Slashdot Mirror

← Back to Stories (view on slashdot.org)

Intel Announces Devil's Canyon Core I7-4790K: 4GHz Base Clock, 4.4GHz Turbo

Posted by timothy on from the let-the-bleeding-edge-do-the-bleeding dept.

MojoKid (1002251) writes "Last year, Intel launched two new processor families based on the Haswell and Ivy Bridge-E based Core i7 architecture. Both chips were just incremental updates over their predecessors. Haswell may have delivered impressive gains in mobile, but it failed to impress on the desktop where it was only slightly faster than the chip it replaced. Enthusiasts weren't terribly excited about either core but Intel is hoping its new Devil's Canyon CPU, which launches today, will change that. The new chip is the Core i7-4790K and it packs several new features that should appeal to the enthusiast and overclocking markets. First, Intel has changed the thermal interface material from the paste it used in the last generation over to a new Next Generation Polymer Thermal Interface Material, or as Intel calls it, "NGPTIM." Moving Haswell's voltage regulator on-die proved to be a significant problem for overclockers since it caused dramatic heat buildup that was only exacerbated by higher clock speeds. Overclockers reported that removing Haswell's lid could boost clock speeds by several hundred MHz. The other tweak to the Haswell core is a great many additional capacitors, which have been integrated to smooth power delivery at higher currents. This new chip gives Haswell a nice lift. If the overclocking headroom delivers on top of that, enthusiasts might be able to hit 4.7-4.8GHz on standard cooling."

157 comments

  1. why get this when Broadwell + new chipsets are com by Joe_Dragon · · Score: 1

    why get this when Broadwell + new chipsets are coming soon?

  2. More useful metrics? by StripedCow · · Score: 4, Interesting

    Why don't we ever read about more useful metrics, such as the amount of (floating-point) operations per second per $ of a given CPU?

    --
    If Pandora's box is destined to be opened, *I* want to be the one to open it.
    1. Re:More useful metrics? by LoRdTAW · · Score: 3, Informative

      Because it doesn't have the wow-wee factor compared to raw clock speed numbers.

    2. Re: More useful metrics? by Anonymous Coward · · Score: 0

      Less glamorous, too technical for mainstream. = less money for company

    3. Re:More useful metrics? by Anonymous Coward · · Score: 2, Insightful

      Maybe because FLOPS hasn't been such a great metric in more than a decade? On a modern architecture, you could estimate FLOPS as some constant times the clock speed, but then you end up with not all operations taking the same number of clock cycles. Or then you have SIMD to consider. Or what if you're interested in a square root function, do you need full precision or some approximation is okay? And heck, that's coming from someone that actually cares about FLOPS... most day to day applications have other bottlenecks.

      But if that's what you wanted, you could just look up the clock speed and divide by the price.

    4. Re:More useful metrics? by excelsior_gr · · Score: 1

      For several reasons: FLOPS is a metric that depends on the algorithm used as well as on the system load. The system load could be minimized by running the benchmark on a bare-bone system multiple times and then averaging the results, but you still get serious variations. A FLOPS benchmark also depends on the RAM-speed and the amount of memory channels on the CPU chip if the benchmark is memory-hungry. Also, are you interested in running a large (parallel) computation over several days or a gazillion of short calculations in parallel? Also, not everyone is interested in the FLOPS performance. Lots of applications are more dependent on the performance of the integer unit.

    5. Re:More useful metrics? by fuzzyfuzzyfungus · · Score: 4, Informative

      Why don't we ever read about more useful metrics, such as the amount of (floating-point) operations per second per $ of a given CPU?

      Because the target market for this thing doesn't consider that a useful metric, and never has.

      For some years now (at least back to the P4 era, if memory serves), Intel has always offered the mad-crazy-overclocker-must-go-faster-edition CPU at the top of their (desktop, sorry Xeon buyers!) price list, usually ~$1,000. This part is always an astonishingly poor value, unless what you want is the fastest x86 money can buy. Most of them go to gamer e-peen setups, they may sell some to compute customers who have some pathologically hard-to-parallelize problem and thus need the fastest single threaded performance they can get, rather than more cores with lower performance per thread but far lower cost.

      If you are actually shopping for CPUs, you probably want something like CPUboss, or CPUbenchmark which allows you to do fairly easy comparisons of performance/price (albeit for performance as measured by one or more general benchmarks, if your workload is somewhat atypical, your mileage may vary).

    6. Re:More useful metrics? by PhrostyMcByte · · Score: 3, Informative

      Why don't we ever read about more useful metrics, such as the amount of (floating-point) operations per second per $ of a given CPU?

      Because most people don't care about these things anymore. Take this from TFS:

      Haswell may have delivered impressive gains in mobile, but it failed to impress on the desktop where it was only slightly faster than the chip it replaced.

      In reality, Haswell had double the FLOPs thanks to the new FMA instructions, near double the integer throughput thanks to AVX2, and a significant boost to multithreaded code thanks to TSX.

      In practice, people saw maybe a 10% speedup in what they actually do. A flops/$ metric would significantly inflate the actual value people would see from these CPUs.

      The thing is, these measurements are either synthetic (who has code consisting of nothing but FMA?), hard and uncommon to use (Integer SIMD is rare and AVX2 has a confusing idea of "lanes" that splits some 256-bit ops into two 128-bit ones), or not on all CPUs (TSX is disabled on their unlocked K line for some reason).

    7. Re:More useful metrics? by Anonymous Coward · · Score: 0

      If this is the case, why don't Intel make their mad-crazy CPU as a dual-core, sacrificing the extra cores for a tiny bit of extra single-thread performance? That gets gamers their best possible gaming (assuming the second core runs everything apart from the game itself).

    8. Re:More useful metrics? by drinkypoo · · Score: 0

      If this is the case, why don't Intel make their mad-crazy CPU as a dual-core, sacrificing the extra cores for a tiny bit of extra single-thread performance? That gets gamers their best possible gaming (assuming the second core runs everything apart from the game itself).

      Since the Xbox 360, gamers have needed quad-core to keep up with console/PC cross-platform titles. The 360 has no OS to speak of, and has three cores, which is how we arrive at the four-core figure. Now that game consoles have said fuck it, we're going to eight cores, gamers will need the same. They won't need more core, though, because the consoles now have relatively significant operating system services which run concurrent with the game and the developers aren't expecting to be able to use the entire machine.

      --
      "You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
    9. Re:More useful metrics? by radarskiy · · Score: 1

      "The thing is, these measurements are either synthetic (who has code consisting of nothing but FMA?), hard and uncommon to use (Integer SIMD is rare and AVX2 has a confusing idea of "lanes" that splits some 256-bit ops into two 128-bit ones), or not on all CPUs (TSX is disabled on their unlocked K line for some reason)."

      Except to the people buying top-bin parts by the thousands that really do need all of these obscure ops.

    10. Re:More useful metrics? by danomac · · Score: 1

      For some years now (at least back to the P4 era, if memory serves), Intel has always offered the mad-crazy-overclocker-must-go-faster-edition CPU at the top of their (desktop, sorry Xeon buyers!) price list, usually ~$1,000. This part is always an astonishingly poor value, unless what you want is the fastest x86 money can buy. Most of them go to gamer e-peen setups, they may sell some to compute customers who have some pathologically hard-to-parallelize problem and thus need the fastest single threaded performance they can get, rather than more cores with lower performance per thread but far lower cost.

      I do a lot of compiling, and I generally build myself a new PC every 8-10 years. Many people I know buy cheaper components only to have to replace them 3-4 times before I replace mine. I did get an EE processor back in 2008, and I'm still using it now and will for the forseeable future. For me, the compilation speed is still very acceptable 6 years in, and it's very possible I won't bother replacing my computer anytime soon - I'll probably get another 6 years out of it. Given my experience with my first EE processor, I'll probably spend the $ again and have it last another 12 years.

    11. Re:More useful metrics? by jones_supa · · Score: 1

      Why don't we ever read about more useful metrics, such as the amount of (floating-point) operations per second per $ of a given CPU?

      Amen, brother. For example, let's take a Pentium 4, 3.0GHz and a Core i7, 3.5GHz. At the same TDP (and all cores utilized), the Core i7 is 28x more powerful than the P4. Even if we compensate the clockspeed to be equal (downclock the i7 to 3GHz), it clearly shows that something like "GHz" is today completely useless for comparing CPUs.

    12. Re:More useful metrics? by theManInTheYellowHat · · Score: 1

      Or floating point operations / watt?

    13. Re:More useful metrics? by geekoid · · Score: 1

      If you know about those things then you know where tt get that info, if you don't know about those thing that will just makes someone eyes glaze over.
      So, give a metric with the widest appeal.

      --
      The Kruger Dunning explains most post on /. http://en.wikipedia.org/wiki/Dunning%E2%80%93Kruger_effect
    14. Re:More useful metrics? by TheRaven64 · · Score: 1

      Can you define what you mean by 'a lot' of compiling? We have some machines where I regularly build LLVM and the FreeBSD base system. They're 32-core with 256GB of RAM, and on them most builds are nice and fast, although it still takes close to two hours to do a universe build (all kernel configs, userland for all supported architectures - the 'quick' sanity test that you haven't broken anything). They're similar specs to the 'beefy' machines in the FreeBSD cluster, which take around 24 hours to build a complete package set (all 24K ports for a single release / architecture).

      I'm not sure what you got 6 years ago, but there are still things where the fastest you can buy it not as fast as you'd like. I'm really interested in better single-threaded performance, because some companies (*cough*Altera*cough*Xylinx*cough*) haven't yet worked out how to multithread their compilers and so it takes an hour and a half to do a synthesis run of our CPU.

      --
      I am TheRaven on Soylent News
    15. Re:More useful metrics? by fuzzyfuzzyfungus · · Score: 1

      If this is the case, why don't Intel make their mad-crazy CPU as a dual-core, sacrificing the extra cores for a tiny bit of extra single-thread performance? That gets gamers their best possible gaming (assuming the second core runs everything apart from the game itself).

      I don't know for certain; but I assume that the volume on these doesn't justify too much customization, just binning and not disabling anything; and that with the 'turbo' arrangement on the newer corei CPUs, you get higher clocks on the cores that are active when some cores are idle automatically.

    16. Re:More useful metrics? by Grog6 · · Score: 1

      This exactly.

      Unless a piece of software takes advantage of the "new" features, it doesn't help.

      The slow adoption of the MMX stuff is a good example; a developer I know rewrote a very complicated piece of code to take full advantage of SSE2 and multithreading, and took a 45 minute computation down to 8 minutes.

      Being an old assembler guy, that really amazed me; kinda like adding a few lines of inline assembler to a C routine, and watching it go zoom.

      Adding cores helps, but the FSB speed is still the main thing to speed up for performance; increasing my memory clock by 20% made a much bigger difference than cranking the core to 4.6GHz on my i7-3930k.

      Cranking the clock AND running at 4.6GHz is very nice; that, and adding another 7970 card to the mix finally fixed the texture popping in Rage, lol.

      --
      Truth isn't Truth - Guliani
    17. Re:More useful metrics? by Anonymous Coward · · Score: 0

      Consoles now have eight cores only if you count AMDs lame approach as true cores. They have eight integer, but only four floating point and they are still much weaker than a quad core i5 or i7 overall.

    18. Re:More useful metrics? by Anonymous Coward · · Score: 0

      That's well and all unless there's some sort of desirable motherboard IO/port improvements that could well come within that time. I've been on a 4-6 year cycle since early 90's and I think the $200-$300 CPU & GPU and $100-$200 motherboard with the most future looking IO in it provide best bang for buck. (that comes to about $200 a year + some spent on HDD's yearly as I don't like agonizing on what to delete vs just buying more storage) I've never needed all the extra IO but considering the price of adding IO in add-on card form and the questionable performance of lower end add-on cards, I think this is a reasonable strategy.

      Currently I have E8500 and I'm trying to figure if there's any compelling reason to wait for Broadwell or Haswell-E. I like the idea of the larger cache in Broadwell - that could allow large NET/Java apps to stay completely in the memory close to CPU if that's how it works (it might not), which might result in visibly better response in heavy apps of that sort. Most likely it will not result in that - but if developers start writing code assuming the CPU has more cache available, you can BET that apps will get much slower in CPUs without that cache. That's makes for compelling case to get Broadwell. It's just a question of whether I can wait for another 8-12 months if it's the desktop I'm wanting to upgrade. (I'm thinking Haswell might be fine for desktop right now, and just wait for Broadwell portables as they'll gain most from the power improvements - the desktops will likely only gain the cache and the effects of that may end up just being things not slowing down few years down the line as said above.)

    19. Re:More useful metrics? by Anonymous Coward · · Score: 0

      http://ark.intel.com/products/80807/Intel-Core-i7-4790K-Processor-8M-Cache-up-to-4_40-GHz

      It hasn't been mentioned anywhere but 4770K specs don't have TSX while 4790K do. Now since CPU-Z does not show if you have TSX or not, the leaked CPU-Z screenshots do not confirm this.

      However from what I've read from forums of people trying to use TSX suggests that unless you are C/C++ library developer or prepared to become one, the gains from having it are more in the "it's nice to have if I'm not upgrading in the next 3+ years".

    20. Re:More useful metrics? by Anonymous Coward · · Score: 0

      Can you define what you mean by 'a lot' of compiling? We have some machines where I regularly build LLVM and the FreeBSD base system. They're 32-core with 256GB of RAM, and on them most builds are nice and fast, although it still takes close to two hours to do a universe build (all kernel configs, userland for all supported architectures - the 'quick' sanity test that you haven't broken anything). They're similar specs to the 'beefy' machines in the FreeBSD cluster, which take around 24 hours to build a complete package set (all 24K ports for a single release / architecture).

      You mention 256GB of ram, which nothing needs to compile, but I wonder, are you using FreeBSD's equivalent of tmpfs (ramdisk, I don't know what it's called) for storage of the build tree? I'm sure that would result in massive speedup over compiling from a on disk/ssd filesystem.

      because some companies (*cough*Altera*cough*Xylinx*cough*) haven't yet worked out how to multithread their compilers and so it takes an hour and a half to do a synthesis run of our CPU.

      Bullshit, Quartus II has had Multiprocessor support since at least version 7.2 (I can't find the exact release they added it), and has been improving P&R speeds every release since then. In the version I am using, 14.1 Web Edition, Multiprocessor support requires Talkback (give back performance metrics in exchange for parallel speedup) and works quite well. You should probably look at buying the subscription version and adding some design partitions if your synthesis is taking an hour and a half.

    21. Re:More useful metrics? by drinkypoo · · Score: 1

      Consoles now have eight cores only if you count AMDs lame approach as true cores. They have eight integer, but only four floating point and they are still much weaker than a quad core i5 or i7 overall.

      I'll grant you the point, it still leaves gamers needing at least a quad-core. Single- or dual-thread performance will simply no longer cut it. PCs are heading towards large numbers of cores, so that's not an especially onerous requirement. It does, however, disqualify dual-core processors from serious game machines.

      --
      "You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
    22. Re:More useful metrics? by TheRaven64 · · Score: 1

      You mention 256GB of ram, which nothing needs to compile, but I wonder, are you using FreeBSD's equivalent of tmpfs (ramdisk, I don't know what it's called) for storage of the build tree? I'm sure that would result in massive speedup over compiling from a on disk/ssd filesystem.

      That's not quite true. For one thing, if your buffer cache is big enough that the entire source tree stays there then it has a noticeable impact on compile time. I tend to put the obj tree in tmpfs when doing a universe build (I don't actually want it, I just want to check everything builds most of the time), and that consumes 50GB. You might also be surprised at how much RAM the compile takes.

      Bullshit, Quartus II has had Multiprocessor support since at least version 7.2 (I can't find the exact release they added it), and has been improving P&R speeds every release since then. In the version I am using, 14.1 Web Edition, Multiprocessor support requires Talkback (give back performance metrics in exchange for parallel speedup) and works quite well. You should probably look at buying the subscription version and adding some design partitions if your synthesis is taking an hour and a half.

      Hmm, is it something that you need to specifically enable? We're taking about an hour and a half with Quartus 13 on these machines. We do pay for the subscription version, although adding more design partitions is not a simple thing to do as our Verilog is generated from a higher-level HDL.

      --
      I am TheRaven on Soylent News
  3. Re:why get this when Broadwell + new chipsets are by binarylarry · · Score: 2

    Better yet, why get this *now* when you can wait til the price drops after the next iteration hits.

    --
    Mod me down, my New Earth Global Warmingist friends!
  4. Meh. by snarfies · · Score: 1

    Still no consumer-level octo-core? I'll continue to stick with my Q9450 then. I'm not willing to spend thousands on a Xeon.

    Before you ask, I do video work as a hobby. I often utilize all four of my cores at 100% capacity for 6+ hours on end when performing filtering and encoding. No, I am not interested in the AMD FX and their half-cores.

    Get it together, Intel. I will buy as soon as you sell.

    1. Re:Meh. by Anonymous Coward · · Score: 0

      What? This doesn't satisfy you?

    2. Re:Meh. by afidel · · Score: 4, Interesting

      Lol, you do realize an i5-5370 is roughly twice as fast at video encoding compared to your old Q9450, right? Just because they haven't double the number of cores doesn't mean they haven't improved performance significantly in the consumer space.

