The Death of Overclocking is Nigh...(we knew it was coming)

Discussion in 'HardForum Tech News' started by Ranger101, Jul 25, 2019.

  1. Ranger101

    Ranger101 [H]Lite

    Messages:
    101
    Joined:
    Sep 11, 2015
  2. RPGWiZaRD

    RPGWiZaRD [H]ard|Gawd

    Messages:
    1,035
    Joined:
    Jan 24, 2009
    Correct me if I'm wrong as I haven't followed the Ryzen launch that so closely but is it even possible you lose performance by manual overclocking since you don't get the whatever high single core boost that it's applying by default in doing so, which leads to some games or whatever applications that don't utilize multiple cpu cores very well benefit more by that couple extra 100MHz by the higher single core boost. Say 4.6GHz single instead of 4.3GHz all core or whatever.
     
    Staples and Ranger101 like this.
  3. Ranger101

    Ranger101 [H]Lite

    Messages:
    101
    Joined:
    Sep 11, 2015
    Yes from what I have read you are 100% on the money, performance can actually regress in apps that don't utilize all cores/threads e.g. games if you manually overclock all cores/threads to say 4.3GHz as opposed to leaving things as they are and having only a couple of cores running at say 4.6Ghz.
     
    Revdarian and Staples like this.
  4. M76

    M76 [H]ardForum Junkie

    Messages:
    9,312
    Joined:
    Jun 12, 2012
    Actually according to gamers nexus's tests cooling is more important than ever, because the Ryzen 3000 series will be faster out of the box without actual overclocking if you lower the temps. It's not just slap some crap on it and it will do stock clocks all the same.
     
  5. Ranger101

    Ranger101 [H]Lite

    Messages:
    101
    Joined:
    Sep 11, 2015
    I get what you are saying, I'm just not too sure that gaining an extra 50-100Mhz or so here and there increases
    performance to a point where it's worth spending the extra $$ on an expensive cooler over the more than
    adequate cooler that is boxed with the processor. Diminishing returns is definitely kicking in so for many
    people, myself included, better to save the money for other upgrades.
     
  6. M76

    M76 [H]ardForum Junkie

    Messages:
    9,312
    Joined:
    Jun 12, 2012
    Instead of more important than ever I should've said for the first time it is important for non-overclockers as well.
     
  7. Ranger101

    Ranger101 [H]Lite

    Messages:
    101
    Joined:
    Sep 11, 2015
    That is a good point.
     
  8. Azrak

    Azrak Gawd

    Messages:
    826
    Joined:
    Oct 4, 2015
    I would argue that non-OCers were never worried about losing 50-100MHz anyway and wouldn't want to spend extra on a better cooler when an adequate one is included the the CPU box.
     
  9. Shikami

    Shikami Gawd

    Messages:
    640
    Joined:
    Apr 5, 2010
  10. Nicepants42

    Nicepants42 Gawd

    Messages:
    768
    Joined:
    Aug 30, 2004
    While I agree with your general sentiment, as someone who is currently running a 3600x on air on the stock cooler, the noise is a huge issue for me. I have enough experience with aftermarket coolers to know that an inexpensive HSF with a copper base, heatpipes and a 120mm fan would be a necessity for me, were I to continue air cooling. The extra single core speed is gravy. Hyper 212's go on sale for $20-$25 all the time.

    Agree with this. Another thing that's been highlighted for me is that the ability to control how the CPU fan behaves is very important. Because of how quickly and frequently the CPU throttles up/down, the stock fan is changing rpm and pitch constantly, which greatly exacerbates the noise issue. Joe Schmoe is gonna want to edit a couple BIOS settings to mitigate this - assuming he knows that's an option.
     
    DanNeely and IdiotInCharge like this.
  11. M76

    M76 [H]ardForum Junkie

    Messages:
    9,312
    Joined:
    Jun 12, 2012
    I believe it's not that they don't want that extra 50-100Mhz, it's that they are afraid to touch OC due to the lack of knowledge and the fact that every manufacturer says it voids the warranty. Getting a better cooler for $20-30 does not require any tweaking and trial / error, and does not void the warranty even in theory.
     
