Ryzen 3000 Series U...
 
Share:

Ryzen 3000 Series User Gaming Benchmarks  

  RSS

balbs10
(@balbs10)
New Member Registered
Joined: 2 weeks ago
Posts: 1
05/08/2019 4:45 am  

Designing a methodology to do theoretical testing is very difficult for gaming GPUs, because not all companies have huge gross margins that allow them to chuck money into theoretical scenarios, like optimizing drivers for the 720p or 480p resolutions for highend gaming GPUs.
 
So, I have used my personal experiences of the way AMD codes and programs its drivers going back to my ownership of an R9 290 4GB in 2013 and then an RX Vega 56 and then an RX Vega 64 and now a Radeon VII.
 
AMD does not attempt to waste a tonne of money on coding and programming for games that are a tad broken or quick asset flips of an older game. Essentially, if a game release is broken performance-wise, AMD will not throw coding and programming hours into an endless money pit endeavour of trying to fix it's FPS performance. Secondly, AMD tends to put coding and programing hours into FPS performance for the resolutions that gamers are using for their GPU for.
 
The first part does mean, there are certain games not worth benchmarking because they are a tad broken or they are just quick asset flips of an earlier release. Benchmarking games that are broken or inferior asset flips does not show what is going on with the performance. And, the 2nd part does mean dropping down to 480p or 720p is largely a fool's errand, because you cannot extrapolate findings on those resolutions when a GPU company is not spending coding and programming budget on them for their higher tier products.
 
Therefore, I decided to re-bench 9 of the 12 games (not CSGO, Strange Brigade and World War Z) with various DDR4 speeds and low latency subtimings at 1080p, but this time dropping the graphical setting down to low or medium, where necessary to simulate the effects of a new AMD GPU running about 30% to 40% faster than the current flagship.
 
Corsair LPX Vengeance DDR4-3000 CL15-17-17 Kit with the XMP Profile will be used as a fixed baseline for comparisons. It will be compared to Teamgroup Dark Pro DDR4-3200-CL14 kit with some low latency subtimings. This will be followed with G.Skill Trident Z DDR4-3866 Kit with low latency subtimings at DDR4-3600 speed and DDR4-3733 speed.
 
11 out of 12 games saw FPS increases with Radeon VII e.g. 91% of games improved. Strange Brigade was left in the sample of the games because there are games that are released that are designed to run well with little regards to a CPU's performance benefits. And, here is a summary of results.
 
The 12 game averages saw a very good boosts to FPS:
Corsair LPX Vengeance DDR4-3000-CL15 TO LLC DDR4-3200 = +4.6%
Corsair LPX Vengeance DDR4-3000-CL15 TO LLC DDR4-3733 = +7.2%
Corsair LPX Vengeance DDR4-3000-CL15 TO LLC DDR4-3733 = +8.1%
The 12 game average minus CPU Boost (-100mhz Cap at DDR4-3600+ Speed):
Corsair LPX Vengeance DDR4-3000-CL15 TO LLC DDR4-3733 = +8.2%
Corsair LPX Vengeance DDR4-3000-CL15 TO LLC DDR4-3733 = +9.1%
 
Top 4 Game Utility Results
Hitman 2016 = +18.9%
Shadow of Tomb Raider = +13.6%
Deus Ex Mankind Divided= +12.6%
Far Cry 5 = +11.7%
 
As can be seen the FPS increase on Ryzen 3rd CPUs do match and sometimes surpass the Ryzen 2nd Gen CPUs with increases in DDR4 speed and with low latency subtimings.
 
Below are all the Game Utility test results!
1920x1080p Settings Targeting 144FPS Gaming
CSGO Default Highest Settings
Corsair LPX DDR4-3000 15-17-17 XMP= 388.3FPS (100%)
Teamgroup Dark Pro DDR4-3200 CL14 Low Latency subtimings= 400FPS (103%) +3%
G.SKill Trident Z Custom Low Latency subtimings DDR4-3600= 403.2FPS (103.8%) +3.8%
G.SKill Trident Z Custom Low Latency subtimings DDR4-3733= 405.6FPS (104.5%) +4.5%
 
2) Deus EX Mankind DX12 Divided Medium
Corsair LPX DDR4-3000 15-17-17 XMP= 140.8FPS (100%)
Teamgroup Dark Pro DDR4-3200 CL14 Low Latency subtimings= 152.4FPS (108.2%) +8.2%
G.SKill Trident Z Custom Low Latency subtimings DDR4-3600= 156FPS (110.8%) +10.8%
G.SKill Trident Z Custom Low Latency subtimings DDR4-3733= 158.5FPS (12.6%) +12.6%
 
3) F1 2019 High Britain Clear Weather
Corsair LPX DDR4-3000 15-17-17 XMP= 234.7FPS (100%)
Teamgroup Dark Pro DDR4-3200 CL14 Low Latency subtimings= 242.7FPS (103.4) +3.4%
G.SKill Trident Z Custom Low Latency subtimings DDR4-3600= 244.3FPS (104.1%) +4.1%
G.SKill Trident Z Custom Low Latency subtimings DDR4-3733= 246.3FPS (104.9%) +4.9%
 
4) Far Cry 5 Normal TAA
Corsair LPX DDR4-3000 15-17-17 XMP= 120FPS (100%)
Teamgroup Dark Pro DDR4-3200 CL14 Low Latency subtimings= 129FPS (107.5%) +7.5%
G.SKill Trident Z Custom Low Latency subtimings DDR4-3600= 133FPS (110.8%) +10.8%
G.SKill Trident Z Custom Low Latency subtimings DDR4-3733= 134.3FPS (111.7%) +11.7%
 
