Intel Launches 8th Generation Core CPUs (anandtech.com)

Reader joshtops writes: Today Intel is launching its new 8th Generation family of processors, starting with four CPUs for the 15W mobile family. There are two elements that make the launch of these 8th Gen processors different. First is that the 8th Gen is at a high enough level, running basically the same microarchitecture as the 7th Gen. But the key element is that, at the same price and power where a user would get a dual core i5-U or i7-U in their laptop, Intel will now be bumping those product lines up to quad-cores with hyperthreading. This gives a 100% gain in cores and 100% gain in threads. Obviously nothing is for free, so despite Intel stating that they've made minor tweaks to the microarchitecture and manufacturing to get better performing silicon, the base frequencies are down slightly. Turbo modes are still high, ensuring a similar user experience in most computing tasks. Memory support is similar -- DDR4 and LPDDR3 are supported, but not LPDDR4 -- although DDR4 moves up to DDR4-2400 from DDR4-2133. Another change from 7th Gen to 8th Gen will be in the graphics. Intel is upgrading the nomenclature of the integrated graphics from HD 620 to UHD 620, indicating that the silicon is suited for 4K playback and processing.

Comments

Lies, damn lies, and benchmarks

[Locke2005]

Intel is claiming a 40% performance gain based on the additional cores. But the fact is that most apps use only a single thread, so with a lower clock rate, most apps will actually run slower! Yes, for the few apps designed to use all available threads there should be an improvement, but only if the memory architecture has enough bandwidth to keep all those CPUs fed with data. The limiting factor in most high performance computing is not CPU cycles, it's delays in getting data to each CPU. In other words memory bandwidth is more important than total available CPU cycles. Also, these processors slow down the clock rate as each additional CPU is utilized to avoid overheating.d Short summary: Your Mileage May Vary.

Re:Lies, damn lies, and benchmarks

[Anonymous Coward]

All modern operating systems are multithreaded and anyone will see huge improvement even in single threaded apps because the OS will be able to allocate other processes and housekeeping chores to entirely different cores. You need to get out of the 1990's my friend and join us here in the 21st. If you make the jump today, you can even see an eclipse!

Re:Lies, damn lies, and benchmarks

[known_coward_69]

this has been the case since they released MMX 20 years ago. Anything coded to use their architecture and special instructions gets a huge boost. everything else no.

welcome to how computers have always worked

Re:Lies, damn lies, and benchmarks

[Anonymous Coward]

It depends.
If you are running a single threaded application and nothing else besides you are unlikely to see any improvement at all by going from 2 cores to 4.
Maybe if your OS is brain dead and running a couple of busy wait loops you might notice something.

Re:Lies, damn lies, and benchmarks

[AmiMoJo]

AMD CPUs have better features anyway. ECC RAM support on the desktop, for example.

Re:Lies, damn lies, and benchmarks

[Anonymous Coward]

what apps are single-threaded these days? sorry if your Microsoft Word experience won't get any better bro

Re:Lies, damn lies, and benchmarks

[alvinrod]

On the other hand, the turbo is faster so unless you're really pegging that single core for a prolonged period of time, the new chip is going to have a small performance edge. Also, the bigger L3 cache will probably be helpful in certain applications.

Re:Lies, damn lies, and benchmarks

[Anonymous Coward]

SolidWorks uses only a single core.

Re:Lies, damn lies, and benchmarks

[Locke2005]

Agreed, any increase in the size of on-chip cache should be a win in terms of getting data into the CPU. Again, I believe the peak boost speed decreases when more CPUs are in use. I assume Intel did a better job of balancing this so there should be some improvement for certain problem sets.

Re:Lies, damn lies, and benchmarks

[Locke2005]

You would see an improvement in going from a single core to dual core, yes. But if your OS is running on one thread and a single-threaded app on another, then not that much improvement, unless your OS actually fully utilizes all but one of the threads -- which it doesn't, due to all the threads time-sharing the same data bus. Again, for most problem sets, data throughput is the bottleneck, not aggregate CPU cycles.

Re: Lies, damn lies, and benchmarks

[Anonymous Coward]

Games.

Re:Lies, damn lies, and benchmarks

[Anonymous Coward]

> the fact is that most apps use only a single thread, so with a lower clock rate, most apps will actually run slower

You're never running a single app. Whatever modern OS you're using is probably heavily threaded, so even if your shitty 1-thread app doesn't benefit, the overall system still will.

Re: Lies, damn lies, and benchmarks

[Anonymous Coward]

In addition to all those cores, in comes complete with a hardware backdoor (IME) that has its own processor to send reports to the appropriate US agency.