    Azrak, blackmomba and IdiotInCharge like this.
  12. Armenius

    Armenius I Drive Myself to the [H]ospital

    Messages:
    18,136
    Joined:
    Jan 28, 2014
    The Death of Overclocking is Nigh on AMD Processors

    FTFY

    Even better, instead of editorializing the title like some clickbait outlet, you could just copy and paste the title of the article:

    Silicon Lottery Launches Binned AMD Ryzen 3000 Series CPUs

    And then add your comment in the first post that you think this is "the end of overclocking."
     
  13. SighTurtle

    SighTurtle [H]ard|Gawd

    Messages:
    1,412
    Joined:
    Jul 29, 2016
    https://www.gamersnexus.net/guides/3491-explaining-precision-boost-overdrive-benchmarks-auto-oc

    So apparently according to GN, using better cooling, i.e lowering temps and allowing the CPU to do all the work from stock, so no modification of any settings whatsoever, and it will auto-clock itself. Sounds pretty good from the standpoint of a consumer who is terrified of stepping into the BIOS. Pretty sure there are more 9900k owners that aren't overclocking their CPUs and therefore are missing out on that wonderful all-core overclock.

    -GN

    I'm confused that article came out in 2015, yet people still need to manually overclock their chips to get the best performance on Intel. If Speed Shift was similar to AMD's Precision Boost, why would people be always seeing better performance than stock on Intel vs a hit and miss with AMD?
     
    Revdarian and kamikazi like this.
  14. SighTurtle

    SighTurtle [H]ard|Gawd

    Messages:
    1,412
    Joined:
    Jul 29, 2016
    Tbf Silicon Lottery basically wrote they were seeing their own death in terms of doing overclocking on chips manually, which you would know if you read the article. With that said, Ranger is incorrect, you still need better cooling, you don't need to fiddle with settings anymore as that day is coming to a end.
     
  15. Armenius

    Armenius I Drive Myself to the [H]ospital

    Messages:
    18,136
    Joined:
    Jan 28, 2014
    I read the article. Just because the business model is no longer sustainable for Silicon Lottery it does not signal "The Death of Overclocking."
     
  16. Shikami

    Shikami Gawd

    Messages:
    640
    Joined:
    Apr 5, 2010
    Overclocking for best performance is relative. Performance is always relative. So, when you say best performance, if overclocking for best performance is to get 5 more seconds or 6 frames from some bench metric that doesn't mean much IMO. Especially, if it costs more, is a lottery, is noisy, crashes, and whatever else. Here is 95 TDP RE-review: https://www.techspot.com/review/1744-core-i9-9900k-round-two/. Its just not that much off in this example, and others will argue the opposite-but there is definitely a wall coming up. Paying for better silicon or modifying (lap) so as to just have a few more clock pulses past a particular bin is not worth it when the silicon is coming to a wall and can automatically do it close to the wall stable. Note, to mention one more time, this is my opinion and I get that.

    CPPC is this (https://patchwork.kernel.org/patch/7316631/):

    CPPC stands for Collaborative Processor Performance Controls
    and is defined in the ACPI v5.0+ spec. It describes CPU
    performance controls on an abstract and continuous scale
    allowing the platform (e.g. remote power processor) to flexibly
    optimize CPU performance with its knowledge of power budgets
    and other architecture specific knowledge.

    This is integral to the modern processors now. Skylake arch and up has had this with Speed Shift but many motherboards did not expose this ability-some did. Now Ryzen 2 supports CPPC

    This is from Anand's Ryzen 2 review:

    Clock Ramping
    For any of our users familiar with our Skylake microarchitecture deep dive, you may remember that Intel introduced a new feature called Speed Shift that enabled the processor to adjust between different P-states more freely, as well as ramping from idle to load very quickly – from 100 ms to 40ms in the first version in Skylake, then down to 15 ms with Kaby Lake. It did this by handing P-state control back from the OS to the processor, which reacted based on instruction throughput and request. With Zen 2, AMD is now enabling the same feature.