      --
      There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
    3. Re:Meh. by bill_mcgonigle · · Score: 1

      Still no consumer-level octo-core? I'll continue to stick with my Q9450 then. I'm not willing to spend thousands on a Xeon.

      You don't need to spend thousands - my last build was a little SOHO Xen server using a Xeon 2620 which is a six-core with hyperthreading. Much to my surprise, HP's benchmarks showed a 30% performance improvement under Xen with hyperthreading in use, so that's 7.8 cores worth, which is pretty good for $400. I used a single-processor mobo, but dual-sockets are available. It's not a linear number cruncher (I have a hot 12-core AMD for that) but if you have a workload that needs parallelism it's a decent value. Sadly, it replaced my AMD Xen server because AMD just isn't stable under recent Xen and there's no PVH support at all.

      --
      My God, it's Full of Source!
      OUTSIDE_IP=$(dig +short my.ip @outsideip.net)
    4. Re:Meh. by Anonymous Coward · · Score: 0

      Not really, since it still gets beaten out by last generation's quad core i7 3770 @ 3.4GHz.

      http://www.cpubenchmark.net/cp...

    5. Re:Meh. by Emetophobe · · Score: 1

      Haswell-E coming out later this year will have 6 or 8 cores (source).

    6. Re:Meh. by Anonymous Coward · · Score: 0

      It gets even better man, the the 3570 is quick sync capable so it's many many many more times faster at encoding when supported.

    7. Re:Meh. by afidel · · Score: 1

      Quick Sync sucks ass, the quality is horrible. If they modified the algorithm to look good it wouldn't be any faster than the best software encoders using the same hardware resources and you can make the software encoders as fast by turning off the features that improve IQ.

      --
      There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
    8. Re:Meh. by DamnOregonian · · Score: 1

      I recently upgraded my Q9650 to a Haswell i7 4770.

      I loved my Q9650. It performed stably and admirably and stayed relevant for a very long time (and is still relevant today), but my 4770 eats its lunch in ways I never imagined when buying it.

      I write a lot of memory-intensive multi-core one-off software, and in some workloads my 4770 is close to twice as fast.

      Anecdotal, and YMMV on real-world workloads, but the generations after Nehalem really are worth the money if you're lifting heavy weight with your CPU.

    9. Re:Meh. by toddestan · · Score: 1

      You might try a newer processor. The latest chips from Intel are considerably faster than the Core 2 chips even if they haven't increased the core count and the clock speeds are more or less the same. Even in the Core i5 line, I would estimate that clock for clock, a fourth gen i5 is 40-50% faster than a first generation Core i5. That's of course, assuming the encoding times are problem for you - if you start it before going to bed and it's done by the time you wake up then it may not matter :)

  5. Can we PLEASE get rid of "turbo" by SpankiMonki · · Score: 1

    Hasn't this tired car analogy run it's course by now? Every time I hear "Turbo!" I think of this guy. JFC.

    1. Re:Can we PLEASE get rid of "turbo" by Anonymous Coward · · Score: 0

      Way back in the 90's some PC's had turbo buttons on the tower. I say bring that back.

    2. Re:Can we PLEASE get rid of "turbo" by Kardos · · Score: 1

      It's not a car analogy any longer. If you find a car with tubro, it's using a CPU analogy.

    3. Re:Can we PLEASE get rid of "turbo" by Anonymous Coward · · Score: 0

      Can we PLEASE get rid of extra apostrophes? Hasn't anyone over the age of eight understood that it's means it is by now?

    4. Re:Can we PLEASE get rid of "turbo" by Anonymous Coward · · Score: 0

      Way back in the 90's some PC's had turbo buttons on the tower. I say bring that back.

      I remember my 8088 in the 1980s ran at 8MHz unless the Turbo button was engaged (then it was a blazing 12MHz!).

    5. Re:Can we PLEASE get rid of "turbo" by Anonymous Coward · · Score: 0

      Hasn't anyone over the age of eight understood that "it's" means "it is" by now?

      FTFY.

    6. Re:Can we PLEASE get rid of "turbo" by allcoolnameswheretak · · Score: 1

      My 486 DX2 had a turbo button which was really useful (turned off) to run all my old 286 games, which were not frame limited.

    7. Re:Can we PLEASE get rid of "turbo" by Anonymous Coward · · Score: 0

      Can we PLEASE get rid of posts that don't apply to the parent? Idiot.

    8. Re:Can we PLEASE get rid of "turbo" by Anonymous Coward · · Score: 0

      Turbo boost, KITT !
      http://media.moddb.com/cache/i...

    9. Re:Can we PLEASE get rid of "turbo" by Jane+Q.+Public · · Score: 1

      Hasn't this tired car analogy run it's course by now? Every time I hear "Turbo!" I think of this guy. JFC.

      It isn't a car analogy. It WAS... back in the late 80s.

      IBM-compatible PC clones came out with a switch in the back to boost the clock speed higher than the "official" IBM speed. This was called "turbo" even though, in practice, people set it there and left it there. Think of it as an early form of overclocking.

      Although, to be fair, Intel's CPU "turbo" is probably a bit more of a car analogy because it's a temporary speed boost, not a permanent one.

    10. Re: Can we PLEASE get rid of "turbo" by Anonymous Coward · · Score: 0

      Right, I remember that. Some programs would run as fast as the CPU would allow without any kind of throttling so you would get like super high-speed motion for some of the games. Like running a video on fast forward. I guess when they built it they thought that it would never run too fast, they were wrong )

    11. Re:Can we PLEASE get rid of "turbo" by Anonymous Coward · · Score: 0

      Sure the name might seem silly, but the feature itself is very good. They could call it something like Thermal Adaptive Automatic Overclocking, but "turbo" is much simpler.

    12. Re: Can we PLEASE get rid of "turbo" by Blaskowicz · · Score: 1

      I remember when you could disable the L1 and L2 caches in BIOS. Made the PC really run like crap, so I think it turned it back on immediately.

    13. Re:Can we PLEASE get rid of "turbo" by nabsltd · · Score: 1

      Although, to be fair, Intel's CPU "turbo" is probably a bit more of a car analogy because it's a temporary speed boost, not a permanent one.

      For Ivy Bridge chips with decent cooling, it's basically permanent. Every Ivy Bridge chip has the base clock speed (the number on the box), the "Turbo" speed (which is available for a single core only), and the "multi-core turbo" that isn't advertised very well, but sites like CPUBenchmark will give you the details.

      For example, I have an E5-2630 (yes, it's a Xeon, but there are fairly equivalent desktop chips) with a base clock of 2.6GHz, a single-core turbo of 3.1GHz, and a multi-core turbo of 2.9GHz. Since the chip stays cool enough, it runs at least at 2.9GHz whenever it needs speed.

    14. Re:Can we PLEASE get rid of "turbo" by canajin56 · · Score: 1

      That's its purpose. It was going to be labeled "Slow" but it sounds cooler if you call it "Turbo" and reverse the On/Off.

      --
      ASCII stupid question, get a stupid ANSI
    15. Re:Can we PLEASE get rid of "turbo" by Jane+Q.+Public · · Score: 1

      Since you brought it up: if you were building a heavy-use machine for code development today, would you go with 6-or-more-core Xeon, or a newer Haswell 4-core chip and board?

      Typically running a code editor, web browser, web / database server, video, and possibly file copies at the same time. Although the web and database severs are almost always single-user at any given time.

      The thing is, it has to have good video capabilities as well as decent number-crunching. A pure server setup would not be appropriate.

    16. Re:Can we PLEASE get rid of "turbo" by Jane+Q.+Public · · Score: 1

      By the way: OP says the i7-4790 was announced today but it has been for sale on Newegg for at least a week.

    17. Re: Can we PLEASE get rid of "turbo" by Grog6 · · Score: 1

      I have a socket 7 mobo working behind me that has 2MB of DIP SRAM in sockets.

      (It has an ISA slot for a very expensive card, and DOS 6.3 with network support :) I also have a EISA bus server for DATA-AQ, lol.)

      --
      Truth isn't Truth - Guliani
    18. Re: Can we PLEASE get rid of "turbo" by Anonymous Coward · · Score: 0

      I sure hope you have a `32 Ford you're workin' on as well...

    19. Re: Can we PLEASE get rid of "turbo" by Grog6 · · Score: 1

      No, all my Fords are modern. :)

      --
      Truth isn't Truth - Guliani
    20. Re:Can we PLEASE get rid of "turbo" by nabsltd · · Score: 1

      Since you brought it up: if you were building a heavy-use machine for code development today, would you go with 6-or-more-core Xeon, or a newer Haswell 4-core chip and board?

      To me, Xeon vs. desktop chip comes down to how much memory you are going to want. Trusting 32GB or more to not have a random memory error isn't something I want to do, so I went with a Xeon and ECC. In addition, if you want more than 32GB, that means an LGA2011 socket for Intel...none of their other sockets have chipsets that can handle more. If you go LGA2011, then your best bet is a workstation board, since you could eventually put a Xeon and 256GB of RAM on it, but could start with a desktop CPU and whatever RAM you can afford. Workstation typically means no on-board graphics and plenty of PCIe slots available.

      Your use case sounds like it doesn't really need the extra CPU threads, but does need disk speed. For that, I say SSD for your boot plus a real RAID controller (preferably with battery backup) with spinning disks for bulk storage and lots of RAM for cache. The same system with the E5-2630 v2 has 64GB of RAM, a 500GB Samsung 840 EVO and five 2TB WD Red drives in RAID-5 connected to an Adaptec 6805 with battery backup. Reads and writes to the SSD and the array are at similar speeds (between 400MB/sec and 600MB/sec), and more than one app hitting the disk at the same time doesn't make things slow to a crawl.

      Essentially, if your disk can't support the I/O for all those services, going from 8 to 12 CPU threads won't really buy you much. You can then add any GPU you want (I have a Radeon R9 270).

    21. Re:Can we PLEASE get rid of "turbo" by Jane+Q.+Public · · Score: 1

      Your use case sounds like it doesn't really need the extra CPU threads, but does need disk speed. For that, I say SSD for your boot plus a real RAID controller (preferably with battery backup) with spinning disks for bulk storage and lots of RAM for cache. The same system with the E5-2630 v2 has 64GB of RAM, a 500GB Samsung 840 EVO and five 2TB WD Red drives in RAID-5 connected to an Adaptec 6805 with battery backup. Reads and writes to the SSD and the array are at similar speeds (between 400MB/sec and 600MB/sec), and more than one app hitting the disk at the same time doesn't make things slow to a crawl.

      I appreciate the advice, but I've been building systems a long time. I was just curious what you thought about processor choices for that use-case.

      The ECC is an angle I hadn't thought of, though. That does makes sense. As for the rest, because I'm probably going with OS X the selection of boards is limited, and so are the GPUs. They're good, don't get me wrong. But there aren't that many of them that work well.

      And I don't use Samsung drives if I can help it. I know Apple uses them a lot, but they aren't really celebrated for their high performance. WD is fine with me and I've had excellent experience with them. Right now I have a good SSD, a secondary HDD, and 2 x 2TB WD Re drives, which is adequate for my needs for now.

      Being a personal-use system (albeit heavy use), I really don't want a 5-drive RAID array spinning all the time. While I have many apps open at a time, they're not usually in heavy use all at the same time (except sometimes playing video). I'm not doing heavy ray-tracing or anything like that on a regular basis. So I'm really not doing that heavy I/O most of the time. Occasionally there will be some heavy database access alongside the video.

      If I need a big drive array I can add it later (the case has room & the PSU is adequate). The SSD and secondary HDD are good for most of my use. The 2TB drives are for storage and backup. They can sleep most of the time.

      Essentially, if your disk can't support the I/O for all those services, going from 8 to 12 CPU threads won't really buy you much. You can then add any GPU you want (I have a Radeon R9 270).

      Yes, well, it looks like I'll stick with the inexpensive Haswell i7 refresh. I might switch when the newer 6- or 8-core units come out later, but maybe not. We'll see. The main reason being, as I say: I may be running many apps simultaneously but very seldom are more than 3 very "active" at any one time.

      Because it's OS X, good Xeon boards with ECC are even harder to find and then set up.

    22. Re:Can we PLEASE get rid of "turbo" by nabsltd · · Score: 1

      And I don't use Samsung drives if I can help it. I know Apple uses them a lot, but they aren't really celebrated for their high performance.

      For spinning disks, maybe not, but for SSDs, Samsung is king, especially their new M.2 XP941 model with 4 PCIe lanes.

    23. Re: Can we PLEASE get rid of "turbo" by toddestan · · Score: 1

      Wow, that's a huge amount of L2 cache on a Socket 7 board. Most came with 256k. Some had 512k. Are you sure it's not a Socket 8 (aka Pentium Pro)?

    24. Re: Can we PLEASE get rid of "turbo" by Blaskowicz · · Score: 1

      Socket 8 had no cache on motherboard at all ;), the CPU virtually had on-die L2 cache. Most Pentium Pro had one 256KB L2 die closely attached to the CPU die, some had one 512K die, and some (atrociously expensive, for high end quad CPU set up) had two 512K dies.
      It's why the socket and CPU are a big rectangle, and it did cost a ton : you would attach L2 and CPU, and then the chip was dead if any of the two was dead.
      Pentium II and first Athlon moved to a cartridge, where the L2 chip(s) were separate from the CPU but geographically quite close. L2 ran at half speed (and third speed on the higher clocked Athlons). That solution was half way between the Pentium Pro and the older cache-on-motherboard scheme. Then integrating L2 right into the CPU die superseded it.

      Some late Socket 7 mobo ("Super 7" for K6/2 and such) did come up with 1024K and 2048K on the motherboard.

  6. Wow by synapse7 · · Score: 1

    Not terribly expensive if this is accurate.

    http://www.newegg.com/Product/...

    1. Re:Wow by Nemyst · · Score: 1

      Yeah, this isn't Haswell-E just yet. As far as I know, that will build upon Devil's Canyon and go up to 8 cores on a single chip, with HyperThreading of course. It should also come with a new platform, chipset and socket to support DDR4 and SATAe. Personally, that's the jump I'm really looking forward to: Intel's first 8-core desktop chip.

    2. Re:Wow by Anonymous Coward · · Score: 0

      You will be able to shave off about $50 from that price if you got it from Amazon. Newegg is a rip-off.

    3. Re:Wow by nabsltd · · Score: 1

      Personally, that's the jump I'm really looking forward to: Intel's first 8-core desktop chip.

      With Google's stats on how often ECC in their servers correct errors, I'm not sure why anyone is anxious for an 8-core chip that can't use ECC memory.

      My guess is that if you need 8 cores, you need more memory, and that means you are more likely to have errors. That plus the fact that the only current Intel desktop chipset supports more than 32GB of RAM has an LGA2011 socket is why I went with a Xeon...if I'm going to pay for a more expensive motherboard and processor, I might as well pay just a little more and get ECC protection on the memory. Also, the Xeon boards support 256GB instead of the 64GB that the desktop board support.

  7. Re:why get this when Broadwell + new chipsets are by aliquis · · Score: 1

    Because this is available now and Broadwell isn't?

    Also when Broadwell is released you know what? There will still be better processors released in the future! So better wait until the very best one is released ..

    Reason to get this is that 1) it exist now and 2) it's better than the last. Number two could actually be an argument to get what was released last if you think that provides a better value.

    Obviously if you already have a decent machine and can wait / have a processor from before the refresh then you likely don't have to rush out and get this one but can wait a little longer. But if you want a PC now then Broadwell isn't an option because it's not here.

  8. NGPTIM? by Anonymous Coward · · Score: 0

    Did we not learn our lesson with "NAPLPS" and "SCSI"?

    1. Re:NGPTIM? by Anonymous Coward · · Score: 0

      SCSI is pronounced as "scuzzy".

  9. Re:why get this when Broadwell + new chipsets are by aliquis · · Score: 5, Funny

    Better yet, why get this *now* when you can wait til the price drops after the next iteration hits.

    If you wait ten years you can possibly get one from someone for free!!

  10. Devil's Canyon? by gstoddart · · Score: 1, Funny

    LOL, I predict all sorts of batshit crazy protests over that one ... ZOMG, teh Intel are teh Satanists.

    --
    Lost at C:>. Found at C.
    1. Re:Devil's Canyon? by L4t3r4lu5 · · Score: 1

      It's why there are no 666MHz parts. Memory, processors, buses... All 667Mhz.

      --
      Finally had enough. Come see us over at https://soylentnews.org/
    2. Re:Devil's Canyon? by gstoddart · · Score: 1

      Nah, they did the original designs on a Pentium 2, and it couldn't do floating point math properly. ;-)

      --
      Lost at C:>. Found at C.
    3. Re:Devil's Canyon? by lagomorpha2 · · Score: 1

      It's why there are no 666MHz parts. Memory, processors, buses... All 667Mhz.

      Well that and 666 2/3 rounds up to 667.

    4. Re:Devil's Canyon? by Anonymous Coward · · Score: 0

      666.6666666mhz

    5. Re:Devil's Canyon? by jones_supa · · Score: 1

      It's why there are no 666MHz parts. Memory, processors, buses... All 667Mhz.

      Try this: round(1000 * 2 / 3)

  11. Re:why get this when Broadwell + new chipsets are by Salgat · · Score: 3, Insightful

    There are always new parts coming soon, you eventually have to pull the trigger and buy at some point.

