Ryzen 3000 series CPUs

Discussion in 'AMD x86 CPUs and chipsets' started by luke o, Dec 27, 2018.

  1. budzilla

    budzilla Member

    Joined:
    May 6, 2006
    Messages:
    2,774
    Location:
    Right There->
    That's what I am banking on. Hopefully the next gen Ryzen will still be AM4 compatible and I have a X570 mobo so I can see what is out then.
     
  2. mesaoz

    mesaoz Member

    Joined:
    Jan 15, 2015
    Messages:
    8,010
    Location:
    South East QLD
    Next one is supposed to be AM4 still, but probably the last one
     
  3. the_pessimist

    the_pessimist Member

    Joined:
    Sep 10, 2008
    Messages:
    3,063
    Location:
    Sydney
    I had a similar issue when i firstly ran the test. I'm now getting about 4.5ghz single core boost on stock air cooling. The unfortunate thing about that is that it doesn't stay there for long as a majority of time it was more like 4.475ghz. Other then the boost clocks performance wise its been awesome.

    Done, i feel like this type of submission should have you including a screenshot for verification.
     
  4. dr_deathy

    dr_deathy Member

    Joined:
    Apr 27, 2007
    Messages:
    2,859
    Not sure what you guys are doing wrong, i see 4.4GHz all the time and 4.6GHz on occasion. I rarely have just one thread tho and not sure why people who do are buy a 12core.

    Both of my X570 machines work as advertised, no issues at all except for a dumb fan curve, which mind you it worked fine i just didnt like it.

    Be good when i get her back to water, once i get a new block/ adapters. Then i might start pushing it.

    Turn power mode to ryzen performance if you want to stop stuffing around as well. Dont cry about volts like a little bitch tho.
     
  5. Daveros

    Daveros Member

    Joined:
    Nov 24, 2009
    Messages:
    85
    Maybe, and yeah, me too. Just seems odd that the only thing I did was swap from a 1700X to a 3900X and the header failed. Remained dead after a BIOS update too. But whatever, you get by with that stuff.

    I’ve seen 4.4 once, but I’m regularly boosting to 4.2-4.3 on single core and 3.9-4.1 on multi. This is with a H100i.

    What really intrigues me is the performance regression with PBO enabled. Occasionally it’ll run a little faster (I got Gooseberry done 4 seconds sooner) but usually I’ll see regression. Odd platform, I need more time with it to actually understand what I’m seeing.
     
  6. 335 GT

    335 GT Member

    Joined:
    Aug 3, 2019
    Messages:
    1,143
    Location:
    W.A.
    I'm happy.
    My 3700x is boosting to the advertised 4.4 on 3 cores when gaming.

    Gigabyte aorus elite wifi x570. Stock cooler.
     
  7. the_pessimist

    the_pessimist Member

    Joined:
    Sep 10, 2008
    Messages:
    3,063
    Location:
    Sydney
    Seems like you're one of the lucky few then because i certainly have never seen anything above 4.5ghz single core and multicore @ 4.1ghz (encoding in Davinci Resolve and Blender r15). This is on stock settings. If you're happy to wait a month + because excessive volts killed your CPU then that's your decision, id rather be able to use mine in the meantime.

    Also i tried ryzen performance and it made no difference.

    Seems like there are background Apps using you cpu. Try running the test with everything disabled and run blender single core and hwinfo only. You might see a jump.
     
    Last edited: Aug 22, 2019
  8. SnooP-WiggleS

    SnooP-WiggleS Member

    Joined:
    Sep 20, 2004
    Messages:
    3,243
    Location:
    Ballarat, Victoria
    On the topic of boost clocks hardwareunboxed have done a good investigation. With the exact same cpu (3800x), cooling, some motherabords will boost above 4.5ghz even, others only hit 4.4ghz. It's not even down to expensive boards vs cheap or brand either, some models are doing better than others.



    Edit: relevent graph
    [​IMG]
     
    Last edited: Aug 22, 2019
    adamsleath likes this.
  9. SKITZ0_000

    SKITZ0_000 Member

    Joined:
    Jul 19, 2003
    Messages:
    1,505
    Location:
    Tas
    If we're just going by HWInfo max mhz (I assume?) I get a clean 4600 on two cores and a mix of 4575/4550 on the rest after leaving it open and using pc for a bit. Good enough for me, it's kind of splitting hairs over 25-50mhz imo. I do understand why people are upset, but AMDs boost is not the same as Intels boost, you can't get blame AMD because it performs differently to what you're used to on team blue. The wording on the box could have been different I suppose, providing it isn't just the motherboards causing the issue (as Hardware Unboxed suggest).

    Haha amen

    It's because all PBO is doing is increasing your power window, you will still scale back clockspeed as the temperature rises, which it will do quicker with more juice.

    On my Alphacool Eisbear AIO, in Cinebench R20 I consistently get ~ 7100cb stock, 7350cb with PBO enabled (max setting), and 7725cb @ 4.35Ghz all core static clocks. Peak temp is ~ 70C.

    I had the same issue with my Zen+ with my air cooler (Dark Rock Pro 4) and PBO and in the end I just set a static clockspeed for the puny 8 core so I didn't have to ramp my fans so agressively trying to chase down the best PBO performance.

    I've also heard that Zen2 boost performs a little like GPU boost, so giving it a +mv offset might increase boost performance, but I'm yet to test this (and am skeptical).
     
    Last edited: Aug 22, 2019
  10. Hotrod2go

    Hotrod2go Member

    Joined:
    Jun 17, 2018
    Messages:
    552
    Location:
    Tas
    You guys with 3000 series, had any luck with manual all core OC's?
     
