or they die. And they won’t get better instantly. Payoff is not instant.

          I think we should be making an effort to keep TR alive. The internet needs at least one good tech review site, and unless you guys know of some other outfit that’s better suited to replace them, we will all be worse off by abandoning this ship.

          Also Scott will be sad.

          /soapbox

          Full disclosure, this had some major edits.

            • NeelyCam
            • 4 years ago

            I see your point.

            If you want to make an effort to keep TR alive, maybe you should be a subscriber…?

            • Arbiter Odie
            • 4 years ago

            Trying to figure out how to justify it to my parents– I’m still a dependent. Maybe both of us should do it…

            EDIT: Done.

            • NeelyCam
            • 4 years ago

            Walking the walk. +1

            • torquer
            • 4 years ago

            I think of it more like supporting vs enabling. TR has had this problem for a long time. Hell I remember complaining about it before I first subscribed and that was a few GPU generations ago if I remember correctly. If your brother is on drugs and keeps asking for money, at some point you stop helping and start enabling.

            TR will not get better by lowering the bar. No one ever gets better when the bar is lowered.

            • Arbiter Odie
            • 4 years ago

            I see your viewpoint, and I do understand. But do you not think the staffing shortage is the current problem? I mean, Scott and the other two guys that wrote a lot of the articles are gone now, so they have to fill those slots. Only 24 hours in a day, and all that.

            • torquer
            • 4 years ago

            No because this isn’t the first time and it isn’t just since Scott left.

            • Meadows
            • 4 years ago

            So TR has cancer. That’s comforting!

        • NeelyCam
        • 4 years ago

        1080 review is “delayed” because NVidia didn’t send TR a part to test.

        What happened to the whole “yes – delay is fine, since the quality of the review is going to be so much better than in other reviews”?

          • torquer
          • 4 years ago

          Other review sites had theirs no more than 10 days prior to launch. Its now been well over 10 days since it launched. Even if they received it on launch day (they actually got it sooner per the previous thread on the topic), its reasonable to say it should have been done.

          I’ve actually never excused away unreasonable delays for a belief in an awe inspiring final product. How many times in life do you get to say to your customers “its okay that its taking 3x as long as our competition because you’ll love the end result”? I’d say anything over a week beyond what other sites are doing is an unreasonable delay.

          I think most of us are willing to give some slack, but this is a pattern for TR, not a one off scenario. Can’t blame this one on Nvidia.

            • Chrispy_
            • 4 years ago

            Aye, it’s the pattern over the course of the last 12-24 months rather than this being an exception because of Nvidia.

            The product has been on store shelves for a whole week now and other reviews appeared all over the web 18 days ago. 18 days ago Jeff confirmed that TR had a 1080 in the lab undergoing testing

            Other sites (eg Guru3D) have already come back for the more in-depth analysis of things like [url=http://www.guru3d.com/articles-pages/fcat-geforce-gtx-1080-framepacing-review,1.html<]frame pacing analysis[/url<] and I've already read 1080 SLI benchmarks in a few places. The TR 1080 review is now so late that even TR's hallmark in-depth coverage of the card might not appease the impatient. Me? I'm not impatient. I won't be buying another GPU until Polaris reviews have come and gone, but I can appreciate why people are frustrated and I can appreciate that people are trying to defend TR because the reviews are usually high enough quality to be worth the delay. How much delay is worth it though? Unless Jeff's working weekends we're not going to see it until next week, which means three weeks. THREE WEEKS on the hottest product of the year, no question of a doubt about that.

            • torquer
            • 4 years ago

            Also, there is no reason they could not at least give a “pre-review” as some sites do to talk about the tech and give some initial findings followed by a more comprehensive analysis.

            I agree with all who say TR does the best reviews. Are they so much better that they justify any length of delay? I doubt it. PcPer, Toms Hardware, a few other sites do reasonably good reviews in much shorter amounts of time. What can you do with an extra 3 weeks that you can’t do in the same timeframe they did?

        • Convert
        • 4 years ago

        Well said. Sadly I too let my membership lapse, not so much in protest, but because TR isn’t producing anything yet that is worth paying for. That may sound harsh, but it’s just a reality. I gladly jumped at the chance to buy a membership when they were first offered because they were producing good content. TR is a business and not even bothering to communicate about the state of affairs shows the income isn’t a problem so I don’t have to worry about it making their lives harder. If they posted, as Scott has before, about having a tough time then I would have kept my subscription active to help them out.

        The whole argument about TR’s reviews being late but being better quality never really held any water to begin with. There are other competent review sites and there was always at least one with more in depth coverage. TR reviews are amazing, don’t get me wrong, they often have something unique to them and I’ve just grown accustomed to liking TR’s way of doing things. Day 0 review releases aren’t important to me as it’s not like I’ll be buying *whatever it is* that day. 0 Day review releases are critical to the news site though to generate all those page hits. Hardly anyone in comparison is going to be reading a review released a month later unless it’s a revolutionary frame pacing style review.

          • torquer
          • 4 years ago

          Anandtech was once a great site. Then Anand left. Now it is a shadow and shell of its former self. If they even do PC hardware reviews, they are often late as well. They were famous for doing great reviews (though often biased in the positive) on Apple products. They just released their iPad Pro 9.7 review a couple of days ago long after it came out.

          The model that worked in the late 90s up until the last couple of years doesn’t work anymore. Money is now being made on youtube (LTT, anyone?) vs the legacy journalistic sites. This is why many of them have moved toward more news agg and click bait and away from the types of reviews and articles many of us older folks loved back in the day.

          Scott and Anand left their cherished sites for a reason. Theres just not enough money to be made. If Scott and Anand were making 6 figures and supporting their families well through income from their sites, would they have moved on to AMD and Apple? Possibly, but its hard to turn a blind eye to whats best for your family when an opportunity like that comes along.

          I hope we aren’t witnessing the end of something that has been a big part of my life for so many years. I fear that TR is just the latest to fall victim to the new economic reality where a shrill voiced Canadian kid can build a mini media empire on youtube while a pure tech site like TR goes by the wayside.

            • Arbiter Odie
            • 4 years ago

            The Anandtech things is spot on. I used to visit them on a daily basis, there were always new and relevant things to read there– especially their SSD stuff. Anand’s Anthology of an SSD was very influential on me, and was actually what started my learning how computers worked.

            Now there’s nothing to go there for. So sad πŸ™

        • dodozoid
        • 4 years ago

        +3
        Edit: It was TR that made me a hardware enthusiast. I was sad when Cyril and then Geof left the site and I almost cried when Scott anounced his departure.
        TR will never be the same, but so far I believe Jeff can make TR great again if gerbils remain loyal. (Any similarity with your future presidents marketing slogan is entirely coincidental)

        Plus this comunity is great and you can get valuable content an deep analysis in the comments…

      • jensend
      • 4 years ago

      I’ll note that the business of collecting data and turning it into graphs can be reasonably well automated. A lot of places had preliminary reviews — data, graphs, initial impressions — up as soon as the NDAs lifted, then filled in more of the editorial content as time allowed.

      I know in the past TR has relied on huge Excel spreadsheets that included everything from the raw FRAPS data to the finished graphs. I wonder how much of the review time is spent tinkering with the spreadsheet; it seemed like a relatively slow and brittle tool for the job. Perhaps doing a little numpy+matplotlib or summat could speed that process considerably. On the other hand, even if they get outside help setting that up initially, they would need to make sure someone on staff could make adjustments in the future.

      One place I’ve found myself visiting a little more frequently as TR reviews have gotten scarcer is computerbase.de. This is partly due to their review quality, partly because [url=http://www.computerbase.de/2016-05/geforce-gtx-1080-test/7/#abschnitt_benchmarks_in_2560__1440_und_1920__1080<]their responsive comparison charts are so handy[/url<], and partly because it's an opportunity to practice my rusty German.

      • ronch
      • 4 years ago

      I think this is the highest thumbed post I’ve ever seen.

      Wake up, TR. Time to Turbo up those cores.

      • MarkG509
      • 4 years ago

      The gerbils have squeaked!

      • Meadows
      • 4 years ago

      As much as I value patience, there comes a time when I upvote comments like this.

        • NeelyCam
        • 4 years ago

        The last news article posted on TR was from three days ago.

          • Jeff Kampman
          • 4 years ago

          Dude, [i<]it's the Monday after Computex.[/i<] I just got home after a day of sleepless international travel and our guys are working on what little is out there. Of all the days to judge us about slow news days, this is not a good one.

            • NeelyCam
            • 4 years ago

            Got it. I keep forgetting the Computex…

            Melatonin is magic when jetlaggin’. I highly recommend it.

      • excession
      • 4 years ago

      I was scared to come back and check how this comment was received. I’m glad I’m not the only one – I’ve also just paid up for my Gold subscription and I suggest you guys chip in too if you want to help.

        • bjm
        • 4 years ago

        Your post hit a much needed point that needed to be made. Thanks for doing it so eloquently, to the point, and without the unneeded sarcasm many of the posts these days are posted with.

      • dpaus
      • 4 years ago

      Let’s be clear: TR has not been a ‘hardware review’ site for years. I absolutely NEED to stay on top of hardware reviews for work, and I stopped coming to (er, ‘living on’??) TR when I monitored the coverage for a few weeks and confirmed that about 30% of all articles were gaming-related (game release rumours, game reviews, etc), that almost every ‘hardware review’ basically consisted of how well that hardware could run games, and that every ‘System Builders Guide’ was strictly about building gaming rigs at different price points.

      If you want to be a gaming site, that’s fine, just be honest about it and change the tagline to ‘PC >Gaming< Hardware Explored’.

      Whatever the future of TR is, I wish you well!

        • Raymond Page
        • 4 years ago

        Do you mind sharing what you /*want*/ to see covered hardware-wise and where the short-falls have been. Of the below four coverage areas, what’s wrong with the hardware coverage, quantity, quality, missed-items, lack of coverage?
        1) System Guides – Build-outs with budget, sweet, hot-rod, god
        2) DIY PC components – cases, power supplies, motherboards, cpus, memory, graphics, HDDs, SDDs
        3) Niche Devices – SFF, HTPC, Silent/Passive, DIY-boards (e.g. Raspberry Pi), Android Tablets, Phones
        4) Accessories – Mice, mice pads, speakers, headphones

        Status:
        1) TR maintains the best system guides currently available.
        2) TR does routine in-depth reviews of cases, motherboards, CPUs, and graphics cards as they become available. If you want more reviews, care to elaborate on what components and price/performance points are interesting?
        3) Not highlighted, but there are some passive cases and devices that get PR coverage (aka news aggregator).
        4) My personal take on accessories is there’s one story – RGB-LEDs have been integrated with independent LED color control to enhance the mood.

        • bjm
        • 4 years ago

        In case you missed it, since as far back as I can remember the forums, the Gaming forum has been subtitled with “How we justify the high-dollar hardware”.

      • f0d
      • 4 years ago

      i wonder how much longer that nano review will be?
      sept 3rd 2015
      [quote<]Looks like I'll have a card to test early next week. May not hit the Thursday launch date with a review, but should be too far off of it. Thanks to everyone for their support.[/quote<] [url<]https://techreport.com/news/28971/wanted-for-review-amd-radeon-r9-nano[/url<]

        • travbrad
        • 4 years ago

        It was so small they couldn’t find it!

        • Meadows
        • 4 years ago

        But that was before all the key people left. Or key person, rather.

        TR hasn’t had a CPU or GPU review in at least 6 months.

          • f0d
          • 4 years ago

          scott left in december
          [url<]https://techreport.com/blog/29390/into-a-new-era[/url<] in september he said [quote<]Looks like I'll have a card to test early next week. May not hit the Thursday launch date with a review, but should be too far off of it. Thanks to everyone for their support.[/quote<] the review of nano wasnt supposed to be too far off in september and as i said scott left in december - thats 3 months before he left i dont want to be a "TR hater" but things have been going downhill for a long while now - either way ill still be here until the end edit: i misunderstood - yes that was before the key person (scott) left, so what would the excuse have been for totally not having that review that was supposed to be fairly soon from the post? again my point has been that TR has been going downhill for a long time now

      • albundy
      • 4 years ago

      you cant tell me that you dont get excited reading threads on pc shipment forecasts and how much money multi-billion dollar companies are making in their most profitable quarters?

      • Jeff Kampman
      • 4 years ago

      Look, I’m not any happier with the pace of feature articles than you, the audience, are. Here’s what we’re doing about it.

      For one, we’ve hired a new full-time staff writer, Robert Wild, to take over Scott’s work as the site’s CPU and graphics reviewer. We’re also developing a talented stable of promising freelance writers in the background to expand our coverage in some critical areas like notebooks.

      The problem is that all of these folks are green, and they need training. The only way to get people into shape for writing feature-length articles is to have them write short-form content like our news posts day after day. Each of those posts requires a significant amount of time in editing, and it takes even more time to generate the kind of detailed feedback necessary to help those writers grow. All of that is time away from writing feature articles. We can’t simply stop publishing news because it’s a major component of our content mix.

      Most of my time since I took over Scott’s job has been spent training, not writing. I would rather that balance be shifted in the opposite direction, but like it or not, those are the demands of running this site at the moment. It’s extremely challenging to have three full-time staffers with over 30 years of combined experience walk out the door in close succession. Rebuilding that depth and breadth of knowledge has been slow, hard going.

      With that background out of the way, here’s a short update on our pending reviews. In all honesty, we could not have produced a GTX 1080 review when we got the card. We weren’t expecting to receive one, and the fact that we did get one created an emergency. Getting Robert trained on frame-time benchmarking was a huge challenge, but it’s done now and he’s hard at work.

      We also weren’t expecting a Core i7-6950X sample from Intel, but one arrived and that was another demand on Robert’s shoulders that we just weren’t prepared for. There was no way either of those products was going to get a launch-day review, or even a couple-days-after-launch review. Sucks, but that’s the way things worked out. Frame-time benchmarking is integral to our CPU reviews, and there was no way Robert could move forward on that project without preparing for the GTX 1080, too.

      I despise writing responses like this one because it sounds like a bunch of excuses and doesn’t actually result in any new content being posted to the site. The bottom line is that it’s still going to be a while before we get the big stuff like the GTX 1080 and the Core i7-6950X reviewed. The good thing is that the groundwork for those reviews has been laid now, and we should be able to respond much more quickly to major hardware releases in the future.

      Anyway, that’s what’s up. If you’ve moved on to reading other sites because they’re producing more timely reviews and/or more of them, I can’t blame you. We have been slow, and it’s been a huge problem for us for quite some time. I’d like to think there’s light at the end of the tunnel now, so hopefully the front page bears that out soon. Thanks for sticking with us if you have, and no hard feelings if not.

        • excession
        • 4 years ago

        Jeff,

        Thank you for your candid and detailed reply. I can only imagine the challenges you guys face, especially when the level of expectation from your readers is so high.

        I should also apologise for the flippant nature of my post; I didn’t envisage creating quite such an uproar.

        Keep it up, we are all excited for TR’s future.

        James

        • dodozoid
        • 4 years ago

        This post-reply would deserve to be made into etc. article so more people would read it. Reason for impatience is IMHO more lack of information than lack of reviews.

          • Duct Tape Dude
          • 4 years ago

          Dodozoid is right, Jeff. Wrap this up into an Etc post.

            • NeelyCam
            • 4 years ago

            I agree.

          • jessterman21
          • 3 years ago

          Yep, came here to post that exact thing.

        • techguy
        • 4 years ago

        Jeff, thank you for taking the time to respond to the concern of many of TR’s loyal readers. I think most of us can appreciate the situation you are in and are willing to give some degree of leniency while we continue to eagerly await the aforementioned reviews. If I could make one suggestion? Would you write this up as a news post so everyone can see it?

        • ludi
        • 4 years ago

        Jeff — this is exactly what we needed to hear. Can you shorten it to an “Etc.” post for the front page?

        • torquer
        • 4 years ago

        I think for those of us, myself included, who have complained and shown concern were mostly looking for something like this. I don’t take this as excuses but rather reasons. You’re aware of the problems, you’ve discussed why they are problems, and you’re talking about what you’re doing to fix it.

        We care about this site. We complain not just to be trolls (well most of us) but rather to inspire change and get a response from leadership. You’ve done that today and I respect you for it. Theres still some life left in this site I think πŸ™‚

        I do agree with the posters below – this should be front page material.

        • Convert
        • 4 years ago

        I can’t even imagine what it must be like trying to rebuild the site from scratch and still keep an eye on future expansion. Good luck Jeff, thanks for all the hard work!

        • Meadows
        • 4 years ago

        Thanks for the update!

        • deruberhanyok
        • 4 years ago

        Thanks for the update Jeff!

        I wasn’t planning on going anywhere – maybe it’s time I subscribed, too. I want to see TR continue to do well. πŸ™‚

        • ImSpartacus
        • 4 years ago

        The candor is refreshing.

        • Sargent Duck
        • 4 years ago

        Thank you for explaining this.

        I, like many others, would think it would be most beneficial to post this on the front page. Us gerbils are an understanding bunch and I can TOTALLY accept what you just said. You just needed to have said it a lot sooner : )

        • chuckula
        • 4 years ago

        Thanks Jeff. We believe! I hope you are over the jet lag from Taiwan too.

        • bjm
        • 4 years ago

        [quote<]I despise writing responses like this one because it sounds like a bunch of excuses and doesn't actually result in any new content being posted to the site.[/quote<] Jeff, I just renewed my subscription based solely on your post (and, in turn, excessive's. I'll upgrade to Gold when I can). I appreciate your honesty and everything you're doing for this site. As others, I did not renew immediately due to the recent events (or lack thereof), but I hope that you find the +s to your post show that your audience can separate lame excuses vs. an honest update on the status of this site. It may have taken excessive's "excessive" post, but in the end, it got the much needed update we needed. All I can say is Jeff, trust that you can be open with your audience, we come here from the truth--even if it's in the form of why new content is not present.

        • LoneWolf15
        • 4 years ago

        Jeff,

        Thank you for the reply, and for your candidness. It’s a lot better response than another company I can think of in recent history.

        I agree with others; just us knowing this is helpful, and would be the kind of post that would be extremely helpful in an “Etc.” because we would know what was going on. Just knowing gives us understanding, as opposed to being in the dark and really makes a difference.

        • Bensam123
        • 4 years ago

        Sounds like a opportunity to do a compilation. 1080/1070, if it’s late enough review the non-Founders Edition cards as well.

        • tipoo
        • 4 years ago

        On the contrary, it doesn’t sound like excuse making. I’m glad to hear steps are being taken to shore up the ship for the future, that’s a great thing to do even if it stalls current reviews. To be honest I was getting a little worried, but it sounds like things should shape up.

        • Mr Bill
        • 4 years ago

        I’m glad I re-upped my subscription so I could give this post +++ (;-)

        • Wonders
        • 4 years ago

        Knowing is half the battle!

        • Cuhulin
        • 4 years ago

        Jeff,

        Thank you so much for writing this. I suspect that all, or close to all of TR’s members can relate to what you said. For one, I will give you all the time you need.

        Your saying it, however, is not making excuses. It is providing information that was needed. Otherwise, minds wonder.

        Thank you again!

        • GrimDanfango
        • 3 years ago

        I can feel the stress in your response, entirely understandable of course…

        Just want to say, don’t let it get you down Jeff – keep chipping away, and things will be running smoothly again sooner or later. Some people might take issue, but I don’t reckon you’re about to alienate your core readership – it took me a long time to settle on Techreport as my primary go-to tech news resource, and it’ll take more than a slow period to drive me away.

        For what it’s worth, one of the other big reasons besides the excellent feature reviews that I’ve kept coming back all these years, is actually *because* of your nicely curated news aggregation… so don’t assume nobody appreciates that side of the site too! πŸ™‚

        • Kougar
        • 3 years ago

        Sure it reads like excuses, but most people understand things do happen and in this instance most of us already know what those things were. Now we at least realize there are additional factors compounding the problem.

        If anything it is refreshing to read your post because A) We know things are in fact being addressed as fast as is reasonably possible and B) Tech Report is actually getting sampled again from the big guys. Last I had read that wasn’t the case, so that’s good news indeed.

        • maxxcool
        • 3 years ago

        Thank you for the transparency! Keep up the good work.

        You still have my SUB unlike others who clearly do not have the patience or attention span of gnats.

        • cobalt
        • 3 years ago

        FYI, I just upgraded back to gold based on this post alone. Agreed with the others — this doesn’t come across like an excuse, it comes across as transparency. It’s greatly appreciated, and I still look forward to reading the articles.

        • kamikaziechameleon
        • 3 years ago

        Even at half speed still the best tech site on the internet, that is why we want your reviews. πŸ™‚

        • BIF
        • 3 years ago

        My support too. Keep on keeping on, Jeff.

        Training new talent is critical. Many companies don’t do it, and eventually they’ll have to pay the ferryman’s painful rates. Or swim.

    • NovusBogus
    • 4 years ago

    Bwahahahahahahahahahahahahaha, they want *how much* for that? I mean, I get that it’s basically an E5 Xeon but still…

      • shank15217
      • 4 years ago

      Its not really a E5 xeon cause it doesn’t support dual socket so Intel just screwed with your wallet. With today’s games using more and more threads, single thread performance isn’t as important.

        • Krogoth
        • 4 years ago

        Not even close. The overwhelming majority of mainstream content operates with one or two threads. There’s a small minority that harness three to four threads.

        The majority of the applications that utilize tons of threads are in the hobbyist and professional realm.

          • ThatStupidCat
          • 4 years ago

          So if you are running 4 applications that use 2 threads each so I assume the cpu know to spread those applications across different threads so they’re all not jammed into the same 2 threads? Pardon the newbish question but you gerbils know a LOT more than I do.

            • brucethemoose
            • 4 years ago

            That’s actually the operating system’s job. Windows is mostly OK at it, but it’s not perfect, which is why some people like to manually assign core affinities to certain programs.

            Google thread or process scheduling if you want to know more.

          • shank15217
          • 4 years ago

          Yea so this cpu is aimed toward what demographic exactly?

            • Krogoth
            • 4 years ago

            Prosumer market for people who need tons of threads but don’t care for ECC.

        • derFunkenstein
        • 4 years ago

        So that’s why, over time, AMD’s CPUs are getting closer to Intel’s in gaming performance.

        Oh, wait…

        • Andrew Lauritzen
        • 4 years ago

        The graphs from games in the various reviews make it abundantly clear: for games that are decently threaded almost all CPUs perform the same (i.e. there’s still plenty of headroom… nothing is really maxing out even 4 cores). For games that are not threaded, the client quad cores (6700K etc) are at the top for obvious reasons.

        i.e. in places where your CPU matters for games, you want frequency and single threaded performance.

        Very different in the professional applications of course, but that’s just another example of the increasing gulf in both hardware and software between client and workstation usages. These chips are the best you can get if you want to blur the edges but I can see it diverging further to the point where it isn’t really possible unfortunately πŸ™

          • travbrad
          • 4 years ago

          Yep the only games I have that don’t run well are games that only use 1-2 cores. Every game that uses 4+ cores runs perfectly fine and is usually bottlenecked by my GPU anyway. MOAR cores won’t help.

          These E CPUs can be a nice option for people who want to play games AND have some other workloads that use a ton of threads, although Broadwell (both E and non-E) is pretty bad for overclocking. Skylake-E will probably be a better “best of both worlds” solution if it can hit clock speed similar to regular Skylake.

        • jihadjoe
        • 4 years ago

        There is an entire family of E5 Xeons that are single socket, though.

        [url<]http://www.cpu-world.com/CPUs/Xeon/TYPE-Xeon%20E5-1600%20v3.html[/url<]

    • ronch
    • 4 years ago

    1. Is this a tick or a tock? Aren’t process shrinks tocks?
    2. 10 cores. Makes me think Intel knows Zen will come in sets of 4 cores so AMD will have to do with 8 cores or add another Zen module for 12 cores, making comparisons a bit tricky. But isn’t AMD planning to go with 16 cores for desktop enthusiast parts?
    3. I don’t know why Intel’s LGA2011 SKUs start with one digit higher than LGA 115x parts. For example, this being a Broadwell 5th gen part, it should start with 5xxx, not 6xxx. Adds more confusion to an already convoluted model numbering scheme.
    4. $1,723. Good golly. If anything, this gives AMD a lot of wiggle room so they don’t get crushed down the pricing ladder should they decide to price lower (which they probably will do).
    5. While I’d rather have TBM run without any software assistance, I think putting in more intelligence via a software application is a logical solution. Not very amazing but logical. It’s also good to see Intel visibly addressing the issue with 2011 parts compared to their mainstream parts.

      • just brew it!
      • 4 years ago

      [quote<]But isn't AMD planning to go with 16 cores for desktop enthusiast parts?[/quote<] I thought it was supposedly 8 cores / 16 threads, but it's all rumors at this point anyway.

      • tipoo
      • 4 years ago

      The tick-tock nomenclature is dead, it’s P-A-O now, Process, Architecture, Optimization…I guess this is an O? Broadwell was already on 14nm.

        • Freon
        • 4 years ago

        Broadwell up and down the vertical is all one step. It’s just as silly to call a Celeron one of these “steps” as it is the -E (HPDT) variant.

        • ThatStupidCat
        • 4 years ago

        I thought it was tick-tock-toe.

          • tipoo
          • 4 years ago

          It damn well should be

          • donkeycrock
          • 3 years ago

          it should be tick-tock-tow

    • raddude9
    • 4 years ago

    The lack of any price drops in the medium-core-count desktop CPUs from intel indicates to me that they don’t think Zen is going to be much of a threat. Which probably mean that Zen is not going to be much of a threat. Screw it though, I’m going to wait for Zen to see if it brings anything to the table before upgrading my 6-core dev machine.

      • ronch
      • 4 years ago

      Or maybe they’re milking the cow as much as they can while Zen still isn’t out then drop prices once Zen lands. All speculation, of course.

      • NovusBogus
      • 4 years ago

      It also indicates that only a tiny fraction of people actually need more than four cores. This is Intel’s version of Titan.

        • Krogoth
        • 4 years ago

        Bingo, chips from Socket 1366 and 2011 families have always been prosumer-tier products.

        Core versions of silicon are just chips that ran too hot/ate too much power for “Xeon” brand.

          • w76
          • 4 years ago

          Wow, the Intel defenders are out strong. I’m a business consumer of various industrial hardware. I still balk at price increases between generations, and still endeavor to avoid getting in to a monopoly-supplier situation, to avoid exactly this sort of exploitation. Doing something for money doesn’t instantly equate to “Oh, okay, rape me please” when we buy something.

      • the
      • 4 years ago

      Zen is going to arrive late this year. They have ~5 months at these prices before competition even arrives. More profit until the need arrives to be price competitive.

      On that note, Intel also has the option of introducing even higher core counts to this ~$1700 price bracket if Zen is competitive.

        • Topinio
        • 4 years ago

        Yep, I remember the domination of Intel’s C2D and C2Q from August ’06 to April ’08 — then when the Phenom B3’s were released in March ’08 the pricing on the Intel chips was slashed and the Core 2 Quad Q6700 went straight from $530 to $266 …

        Anyone who built a C2Q machine in late ’07 was laughing when the initial Phenoms were released and weren’t great, but not so much by the spring. So I’m going to wait for Zen / Summit Ridge this time around, and possibly for the errata release too πŸ˜‰

      • derFunkenstein
      • 4 years ago

      There’s plenty of time once Zen is actually in volume production to worry about it. If I was Intel, I wouldn’t drop prices in anticipation of anything – get all the money you can while you can.