Re:Lies, damn lies, and benchmarks

[ShanghaiBill]

All modern operating systems are multithreaded and anyone will see huge improvement

No, you will see tiny improvements. OS overhead is usually less than 10%, and often less than 2%. So adding a second core will give you a slight boost in performance for a single threaded app. Additional cores after that will get you nothing.

Re:Lies, damn lies, and benchmarks

[Anonymous Coward]

I care more about reliability, and on that front I can only draw from my own experience. Being a former Amiga user, I finally migrated to the PC world with a self-built AMD-K6 cpu. I got royally fucked because I was all into playing games like Quake that my Amiga could never play, only to later learn that AMD had a shitty FPU. My next PC was AMD Athlon based. But, after this I worked a few years as a PC repair tech. I saw countless fried CPUs in my day.... ALL of them being AMD. Since then, I went Intel and never went back. My opinion of them slipped further when they acquired ATI, who I consider to be a loser in the field. I've been nursing the same Core 2 Quad chip I bought back in 2010. I'm due for a new machine, but I just don't feel confident about AMD, at all. Give me some hard data to convince me.

Re: Lies, damn lies, and benchmarks

[EvilSS]

Lots of games are multi-threaded these days.

Re:Lies, damn lies, and benchmarks

[lexman098]

Don't let the AMD circle-jerkers hear you talking about single threaded performance. Didn't you know Ryzen is The Second Coming and Intel will be out of business in a matter of months?

Re:Lies, damn lies, and benchmarks

[mwfischer]

not really. see this article from 2010

http://www.javelin-tech.com/bl...

Re:Lies, damn lies, and benchmarks

[Anonymous Coward]

It's a fact. Single core. That's why buying a 8 core processor does not matter as much as speed and RAM.

Re:Lies, damn lies, and benchmarks

[Khyber]

"All modern operating systems are multithreaded"

And by default all modern OSes run everything (or try to) across every core, which quite often fucks up performance of single-threaded applications.

Re: Lies, damn lies, and benchmarks

[Anonymous Coward]

Everything where one core isn't enough is multi-threaded.
Compiling, media conversion, gaming.
Well, not every program in those sections, there are always badly coded examples.

Almost every problem that is considered "heavy" is really just the same operation on a large number of items.

Gaming is such a bad example too considering how much it is locked to a very GPU with a silly number of cores.

Re: Lies, damn lies, and benchmarks

[Anonymous Coward]

And not just games - browsers are, for instance (IF I Recall, at least), turning to a model where each thread is (its own thread, or its own process - my memory is fuzzy here, mind you).

Theoretically, wouldn't more cores/threads allow you to run more multi-threaded apps "at once" without worry about them being put onto a waiting queue, waiting for free processor cores to run them, and allowing them to complete their task faster than if they had fewer cores/needed to wait with some frequency?

Re: Lies, damn lies, and benchmarks

[cfalcon]

Many games are multithreaded, but there's still a wide difference in how they work. Some games have two threads, and one usually ends up doing a bunch of work- and of course, most processors handle two threads just fine. Few games are meaningfully split into like thirty threads or whatever and will scale evenly with processor speed or processor core count, and some games have an absolutely trivial amount of CPU work anyway. It's pretty variable. Certainly you wouldn't chase core count if gaming was your primary driver for chip purchase, you would chase single threaded performance as long as you weren't dropping all the way down to two cores or something.

Re:Lies, damn lies, and benchmarks

[Anonymous Coward]

ATI, who I consider to be a loser in the field.

Accurate, but when it's a two person race, loser doesn't carry accurate weight.

Say, rather, useful idiot. Much like AMD - a match made in heaven, to be sure.

Both Intel and nVidia have to keep AMD around, because competitive genocide is frowned upon.

So, AMD gets to live. And occasionally catch up. And Intel and nVidia both pull whatever shit fell on the floor, dust it off, and bitchslap AMD's forward momentum. Not enough to kill, just enough to say, "Oh, see, we're not a natural monopoly. Consumers can (stifled laughter) always go with AMD!"

Re: Lies, damn lies, and benchmarks

[ChrisMaple]

Code an engineer writes for use once or a few times will be single-threaded unless the compiler can automatically convert it to multi-threaded. Write once, start it at 5 PM, hope it's done when you return to work the next morning.

Re:Lies, damn lies, and benchmarks

[Anonymous Coward]

Noob alert.

Fire up 'ps aux' on a unix machine and you will see hundreds of processes and threads. The idea of "most apps use only a single thread" died sometime 15 years ago. Any app that runs on Windows, macOS or Linux uses dozens, if not hundreds of threads. Just because the programmer doesn't explicitly schedule threads, doesn't mean there aren't dozens running (like, network access, disk access, memory management, keyboard input, mouse input, sleep mode, etc...)