    David_McAfee-Next_Horizon_Gaming-3rd_Gen_Ryzen_06092019-page-008%20-%20Copy%20%282%29_575px.jpg

    AMD already has sufficiently more granularity in its frequency adjustments over Intel, allowing for 25 MHz differences rather than 100 MHz differences, however enabling a faster ramp-to-load frequency jump is going to help AMD when it comes to very burst-driven workloads, such as WebXPRT (Intel’s favorite for this sort of demonstration). According to AMD, the way that this has been implemented with Zen 2 will require BIOS updates as well as moving to the Windows May 10th update, but it will reduce frequency ramping from ~30 milliseconds on Zen to ~1-2 milliseconds on Zen 2. It should be noted that this is much faster than the numbers Intel tends to provide.

    The technical name for AMD’s implementation involves CPPC2, or Collaborative Power Performance Control 2, and AMD’s metrics state that this can increase burst workloads and also application loading. AMD cites a +6% performance gain in application launch times using PCMark10’s app launch sub-test.
     
    Last edited: Jul 25, 2019
    Revdarian, GoodBoy, Mav451 and 2 others like this.
  17. Not the end of OCing at all. AMD has for a while now pushed it's silicon to the max leaving little on the table for OCing. Also, the gains people are seeing from OCing is all core, but people are comparing them to single core stock boost clocks and saying you only gained 50mhz, when you gained far more than that, as it's across all cores. And when talking about stock, under an all core load, most of the reviews show the clocks taking a pretty big hit, as the stock cooler cant keep up and it throttles back, meaning aftermarket cooling is even bigger of a deal for even non-OCers.

    So everything is the same, if you are someone just buying an appliance, you buy it and don't think about it and get whatever performance it gives you. If you are looking for that extra bit, then you will be investing into a quality cooler and testing OCs.
     
    Baenwort and IdiotInCharge like this.
  18. MacLeod

    MacLeod [H]ardness Supreme

    Messages:
    7,601
    Joined:
    Jul 28, 2009
    I was kind of thinking the same thing when I first read Zen2 reviews. I stopped overclocking my GPU's a few years ago because it wasn't worth the extra effort or heat for 5% performance gains. My current 1600x is overclocked to 4 Ghz and is working well but if the new chips boost to 4.6 but are usually only stable to 4.3 on manual overclocks over all cores, I will most likely leave everything set at default settings when I upgrade to a 3600x in the next few weeks.

    As much fun as I've had overclocking over the last 10 years or so, I'm ashamed to admit the thought of buying a quality component, dropping it in and letting it do the work does kind of appeal to me but then maybe I'm just getting lazy in my old age.
     
  19. Dan_D

    Dan_D [H]ard as it Gets

    Messages:
    54,449
    Joined:
    Feb 9, 2002
    That's exactly what happens.
     
    AceGoober and DrLobotomy like this.
  20. fullvietFX

    fullvietFX [H]ard|Gawd

    Messages:
    1,786
    Joined:
    Sep 1, 2004
    I used to love overclocking when I was younger. I'm too lazy now though and just want things to work. However I'm still going to buy higher-end mobos and coolers just because.
     
  21. KingerXI

    KingerXI Gawd

    Messages:
    700
    Joined:
    Mar 31, 2009
    All I had to do was click on a big red button in the BIOS of my MSI Z270 board, and it changed all the settings I needed and put the chip at 4.8 GHz vs. the stock 4.2 GHz. It has worked flawlessly, never a hiccup, and runs nice and cool with the Hyper 212 cooler. Why not do it when it is that easy?
     
  22. Bowman15

    Bowman15 [H]ard|Gawd

    Messages:
    1,238
    Joined:
    Apr 7, 2015
    Did you ever check the voltages after you did that? I bet the mobo is supplying some outlandish voltages....:smuggrin:

    Not that there is anything wrong if it's nice and stable for as long as you need it.
     
  23. KonaKona040

    KonaKona040 n00b

    Messages:
    19
    Joined:
    Feb 6, 2019
    Maybe not on it's own, but it's obvious that BOTH AMD and Intel aren't able to just leave clock speed on the table like they have in the past. As soon as AMD started being competitive Intel went from selling chips with 500-1000mhz of headroom to selling chips with more like 100-200mhz of headroom. I wouldn't be surprised at all if Intel sorts out their own PBO like feature that makes manual overclocking not really worth the effort for most workloads.