  12. NGPTIM is rather clumsy.. by kheldan · · Score: 1

    I propose they refer to their Next Generation Polymer Thermal Interface Material as 'NexGen Poly TIM'.

    --
    Are YOU using the TOOL, or is the TOOL using YOU? Think about it!
    1. Re:NGPTIM is rather clumsy.. by wonkey_monkey · · Score: 1

      "Special Polymer Used in New Komputers."

      --
      systemd is Roko's Basilisk.
    2. Re:NGPTIM is rather clumsy.. by Anonymous Coward · · Score: 0

      Making "Next Generation" part of the official name for it is stupid as hell if they are indeed really doing that.

    3. Re:NGPTIM is rather clumsy.. by Anonymous Coward · · Score: 0

      I suppose though that a marketing person thinks it's brilliant because then their TIM can always sound ahead of the curve.

  13. Re:why get this when Broadwell + new chipsets are by Anonymous Coward · · Score: 2, Funny

    I'm waiting for the Core i7 4860DX/2 66GHz.

  14. Re:why get this when Broadwell + new chipsets are by kheldan · · Score: 5, Insightful

    Some of you early-adopters may laugh at this, but this has been my upgrade strategy for decades now and from a bang-for-the-buck perspective it's extremely effective.

    --
    Are YOU using the TOOL, or is the TOOL using YOU? Think about it!
  15. About time... by Lumpy · · Score: 1

    A lot of us really do need powerhouse laptops and I really could use a 4ghz-5ghz boosted quad i7 laptop. I really hope we get some real performance in the upper end soon.

    --
    Do not look at laser with remaining good eye.
  16. Re:why get this when Broadwell + new chipsets are by Anonymous Coward · · Score: 1

    I'll only buy one if it comes in a case with a connected and operational 'TURBO' button

  17. Possibly more interesting by gman003 · · Score: 2

    Also announced were an i5 and a Pentium-branded Devil's Canyon processors. All three have the same TIM upgrade and overclocking focus. The i5-4690K is similar to the i7-4690K, dropping hyperthreading, a bit of cache and some stock clock, but for $100 cheaper ($242 instead of $339, if reports are accurate).

    The really interesting one is the Pentium G3258. Two cores, no hyperthreading, but with an unlocked multiplier, for $72. If you care more about single-threaded performance than multi-threaded, this might be a very cool thing. Buy one, and a good aftermarket cooler, and overclock it into the 4GHz range. If your load is mainly single-threaded, like far too many games are, that can give you the same performance but be much, much cheaper.

    1. Re:Possibly more interesting by Anonymous Coward · · Score: 0

      I have a Core i7 950 from ages ago, and overclocked it to 4.5 ghz. It still benchmarks faster than most of what Intel is releasing these days, so I just sit waiting for something better. It seems that the desktop CPUs have stalled out, which is very depressing. I want something faster, Broadwell would probably only double my CPU perf at best.

    2. Re:Possibly more interesting by Anonymous Coward · · Score: 0

      kinda true about games being largely single thread. lots of mulit-thread games use a main thread with a couple side threads doing small ancillary stuff.

      but today, having a CPU that can do more than 2 threads can help quite a bit with a lot of games (particularly AAA titles), and with "next-gen" consoles both packing 8-core x86 CPUs... multi-threading will likely only become more prevalent, or at least better done.

      so in summary, buying a single or dual core CPU for gaming is kinda shooting yourself in the foot.

  18. *sigh* by Anonymous Coward · · Score: 0

    And I'm too poor to upgrade from my 1.6GHz Pentium Dual Core. :

    1. Re:*sigh* by jones_supa · · Score: 1

      Keep an eye on the Core i3 lineup, they have a good price/performance ratio, and a low power consumption as a bonus.

  19. Re:why get this when Broadwell + new chipsets are by Anonymous Coward · · Score: 1

    Not really. With my Difference Engine I only get around $0.10 per instruction since running the waterwheel ain't free.

  20. Why wait for this when by Tmackiller · · Score: 1

    Quantum computing is just around the corner. -_-

    --
    sudo apt-get install sl && sl
  21. Wow...? by Anonymous Coward · · Score: 0

    This is literately the first time in years that Intel has released a CPU that works with the existing motherboard. Ha (No Ivy/Sandy Bridge does not count as it replaced a previous socket type as well as replacing a chipset.)

    The TIM improvement might make it worth an upgrade if you previously had the i3/i5 model that fits this socket, otherwise the performance is merely incremental, and of no interest to users who actually use the virtualization features.

    As I've said before, Intel has hit the wall, there are one or two die shrinks left, but these are going to come at enormous cost, so it's better for Intel to not even bring it out until a competitor (eg AMD) moves to the die size they're already on (currently at 28nm where as Intel is at 22nm) or leapfrogs it. Nobody wants to shrink the die further because it will be less profitable in the long term, and further shrinks will be more error prone. As it is the current generation of CPU's have questionable reliability for use in servers because of quantum tunneling randomly happens due to weak spots in the die at high thermal levels.

    Likewise NAND memory lifespan is reduced with every die shrink, who's to say that CPU lifespan isn't also affected. We just replace our electronics faster than we've had enough time to test it. All these smartphone/tablet devices are going to see a lot of disposal in the next 5 years due to the NAND wearing out or the CPU/RAM failing. Nevermind the battery.

    I have faith in Intel to not screw things up, much more than AMD (who somehow went back to the "weaker FPU" model of competition's of the unreliable K6 era for some boneheaded reason.) If I buy a CPU I want a clock-for-clock performance match between the ALU, FPU and other core technology (eg onboard GPU), and I'm not willing to trade power consumption/thermal headroom just for a cheaper part. If AMD can't compete at the high end then maybe they shouldn't compete at the high end and instead aim at the general consumer parts that go into Dell's instead of Apples.

  22. some times are better than others by Chirs · · Score: 1

    If you know that a new model is coming out in the very near future, then it might be best to wait. Either you get the new model, or else you can sometimes pick up the old model for cheaper.

    1. Re:some times are better than others by Lost+Race · · Score: 1

      That is always true. The longer you wait, the cheaper it gets. If you wait forever, it costs nothing.

      If you need a job done now then get the cheapest equipment that can do the job. If you don't need it now then wait; the price of computer equipment only goes down.

  23. Crappy TIM Application for IVB and HW by WhoBeDaPlaya · · Score: 1

    Thanks Intel for shafting your early adopters and denying that there is a problem with the IHS assembly process for IvyBridge and Haswell. The TIM that they used is actually quite good, it's just the spacing inconsistency between the die and IHS during manufacturing that is the issue.

  24. Wow! .6 GHz Faster Than 3 Year Old 2600k! by BrendaEM · · Score: 0

    I better pry my chip out right after I finish this!

    BTW, my 2600k will overclock to 4.2GHz on air, but I don't do it--because that speed difference is so incremental, and so is Intel's progress.

    The problem is: many program usually are largely single threaded, such as CAD and meshing operations for many other 3D applications.

    --
    https://www.youtube.com/c/BrendaEM
    1. Re:Wow! .6 GHz Faster Than 3 Year Old 2600k! by imashination · · Score: 3, Insightful

      3D Animator here. I made the same mistake as you, thinking my 2.8GHz i7 920 (overclocked to 3.7GHz) would be as fast a current 3.7GHz i7. Each new generation of i7 has been ~5% faster clock for clock. For example my new i7 laptop at 2.6GHz is roughly the same speed as my desktop at 3.7GHz in both single and multi threaded tasks.

    2. Re:Wow! .6 GHz Faster Than 3 Year Old 2600k! by WhoBeDaPlaya · · Score: 1

      C2 45nm -> Nehalem (+10%) -> SandyBridge (+10%) -> IvyBridge (+3%) -> Haswell (+10%)

    3. Re:Wow! .6 GHz Faster Than 3 Year Old 2600k! by geekoid · · Score: 1

      You don't really understand how chips work or what power means, do you?

      --
      The Kruger Dunning explains most post on /. http://en.wikipedia.org/wiki/Dunning%E2%80%93Kruger_effect
  25. In other news... by Torp · · Score: 4, Insightful

    My 2 year old Ivy Bridge Core i7 is fast enough, and will be fast enough for the foreseeable future with no overclocking. Neither Intel nor AMD will get any money from me for at least 3 more years ;)
    Last time an overclock was meaningful for me was when I had a Pentium 1 at 233 Mhz. The bus was 66 Mhz, and that was the ram speed as well. Upping it to 250 Mhz on a 100 Mhz bus (remember back when multipliers went in 0.5 steps?) speed greatly improved the overall responsiveness of the system.