Re: Lies, damn lies, and benchmarks

[Anonymous Coward]

They have lower base clock but higher boost. If you're really only using a single threaded app, maintaining boost clock rate shouldn't be a problem.

Re:Lies, damn lies, and benchmarks

[Anonymous Coward]

I like AMD and want them to succeed too, but I've never found an AMD system as reliable as Intel. I had a K6 system too, and the VIA chipsets were just not as good as the Intel 440 chipsets. The Athlon XP was a great chip, but was hampered by shitty VIA chipsets for almost its entire production run. nVidia finally saved the day with the nForce chipsets, which are the only decent chipsets for Socket A. Based upon this I built some Athlon 64 systems on the later nForce chipsets and found them to be absolute garbage. The last time I gave AMD a chance was with the Phenom II and one of their own chipsets. However, after their buggy SATA implementation corrupted the harddrive a couple of times I was done. The most interesting part of that is I swapped the motherboard and CPU with Intel, but kept the AMD graphics card I had because I didn't have the money to replace that too. With this card I would get a BSOD from the graphics driver every week or two, but the same card using the same driver on an Intel platform it never BSOD'd again. So even their own GPU's work better on Intel. Go figure.

So, in fewer words

[Anonymous Coward]

Intel has new labels for the same processors at the same price. AMD must really have Intel executives crapping their pants.

Re:So, in fewer words

[Daemonik]

This is why I love AMD. Intel and nVidia start getting complacent then AMD punches them in the face until they catch up again.

Re:So, in fewer words

[Bing Tsher E]

Agreed. I am glad that AMD exists, and I am glad that somebody else is willing to buy the AMD 'stuff.'

Re:So, in fewer words

[rock_climbing_guy]

Amen to this. I remember years ago when AMD was considered a joke and they suddenly burst onto the scene with their "Athlon" line of processors.

Prior to this, Intel would charge north of $1000 for their top-of-the-line CPUs and AMD forced them to sell for more competitive prices.

Re:So, in fewer words

[Bing Tsher E]

The punchline to the joke was the first time I visited my brother-in-laws house after he had bought an Athlon box. It was so loud I thought maybe he was running a Rocket Simulator on it.

Re:So, in fewer words

[Ost99]

Going for 2 core to 4 cores (U-series laptop processors) and from 4 to 6 (desktop) doesn't seem to be just relabeling.

Re:So, in fewer words

[EvilSS]

This is why I love AMD. Intel and nVidia start getting complacent then AMD punches them in the face until they catch up again.

LOL yea that Vega release was a real big blow to NVidia.

Re:So, in fewer words

[Anonymous Coward]

Noting, of course, that the processor itself makes little noise.

Re:So, in fewer words

[Hydrian]

I agree that the AMD's k5 were crap. But AMD was good even before they released the Athlon line. My AMD 66 MHz DX2 did loops around my friends Intel Pentium 100Mhz.

Re:So, in fewer words

[rock_climbing_guy]

I've been surprised at how awfully noise stock coolers are these days. If they're going to give you such a crummy cooler, they ought to simply save you a few $$$ and not include one since you're going to throw it away and buy a better one.

Flailing in failure.

[Gravis Zero]

Intel has lost the crown for performance, never had the crown for being low power and has even discarded all attempt to enter the IoT market. It seems like all these releases are Intel's attempt at throwing everything at the wall and hoping something sticks. Meanwhile, I wonder how much cash they are doling out to prevent people from selling systems with AMD chips.

Re:Flailing in failure.

[KingMotley]

The what? AMD doesn't have anything available that competes at the high end performance wise, and hasn't for a very long time. As for low power, except at the extremely low end, intel chips require a lot less power than their AMD equivalent at the same performance level.

Re:Flailing in failure.

[Anonymous Coward]

The high end is all about price/performance, and AMD is eating Intel's lunch at the moment.

Re:Flailing in failure.

[Anonymous Coward]

"AMD doesn't have anything available that competes at the high end performance wise" You don't know what you're talking about, again. Intel fanboy needs a good cry.

Re:Flailing in failure.

[Anonymous Coward]

Threadripper competes nicely on the high end...or didnt you notice? Sure, at 4k resolutions with a GTX1080Ti, you will get 3-4 fps more out of Intel's chips, but in all other applications, Threadripper mops the floor with Intel. Fanboys are sad, and just a little bit pathetic.

Re:Flailing in failure.

[Anonymous Coward]

Intel's unlocked CPUs can still be raised closer/easier to 5 Ghz than Ryzen can. That's important for some applications, including maxing framerate on games, but having more cores will help as well.

Which is why Intel is moving their i5s and i7s up to six-cores. They probably had this ability years ago, but refused to do it to maximize profits (if they had, AMD would probably be a distant memory right now). But I'd be willing to buy AMD just out of spite.

Hell, I'd advocate the government giving priority to AMD CPUs just to rebalance the competitive market. Intel screwed with AMD enough in the past that it's justifiable. Their APUs should be good enough for the legions of office drones in the bureaucratic ranks. Higher-core count desktop and server CPUs should work fine in the roles they are suited to.

Re:Flailing in failure.

[Bert64]

It doesn't matter, high price and high performance ends up in an ever shrinking niche - especially since processors have been powerful enough for 99% of users for many years already.

Intel never made anything that was performance competitive with Alpha, MIPS, POWER, HPPA or SPARC in their heydays, Intel were just much cheaper. ARM and AMD don't need to produce the best chips, just good enough chips that are cheaper.

Intel have lost the lowend to ARM, and the lowend is where the volume is.

Re:Flailing in failure.

[Gravis Zero]

AMD doesn't have anything available that competes at the high end performance wise, and hasn't for a very long time.

Umm... did you wake up from a coma recently?

As for low power, except at the extremely low end, intel chips require a lot less power than their AMD equivalent at the same performance level.

This isn't an AMD vs Intel argument, I'm talking about ARM. Intel has tried and failed to enter the cell phone market several times. ARM chips own the low power market and Intel doesn't have a single ARM chip.

Re:Flailing in failure.

[EvilSS]

No, it's not. High end is all about performance, damn the price.

Re:Flailing in failure.

[Anonymous Coward]

The high end is all about price/performance, and AMD is eating Intel's lunch at the moment.

The high end has never been about price. Citation?

Re:Flailing in failure.

[EvilSS]

AMD doesn't have anything available that competes at the high end performance wise, and hasn't for a very long time.

Umm... did you wake up from a coma recently?

As for low power, except at the extremely low end, intel chips require a lot less power than their AMD equivalent at the same performance level.

This isn't an AMD vs Intel argument, I'm talking about ARM. Intel has tried and failed to enter the cell phone market several times. ARM chips own the low power market and Intel doesn't have a single ARM chip.

AMD has nothing that can out-compete Intel on single thread performance, and even for multi-threaded performance ThreadRipper only beats Intel high end CPUs for extremely multi-threaded applications like rendering apps.

Re:Flailing in failure.

[KingMotley]

Umm... did you wake up from a coma recently?

No, have you not taken off your AMD rose colored glasses lately? Name your segment (IoT, laptop, desktop, or server) and the fastest performing AMD chip, and I'll respond with an Intel one that outperforms it. Sure, if you limit yourself based on price, then AMD competes, but we aren't talking about a performance/price comparison, it was is the "crown for performance" as YOU specified.

Re:Flailing in failure.

[unixisc]

Why doesn't Intel upgrade i5 to 4 cores and i7 to 6? And keep the hyperthreading to i7?

Re:Flailing in failure.

[Billly Gates]

Facepalm .... FYI AMD rehired the Alpha CPU and AthlonXP architect in 2014 and created a brand new architecture from scratch.

Spend 10 seconds and google "Ryzen CPU". Ryzen had embarrased Intel this spring and ran less watts than the i7 7700K. No you did not misread my last sentence.

Seriously watch the video if you are into computers?

Re:Flailing in failure.

[Billly Gates]

Keep in mind Intel OCed the i7 7700K waaay overboard out of fear of AMD. It gets up to 90C easily without liquid cooling at 5 ghz for crying out loud! Also the first reviews of Ryzen had ram at 2600 mhz and had bios issues.

The silicon used on Ryzen is more power efficient and therefore has a limit between 3.8 ghz and 4.2 ghz. Basically a Ryzen 1800X is an 8 core 16 thread i7 4790K Haswell. It is not that far behind and of courtse a Ryzen 1800x will cream an i7 7700k for multithreaded performance with no threadripper.

FYI I am typing this on an i7 4770K OC to 4.0 ghz and see no reason to upgrade as it is fast enough so I am no AMD fanboy or even use their CPUs at the moment. Sorry AMD Ryzen is most certainly a high end chip prosumer chip as far as I can see and even in games is at least in the same ballpark as Intel. Doom, Crysis 3, and Dues Ex run faster on AMD. True Grand theft auto runs better on Intel which every review and Battlefield 1 but overall. It is a good architecture I can see. If I were to buy a new CPU today I would be an idiot to spend $$$$ on a freaking 4 core cpu in 2017. My phone as more cores

Re:Flailing in failure.

[Anonymous Coward]

Are you too fucking stupid to read?! He told you TWICE that he's not even talking about AMD.
Not every word that starts with an A is AMD.

Re:Flailing in failure.