5) Forza Horizon 4 Medium
Corsair LPX DDR4-3000 15-17-17 XMP= 192.6FPS (100%)
Teamgroup Dark Pro DDR4-3200 CL14 Low Latency subtimings= 194.2FPS (100.8%) +0.8%
G.SKill Trident Z Custom Low Latency subtimings DDR4-3600= 195.4FPS (101.4%) +1.4%
G.SKill Trident Z Custom Low Latency subtimings DDR4-3733= 195.9FPS (101.7%) +1.7%
 
6) Hitman DX12 Medium
Corsair LPX DDR4-3000 15-17-17 XMP= 141FPS (100%)
Teamgroup Dark Pro DDR4-3200 CL14 Low Latency subtimings= 156FPS (10.6%) +10.6%
G.SKill Trident Z Custom Low Latency subtimings DDR4-3600= 165.4FPS (117.3%) +17.3%
G.SKill Trident Z Custom Low Latency subtimings DDR4-3733= 167.7FPS (118.9%) +18.9%
 
7) Middle-Earth Shadow of War Medium + HD Textures
Corsair LPX DDR4-3000 15-17-17 XMP= 179.3FPS (100%)
Teamgroup Dark Pro DDR4-3200 CL14 Low Latency subtimings= 187.7FPS (104.7%) +4.7%
G.SKill Trident Z Custom Low Latency subtimings DDR4-3600= 190.3FPS (106.1%) +6.1%
G.SKill Trident Z Custom Low Latency subtimings DDR4-3733= 192.3FPS (106.7%) +6.7%
 
8) Rise of the Tomb Raider DX12 Medium
Corsair LPX DDR4-3000 15-17-17 XMP= 180.3FPS (100%)
Teamgroup Dark Pro DDR4-3200 CL14 Low Latency subtimings= 189FPS (104.8%) +4.8%
G.SKill Trident Z Custom Low Latency subtimings DDR4-3600= 195.9FPS (108.6%) +8.6%
G.SKill Trident Z Custom Low Latency subtimings DDR4-3733= 198.6FPS (110.1%) +10.1%
 
9) Shadow of Tomb Raider DX12 Low
Corsair LPX DDR4-3000 15-17-17 XMP= 145FPS (100%)
Teamgroup Dark Pro DDR4-3200 CL14 Low Latency subtimings= 154FPS (106.2%) +6.2%
G.SKill Trident Z Custom Low Latency subtimings DDR4-3600= 162.3FPS (111.9%) +11.9%
G.SKill Trident Z Custom Low Latency subtimings DDR4-3733= 164.7FPS (113.6%) +13.6%
 
10) Strange Brigade DX12 All Settings at High
Corsair LPX DDR4-3000 15-17-17 XMP= 185FPS (100%)
Teamgroup Dark Pro DDR4-3200 CL14 Low Latency subtimings= 185FPS (100%)
G.SKill Trident Z Custom Low Latency subtimings DDR4-3600= 185FPS (100%)
G.SKill Trident Z Custom Low Latency subtimings DDR4-3733= 185FPS (100%)
 
11) The Division 2 DX12 Medium
Corsair LPX DDR4-3000 15-17-17 XMP= 191FPS (100%)
Teamgroup Dark Pro DDR4-3200 CL14 Low Latency subtimings= 194.7FPS (101.9%) +1.9%
G.SKill Trident Z Custom Low Latency subtimings DDR4-3600= 196FPS (102.6%) +2.6%
G.SKill Trident Z Custom Low Latency subtimings DDR4-3733= 196.6FPS (103%) +3%
 
12) World War Z Vulkan Ultra 100% Scaling
Corsair LPX DDR4-3000 15-17-17 XMP= 192FPS (100%)
Teamgroup Dark Pro DDR4-3200 CL14 Low Latency subtimings= 201FPS (104.7%) +4.7%
G.SKill Trident Z Custom Low Latency subtimings DDR4-3600= 209FPS (108.8%) +8.8%
G.SKill Trident Z Custom Low Latency subtimings DDR4-3733= 211FPS (109.9%) +9.9%
 
Notes:
All Ram validated as "error free" with Memtest Pro 100% XMP and 200% for Low latency subtimings.
Aida64 Latency Test (1st run discarded and 3 run average):
Corsair LPX DDR4-3000 15-17-17 XMP 1.38volts= 76.7ns
Teamgroup Dark Pro DDR4-3200 CL14 Low Latency subtimings 1.43volt = 71.3ns
G.SKill Trident Z Custom Low Latency subtimings DDR4-3600 1.47volts = 66.5ns
G.SKill Trident Z Custom Low Latency subtimings DDR4-3733 1.484volts = 64.9ns
Screenshots of subtimings used are on this link:
https://imgur.com/a/JwGyZGQ
 
I had no issues getting DDR4-3733 working on the Ryzen 7 3700X with 1 to 1 divider and the memory controllers on this SKU look spot-on for maxing out gaming FPS.
 
Test Systems Configuration
Ryzen 3700X PBO + Enhanced XFR + AutoOC200mhz
7 PC case fans and one 240mm AIO for the CPU.
Gigabyte Auros X470 7 WiFi (BLCK 100.7) with onboard Audio used.
Sata SSDs.
Corsair 850Watt Platinum PSU 94% Efficiency.
Windows 10 "turn on fast startup" disabled.

This topic was modified 2 weeks ago 2 times by Kirk Johnson

muziqaz and Monster liked
Quote

Do NOT follow this link or you will be banned from the site!
  
Working

Please Login or Register