      • tipoo
      • 4 years ago

      Why would they cut it half a year in advance? If they need to, they could slash prices the week of.

      • travbrad
      • 4 years ago

      I don’t expect Zen to bring much of a challenge to Intel (I’d love to be wrong though), but even if it did there would be no reason for them to drop prices in anticipation of a product that isn’t available yet. They could always drop prices if/when that happens.

    • DrDominodog51
    • 4 years ago

    Wake me up when Skylake E pushes 10 cores down to $1000.

      • ronch
      • 4 years ago

      You’ll be sleeping forever if that’s what you want. :-p

      • Freon
      • 4 years ago

      Intel will never do that without competitive pressure.

        • Krogoth
        • 4 years ago

        without any killer mainstream application for that matter.

          • w76
          • 4 years ago

          I haven’t seen hard numbers, just commentary based on anecdotal evidence, but I’ve been under the impression for a year or so that soaring Twitch-like services and endless armies of new at-home YouTube (wannabe) stars has fueled a lot of purchases of systems with cores to spare — either to speed video work by itself or to multi-task other broadcast apps while gaming or doing other things. In fact, that’d be my first guess as to why the astronomical price; Intel knows that market will pay for the best, and there is simply no alternative whatsoever.

            • travbrad
            • 4 years ago

            That doesn’t make much sense for twitch streaming since you can just use the encoders on your video card with virtually no performance hit. For youtube it would make sense though, to get your locally recorded files down to a size that is practical to upload.

            • w76
            • 4 years ago

            It’s purportedly the YouTube crowd that’s soaked up all the older dual Xeon boards that enjoyed popular news coverage earlier this year/last year. As for game streaming, I know someone that streams EveOnline and he throws cores at the problem, but he’s running a fancy overlay over the screen that’s broadcast (news stream that hides in-game local chat, video in another corner, etc). No idea how common that is, though, outside of Eve; I’m still at the get-off-my-lawn stage with Twitch.

      • sleeprae
      • 4 years ago

      Agreed. I had actually persuaded myself to go for the $1000 SKU with Broadwell-E, which is why I went with the 5820K HW-E on my current build to hold me over. However, that was based on the expectation that I was buying the premier part. There’s no way I’m going to spend $1k on a mid-range SKU. Accordingly, I’ll probably use the 6850K to upgrade my 5820K.

      So, to Intel, your idiotic pricing at the top will result in less revenue from me.

        • cobalt
        • 4 years ago

        Why even upgrade from a 5820K, then? Seems like a minimal benefit from reviews.

          • sleeprae
          • 4 years ago

          No good reason. I just want it. πŸ™‚

          With multiple GPUs, RAID cards, and NVMe drives, I would like the extra PCIe lanes.

        • BestJinjo
        • 4 years ago

        6850K is not an upgrade to a 5820K. Unless you have 1080 SLI already and a 1440 100-165Hz or a 4K monitor, spend the $$ on something useful for a superior gaming experience. Besides, 6700K 4.7Ghz will outperform the 6850K for games, while still having sufficient lanes off the Z170 chipset for RAID M.2 32GB/sec SSDs. Plenty of boards can do this such as MSI M5 or Gigabyte Z170X-Gaming 5 without breaking the bank.

          • sleeprae
          • 4 years ago

          Oh, I agree, it isn’t needed. I have several Skylake/Z170 systems as well. My two X99 boards both have 7 x16 slots (two PLX switches) so the 28 vs. 40 lanes doesn’t really matter either.

          I’m not actually a big gamer, but with GPU, multiple RAID cards, NVMe drives, and 10GbE NICs, I do use more lanes than average–which is why my main rig is HEDT, even though it’s been a generation behind for some time.

          I still want it though. I made the decision long ago that simply wanting it was sufficient justification for me. πŸ™‚

Pin It on Pinterest

Share This