[Anonymous Coward]

Word is they're giving the i3 4 cores, although it will likely drop hyperthreading. It does seem odd that hyperthreading is being regarded as such a premium feature. This may of course not be market driven, it could be based on the binning process in fabrication.

Re:Flailing in failure.

[ChrisMaple]

Do an honest survey of multiple tests reported on internet sites. Ryzen's base clock is lower, ryzen has lower IPC, ryzen can't be overclocked as far whether on air, water, or LN2.

AMD has greatly improved performance, so much so that in very heavily multi-threaded applications where AMD's core count advantage can be used, AMD is faster on about half of the applications. Good for AMD, I like seeing the company successful. Nonetheless, the Ryzen line is not as good as Intel's products.

Re:Flailing in failure.

[Billly Gates]

Do an honest survey of multiple tests reported on internet sites. Ryzen's base clock is lower, ryzen has lower IPC, ryzen can't be overclocked as far whether on air, water, or LN2.

AMD has greatly improved performance, so much so that in very heavily multi-threaded applications where AMD's core count advantage can be used, AMD is faster on about half of the applications. Good for AMD, I like seeing the company successful. Nonetheless, the Ryzen line is not as good as Intel's products.

I disagree as not as good.

4 cores for an expensive CPU is not a good product in 2017. Again I say this as an Intel user if I were to buy today. Google threadripper and also google the i7 6800 series which perform about as bad in games and worse than the lower core i7 7700K for IPC per core. I don't see people saying the i7 6850 sucks comments or they are not good? Sure both Ryzen and a i7 6900/6800 can do the heavy workloads and game. Ryzen is half the cost of these exotic intel CPUs.

I guess it is the definition of good. I view them as different markets. For the low and high end AMD wins while the 7700K wins for people who just game. This also begs the question who just buys an i7 for gaming? It was meant as a workhorse. The i5 and Ryzen R5 is quite competitive. Gamers forget it is the GPU almost always rather than the CPU for the bottleneck.

My PC right now running Win10 pro has 1600 threads and 131 processes with just Chrome with 5 tabs, Outlook, several word documents and a single Excel file. I think the case for more cores is becoming more and more important for user experience as even my phone as more cores than the i7 7700K.

Re:Flailing in failure.

[KingMotley]

I may have not misread your last sentence, but it is also not true. The 1800X uses about 35%-40% more power than the 7700K. Go check any review where they actually measure the power draw.

Re:Flailing in failure.

[KingMotley]

I tend to agree with your general line of thinking, but we aren't there yet, and probably won't be for at least 3-5 years.

AMD with $1.1b in revenue vs intel with $14.8b in revenue in the most recent quarter.
AMD with $25m in profit, Intel with $3,500 in profit.

Volume doesn't really mean much if you are losing money on each and every one of them.

Re:Flailing in failure.

[Billly Gates]

I may have not misread your last sentence, but it is also not true. The 1800X uses about 35%-40% more power than the 7700K. Go check any review where they actually measure the power draw.

Video above. Wattage of of 7700K is slightly less.

As you can see at idle the i7 770K is 1.5 whole watt less than the Ryzen. When cranked up AMD wins in the AutoCad test in power consumption over Intel. In torture loop in the end Intel wins but AMD runs cooler.

I may add Tomshardware has been accused for 15 years now of being in the pockets of Nvidia and Intel and many would say biased in the link above.

They are pretty close as AMD measures power consumption differently. Some tests AMD is ahead. Go look at the crappy FX series? I am no AMD fanboy and the FX series and my underwhelming Phenom II that I had before it made me switch to Intel. AMD fired all the good architects last decade to cut costs and brought in H1b1 to do the job. Big mistake. FX bulldozer was terrible!

But anyway in the past decade it was ATI graphics that kept them afloat after screwing up their CPUs. Now we see the opposite and I hope they remain in business.

Re:Flailing in failure.

[Gussington]

Intel have lost the lowend to ARM, and the lowend is where the volume is.

I don't know the economics of CPUs, but as with Apple v Android, margin is more important than volume.

Re:Flailing in failure.

[Gravis Zero]

LOL! The AC above nailed it.

Re: Flailing in failure.

[Anonymous Coward]

Any measurement of power use is meaningless (except for peak cooling design) unless the amount of work is also measured. As an example, if the i7 completed the same amount of work in 75% of the time, then it spends 25% of the amd's processing time at the idle power rate. Such a test tends to bias against faster processors.

Re:Flailing in failure.

[Anonymous Coward]

Intel never made anything that was performance competitive with Alpha, MIPS, POWER, HPPA or SPARC in their heydays, Intel were just much cheaper.

This is probably true until about y2000 (NetBurst), but barely.