    Yeah sure, you'll still be able to manually overclock and/or undervolt if you want. Intel would still love to charge a premium for those extra few hundred megahertz I'm sure. But I don't see a problem with new chips being able to hold themselves at their own limits. It's not like overclocking still has the technical or romantic appeal of back when you could buy chips and run them at twice their rated clock speed, at best it's eeking out maybe low double digit performance gains at the expense and cost of more power and heat. Overclocking won't "die", but it also doesn't mean the same thing it did in the past.

    Better off benchmarking old hardware if you want overclocking to be "fun".
     
    Revdarian and wolfofone like this.
  24. Ranger101

    Ranger101 [H]Lite

    Messages:
    101
    Joined:
    Sep 11, 2015
    This was my first thread so apologies for any offence caused by not following standard protocols or the manner in which
    I phrased the title.

    I own a 4790K myself, it is overclocked and I have derived a lot of pleasure from overclocking myself going back
    to the days of the early Athlon Cpus.

    That being said, from what I have read, as silicon technology is being pushed to it's limits, companies like AMD and Intel can ill afford to
    leave "free" performance via overclocking on the table, even less so when competition between the two is fierce.

    Consequently overclocking headroom is decreasing on both AMD and Intel processors - sad but I think it's true.

    Thank you for the fix and the advice.
     
  25. Lumpus

    Lumpus Limp Gawd

    Messages:
    259
    Joined:
    Sep 2, 2005
    This... oh so much!
    /going to heavy water cool my next build (mostly) just for the thermals
     
  26. ChadD

    ChadD [H]ardness Supreme

    Messages:
    4,130
    Joined:
    Feb 8, 2016
    If your choice is... spend 200-300 bucks on CPU + 100 bucks on waterblock Vs say spend 400 bucks on bigger badder CPU with better stock cooler. I think the choice is clear that a cheaper CPU+expensive cooling is mostly fail.

    Your right of course on the high end. Ya if you have no constraints one your budget and you already have a sweet GPU, 32gb or more of fast ram, a top end SSD... then ya get a sweet CPU 3900, 3950 or whatever else comes along in that range. Fit it with a kick ass water cooler and enjoy nice long sustained boost speeds. (even if 99% of the time the CPU won't be doing anything that requires it to boost any longer then the stock cooler could probably do anyway) Of course if you are not rocking a highend samsung/corsair/wd M2... a 5700/2070 or better GPU.... 16gb+ of 3200-4000 mhz ram... and or your still on a 75hz 1080p monitor. Perhaps spending a ton of your budget on a high end cooling setup is not the greatest return on investment. (gone are the days where spending a hundred bucks on a high end cooler would save you double that on the CPU).

    I'm torn on AMDs boosting tech... on one hand it's insane to get out of the box high end overclocking performance on demand... on the other hand all of the joy of squeezing every drop out of the cheap stuff has been tampered down. lol

    PS... having said all that the pricing on lowerend basic water block systems has gotten pretty damn cheap... so cheap it seems to me at least that high end fans seem a waste of money. For 50-70 bucks these days you can get a pretty darn decent AIO CPU block. Can always argue noise levels and make yourself feel good about your purchase. ;)
     
    Last edited: Jul 25, 2019
  27. KingerXI

    KingerXI Gawd

    Messages:
    700
    Joined:
    Mar 31, 2009
    The voltages are nice and low.
     
  28. Kardonxt

    Kardonxt 2[H]4U

    Messages:
    2,924
    Joined:
    Apr 13, 2009
    I liked the excitement of getting top of the line performance out of a low \ midrange parts.

    Busting your balls to get an extra 200 mhz for nearly no tangible benefits just isn't exciting to me anymore. I'm impressed with the patience reviewers like Dan D. have in recent reviews, trying every which way to squeeze water from a rock. I always thought it would be fun to be a reviewer, but I don't have that kind of patience lol.

    I was hoping the 3800x would have some impressive OC headroom without the power constraints of the 3700x and the extra cores of the 3900x. Unfortunately that doesn't seem to be the case.
     
  29. Nicepants42

    Nicepants42 Gawd

    Messages:
    768
    Joined:
    Aug 30, 2004
    Agree with general sentiment but there's zero reason for stock coolers when good ones are not expensive.

    If your HSF budget is $50-$70, high end air will always be preferable in my book, simply due to number of cut corners involved in $70 water loops, and the potential for reliability and maintenance problems. If we're talking bang-for buck value, putting the GPU (only) under water is probably going to yield more benefit than water cooling the CPU. But when GPU water kits cost at least as much as the price gap to the next tier....
     
    ChadD likes this.
  30. Grimlaking

    Grimlaking 2[H]4U

    Messages:
    2,857
    Joined:
    May 9, 2006
    I don't think it is the death of overclocking. We will still find memory tweaks and settings to eek out the best performance. Ways to make our video cards better and faster. (Lower models with bio flashes to make them operate as newer models perhaps.) And many others. Just over clocking at the CPU is becoming a net neutral thing and your gains will not be that great.
     
  31. Dan_D

    Dan_D [H]ard as it Gets

    Messages:
    54,449
    Joined:
    Feb 9, 2002
    Everyone hoped for more overclocking headroom. The reality is, we weren't going to get that. What we got is pretty much exactly what I expected. I expected a small clock speed improvement over Zen + and that's what we got. All of the previous Zen chips hit the same basic clock speed wall. From what I've seen so far, the absolute wall is about 200MHz higher than it was with Zen or Zen+. Even then, you can only do that on some CCX complexes. On all cores, you are limited to the same basic 4.3GHz on all the parts with rare exceptions. That's not much over the previous generation.

    As for being a reviewer, it can be fun if you have the right disposition for it. I have a passion for hardware which my career in IT doesn't really let me leverage too much. You also have to have a bit of scientist in you in wanting to test, collect data and share it. There is also a masochistic streak required as you give up sleep and free time to do the work. A lot of people think it will be all fun and games with free hardware raining down. They hear game testing and think they'll get paid to play games all day. The second part is true, but most people quit by their second or third article. I watched probably a dozen or more reviewers come and go over the years at HardOCP. They realize its a ton of work and the low pay and promise of free hardware isn't even enough to retain them. As an example: There is a good 50 hours in that Ryzen 9 3900X review. My upcoming article on the MSI MEG X570 GODLIKE, will be the longest review I've ever done on a board. It's without a doubt the most tested motherboard I've done. I test the crap out of them, but I've used it for a number of articles and data gathering. I'm going to continue to do so for a while too. I've tested two processors on it and beat on it with different RAM, BIOS revisions, etc.

    Another thing you don't see too much of in the reviews are the problems we encounter. I started my comparison testing on an ASUS Maximus XI Formula. I had a hose split at the fitting and it spewed non-conductive coolant all over the video card, in the CPU socket and all over in general. So I ended up having to replace that board, hoses on the cooler and all the coolant. A bigger mistake there was that I had used some of that garbage Thermaltake coolant I bought locally that's kind of silvery. It was supposed to be a bright orange color, but in those old hoses it was a gross flesh color. Anyway, while being non-conductive, whatever was in it created this crusty film that I couldn't completely clean off. It got into the memory slots and that board is basically bricked. At best I can get a single DIMM slot working.

    Then there is the fact that I heard about the scheduler changes, so all the AMD stuff had to be retested. Of course since Microsoft released a new build, I had to keep it apples to apples, so the Intel system got retested. I actually planned to have more in that article but cut stuff out for time due to all the issues I had. Oh, and I had a PSU that randomly shut down, the Maximus XI APEX I got from ASUS for review that I ended up testing on had to be replaced due to some weird issue I won't get into.

    So being a reviewer can be awesome, but it's a crap ton of work. On the upside, if I want to know something, I generally have the extra contacts, or the hardware to find out. Free hardware doesn't hurt either, but allot of times, I don't get to use it personally until after I'm done with it on the test bench.
     
  32. Hagrid

    Hagrid [H]ardForum Junkie

    Messages:
    8,389
    Joined:
    Nov 23, 2006
    Even if I am running stock, I still put a much better cooler on than needed. Noise and no worrying that it's getting close or not.
    OC is more done by the enthusiasts which is quite a few people here. Now in games with a 6-8 core are you going to see huge benefits? Maybe at 1080p.
    If you do work with your system, then it might help a lot. Very subjective on your needs.
     
  33. Dan_D

    Dan_D [H]ard as it Gets

    Messages:
    54,449
    Joined:
    Feb 9, 2002
    System tuning won't ever stop. The problem is that overclocking as we know it will die off or get minimized because Intel and AMD are running their silicon pretty close to the edge of its capabilities. It will take a major breakthrough in chip design to allow us to see clock scaling again like we've seen previously.
     
    Revdarian, DustySilicon and Chimpee like this.
  34. Grimlaking

    Grimlaking 2[H]4U

    Messages:
    2,857
    Joined:
    May 9, 2006
    We're a long way away from a new Pentium Celeron candybar overclocking era! :)
     
  35. M76

    M76 [H]ardForum Junkie

    Messages:
    9,312
    Joined:
    Jun 12, 2012
    You don't have to spend $100 on a cooler that's better than stock. You can get pretty decent tower coolers for $30-40. Stock vs custom waterloop are not the only options.
    Besides I've already ordered a 3900X I can't spend more on a CPU. Well, I could but it wouldn't be better at all. As 3900X is the best money can buy right now for multi threaded applications as it seems. But I don't plan to spend anything on a cooler I have leftovers.


    That happened long ago. I think the last CPU I enjoyed OC-ing was my Sandy Bridge HEDT I7. Everything since was totally underwhelming and not rewarding at all. Especially GPUs. The last one I could get a decent OC out of was the GTX660 SLI setup I had. The 290x was a furnace at stock. The 980TI was barely stable on stock clocks, and I can't even be bothered to OC my 1080s.
     
    Revdarian and ChadD like this.
  36. Mchart

    Mchart 2[H]4U

    Messages:
    3,361
    Joined:
    Aug 7, 2004
    The cooler that comes with the 8 core Zen2 parts is pretty darn good. You'd want more for the 3900x/3950x to achieve decent clocks, but the average user will be fine with the wraith prism.
     
    Revdarian and dragonstongue like this.
  37. dragonstongue

    dragonstongue 2[H]4U

    Messages:
    3,121
    Joined:
    Nov 18, 2008
    for a while now? I am pretty sure that for the most part up till Ryzen there was actually a pretty decent range of overclocking that was able to take place. I know my 955 can do 4.35+Ghz (just need to slap on my new Gammaxx 400 or update from the hyper 212+ one more notch to keep below 54c, I got a solid chip IMO
    Bulldozer etc were also "known" if not needed to overclock to rat snot to get the most out of them, am pretty sure mobile anything never counted for overclock/declockability

    ^.^
     
  38. lcpiper

    lcpiper [H]ardForum Junkie

    Messages:
    10,536
    Joined:
    Jul 16, 2008
    I'm not a Ryzen guy, but I was going to say that I do watercooling for the better cooling and quieter environment anyway. I haven't needed an overclock since the Northwood chip came out.
     
    IdiotInCharge likes this.
  39. nutzo

    nutzo [H]ardness Supreme

    Messages:
    7,379
    Joined:
    Feb 15, 2004
    As someone who has been overclocking for over 35 years, getting lazy in my old age is part of the problem. :p
    However, it's mainly the diminishing improvements when overclocking the current generation of chips.

    Just some of the improvements I used to see:
    286-6 to 286-8, 30% improvement
    486-25 @ 33mhz, 32%
    Celeron 300a from 300 MHz to 450 MHz, or a 50% improvement
    Athlon slot A from 700 MHz to 900 MHz or a 28% improvement
    P4 3.0 ghz to 3.6 Ghz, 20% improvement.

    With the latest chips, any improvement is minimal, and I'm now more concerned with quieter low speed fans.
    There's also the fact that CPU's have improved so much, it's rare when I actually push my slightly overclocked i7-7700K.
    I'm actually running my current CPU slightly slow than I've tested it at, because the added heat & noise isn't worth the extra 3% speed improvement under full load.
     
  40. viivo

    viivo [H]ard|Gawd

    Messages:
    1,245
    Joined:
    Sep 7, 2005
    There's nothing sensible about an $1100 GPU that's only marginally faster than GPUs costing 70% less.
     
    dgz likes this.