• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

3800x vs 9900k

Robert896r1 said:
You're getting score regression which could also be due to run variance.

I'd run Geekbench 3 or 4 to compare memory and fabric impacts. It'll be more consistent than 3dmark.

I've read through that GB guide. Really interested to see how many people will run their recommended voltages. Should be interesting.
Click to expand...

The timings are not as good at IF1900

Geekbench 4 at IF 1900
https://browser.geekbench.com/v4/cpu/14771477

Single-Core Score
6241

Multi-Core Score
39787

Geekbench 4 at IF 1867
https://browser.geekbench.com/v4/cpu/14771461
Single-Core Score
6230

Multi-Core Score
39387

9900KS vs 3800x
aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS82L0EvODQ1MjE4L29yaWdpbmFsL1J5emVuLTctMzgwMC1YLXZzLUNvcmUtaTktOTkwMEtTLlBORw==


So in Geekbench 4 I am faster both in ST and MT than a 5GHz all core 9900KS. The Ryzen 7 3800X system was using memory clocked at 3,460 MHz while the Core i9-9900KS system was strangely utilizing memory running at 2,134 MHz.

3800x with the same memory

https://browser.geekbench.com/v4/cpu/13567135

Multi-Core Score
34059

https://www.tomshardware.com/news/ryzen-7-3800x-vs-core-i9-9900ks,39872.html

https://www.notebookcheck.net/Intel...7-3800X-in-Geekbench-comparison.427411.0.html

Corsair Vengeance 3200Mhz 16GB
Intel Core i9 9900K @ 5050Mhz 101 x 50 1.34v
October 19, 2018
https://www.overclockersclub.com/reviews/intel_core_i9_9900k/3.htm

Geekbench 4
https://www.overclockersclub.com/reviews/intel_core_i9_9900k/5.htm

Single-Core Score
6594

Multi-Core Score
37745

With IF 1900MHz
https://browser.geekbench.com/v4/cpu/14130048

Single-Core Score
6171

Multi-Core Score
40396

Looks like at 5GHz all cores the 9900k and 3800x with 2134MHz RAM are the same both ST and MT. With better RAM the 3800x pulls ahead of the 9900k with MT.

graph_3.png



graph_2.png


Note in the above two images: The Ryzen benchmark was only using DDR4-2133mhz memory while the i9 was using DDR4-2666mhz. After ABBA BIOS Ryzen does 6241 ST and memory appears to have little effect.

3800x @ 5.05GHz IF 1800

GB4_6618SC_42300MC-1024x639.jpg
 
Last edited:
Robert896r1 said:
You want to be a bit careful with the review stuff. A lot of them just bump up the clock but leave the uncore/cache at default (43x). When you run too far out of sync, it starts hurting you notably. uncore helps a lot with RAM latency also which in turn helps with overall performance.

Unfortunately, uncore is tied to vcore which means you can run into a scenario where you need more vcore just for uncore to go higher which is generally not worth it on it's own.

https://browser.geekbench.com/v4/cpu/14771538 that's mine after closing out steam and other apps.
Click to expand...

You have an epic RAM overclock as well, even better than mine. 1800MHz = 3600CL15 Reduce your ram to defaults at 2666-3200 and that score will hopefully match the benchmarks.

That is a damn good overclock you have. RAM and CPU.
 
Last edited:
Running Aida64 at stock core clocks, IF 1900 and the vcore is constant 1.408 volts. Clock is approx. +/-4.2GHz. LLC is 3 which is auto in bios. vcore seems very high with ABBA or these chips can take a lot of voltage?
 
Last edited:
102152.png


9900k stock with 2666MHz its default RAM. Average FPS 109.9
3800x stock with IF1900/3600 RAM (3800 tighten timings), AMD recommended for reviews DDR4 3600. Average FPS: 148

 
Last edited:
Last edited:
peter224722 said:
Wow a 9900k is slower on a graphed bottlenecked (RTX 1080), than a AMD 3800x with a faster RTX 2080 which is said to be 30% plus faster.

What are you saying ?
Click to expand...

Miss the next post were it's faster than a 9900k with a 2080ti as well? A 9900k all core's 5GHz with 3600CL16 RAM. A 15fps lead against a 2080 ti. Or 12%.
 
Last edited:
Robert896r1 said:
Exactly this. This is why synthetics and dedicated benchmarks are more usable in comparing across system due their linear scaling than workloads such as games which can be impacted by a number of external factors (drivers, game version, control panel settings, AA settings etc). In Tomb raider you can set highest and then adjust AA on a different screen and change the results alone just on that. I have SOTTR and used it for testing my GPU a lot. The game is fun also.
Click to expand...

The settings are the same but the graphic card is different. The 3800x system is faster, this is 3800x 2080 vs 9900k 2080ti both overclocked.

The only exception so far for the 3800x being faster is WoT EnCore. A game optimised by their friends at Intel.

I am carefully picking like for like benchmarks. Or reasonable comparisons. It's hard, its like the reviews are making sure you can't fact check them. Lot's of custom benches they can only run.
 
Last edited:
Gears of War 5

Core i9 9900K
Z390 (ASRock Tachi Ultimate)
32 GB DDR4 3200 MHz CL16
NVMe M.2. SSD WD Black
https://www.guru3d.com/articles_pages/gears_of_war_5_pc_graphics_performance_benchmark_review,3.html


1080p settings as per the website.



2080 ti 135fps
Titan Xp 120fps
2080 super 119fps
2080 111fps

3800x and 2080.



Average fps 123.5FPS
GPU bound 99.90%
CPU bound 0.10%

This is an AMD sponsored title.

See this video which shows a 9900k is CPU bound 10.09%. 24gb Ram / RTX 2080 OC 8gb / i9900k 5.0 GHZ I love the RAM amount.
https://youtu.be/tUZzOchs8PY?t=216
 
Last edited:
Looks like the 3800x under water with IF1900 and 3600 RAM OC'ed to 3800 CL16 with tightened timing is faster than a 9900k in games.
 
Rroff said:
Seems a bit of a jump to me if your claiming you are getting an increase in performance in Shadow of the Tombraider of ~33% just from water cooling and tuning the RAM/IF.
Click to expand...

What else would you think it was? It's what I am getting with the ABBA BIOS, stock cores (no pbo or anything else) and just the RAM overclocked. Note the issues of RAM latency are greatly reduced. Maximum latency I have seen can be 80ns but with this OC its as low as 60ns in userbench.

I 'think' the memory latency is the big issue of chiplet based CPU's. The more you reduce latency the better. I believe Intel can get as low as 45ns. My 4930k is 61ns.

https://uk.crucial.com/gbr/en/memory-performance-speed-latency

true latency (ns) = clock cycle time (ns) x number of clock cycles (CL)

The effect can be seen on Intel cpu's as well.

https://www.reddit.com/r/intel/comments/82jaqg/ram_latency_effect_on_gaming_performance_on/

Here on an Intel cpu there is a 27% increase in AotS going from 68.2ns to 46.5ns. So why do you find it hard to believe? It's exactly what you should expect.



Metro Redux with latest RAM timings.

Average 204fps.

Metro Exodus with latest RAM timings.


  • Average Framerate (99th percentile): 65.30
  • Max. Framerate (99th percentile): 90.33
  • Min. Framerate (99th percentile): 38.14
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom