AMD Ryzen For Simracing?

[Fi22le]

The Ryzen 3000 series responds best by setting the RAM frequency manually with tighter timings and letting the CPU self boost.
The cooler you keep it, obviously the better it performs.
I don't use any of the other features....no PBO ...no XMP.
I've set my cheap TimeTec 3600 to a fixed 3600 and tightened timings to 15, 19,19 19, 36 from the stock 19, 22, 22, 22, 42.
The performance difference is remarkable.
No temperature or stability issues whatsoever...and extremely smooth reliable gaming in everything.
No sure why guys are still manually 'hard' overclocking these processors for indiscernible gains.

Yep. Ryzen is not for overclockers. I’ve also trimmed my memory timings, but even that makes a pretty negligible difference.

When I switched from a lifetime of intel CPU’s, I was a bit apprehensive, but Ryzen has literally been a game changer. Can’t recommend them enough.

 

[Case_]

So, yeah, I've joined the 3600 crowd. Will have some benchmarks soon and comparison with the 2600 (honestly Rasmus, this time I will!).

Nice improvement. Holy crap the temperature and voltage fluctuations, though.

 

[Terry Rock]

So, yeah, I've joined the 3600 crowd. Will have some benchmarks soon and comparison with the 2600 (honestly Rasmus, this time I will!).

Nice improvement. Holy crap the temperature and voltage fluctuations, though.

They're both excellent CPUs. Ryzens may not have the highest clocks compared to Intel but the work done per cycle is so significant, they really don't have to. I love smooth, consistent frames and that is what I've received throughout my time with first the 2600x and now the 3600x. The Nvidia 450.12 driver on any of their GPUs in combination with either of those processors is a winning combination.

 

[Case_]

OK, I think I'm starting to see the appeal of tinkering with the Zen 2 CPUs. It's not as much to get more performance (because frankly, the PBO does its job perfectly fine), it's more to get the annoying swings in voltage and temperature under control.

So I started tinkering a bit, got my all core PBO values as a start - 3975 MHz, 1.238 V. Slowly worked myself up to 4300 MHz, so 100 MHz over the maximum single core boost, with no issues.

So then I started lowering the voltage...and honestly, I'm absolutely amazed. I managed to get all the way to 1.15 V without touching the frequency. Only then Cinebench crashed on me. So I'm currently at 1.163 V, which is 1.125 V with VDroop under full load. Still at 4.3 GHz all core. And I still can't believe it actually runs at least semi-stable. I'm saying semi-stable because while it seems perfectly stable so far, I haven't done any longer stability tests, I'm just checking the performance with repeated runs of Cinebench - I didn't originally plan to do an "overclocking" session, I just thought I'll quickly see how it goes

The best thing, though - my temperatures dropped by good 10 degrees compared to how they were with PBO. And I think I'll get them lower still, because I moved the fan curves so the CPU ran a bit hotter in an attempt to get the annoying fan stepping up and down on idle under control. So now I might be able to get it back where it was.

 

[Fi22le]

OK, I think I'm starting to see the appeal of tinkering with the Zen 2 CPUs. It's not as much to get more performance (because frankly, the PBO does its job perfectly fine), it's more to get the annoying swings in voltage and temperature under control.

So I started tinkering a bit, got my all core PBO values as a start - 3975 MHz, 1.238 V. Slowly worked myself up to 4300 MHz, so 100 MHz over the maximum single core boost, with no issues.

So then I started lowering the voltage...and honestly, I'm absolutely amazed. I managed to get all the way to 1.15 V without touching the frequency. Only then Cinebench crashed on me. So I'm currently at 1.163 V, which is 1.125 V with VDroop under full load. Still at 4.3 GHz all core. And I still can't believe it actually runs at least semi-stable. I'm saying semi-stable because while it seems perfectly stable so far, I haven't done any longer stability tests, I'm just checking the performance with repeated runs of Cinebench - I didn't originally plan to do an "overclocking" session, I just thought I'll quickly see how it goes

The best thing, though - my temperatures dropped by good 10 degrees compared to how they were with PBO. And I think I'll get them lower still, because I moved the fan curves so the CPU ran a bit hotter in an attempt to get the annoying fan stepping up and down on idle under control. So now I might be able to get it back where it was.

As a 3600 owner that has done very little tinkering (PBO and RAM timings only), I'd be interested to hear where you end up with your settings. I originally subsribed to the idea that Ryzen 3000's are so good, you just fire and forget, but if there's a significant difference in temperatures to be had, I'm all ears.

I'm running a Noctua NHD-15S (overkill I know), so it's cooling just fine, but it would be good not to generate the heat in the first place (especially as summer is coming).

 

[Dean Riley]

worth going from a 2600 to 3600 or waiting till new CPUs?

 

[John-Eric Saxén]

Yep, the fan curves are ridiculous (obviously you won't notice this as much if you have a large custom cooler, but my old Hyper 212 EVO is not compatible with AM4 and the bundled cooler is good enough for me in terms of cooling capacity). I seriously wonder who thought the best idea was to have the fan spinning up every time you open a new tab or refresh a website. They must have done no testing with actual users, because it's obvious to everyone that a more constant higher dB sound is much less irritating than sound levels going up and down all the time. What's worse, the fan dynamics are far too slow to react precisely to the load changes, so by the time the fan has spun up, the CPU might already have throttled back down, making it a completely useless exercise.

Anyway, rant over, luckily the bad things about 3rd gen Ryzen ends there. My MSI B450 board does provide fan curve adjustment, though not the most intuitive one, but with one minute of tweaking it got much better than original. Haven't tested tweaking voltages yet.

 

[Terry Rock]

Yep, the fan curves are ridiculous (obviously you won't notice this as much if you have a large custom cooler, but my old Hyper 212 EVO is not compatible with AM4 and the bundled cooler is good enough for me in terms of cooling capacity). I seriously wonder who thought the best idea was to have the fan spinning up every time you open a new tab or refresh a website. They must have done no testing with actual users, because it's obvious to everyone that a more constant higher dB sound is much less irritating than sound levels going up and down all the time. What's worse, the fan dynamics are far too slow to react precisely to the load changes, so by the time the fan has spun up, the CPU might already have throttled back down, making it a completely useless exercise.

Anyway, rant over, luckily the bad things about 3rd gen Ryzen ends there. My MSI B450 board does provide fan curve adjustment, though not the most intuitive one, but with one minute of tweaking it got much better than original. Haven't tested tweaking voltages yet.

I use DC instead of PWM for fan control on my B450 Gaming Pro Carbon AC.
I don't get that constant spin up and down when loading the processor.
You could have purchased an adapter bracket for that H212+ to allow for AM4 use.
I did just that at a cost of $7.00....I've also modified my cooler with some helicopter speed tape to allow the airflow to pass straight through the fins.
It has dropped the overall temperature by 8 degrees over the stock configuration.

CoolerMaster RR-ACCY-AM4B-R1 AM4 BRACKET for Hyper 212X, Hyper 212 EVO, and MasterAir Pro 3/4,

 

[Ruttman98]

Linus is high on the new AMD Ryzen 3s:

"Even USED Intel CPUs aren’t a good value anymore… - Ryzen 3 3100 & 3300X" -

$100 to $120 USD MSRPs make the $280 USD I shelled out on a Ryzen 5 3700X look like I went overboard on my CPU.

 

[Case_]

worth going from a 2600 to 3600 or waiting till new CPUs?

As mentioned, I have just done exactly that, and my answer would be...depends on what you expect and need. The performance boost is there, but the upgrade also costs a fair bit of money, so...yeah.

I will hopefully put together the comparison video for framerate in sims over the weekend (though I will need to record some commentary for that, which for me is always the thing that gets in the way the most). But I have the benchmarks all recorded, so it really is just to put it all together and do the commentary, hoping it will be at least somewhat comprehensible...

 

[Terry Rock]

worth going from a 2600 to 3600 or waiting till new CPUs?

AMD has stated that the new 4 series CPUs will not be compatible with B350, B450 or X470 boards.
Unless you have a 5-series motherboard, you'll be limited to what is currently out there already.
If you do moderately threaded task, the 2 series to 3 series switch is good but only if you have a buyer to offset the purchase cost.

 

[robt100]

They say its down to the 16mb BIOS....but MSI's 'MAX' series have 32mb BIOS, so you never know, it might work

 

[Terry Rock]

They say its down to the 16mb BIOS....but MSI's 'MAX' series have 32mb BIOS, so you never know, it might work

According to HU...the official word from AMD, is it has no plan of releasing any of the BIOS coding software to board partners to make it happen.
If they're going to try to make it happen, they'll have to come up with the coding on their own.
Go straight to 3:35 of video.

 

[Bob Gnarley]

.