In the heydey of SPARC, the professor with a SPARC workstation on his desk would have a slower computer than the one next door with an x86 workstation running BSD/OS or Linux, though the one with the SPARC would probably be happier because of all the stupid problems with PCs back then. The low-end designs of these exotic CPUs were beautiful, especially HPPA and SPARC, but because of component count, not performance. They were similar to the original 68k all-in-one Mac. There was an all-in-one SPARCstation, too, but the IPX and the B712 which did take external monitors were both simple, feature-complete machines in the style of the original Jobs 68k Mac. For example B712 held the hard disk in place using clips and styrofoam. The CPUs in these machines were physically and internally very different from the ones in the servers, more so than any Intel CPU, more like the difference between Intel Atom and Intel's normal CPUs, or the difference between an Arm chip in a phone and an Arm microcontroller. Intel gave you their best game on the desktop, while all these vendors kept one hand tied behind their back on the desktop (albeit at considerable savings in watts).

The faster computers based on these old exotic chips relied on:
  - many dies, usually about four to sixteen, running a single kernel
  - huge L2 caches

to deliver the edge over Intel. The single-core performance modulo cache, even with the other hand untied from behind their back, wasn't that much different, but x86 chips had tiny L2 cache to keep the cost down and no multiprocessor features, eventually graduating to very limited multiprocessor features (two to four dies).

The Intel instruction set has always been terrible, but the exotic CPUs were hog-tied by the need to use a special compiler to get reasonable performance out of them. Each vendor charged >$1k per "seat" for HP ACC or SunPro, and programmers hated the compilers because many gcc features were missing. On Intel, gcc was about as fast as anything else. On the exotic RISC architectures, it could be as bad as half speed.

In particular, from the early days MIPS required fancy instruction scheduling to get good performance (or even correctness---MIPS = "multiple instruction pipelines without interlocks," and the S stands for "without interlocks"), and gcc was barely able. Alpha implemented division in software, and Digital's patented/closed algorithm was twice as fast as the one in GNU libm. Partly they were fucked because like Hillary they could not get out of their own way, contribute to gcc, share their basic math library, to promote their own CPUs. The stupid neckties were too obsessed with charging people both on the way in and the way out of the funhouse.

I hate Intel, but the big Unix vendors were kinda dicks, too, and I kinda miss them but kinda don't.

Re:Flailing in failure.

[unixisc]

So if hyperthreading on an i7 doesn't work, is it binned down to an i5?

Re:Flailing in failure.

[EvilSS]

I never said it didn't have a place in high-end PC builds, it certainly does. But the GP's assertion that "Intel has lost the crown for performance" is absurd. That (and his attempted defense of it) is what I replied to.

Not true

[Anonymous Coward]

It's next to impossible to make an application that doesn't call some OS function that casts off a thread these days.

In the Android world, 50+ threads are quite common on bigger apps, even on Windows, you'll get 5-6 threads per application running.

Here's the thing, these Intel processors are shit. On real world they run slower than ARM per core, and 8 cores are the norm now on smartphones and tablets, yet Intel is still trying to push 2-4 core chips whose 'speed' is a particular optimization for a particular benchmark application.

Intel need to up their game drastically here.

They lost their dominance in volume a long time ago, they lost the lead in $$$ recently, they lost the performance lead to AMD, at some point they have to stop polishing these turds and move forward.

Re: Not true

[Anonymous Coward]

Wtf are you talking about, most Windows APIs don't spawn threads. Windows has a thread pool you can use if you want to but low level function don't.

Re:Not true

[toddestan]

What are you talking about? In the "real world", an Intel (or AMD) x86 core stomps all over ARM. An i7 absolutely crushes those 8 core ARM chips in your phone. The fastest ARM chips perform on the level of a Pentium 4 from 15 years ago, and most of them are more like a Pentium III.

Of course, the ARM chips also sip power which is why they are used in tablets and phones. So really different chips for different purposes.

In Windows, here's direct control of it... apk

[Anonymous Coward]

See subject: Increase cpu core count @ hardware level (OS can use it for starters ala this in Windows for example) using regedit.exe:

Windows Registry Editor Version 5.00

[HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Session Manager\Executive]
"AdditionalCriticalWorkerThreads"=dword:00000008
"AdditionalDelayedWorkerThreads"=dword:00000008

* I.E. - How much extra cores will help BEYOND today's CPUs for the OPERATING SYSTEM itself (in Critical Worker Threads) in juggling threads in itself & for other processes (in Delayed Worker Threads) per https://msdn.microsoft.com/en-us/library/cc615012(v=bts.10).aspx/

Here I use 8 for an Intel Core I7 as shown above (both in 1st a 920 & currently a 4790k, since they're quad core (& hyperthreaded) & it was lesser based on physical core count of earlier systems I had (this setting has been around since, iirc, Win2k (correct me IF I am off/wrong - it's been SO long since then)...

(Those are settings in WINDOWS you can adjust to take advantage of added cores as you upgrade to CPUs w/ more cores, for example).

ANYTHING/EVERYTHING, in theory, gains there alone (less "process scheduler thrashing" in other words) - I don't care so much about applications/programs (they are probably written to their practical limits anyhow as to what threadwork will gain them) but again, MORE about how the OS will utilize them (per the 2 TUNABLE PARAMETERS in the .reg file I note above as a way to REALLY use the extra cores, almost guaranteed - Windows allows it, not sure of other OS like *NIX based ones).

APK

P.S.=> The rest will be done @ compiler level (already good, only depends on HOW you can leverage it OR if internal-to-program itself datasets AND PROCESSES (imo, a Gannt chart illustrates this well) allow for it - not all do) & it's always that way, pretty much - hardware 1st, software catches up (& it does, mostly inefficiently @ 1st, sucking up the CPU cycles/efficiencies gained)... apk

Re: In Windows, here's direct control of it... apk

[Brockmire]

That isn't increasing core count, you fucking moron. It's just a tunable setting for how aggressive to set something. Set it too high or too low is worse than it auto tuning. Also, that's recommendations for running a specific business app on a server OS. I must have missed where the topic is xeons and not desktop CPUs.

Still waiting for a fix Intel

[Anonymous Coward]

I still dont know if its safe to turn on Hyperthreading on my Kaby Lake processor. Do I get a motherboard bios update or something from Intel to install or what?

https://m.slashdot.org/story/327973

with more pci-e lanes at ALL Levels

[Joe_Dragon]

with more pci-e lanes at ALL Levels.

intel needs more pci-e lanes with no cut down

[Joe_Dragon]

intel needs more pci-e lanes with no cut down cpus in an socket with lanes disabled

Moore's Law over, according to Slashdot

[rbrander]

"Obviously, nothing is free" [so speed had to go down to pay for other improvements].

I assure you, sir, that in the past, we got stuff for free all the time. Basically, every new generation had more complex circuitry (oh, man, that jump from 16-bit to 32!) with more instructions and a drop in cycles/operation, **A*N*D** the chip ran faster.

The curve has been bending for some time, of course - I read these things because I got a high-end i7 in 2013 that had come out in 2012, (i7-3930K CPU @ 3.20GHz) and I'm still not sure if I would *notice* the speedup if I bought a 2017 system to replace it. A gamer friend tells me I will notice, but only if I get the latest thing in RAM and the latest thing in SSD disks, each on the latest thing in buses. All of that together will not double the performance of my early-2013 purchase, and "double" used to be every couple of years.

But, anyway, it was that "of course" that got to me. It means that the psychology has changed; the lack of automatic silicon progress has been accepted at a deep level, and people are planning around an era of Limits To Growth. "Moores Law" as a *social* era, has ended. We no longer expect next year's progress to solve this year's problems. We'll have to make some Hard Choices, give up something, to solve a resource lack.

Re:Moore's Law over, according to Slashdot

[Thanatiel]

You will not notice anything using a recent CPU, the gain is insubstantial.
As you said : the RAM, the graphical card and the disks are what matters most.

Re:Moore's Law over, according to Slashdot

[BlackSupra]

with more instructions

This is the truth! Instructions like AVE: https://en.wikipedia.org/wiki/...

Re:Moore's Law over, according to Slashdot

[rbrander]

Yow, those ARE cool! And in my 5-year-old CPU.

Oh, yeah, my rant skipped another thing. When it was still debatable that Moore's was over, one counterargument was that maybe GHz had stopped climbing, but we were getting more and more cores! I see the new line also quits at 6 cores, maybe you can get 8 later at high prices. Exponential increase in core-count has also stopped dead. (Just as well; Amdahl's Law says it wasn't a really productive direction, going to 16 and higher...)

Re:Moore's Law over, according to Slashdot

[Gussington]

...and I'm still not sure if I would *notice* the speedup if I bought a 2017 system to replace it. A gamer friend tells me I will notice, but only if...

I still have my i7-3770 (3.4GHz) in my gaming PC because CPU is not a major factor. SSD and GPU makes most of the difference these days.
A new CPU also means a new motherboard and new RAM. An expensive upgrade for probably 10% difference...

Eh...

[XSportSeeker]

I honestly would've liked to see Intel keeping Atom development going to see next generations to Cherry Trail... low powered small PCs seemed to have a good future there if only Atom kept going for some more years.

Re:Eh...

[AbRASiON]

New Atom chips just came out.
https://ark.intel.com/products...

