1600 to 3600 worth it in my situation?

Discussion in 'AMD Processors' started by mazeroth, Jul 10, 2019 at 9:47 PM.

  1. mazeroth

    mazeroth Limp Gawd

    Messages:
    242
    Joined:
    Oct 2, 2015
    I have a Ryzen 1600 running stock cooler and clocks (I know, I know...I'm an old-school overclocker that just got lazy on this build) with a 1080 Ti. 95% of the time I'm playing VR games like In Death, Skyrim, iRacing etc. I used a Rift CV1 since launch and just upgraded to the Rift S. Life is good.

    When playing, framerates seem very stable. I keep task manager open and sometimes peek out to see what % utilization my CPU and GPU are at. Typically, my CPU is hovering around 30-40% and my GPU around 70%, on average. Now, watching all these 3000 series reviews, they show average framerates along with 1% and 0.1%. The latter is a better test of how stable a game plays, and if there will be any dips (I'm sure you guys know this!). If none of my CPU cores ever even come close to 100%, does it make any sense to upgrade to the 3600? Does a core have to hit 100% to cause stutter, or can it happen below 100%?

    I honestly can't tell if my system is dropping below 80 fps on the Rift S. I've tested reprojection before, using the Oculus Tray Tool, and it's very apparent when it happens to me.

    Long story short, do you think it's worth it for me to upgrade from my 1600 to a 3600? On paper and from the reviews, it's a big step up in performance. However, I feel like my system is chugging along just fine.

    Thanks!
     
  2. mnewxcv

    mnewxcv [H]ardness Supreme

    Messages:
    6,391
    Joined:
    Mar 4, 2007
    you may have answered your own thread. The 3600 is a big improvement in terms of single core and IPC performance. However, you are happy with what you have. Wait until you aren't.
     
  3. DrLobotomy

    DrLobotomy [H]ardness Supreme

    Messages:
    5,589
    Joined:
    May 19, 2016
    If it ain't broke, don't fix it.
     
    Mav451, DrDoU and jmilcher like this.
  4. buttons

    buttons [H]ard|Gawd

    Messages:
    2,027
    Joined:
    Oct 12, 2011
    You could do exactly what i did. tweak the crap out of your 1600 so the 3600 will be less appealing.

    I got my single threaded performance 17% above stock which helps FPS a fair bit. overclocking memory for ryzen seems to make the biggest difference. My setup for gaming is 3.7ghz all cores, smt disabled, memory timings set to stilts recommended using safe settings from ryzen dram calculator (3000 CL16)

    Edit: If your running an application or game that loads up just one core then it could cause a stutter if that core cant keep up.
     
    Last edited: Jul 10, 2019 at 10:14 PM
    Mav451 likes this.
  5. ManofGod

    ManofGod [H]ardForum Junkie

    Messages:
    10,328
    Joined:
    Oct 4, 2007
    I have my 1600 running at 4.0 Ghz all cores and SMT enabled. Why do you disable the SMT?
     
  6. Vengance_01

    Vengance_01 [H]ardness Supreme

    Messages:
    5,715
    Joined:
    Dec 23, 2001
    If your happy with frame rates as other said don't break what's not broken. VR is heavy GPU anyways. Just o/c your current chip if you want to see what kind of boost you might be looking at first.
     
  7. buttons

    buttons [H]ard|Gawd

    Messages:
    2,027
    Joined:
    Oct 12, 2011
    i like to play a game with garbage optimization. (mechwarrior online) I also saw some benchmark improvements in single threaded tasks (its currently only disabled in my game mode profile)
     
    ChadD likes this.
  8. ChadD

    ChadD 2[H]4U

    Messages:
    3,929
    Joined:
    Feb 8, 2016
    SMT can degrade performance in some games. Not just old titles either some fairly recent titles run better with SMT off. In general it's going to be a wash outside of some specific titles. There are a few cases where keeping everything on the 6 or 8 physical cores will gain you 10-20%.

    The truth is in games SMT threads are mostly useless. They are nice for showing lower CPU usage in your benchmark... but the processor running 16 threads at 40% vs 8 at 80% is going to get just as hot. The FPS will be nearly identical... mostly. Some engines like the source engine run best with SMT off.

    I agree with you, I would never turn it off either. I think a few titles like Final Fantasy where SMT harms performance there is even a launch option that disables it in game rather then system wide.
     
  9. buttons

    buttons [H]ard|Gawd

    Messages:
    2,027
    Joined:
    Oct 12, 2011
    Our intern brought in his computer to work today. I tested my Mechwarrior online game briefly on it.