 

[Durge Driven]

I always sell my CPU\Mobo\Ram as going concern
3900X do me for at least 2 years more
Can't see ever upgrading my 3200G, no point

 

[Durge Driven]

Dram Calculator for Ryzen v1.7.1

Download DRAM Calculator for Ryzen v1.7.3

You can download Ryzen DRAM calculator helps with overclocking your memory on the AMD Ryzen platform. It is a calculator that will make it easier to overclock the memory. It works with all kinds of m...

www.guru3d.com

• Reworking of all presets for Samsung b-die, Micron e-die and Hynix CJR(DJR). Improved support for configurations with 4 or 8 RAM modules.
  [*]R-XMP and Calculate EXTREME buttons removed.
  [*]Added new Help and my reviews section, which provides links to my materials in addition to reference information.
  [*]DRAM PCB Revision - select PCB RAM for more accurate calculation of timings. In most cases it is recommended to use "A0" for better compatibility.
  [*]The "Compare timings (ON/OFF)" button has received updated functionality. It can read not only the timings, but also show the state of the timings in a color. Your personal assistant for creating individual presets.
  [*]Added the "New version?" button. - button allows you to visit the product homepage and read the materials regarding DRAM Calculator for Ryzen™, check for updates or ask any question.
  [*]Error fixes and some graphical adjustments.

 

[Durge Driven]

AMD Ryzen Master v2.2.0.1543

2.2 PDF:

https://www.amd.com/system/files/documents/ryzen-master-quick-reference-guide.pdf

DOWNLOAD:

https://www.amd.com/en/support/cpu/amd-ryzen-processors/amd-ryzen-9-desktop-processors/amd-ryzen-9-3900x

 

[Locoblade]

OK, I think I'm starting to see the appeal of tinkering with the Zen 2 CPUs. It's not as much to get more performance (because frankly, the PBO does its job perfectly fine), it's more to get the annoying swings in voltage and temperature under control.

So I started tinkering a bit, got my all core PBO values as a start - 3975 MHz, 1.238 V. Slowly worked myself up to 4300 MHz, so 100 MHz over the maximum single core boost, with no issues.

So then I started lowering the voltage...and honestly, I'm absolutely amazed. I managed to get all the way to 1.15 V without touching the frequency. Only then Cinebench crashed on me. So I'm currently at 1.163 V, which is 1.125 V with VDroop under full load. Still at 4.3 GHz all core. And I still can't believe it actually runs at least semi-stable. I'm saying semi-stable because while it seems perfectly stable so far, I haven't done any longer stability tests, I'm just checking the performance with repeated runs of Cinebench - I didn't originally plan to do an "overclocking" session, I just thought I'll quickly see how it goes

The best thing, though - my temperatures dropped by good 10 degrees compared to how they were with PBO. And I think I'll get them lower still, because I moved the fan curves so the CPU ran a bit hotter in an attempt to get the annoying fan stepping up and down on idle under control. So now I might be able to get it back where it was.

I'm in a similar boat, having built my new gaming rig a couple of weeks ago with a 3600, MSI B450 Carbon AC, 240mm AIO cooler and 3600Mhz RAM. Having got the RAM running to spec I was a little disappointed in the stock "auto" figures when running Cinebench R20 as the CPU was only hitting ~3950 all core at ~75c temp, and although single core runs would touch 4200 it didn't seem to stay there for more than a split second before settling around 4150-4175. I believe its just regular Precision Boost (not Overdrive) that's lifting and dropping the clock speeds during this process so in theory enabling PBO should push things a little further but in fact I found it made sod all difference whether it was on or off, much like what Gamers Nexus etc found in their reviews.

I then tried a bit of manual all core overclocking and without much effort got it Cinebench stable at 4200 at 1.15v dropping temps by 10c in the process. That improved the all core benchmark by about 10% and single core benchmark by about 3%, although it wouldnt run Prime95 for more than a minute or two without one core erroring and stopping the run, so it's not 100% stable and perhaps needed a bit more voltage to achieve that.

Despite all that though I'm still a bit concerned about all the conflicting information regarding running these CPUs in a fixed overclock/voltage state and as the biggest gains of overclocking it are multithreaded performance that I'll rarely utilise fully in a gaming system, I'm still not sure whether to go with the overclock or just leave PB and PBO etc to do it's thing. Thinking logically though you'd have thought if you can reduce temperatures and keep voltages significantly below their peaks in auto then is that really going to be detrimental to the CPU?

 

[Case_]

I'm not really worried about running my CPU 10+ degrees colder at a noticeably lower voltage than PBO uses even in all-core worst case scenario. I don't think I should be.

As for stability testing, maybe try some x264 encoding. My settings passed repeated Cinebench runs and maximum Prime 95 stress test, but crashed consistently when I tried to re-encode some video in Avidemux, so I had to up the voltage a bit.

 

[Locoblade]

Yeh I think the argument against is that for the majority of time when the cores are idling and ramped down they're often subjected to voltages under 1v, and the peaks of 1.3-1.4v triggered by PBO are only for fractions of seconds at a time whereas overclocking and fixing it at say 1.2v means the CPU is seeing higher voltages all the time even though the peak voltages aren't as high. I'm still of the opinion though that heat output will be directly linked to how much stress the CPU is under so if overall temps are similar at idle and lower under load then surely it's under less stress overall.