The 4 core and weaker ones sound like what you're after. I suspect we'll see some very competent little HTPC boxes soon.

Becoming a better communicator

[Futurepower(R)]

APK, It seems that you have something important to say, but I don't understand most of it. I suggest you read Confessions of an Advertising Man by David Ogilvy. That book was very helpful to me.

Re:Mr. Pot calling a kettle black hypocrite

[Anonymous Coward]

See subject: SQLite had a bug in it for a decade or more https://it.slashdot.org/story/17/05/13/0113255/google-found-over-1000-bugs-in-47-open-source-projects/ & it's written in C so buffer overflows might still be present in it too.

Regarding threadwork in my program?

It's done how it is for "full bore speed ahead" in it's longest routine (which I improved by 66% w/ the test model I have here I'll release soon) - putting it onto a backgrounded thread would SLOW it & VCL it's done on ISN'T GUARANTEED THREADSAFE as well.

Thus, I won't do that.

HOWEVER - It does excellent concurrent ops on threads for reverse DNS of favorite sites (up to 5 extra when needed threads) though.

As far as the registry settings I posted are on topic and work for added cores!

* By the way/again: You just got blown away so get on topic as was suggested instead of being a butthurt pest you are.

APK

P.S.=> I also wrote functionality in my program BY HAND that does what SQLite would've for it, so I have control over it & could patch immediately - I couldn't if I used SQLite & what I wrote does what SQLite does for this program too (so I don't need it OR BUGS IN IT EITHER) & that code of mine does what roughly 14++ *NIX commands do!

How about you? Zero is what you are YOU UNIDENTIFIAIBLE do-nothing "ne'er-do-well"... apk

I don't see you do better talker... apk

[Anonymous Coward]

Threadwork in my program is done how it is for "full bore speed ahead" in it's longest routine (which I improved by 66% w/ the test model I have here I'll release soon) - putting it onto a backgrounded thread would SLOW it & VCL it's done on ISN'T GUARANTEED THREADSAFE as well.

Thus, I won't do that.

HOWEVER - It does excellent concurrent ops on threads for reverse DNS of favorite sites (up to 5 extra when needed threads) though.

* By the way/again: You just got blown away so get on topic as was suggested instead of being a butthurt pest you are.

APK

P.S.=> I also wrote functionality in my program BY HAND that does what SQLite would've for it, so I have control over it & could patch immediately - I couldn't if I used SQLite & what I wrote does what SQLite does for this program too (so I don't need it OR BUGS IN IT EITHER) & that code of mine does what roughly 14++ *NIX commands do!

How about you? Zero is what you are YOU UNIDENTIFIAIBLE do-nothing "ne'er-do-well"... apk

Opinions vary (same post modded up before)

[Anonymous Coward]

See subject: Others disagree as the very nearly same post almost verbatim is upmodded here on /. before https://hardware.slashdot.org/comments.pl?sid=10960603&cid=54957501/

APK

P.S.=> You might want to get out your "HOOKED ON PHONICS" remedial reading lessons... apk

Additionally, /.ers quoted disagreeing w/ you

[Anonymous Coward]

I'm going to continue using the Host File Engine. Your software is well written, functional. The Host File Engine performs exactly as promised by mmell

his hosts program is actually pretty good by xenotransplant

his hosts tool is actually useful for those cases in which one does indeed want to locally block stuff outright while consuming minimum system resources by alexgieg

(APK's) work, I've flat out said it's good by BronsCon

I've tried his hosts file generating software. It works by bmo

APK your posts on this & the hosts file posts, and more, have never been in error &/or bad advice by BlueStrat

Your premise that hostfiles are a good way to deal with advertising & malvertising is quite valid by JazzLad

I like your host file system by Karmashock

* It's recommended/hosted by Malwarebytes' hpHosts!

APK

P.S.=> See subject & quoted /.ers above - now, show us you've done better earlier than I - fact: YOU CAN'T & you know it... apk

"QUAGMIRE" Lol... apk

[Anonymous Coward]

See subject dumbass - it's increasing how you take advantage of cores counted for BOTH the OS & the applications it juggles!

It was a PLEASURE blowing YOU away easily on threadwork my program does https://it.slashdot.org/comments.pl?sid=10606043&cid=54414317/ & also on SQLite's bugs https://slashdot.org/comments.pl?sid=10606043&cid=54414369/

(Given those facts in those links? YOU are the UNIDENTIFIABLE WORM that's been stalking me here & YET AGAIN YOU kept up on those errors of yours I have just CRUSHED YOU ON YET AGAIN - seeing as how you forgot to post anonymously this time, lol!)

APK

P.S.=> You're "the expert" here, QUAGMIRE? Compared to me & things I've done in the art & science of computing, you're far from even an AMATEUR... apk