    His setup all default settings
    Asus AB-350
    Ryzen 5 3600
    4x8gb corsair 3000 CL16 set to DOCP settings
    Evga AIO cooler
    EVGA 1080ti

    CPU-Z single threaded bench. r5 1600 100%, my 1700 @ 3.7ghz 117%, r5 3600 141%
    MWO Online max fps in testing grounds: My 1700 is 200-210 fps. His 3600 was 245-255fps. We both have 1080Ti i am running 3000 CL14 memory. His clocks / power / temps behave a little qwerky. Computer runs fine, but his benchmark scores are not consistent from run to run. My machine might deviate 1-3 points per run. His will deviate 30-40 points.

    Conclusion:
    I am totally convinced I need a 3700X cpu, X570 board, 3600 CL16 memory now.
     
    Mav451 and Sulphademus like this.
  10. ManofGod

    ManofGod [H]ardForum Junkie

    Messages:
    10,328
    Joined:
    Oct 4, 2007
    I have my Ryzen R5 2600 running at 4.1 Ghz with my Vega 56 flashed with the 64 bios. (3600 ram running at 3200 and tighter timings.) I also have my 1600 running at 4.0 GHz with my RX580 and my 1700 running at 3.8 Ghz with my RX 570. I figure my systems are pretty much balanced out as best I can get them.

    That said, for your system and what you do, I would say the upgrade is a really good idea. By the way, I do not have the game but, I am curious, is there a way I can benchmark that game without buying it? Also, what resolution and settings is the game set at please? Thanks.
     
  11. buttons

    buttons [H]ard|Gawd

    Messages:
    2,027
    Joined:
    Oct 12, 2011
    I didn't mean to Hijack op thread. Ive been on this forum and reddit, pressing F5 non stop waiting for actual people to post their impressions. Mechwarrior online is free to play. mwomercs dot com. I pre-ordered MW5 which launches in Sept. My kids FX6300 does all i "need" at the moment, but when i get back into Video encoding the 3700x would be a nice jump.
     
  12. ManofGod

    ManofGod [H]ardForum Junkie

    Messages:
    10,328
    Joined:
    Oct 4, 2007
    Honestly, I do not think you hijacked it at all, your example definitely gives a good idea of what to expect with the upgrade that the OP asked about. I will run the benchmark and post the results here as well, for my 2600. (What resolution and settings, by the way? Thanks.)
     
  13. buttons

    buttons [H]ard|Gawd

    Messages:
    2,027
    Joined:
    Oct 12, 2011
    If you are going to test Mechwarrior online the settings we used were 1080p - highest graphics preset. Testing grounds for single player and then mining collective map. Each map has different FPS results. F9 is the hot key to enable FPS counter, for multiplayer where it matters the most i really need to run a utility in the background to capture that data.
     
  14. ManofGod

    ManofGod [H]ardForum Junkie

    Messages:
    10,328
    Joined:
    Oct 4, 2007
    Well, I would say I was anywhere from 160 to 201, give or take. Since I do not have a Ti, you will have to tell me if that sounds about right to you or not. (Ryzen 2600 at 4.1 Ghz.)
     
  15. somebrains

    somebrains Gawd

    Messages:
    570
    Joined:
    Nov 10, 2013
    I'm beating my 2600 right now like it owes me $.

    It's the upside to $200 CPUs all over the place, if I break it then I'll find a 2700x to flog.

    If I break one of those, 3600.
     
    buttons likes this.
  16. ManofGod

    ManofGod [H]ardForum Junkie

    Messages:
    10,328
    Joined:
    Oct 4, 2007
    I did the same to mine but, it just will not go beyond 4.1 Ghz, no matter what I do or what board I install it in.
     
  17. somebrains

    somebrains Gawd

    Messages:
    570
    Joined:
    Nov 10, 2013
    That's where I'm at.
    Arguably stock 3.7 boost at 1.1xx is better than 4.1 all core at 1.3v bc of the jump in heat.

    I just wanted to game all day in a sauna apparently bc I'm not getting framerate that's making this all worth it.
     
  18. mazeroth

    mazeroth Limp Gawd

    Messages:
    242
    Joined:
    Oct 2, 2015
    I appreciate all the feedback. However, would you guys mind answering this one question?

    "If none of my CPU cores ever even come close to 100%, does it make any sense to upgrade to the 3600? Does a core have to hit 100% to cause stutter, or can it happen below 100%?"
     
  19. buttons

    buttons [H]ard|Gawd

    Messages:
    2,027
    Joined:
    Oct 12, 2011
    My 1700 died tonight. System wont boot at all, unseated and reseated everything, removed bios battery and waited.... Machine turns on , motherboard shows CPU fault led. All i did was change my memory from 1500 to 1600 mhz from master ryzen app. never seen something so trivial cause this. Now im stuck between rock and hard place. i cant order a 3000 cpu / motherboard because 1700 wont work in a x570 and a 3000 wont work in my x370 without that beta bios update. When i first built the PC last year it seemed really qwerky but has been rock solid stable last 7-8 months. not sure if something finally gave up.

    I cant decided if i want to swing by microcenter and get a cheap 2700 to have spare parts to swap and test before I deal with RMA
     
  20. ManofGod

    ManofGod [H]ardForum Junkie

    Messages:
    10,328
    Joined:
    Oct 4, 2007
    What do you mean, in all seriousness, about a 1700 not working in an X570 motherboard?
     
  21. ManofGod

    ManofGod [H]ardForum Junkie

    Messages:
    10,328
    Joined:
    Oct 4, 2007
    Honestly, I have no idea how to answer this. When you overclock the 1600, does your minimum and average fps go up in games? Does your cpu usage go up when you overclock, even a mild one?
     
  22. somebrains

    somebrains Gawd

    Messages:
    570
    Joined:
    Nov 10, 2013
    Depends on the game.
    If it's GPU bound at 1440p, burn hours of your life crashing your box until it's stable and you see minimal gains like 135-145fps. It tightened up. Or you get higher peak fps like to the 161fps limit I set so my 165hz panel doesn't overshoot.

    If you plays BRs like Pubg, Apex, Fortnite, Blackout or a build like BF5 then the unoptimized POS maps will eat 40-50fps depending on where you are looking. You can kinda tell, but not really

    If you play CPU bound games I argue that you'd want a 5ghz i7. I've yet to see anything that wants 12+ cores vs 8cores at 5ghz.

    Are the x570 boards only backwards compatible with Zen+? Thoutht that's why guys are selling 1700x's for less than $80

    A cpu is measured by more than just it's thread count and frequency. So for common workloads a 3600 on paper seems to be better than a 2700x until you start getting into content creation or compiling monolithic code.

    Depends what you are doing with it.
    $100 used 2600 is a thing right now.
    2700x guys don't seem to want to see the MC $200 price and work down to $150 shipped yet.

    I don't see anything wrong with buying a new 2700/x from MC and leaving everything else alone.
     
    Last edited: Jul 12, 2019 at 12:54 AM
  23. funkydmunky

    funkydmunky 2[H]4U

    Messages:
    2,295
    Joined:
    Aug 28, 2008
    Try an app called fpsVR. It is like $4 and shows you in game with a small display that shows your CPU and GPU stats on the fly. All kinds of cool stats and shows at a glance if you drop bellow a threshold. It is a must have IMO. I would be flying blind without it. Adjust game settings at see the results in performance as you play. Really helped me with IL-2 and Project Cars2.
    As far as upgrading I am in a similar situation. R7-1700X@3.8 SMT disabled. My RTX-2080 with a high OC always seems to be my bottleneck. I did the max memory OC on the card first as that had the biggest impact on minimum fps. I then OC'ed the core only slightly as I found it would go higher but it started to reduce my min fps. Not ideal. I am running a Odyssey+ which has the same rez as your Rift-S which is very high. Although I am trying to stay above 90fps and (I believe the Rift-S only needs 80?) it was suggested to me that the R5/7-3000's might help with my minimum fps. I don't know if true as it is hard to get real VR benchmarks. Especially since the [H] has shut that down :(
     
    Last edited: Jul 12, 2019 at 1:32 AM
  24. somebrains

    somebrains Gawd

    Messages:
    570
    Joined:
    Nov 10, 2013
    ^ bust out a credit card and start screen recording all your test output.

    Hopefully you'll find a retailer that won't ban you or hammer you with return fees. Best Buy should be good for that. They're trying to be CompUSA now.

    Upload to YouTube.
    Make some view $.
     
  25. funkydmunky

    funkydmunky 2[H]4U

    Messages:
    2,295
    Joined:
    Aug 28, 2008
    From what I have read lately, get the OC to a stable 4GHz or close, but more importantly tighten the RAM and be happy (3200MHZ CL-14). At VR rez it matters little. The good old Zen1 does a good job at coming very close in these situations.
    Personally I was hoping it would be worth the upgrade. For me no. Maybe Zen 2+ will hit some high clocks and I'll make the jump in a year. For now just tweak and enjoy ;)
     
  26. sabrewolf732

    sabrewolf732 2[H]4U

    Messages:
    3,862
    Joined:
    Dec 6, 2004
    I went 1600 to 3600, it's a big improvement in terms of minimum fps (I mostly do overwatch apex etc)
     
  27. 123Lanoix

    123Lanoix 2[H]4U

    Messages:
    3,778
    Joined:
    Nov 13, 2005
    Yah i saw a big improvement also in Overwatch going from 2600 oc to 4.1 ghz to 3700x stock