    --
    I apologize for the lack of a signature.
    1. Re:In other news... by jandrese · · Score: 1

      The sad thing is that my positively ancient Core2Duo 2.4Ghz is still plenty fast for most tasks these days. This isn't the old days where upgrading the CPU every couple of years gave noticeable improvements to the OS and generic applications like word processors and web browsers. Worse, application developers have been focusing on improving speed (especially on browsers) instead of just jamming in more features. Its getting harder and harder to find motivation to upgrade. We really need some CPU dependent killer app to convince people to get back on the upgrade treadmill. Maybe some sort of AI?

      Or maybe it's just fine that even cheap chips are plenty powerful for most people and we can stop worrying about it. Sort of like how even cheap cars can hit highway speeds with little effort these days, and do so with good safety margins, good gas mileage, and little maintenance. Enthusiasts will complain that they have no soul, but for regular people they're perfectly good and free up resources for other endeavors.

      --

      I read the internet for the articles.
    2. Re:In other news... by Torp · · Score: 1

      Yeah, my needs are already above the norm. I could do with your core 2 duo for about 90% of my stuff. Heck, I could do with my 2009 macbook white for about 90% of my stuff.

      --
      I apologize for the lack of a signature.
    3. Re:In other news... by Anonymous Coward · · Score: 0

      Javascript App...

    4. Re:In other news... by jandrese · · Score: 1

      All major browser manufacturers are putting an enormous amount of effort into making Javascript fast. Every year it gets more and more efficient. Already you can do quite complex tasks on a browser without noticeable slowdown (see: Google Docs). As long as you aren't doing something silly like ray tracing with Javascript, even my 8 year old chip is more than enough.

      --

      I read the internet for the articles.
    5. Re:In other news... by nhat11 · · Score: 1

      Not for a laptop. I have a 3rd gen i5 and because of the amount of software security and applications running, I can't imagine having this run on a C2D

    6. Re:In other news... by raxtich · · Score: 1

      Agreed. I also have an i7 Ivy and have zero complaints about performance. With 16GB of RAM and an SSD drive, I leave everything I use on a regular basis open all the time (Photoshop, Steam, Aptana, iTunes, and Firefox with numerous tabs opened). I just works, with no lag or hiccups.

    7. Re:In other news... by Anonymous Coward · · Score: 0

      We really need some CPU dependent killer app to convince people to get back on the upgrade treadmill.

      Is this you again, Brian Krzanich?

    8. Re:In other news... by Anonymous Coward · · Score: 0

      I had a E8400 Core2Duo, and most of the time any sort of operation such as updating Windows or other menial program tasks pushed the CPU to 100% utilization. The CPU was overclocked as well, though can't remember how much. I was being bottlenecked.

      So I bought a i5-2500k, overclocked to 4,4GHz and now I don't have anything to do on a regular basis that will max out utilization. It basically sit idle for most of the time. I'm confident I can keep this chip until electromigration kills it.

  26. Re:why get this when Broadwell + new chipsets are by InvalidError · · Score: 2

    When Ivy Bridge came out, popular Sandy Bridge chips' prices went up and Intel also bumped Ivy Bridge chip prices up $10-20.

    When Haswell came out, many of the more popular Ivy Bridge chips went up $10-20 again.

    How many years has it been since the last time Intel made major price cut announcements after introducing newer higher-end models within a product line or even introducing a new product line? I do not remember reading about such announcements in over five years; instead of slashing prices, Intel simply discontinues models altogether. If you want to buy Intel chips at prices significantly below launch prices, you have to either buy second-hand or find a vendor who has surplus stock they need to get rid of. (Or buy from Microcenter which appears to have some sort of sweetheart deal with Intel for unlocked chips.)

  27. TIM? RLY? by DarthVain · · Score: 0

    One of the two new "features" is basically TIM that doesn't suck so much?

    A) The first thing than an OC does is wipe whatever the fsck is on there off.
    B) Quality stuff is literally 4$ a tube, and per application is measured in pocket change. Is that significant to the cost of a high end CPU?
    C) Many new components like aftermarket HS and water blocks now come with quality TIM, and not just silicone sludge.

    1. Re:TIM? RLY? by Torp · · Score: 1

      They're talking about the *other* layer of TIM, between the metal capsule that you wipe off and the actual CPU core. Can't access it without a razor blade.

      --
      I apologize for the lack of a signature.
    2. Re:TIM? RLY? by Hodr · · Score: 1

      You may be confused. They aren't referring to the white silicone paste or the sticky pad that is normally use to attach the heat spreader (top of the CPU you normally see) to the OEM heatsink/fan.

      They are talking about the material used to attach the heat spreader to the actual CPU core. Unless you are prying the metal cap off the top of the CPU you aren't replacing this existing TIM with your $4 arctic silver compound.

    3. Re:TIM? RLY? by DarthVain · · Score: 1

      Ah. Obviously didn't read the article.

      Though I remember people lapping CPU's back in the day, though again that was more to level and ensure a good mating with the HS. Thought I have seen some that got the metal pretty thin as well almost exposing the core.

      However the whole conversation is a bit moot. By Intel "targeting" the OC you pretty much eliminate the purpose for doing so in the first place. The whole idea was you take a cheap chip, and OC it to something much better. Now the chips you can OC are more expensive anyway...

    4. Re:TIM? RLY? by sexconker · · Score: 1

      Ah. Obviously didn't read the article.

      Though I remember people lapping CPU's back in the day, though again that was more to level and ensure a good mating with the HS. Thought I have seen some that got the metal pretty thin as well almost exposing the core.

      However the whole conversation is a bit moot. By Intel "targeting" the OC you pretty much eliminate the purpose for doing so in the first place. The whole idea was you take a cheap chip, and OC it to something much better. Now the chips you can OC are more expensive anyway...

      What? "Back in the day" we had exposed cores. Enthusiasts would lap the heatsink to a mirror shine, meticulously apply a thin, even coating of goo to the CPU core, and pray as they clamped/bolted/screwed/whatevered the heatsink on.

      Then Intel started putting heat spreaders on their shit. Some people removed the built in heat spreaders in an attempt to get better cooling by removing the middle man. Usually it made no real difference. AMD followed suit a few years later.

      Today, only an idiot would remove the IHS from a CPU in an attempt to get better cooling. The installation at the factory is far better than anything you'll do yourself. Intel's last generation of CPUs was an outlier - they did such a shitty job of capping the CPU that people desperately tried to improve the cooling by going back to a decade-old trick. Regardless of what you tried, though, the overclocking capabilities of that generation were pure shit.

      I wish Intel/AMD would just offer an SKU with a factory-mounted permanent heatsink. I also wish phone manufacturers would offer an SKU with a factory-installed screen protector. I sure as fuck don't enjoy buying a 3-pack on Amazon and fucking up the first one, getting the second one kind of okay, and then not being able to decide to try to do it better with the third one or not. For heatsinks, I don't enjoy the added cost of 3rd party heatsinks, fans, goo, etc. While installation is simple, I would appreciate the extra efficiency (in cooling and cost) I could get if the manufacturer did that shit at the factory.

    5. Re:TIM? RLY? by DarthVain · · Score: 1

      It is probably a protection thing for transport.

      Not really the same but close, you can get certain form factors with the CPU soldered to the MB... All CPU come with a basic HS, and some "goo".

      Thought it is an interesting question. It is well known that CPU *REQUIRE* a heat sink, yet I have never seen any integrated into the actual design... perhaps it just doesn't work very well if they have tried it at some level...

  28. Re:why get this when Broadwell + new chipsets are by drinkypoo · · Score: 4, Insightful

    Some of you early-adopters may laugh at this, but this has been my upgrade strategy for decades now and from a bang-for-the-buck perspective it's extremely effective.

    It's not just about the bang, either, but about the boon or the bane. If you wait for a while, you get to see whether something has massive fail built into it. When I buy based on hope I usually fail. When I buy based on what seems to have held up, I am usually happy. Leaving time for the 1.1 or 2.0 rev motherboard and some bios updates, and for some video driver updates, really improves system stability.

    --
    "You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
  29. Re:why get this when Broadwell + new chipsets are by timeOday · · Score: 2

    I've been waiting years for the quad-core 65W Socket 775 CPUs to come down. The Q9550S was released in January of 2009 and still goes for over $200. Meanwhile the Core 2 Duo in the system I would upgrade is working fine; I just check ebay a couple times per year to see if I can pick up a quad core for cheap, but no.

  30. NGPTIM by Anonymous Coward · · Score: 0

    Klaatu Barada N...

    Necktie... Neckturn... Nickel... It's an "N" word, it's definitely an "N" word!

    Klaatu... Barada... NGPTIM

    *pause*

    Okay then... that's it!

  31. 4 GHz by ilsaloving · · Score: 1

    I think the most significant thing is that they managed to break the 4GHz barrier. Is this a one off, or did they finally come up with the technology required to make >4GHz chips a standard thing from now on?

    1. Re: 4 GHz by Anonymous Coward · · Score: 0

      Imagine an of to 5ghz. That will be cool. I think a couple years ago they had to run liquid nitrogen to get a CPU high.

    2. Re:4 GHz by Anonymous Coward · · Score: 0

      I'm running my i7 at 4.5 ghz right now. Many of them worked fine at that speed, and I'm fairly sure they released prior out of the box Intels that were 4 ghz+.

    3. Re:4 GHz by radarskiy · · Score: 1

      "they managed to break the 4GHz barrier"

      Dell sold a Presler-based system at 4.25 GHz in 2006: http://anandtech.com/show/1916...

    4. Re:4 GHz by Anonymous Coward · · Score: 0

      Doesn't AMD sell a chip that's 4+GHz stock?

  32. Re:why get this when Broadwell + new chipsets are by gstoddart · · Score: 2

    Yup. I usually buy somewhat trailing edge stuff myself.

    A friends aunt once asked what kind of computer she should buy, and she said "you're going to tell me to buy the most expensive computer around, aren't you?". To which I said "nope, I'm going to tell you to buy the cheapest machine you can find which can have its memory upgraded to at least double what entry level is, because you don't need anything faster, but more memory is always better".

    Several years later, she was still happily running the same machine, and never once found it too slow. She had passed on the advice to several friends, all of whom felt they ended up with a better computer as a result.

    I know for me, I'm pretty much never CPU bound, and haven't been in quite some time. Memory, however, is something you should buy as much as the machine can hold.

    I know some people will need this much CPU power, but for most people I suspect CPU speed hasn't really mattered in years. The early adopters just ensure that in a year or so we'll all be able to buy it for half the cost.

    --
    Lost at C:>. Found at C.
  33. Not much in the last year by Anonymous Coward · · Score: 0

    What were they doing for a whole year? Just give us 512 bit AVX on 8 core + HT already!

  34. Re:why get this when Broadwell + new chipsets are by NJRoadfan · · Score: 1

    Sold listings on ebay point to an average selling price of $125-150 for the Q9550S, some of which came with motherboards. The 45nm Yorkfield chips didn't seem to be very popular and were released quite late. Most folks went with Nehalem CPUs, so the low number of C2Qs from that era likely keep the prices high. The 65nm Q6600 was far more popular, easy to overclock, and can be had for cheap. I have a X38 based system I would like to upgrade myself, but with the high cost of CPUs and DDR2 memory, its likely to get replaced completely at this point.

  35. Re: why get this when Broadwell + new chipsets are by Anonymous Coward · · Score: 0

    The 3rd reason which is very important is : revenue from these new products fund the company for r&d, which is extremely expensive. I'm not saying everybody should always like cutting edge the market finds a way to balance out - most buy newest releases like in oem computers, some buy older stuff.

    Could you imagine if everybody always waited until the latest version to buy the earlier version. We would never advance. So we need people to buy the latest stuff we need people to buy the older stuff.

  36. Intel's catching up in the clock wars! by OrangeTide · · Score: 1

    "It was released on June 8, 2007 at speeds of 3.5, 4.2 and 4.7 GHz" -- POWER6

    --
    “Common sense is not so common.” — Voltaire
    1. Re:Intel's catching up in the clock wars! by TechyImmigrant · · Score: 1

      High clock speed is a poor trade off these days. That's why the clock speed wars ceased.

      If you see an oddball chip touting a 50% clock speed improvement over the current top end mass market CPUs you know you're going to need a power station and a refrigeration plant just to run the thing.

      --
      I should use this sig to advertise my book ISBN-13 : 978-1501515132.
    2. Re:Intel's catching up in the clock wars! by AnontheDestroyer · · Score: 1

      High clock speed is a poor trade off these days. That's why the clock speed wars ceased.

      It is? You don't say!

    3. Re:Intel's catching up in the clock wars! by Anonymous Coward · · Score: 0

      Yea, but when you have a linear problem to solve, nothing can do the job like a high clockspeed CPU. And linear problems exist, even if you have a strategy for partitioning a linear problem at massive efficiency losses such as speculative precomputation, you still need a fast CPU at the top of the tree to run the toplevel linear algorithm.

      Remember the most efficient algorithm isn't always the fastest. You can make the critical path slightly shorter by doing much more work off to the sides, but the total work done is much larger. This is the case for GPU accelerated raytracing, A GPU cannot branch each datapath, so it speculatively computes a bunch of paths which are not taken, and selects the result after it knows which path should have been taken. This is clearly much less efficient than the linear algorithm for raytracing, but has a usefully shorter critical path and is thus preferred by users.

      Other problems just don't have a reasonable paritioning strategy, the one I can think of off the top of my head, which is not very useful, is cipher feedback (C(n+1) = S(C(n)) xor P(n+1)). you can't begin calculating S(n+1) until you know C(n), so there is no partition strategy, and if there was then S() is not a secure cipher. It's not a useful example because CFB is easily replaced with CCM (C(n) = S(n+x) xor P(n), where x is a nonce), where S(n+x) can be calculated in parallel.

      Needing a power station and a cryogenic plant just to run a CPU isn't such a bad thing if it's sitting atop a very expensive pyramid of precomputation. If it halfs your critical path, it doubles the value of your entire precomputation.

      --
      C(n) is the ciphertext at position n
      P(n) is the plaintext input at position n
      S() is a symmetric cipher encryption

    4. Re:Intel's catching up in the clock wars! by TechyImmigrant · · Score: 1

      But if the computation takes more than two years, you're better off waiting for faster silicon.

      CCM is passe. OTR mode is where it's at.

      --
      I should use this sig to advertise my book ISBN-13 : 978-1501515132.
  37. Re: why get this when Broadwell + new chipsets are by HornWumpus · · Score: 1

    It's similar to brand new car buyers. Without them there would be no used market. But damn they are stupid as rocks and buy far to many automatics.

    --
    John McAfee 'It was like that time I hired that Bangkok prostitute; to do my taxes, while I fucked my accountant'
  38. Re:why get this when Broadwell + new chipsets are by timeOday · · Score: 1

    But that Q6600 sucks down 105W / 155W Max, so I am loathe to put it in my home server (which runs all the time) - on top of the electrical costs, it would be a gamble on the PSU and cooling in that box... and what is a more annoying waste of time than an "almost stable" system?

  39. Re:why get this when Broadwell + new chipsets are by davester666 · · Score: 2

    Or, with just a small amount of risk, you can get one for free soon after they are released...

    --
    Sleep your way to a whiter smile...date a dentist!
  40. Re:why get this when Broadwell + new chipsets are by Blaskowicz · · Score: 1

    You could get a low end motherboard and a Haswell Pentium G : dual core and dual thread, but the same multithreaded performance as that Q9550S roughly, for less power use.

  41. Re:why get this when Broadwell + new chipsets are by K.+S.+Kyosuke · · Score: 2

    Quite possibly, but by then, the memory modules will cost you an arm and a leg! (Or an x86 and a leg?)

    --
    Ezekiel 23:20
  42. Re:why get this when Broadwell + new chipsets are by Anonymous Coward · · Score: 0

    Why would you want that one? I'm sure it will only support 640 GB of main memory.

  43. Re:why get this when Broadwell + new chipsets are by K.+S.+Kyosuke · · Score: 1

    Uh...go buy a Pentium G3420, or its successor, but not before the machine breaks down? For the money it costs, it seems to me you're solving a non-existent problem. (The new server boards with Atom C2750 are nice for multi-threaded workload in a similar performance range, with very little power, though. That depends on what you're running, of course. More expensive though, but you get ECC and proper server interfaces at least.)

    --
    Ezekiel 23:20
  44. So is my Core 2 Q6600 obsolete now? by AnontheDestroyer · · Score: 1

    Tell me the truth. I think I can handle it.

    1. Re:So is my Core 2 Q6600 obsolete now? by Anonymous Coward · · Score: 0

      no, it's ok.

      footnote: i still keep my q6600 in my desk. just the cpu, that is. it's hard to abandon such a loyal friend. i still remember the time we ran at 4.2ghz together for a few weeks. the internet seemed so much nicer back then.

    2. Re:So is my Core 2 Q6600 obsolete now? by raxtich · · Score: 1

      I passed my Q6600 to my wife's computer. She somehow manages to get all her work and gaming done with no complaints on that "obsolete" CPU.

    3. Re:So is my Core 2 Q6600 obsolete now? by Anonymous Coward · · Score: 0

      My previous machine was a Q6600 with 8GB of DDR2. A year and a half ago I built a new machine with an i7-3770 (no suffix) with 16GB of DDR3. I still have the old machine, but it's powered off now. The new machine's CPU is 2x faster on single threaded benchmarks (3x faster multithreaded; this is to be expected, since it has 1.5x higher clock rate and 2x as many threads), and its DDR3 RAM is 2x faster than DDR2. Also, the new CPU uses ~25% less power.

      p.s. Your Q6600's motherboard probably doesn't do SATA3 (2x faster than SATA2) or PCIe x16 3.0 (2x faster than PCIe x16 2.0) or USB3 (10x faster than USB2), so you will also at least double the performance of every bus in your computer by upgrading.

  45. Market segmentation by ponos · · Score: 1

    Sounds like a really cool product. It's a pity intel has over-segmented their product line and I can't get this chip with ECC. That would be cool. In fact, ECC should some day become a standard feature. But that's just wishful thinking.

  46. Re:why get this when Broadwell + new chipsets are by aliquis · · Score: 1

    Quite possibly, but by then, the memory modules will cost you an arm and a leg! (Or an x86 and a leg?)

    You'd likely get at least half the large components of the computer together with it (motherboard, ram and possibly graphics card.)

  47. Re:why get this when Broadwell + new chipsets are by aliquis · · Score: 1

    Or, with just a small amount of risk, you can get one for free soon after they are released...

    How you figure?

    Steal one?

  48. Pffft by Anonymous Coward · · Score: 0

    My 2500k still does 5.0Ghz on air, and only lacks in apps which really (really really) need hyperthreading. Plus, I bet the max headroom on these chips is barely different (if at all) from the 3770k or 4770k. I wouldn't bet on better binning, they all seem to perform pretty close to the same.

  49. Tick/Tock/Wow, new capacitors by justthinkit · · Score: 1

    So the new cycles are: Tick/Tock/Wow, new capacitors.

    The article, amazingly poorly written by the way, mentions exactly zero changes to the chip itself.

    We are given "new caps, new glue, and you can overclock it for more performance".

    I'm off to ask my doctor to start me on Prozac...

    --
    I come here for the love
  50. Re:why get this when Broadwell + new chipsets are by Grog6 · · Score: 1

    "640GB is enough for anybody." - Bill Gates III , 2064 :)

    --
    Truth isn't Truth - Guliani
  51. Re:why get this when Broadwell + new chipsets are by Anonymous Coward · · Score: 0

    The single threaded performance of 4970K @ 5.0 Ghz is approx double that of E8400 in worst case scenario based on extrapolations from 4770K and from looking at countless benchmarks over the last couple years. If your app can utilize more than two cores efficiently but can't be ported to use GPGPU algos (which can give ~10x gain on $200 GPU over same algo on E8400), that would mean around ~6x speedup over E8400.

    I can't speak for your app, but I have a C# program I've written that assuming I can't bother to optimize it by rewriting, I'm estimating that going to Q9550S from E8500@3.8 would at best being 1.3 times current perf as it's parallizable in rather minimal way without major rearchitecting. If I upgrade to 4970K, that would cost about $600. That is interesting for following reasons: With no changes it would mean approx 2-2.5 times current perf. Since each core in 4970K is roughly 2 times faster than single thread of C2D @3.8 with no changes, that would leave me with "$400" worth of motivation to improve the app to better take advantage of the new cores. The issue here is : is $400 enough to motivate that? Perhaps it would make more sense to get Haswell-E and spend $1500 to have ~1300 reasons to take advantage of 8 cores instead of just 4 but at that point one would have to also look at whether the workload is suitable for eg. Nvidia Titan. I already have a GTX 760 so if I really cared, I could learn enough Cuda to make tests to see what kind of improvement Titan might give, and then base the choice between Titan and Haswell-E on that. But at this point I suspect the problems of parallelizing the app could pose too much difficulty. The easiest path to most gain is thus to just get 4970K now and change it to Broadwell K later.

    Now - If you had 2x00K, upgrading now (before say Broadwell/Skylake which bring more significant improvements) would be questionable unless you know you either run apps needing some of the newer instructions or use things that Haswell is specifically good at, like emulations/virtualization stuff. And if you really really care about virtualization stuff (need to passthrough hardware to a VM), the just announced K models that lack VT-d are not as tempting. Though on Windows desktop I think limited if any good support for using VT-d (vmware and hyper-v do not support it for GPU/storage passthrough atleast in a documented way).

  52. Re:why get this when Broadwell + new chipsets are by Anonymous Coward · · Score: 0

    I happen to be in the market for a new LGA 1150 motherboard for an i5. Does this mean I should hold off from the new Z97 motherboards, and stick with the more time-tested Z87 boards? The new Z97 boards are only marginally more expensive than the older Z87, making them temping as future-proofing.

    There are rumors that the Z87 boards will support the Devil's Canyon cpus via bios update. Any news on this front?

  53. Possibly more interesting by Anonymous Coward · · Score: 0

    TL;DR instead of getting the Pentium, I'd argue to loan money if you have to upgrade and get atleast the 4690K/4790K or wait for Broadwell/Skylake replacement for that at the same perf level or better.

    I'd look it like this (when everyone got E8400/Q6600 I got E2140 then E8500 used later, I saved some $100-150 over last 6 years but gained a lot of hassle and agony over perf - infact given all the agony I'm tempted to look at Haswell-E now too, even if that cost $1500 - there's nothing worse in the PC tech than having to agonize over upgrading a perfectly working system prematurely in my experience over past couple years still being on C2D while apps are written targeting atleast 2500k)

    You can upgrade memory easily. You can't upgrade MB without tearing things apart and risking DRM apps breaking and other possible pain that I won't go into.
    You can upgrade CPU relatively easily but as I said above, it doesn't really make sense - if you have a budget, save up instead of going for the sub-$200 cpu. Been there with the sub-$100 cpu and regretted it couple years in.

    The Pentium lacks a bunch of new ISA that compilers may start to opt in. So when software update comes, it may be that you end up on the slow path that the dev isn't really bothering to test in worst case as devs generally get the highest end kit to speed up their compiles. I disagree with this practise (devs should target the medium perf hw instead of top end or lowest end). And as devs are using 8 cores they have incentive to make most of that, so if they use even just 3 cores, that 2 core one will start to trash as the app was written expecting to use 3 cores (using 3 cores efficiently is easy for almost any type of app, more than that starts to get algo-specific).

    Summary: The Pentium is just for those who seek the highest clocks and idiots like myself who hoped to save $ but gained poor perf prior to my planned upgrade.

  54. 4790K may actually have TSX! by Anonymous Coward · · Score: 0

    http://ark.intel.com/products/80807/Intel-Core-i7-4790K-Processor-8M-Cache-up-to-4_40-GHz

    According to Intel 4970K has TSX while 4770K does not. Recent CPU-Z does NOT show this from what I know, so those screenshot leaks are not necessarily accurate.

  55. Imagine a Beowulf cluster of these things.... by spike2131 · · Score: 1

    Yeah, I went there.

    --
    SpyDock: Scientific Python in a Docker container
  56. Re:why get this when Broadwell + new chipsets are by trev.norris · · Score: 1

    I can see this being true. The only time I max out my i7 is when I'm compiling a program. Other than that it pretty much sits 90% idling. Though, a nice SSD upgrade can also have much greater impact on performance than a CPU upgrade.

  57. You don't know how money works do you? by BrendaEM · · Score: 1

    For a person owning a single workstation power consumptions means little.

    Even if you are an intel fanboi, look at the single core performance, in a comparable test.
    http://www.tomshardware.com/ch...

    I don't thing the newer chips will be much faster at multi-core raytracing either.

    --
    https://www.youtube.com/c/BrendaEM
  58. Re:why get this when Broadwell + new chipsets are by toddestan · · Score: 1

    I would say this was true a few years back, when memory was expensive and you generally never had enough to keep your OS from having to hit the swapfile/pagefile. Nowadays, memory is cheap and with "entry level" machines coming with 4GB, that's more than enough for most anyone who isn't going to push the machine hard. Someone else mentioned a SSD - that's probably the one performance upgrade that most anyone is going to notice, though if someone was concerned primarily about stability and longevity I would have a hard time recommending a SSD. I'd probably tell them to get a fairly basic machine and put the extra money towards a nicer screen.

  59. Re:why get this when Broadwell + new chipsets are by toddestan · · Score: 1

    Those chips have been long out of production, so you're either looking at a used chip or maybe the occasional new-old stock. The problem with trying to buy the high-end chips from older sockets is that there is a fairly good sized number of people after those chips who want to do an easy upgrade, or perhaps have a computer where the OEM Windows license is tied to the motherboard. This usually keeps the "top dog" at around $100-$200 for a surprising long time. A hint though, is to search for another model that's a few steps back from the top end. They usually go for considerably less, and that could get you into a quad core of some sort.