  11. budzilla

    budzilla Member

    Joined:
    May 6, 2006
    Messages:
    2,774
    Location:
    Right There->
    That's what I am banking on. I set up my PC last night. It swiched on but didn't have time to install windows or connect the monitor. I'm pretty excited about the weekend :D
     
    mesaoz likes this.
  12. SKITZ0_000

    SKITZ0_000 Member

    Joined:
    Jul 19, 2003
    Messages:
    1,505
    Location:
    Tas
    I tested this last night, at least for my board it makes no difference. +50mv and +100mv didn't change boost behavior, was margin of error difference in score (lower). I also tried -50mv and again, margin of error difference in score (again lower but a bit more this time).

    On the 3900X I really don't think it's worth it, unless you're mostly doing heavily threaded workloads and PBO is winding too much out of your clocks (perhaps on air).

    For me though, with PBO+AutoOC after my final daily memory tweaking I get 7400cb (image attached). That's with cores averaging 4150-4175 as reported in HWInfo during the run.

    The best attainable daily clocks I can achieve is somewhere around 4.25Ghz (~7550cb) (though not heavily stress tested, as it's not worth it) which is ~ 2% increase in all core performance, but at a loss of >5% performance in lightly threaded workloads.

    I can see that all core OC might be worth it on the 8 core chips, and definitely on 6 core chips like the 3600, but not so much for the 3900X as the disparity between all core and light load boost is so significant.

    Another thing to note is this:

    and this

    Something to be aware of if you want to be on the safe side long term, especially if you leave your PC on 24/7 and load it up for long periods of time like me.
     

    Attached Files:

    GoneFishin22 and Hotrod2go like this.
  13. mesaoz

    mesaoz Member

    Joined:
    Jan 15, 2015
    Messages:
    8,010
    Location:
    South East QLD
    I'm using an offset of -0.1 and it increased my automatic all core boost by almost 100mhz.. made no change (positive or negative) to single core boost though
     
  14. SKITZ0_000

    SKITZ0_000 Member

    Joined:
    Jul 19, 2003
    Messages:
    1,505
    Location:
    Tas
    On your 3900X? Interesting...

    -100mhz offset on my 2700X gave me some tidy improvements, but definitely not on the 3900X.

    This reddit post is interesting and lines up with the slight drop in performance I had.

    https://www.reddit.com/r/Amd/comments/cc19mm/how_voltage_affects_performance_on_a_3700x/

    Perhaps it's down to cpu binning and temps? I certainly don't have a golden sample.
     
  15. mesaoz

    mesaoz Member

    Joined:
    Jan 15, 2015
    Messages:
    8,010
    Location:
    South East QLD
    My board is pumping high volts into my 3900x for no good reason, that's why I have the offset in place. WITH the offset in place it's still feeding 1.36v. MSI plox release b450 update

    Load temps under water with precision boost turned on hit the mid 80's with default voltage control, and are a full 10c lower with the -0.1 offset applied. Shit's whack
     
    SKITZ0_000 likes this.
  16. SKITZ0_000

    SKITZ0_000 Member

    Joined:
    Jul 19, 2003
    Messages:
    1,505
    Location:
    Tas
    Yeah that is whack, mine runs ~ 1.33v all core load with PBO x10, peaks @ 70C. I do have my LLC set to 5 (the most relaxed) which would decrease the load voltage.
     
  17. Hotrod2go

    Hotrod2go Member

    Joined:
    Jun 17, 2018
    Messages:
    552
    Location:
    Tas
    Yeah, I've heard of that Stilt guy b4, he certainly knows his stuff. It's like what I was saying earlier in this thread.; the laws of physics with respect to the flow of electrons (aka electricity) are severely challenged the smaller they go with nm manufacturing of silicon for computers.
     
  18. mesaoz

    mesaoz Member

    Joined:
    Jan 15, 2015
    Messages:
    8,010
    Location:
    South East QLD
    MSI have a b450 bios update out, only patch note is memory compatibility and it's a beta release.
     
    Hotrod2go likes this.
  19. wintermute000

    wintermute000 Member

    Joined:
    Jan 23, 2011
    Messages:
    2,563
    Yeah it's a bit iffy. Even on my 3600 with a deepcool gammax ( cm212 class cooler but still annihilated wraith stealth) I idle at 35 but boost high 70s with 0.02 offset. Without the offset it's like 5 degrees worse. Funnily in BIOS it says it idles at 30 degrees.... 1.4 volts happening at will, and all of this clouded by confusion over observer effect / different measurements to traditional / broken tools

    Ultimately though seems plenty fast and no BSODs so not sure there's anything to complain about lol
     
    Last edited: Aug 24, 2019
    mesaoz likes this.
  20. Zenskas

    Zenskas Member

    Joined:
    Jan 27, 2010
    Messages:
    1,886
    Location:
    Tas
    That is where I am sitting on the matter right now. Checked my volts/temps once when I first put it together, since then it has been fast, stable and hasn't affected the noise of the rest of the system so I'm fairly confident it isn't heating everything else up.
    Single threaded performance is great, as is multithreaded. Even in fast paced "esport" games I've noticed a big difference in minimum frame rates and stutter, coming from a strong Haswell Xeon 4C/8T chip with very similar boost clocks. Perhaps the extra couple of cores and that massive cache are helping out in that area.

    Now I have a CPU that is already fast, but should only get faster. Where the OS, firmware, and other software all need to catch up to make full use of the hardware - not such a bad problem to have!
     
    335 GT likes this.

Share This Page

Advertisement: