[REVIEW] Should you switch from first or 2nd gen Ryzen to 3rd gen?

Skiiwee29 · August 6, 2019

In all fairness, this is a loaded question to begin with. The Ryzen family of CPUs has been a god send not only for the consumer, but has also breathed some much needed energy back into the PC space as a whole. It has truly RYZEN'd to the occasion... (Ok, bad puns out of me now).

If you have a 1st gen Ryzen CPU, should you upgrade to 3rd gen? I personally just did this today myself going from my 1800x that I was easily able to overclock to 4ghz at 1.35v all day long without breaking a sweat and never breaking 70c on temps. I game, I stream, I watch youtube, scroll through facebook, twitter, instagram etc... so not really a large work load beyond the gaming/streaming combination. Today arrived my Ryzen 7 3700x that I purchased yesterday. Before swapping it out, I ran some benchmarks to be able to have a quantitative comparison to be able to look at and gauge just how good of a CPU upgrade it was. Gaming wasn't as big of a change or increase in performance as I was hoping for, but gains are still great non the less.

To start this off, here is a look at the full system used. Everything was exactly the same between them with the exception of the CPUs. The 1800x was overclocked as I had used it for the past 2 years to 4ghz at 1.35v on my Asus Crosshair VI Hero motherboard. My 3700x is all at stock except the voltages which I set to 1.35v as well, which is how I intend to use the 3700x daily anyways since there is little to no headroom for overclocking on the chips due to the aggressive nature of Precision Boost Overdrive from AMD.

Asus Crosshair VI Hero Bios 7201

16gb Gskill Flare x at 3200mhz 14 14 14 34

EVGA RTX 2080ti Black edition Stock clocks and memory, only increased Power and GPU temp target.

Nvidia drivers 431.36

Windows 10 Pro Version 1903

1800x CPUz

Spoiler

3700x CPUz

Spoiler

First the gaming benchmarks. I admit they are rather small in comparison as I didn't really want to spend all day long benchmarking games.

Firestrike Ultra	Total	CPU	GPU
1800x	8066	20814	7967
3700x	8245	22918	8108
Difference	179	2104	141
% change	2.17%	9.18%	1.74%

Timespy Extreme	Total	CPU	GPU
1800x	5826	3537	6578
3700x	6269	4437	6763
Difference	443	900	185
% Change	7.07%	20.28%	2.74%

Ghost Recon	Average	Minimum	Maximum	– Ultra Preset 2560x1440
1800x	68.26	57.84	77.84
3700x	71.1	58.88	79.52
Difference	2.84	1.04	1.68
% Change	3.99%	1.77%	2.11%

Division 2	FPS	Ultra Preset 2560x1440
1800x	84
3700x	89
Difference	5
% Change	5.62%

Rise of the Tomb Raider	FPS	Very High Preset 2560x1440
1800x	121.16
3700x	142.05
Difference	20.89
% Change	14.71%

Not bad. In general, games saw an increase in FPS by about 1-5% with an extreme case in Rise of the Tomb Raider and a WHOPPING 15%!!! Firestrike and timespy show a good chunk of improvement in there CPU scores that brought up there scores a good deal as well!

Lastly, we have the more productivity centered... if you do productivity work with your Ryzen 1st gen, based on this limited test results, I would say its a HUGE upgrade for you.

Cinebench R15	Single Thread	Multi Thread	Multi Thread Time
1800x	161	1689	25.38
3700x	199	2044	21.03
difference	38	355	-4.35
% change	19.10%	17.37%	-20.68%

Cinebench R20	Single Thread	Multi Thread	Multi Thread Time
1800x	397	3785	82.74
3700x	481	4562	68.85
Difference	84	777	-13.89
% Change	17.46%	17.03%	-20.17%

LOOK AT THOSE MASSIVE Gains!! My lord!! 20% in each in there multi threaded times and 17-19% faster multi threaded AND single threaded work loads!! AMAZING!!!!

Conclusion time. Should you upgrade from a 1st gen Ryzen to 3rd gen for gaming? My answer would be no, its not worth the cost. I personally only did it as I am re-purposing my 1800x into a dedicated streaming rig so it made a little more sense. If you do productivity with your 1st gen Ryzen, then the answer from me is an astounding YES YES YES!!! Upgrade as soon as you can. These are some massive gains I saw, again in a limited testing with Cinebench only, so your work flow will vary so you should try and look for those figures and numbers and see what others are reporting as well.

I hope you guys liked my first ever written review on anything. Please let me know what you think.

porina · August 7, 2019

Nice, I've been too lazy after doing similar e.g. dropping a 3600 in place of a 2600. I'm too lazy to swap back now, but might still do a vs. Intel comparison some time.

Interesting choice of 1.35v though, I have concerns that might cause problems with the boost feature. It likes higher voltages with fewer cores, and adjusts voltage as needed with actual clock, so you might be running higher than needed in some load scenarios.

Also to my understanding, the stock boost mechanism is PB2, not PBO which is an auto-overclock option for X570 only.

Skiiwee29 · August 7, 2019

3 hours ago, porina said:

Nice, I've been too lazy after doing similar e.g. dropping a 3600 in place of a 2600. I'm too lazy to swap back now, but might still do a vs. Intel comparison some time.

Interesting choice of 1.35v though, I have concerns that might cause problems with the boost feature. It likes higher voltages with fewer cores, and adjusts voltage as needed with actual clock, so you might be running higher than needed in some load scenarios.

Also to my understanding, the stock boost mechanism is PB2, not PBO which is an auto-overclock option for X570 only.

When I had Aida64 running, I was hitting 4.1ghz on all cores. Ill see if setting voltage back to Auto makes any difference

WIth Voltages at auto, it was still going to the same 4.1ghz in Aida stress test. I did notice that the load voltages went from 1.45v down to 1.35v as well during load, which is mentioned in an article I read yesterday as well.

porina · August 7, 2019

30 minutes ago, Skiiwee29 said:

WIth Voltages at auto, it was still going to the same 4.1ghz in Aida stress test. I did notice that the load voltages went from 1.45v down to 1.35v as well during load, which is mentioned in an article I read yesterday as well.

I can't remember exact values, but I find voltage tends to lower with more cores loaded. I think the higher voltage is needed to get higher clocks stable when fewer cores are running, hence leading to the "high voltage on idle" reports previously. That was more my concern, that fixed 1.35v might be less stable with fewer cores active if it really needs to higher voltages to run at the higher (fewer) core clock.

Skiiwee29 · August 7, 2019

33 minutes ago, porina said:

I can't remember exact values, but I find voltage tends to lower with more cores loaded. I think the higher voltage is needed to get higher clocks stable when fewer cores are running, hence leading to the "high voltage on idle" reports previously. That was more my concern, that fixed 1.35v might be less stable with fewer cores active if it really needs to higher voltages to run at the higher (fewer) core clock.

I'll keep that in mind, today is when I will be putting it through its paces fully and throwing the gambit of what I do at it since I spent all day yesterday with the few benchmarks and building up my 2nd dedicated rig and redoing all the wiring around my desk and stuff.