• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

How is Intel Beating AMD Zen 3 Ryzen in Gaming?

W1zzard

Administrator
Staff member
Joined
May 14, 2004
Messages
29,001 (3.75/day)
Processor Ryzen 7 5700X
Memory 48 GB
Video Card(s) RTX 4080
Storage 2x HDD RAID 1, 3x M.2 NVMe
Display(s) 30" 2560x1600 + 19" 1280x1024
Software Windows 10 64-bit
Our Ryzen 5000 Zen 3 launch day reviews saw unexpected gaming FPS results many questioned. In this article, we will investigate these results in more detail, and do more testing to figure out what is going on. The results are surprising and set things right in the battle of AMD vs. Intel.

Show full review
 
Deversification will only get deeper as time goes on, I'm afraid, my friend.

  • There will be people who want Zen 3 based GPU tests, because for their scenario it makes the most sense
  • There will be people who want Intel based GPU tests, since globally it is still very common
  • There will be people who want tests with slower and more common memory speeds, and those who want with high speeds and tight timings
  • There will be people who want CPU tests with an RTX 3080 because it is the most advanced NVIDIA GPU the common can afford, but others who want an RX 6800 XT one because it represents a fair alternative
  • There will be those who want GPU tests on both Intel and AMD systems, since RDNA 2's SAM feature is AMD exclusive, but you can't exclude most users who don't have the option to use SAM since they have already purchased older hardware
  • RTX ON? RTX OFF? Smart Cache OFF? Smart cache ON? What about 8 core vs 10 cores vs 12 cores? And Memory types and configuration? why not 4 DIMMS? why not 2 DIMMS?
  • What about overclocking? And what about manual memory tuning? will you be tuning each system to its own capabilities? And what about cooler use? Air, WC, or maybe just stock?
With how many possible branches to base a testing platform on, you just can't satisfy everyone. Makes sense.
I'd advise the little I can here, and say, go with the gut feeling of what can serve the common viewer the most benefit. At times, Its ok to serve a piece of information knowing not everyone can benefit from it, but at least a sizeable amount of readers can.

Extrapolate, my friends... extrapolate.
 
Last edited:
I would love to see a 5700 or 5700XT added.
 
The quality of these tests/benchmarks is over 9000
 
Why would you get flack for going with 3600MHz memory? It is after all what AMD guarantees, not 3800MHz or 4000MHz where it's up to you to see if you can hit it or not.
In all fairness here, I don't think you should even use Samsung memory, but maybe Hynix or Micron instead, as it's generally a lot more affordable.

That said, it seems like 3200MHz CL14 with "tuned" settings can perform just as well on Ryzen as 3600MHz CL16, but I have a feelin 3200MHz CL14 costs more and in general is Samsung B-dies.
 
Last edited:
@W1zzard Great article, really interesting. I have one suggestion though: might PCIe 4.0 support on Zen 3 and Ampere be part of what makes up the difference? It would be really interesting if you re-ran the numbers for one of your test scenarios (3070 or 3090) but limited the PCIe link to either 4.0 x8 or 3.0 x16. I'm obviously not saying I think this is the magic bullet that explains this, but at least it's another interesting variable to look into.
 
Far Cry 5 results are really interesting.
5900x trails 10900K at all resolutions but at 4K, which should not be cpu bound at all and should usually show no difference, Zen3 overtakes 10900K
As a matter of fact in almost all 4K tests 5900X is faster, measurable albeit not big.
 
memory speeds and pcie 4.0 are the cause here.
In any case, mighty impressive how much it took AMD to finally overtake Intel's 2015 CPU.
Too bad Intel isn't doing anything in this space...
 
Interesting review, sure.

I supose that you already know: Robert Hallock told in social networks that Infinity Fabric working at 2 GHz would be easier yo achieve with future newer BIOS so maybe your efforts now retesting all could be innacurate in a future not long ago.

Forums are really actives with a lot of people testing their new Zen3 but also I see some frustration due instability at certain memory speeds, new beta BIOS solves this and runs faster, but got problems on other areas like USB ports, so some people decided to stop testing their OC and memory speed until next stable BIOS version arrives.

I'm still waiting for you habitual and really interesting memory speed comparison on Zen 3 to know exactly where is the best performance and where the best perf/price, but probably those test will wait until next BIOS are released and this platform gets more mature.

Thanks for your time and for your honesty considering that you could did some wrong
 
Last edited:
memory speeds and pcie 4.0 are the cause here.
In any case, mighty impressive how much it took AMD to finally overtake Intel's 2015 CPU.
Too bad Intel isn't doing anything in this space...

I'd say is not too bad for a company that was on the brink of bankruptcy and is 1/10th the size of Intel.
 
Kudos to the author for responding so positively to feedback. For CPU focused testing I would suggest reporting 1% or 5% lows as well. One can argue that they’re far more useful and informative than average fps.
 
one thing not mentioned here, is AMD = 0/null amount of security issues. Intel however... lmao...
Those issues should be already fixed/mitigated through the use of Windows 20H2 and whatever BIOS updates may provide. So, in theory, W1zz is already testing with all the security patches applied
 
Amazing article, good job.

My interpretation is that right now amd or intel wont make a difference because everyone will be GPU limited. But as time goes by, GPUs will become more powerful, the point at which the games operate will move increasingly to the left in the "GPU/CPU limited" axis, have an increasinlgy higher chance in operating in the range where AMD has a significant advantage.

But, it will not be any time soon.

So for people that update often, either CPU will be about equal (well I guess thats almost always the case?). For others that like to keep for as long as possible (7+ years), AMD CPUs should be seriously considered.

EDIT: Oh, and 3600 C16 is perfectly fine. Beyond that its hard to find and too expensive.
 
How about a 2 vs 4 sticks memory comparison ?
I would love to see a 5700 or 5700XT added.
would shed a little more light on whether Ampere is the solution, or Turing is the problem.
Great ideas, keep them coming, I'll test them all over the weekend

Edit: note to self, from Jonny via email, force Zen 3 + Ampere to Gen 3 to more clearly see PCIe 3 vs 4
 
Last edited:
Those issues should be already fixed/mitigated through the use of Windows 20H2 and whatever BIOS updates may provide. So, in theory, W1zz is already testing with all the security patches applied

nah I didn't mean that homie. i just meant Lisa Su so ninjutsu she doesn't need security patches. get skill Intel executives ~ long live the engineers ~
 
Really. It's measly single digit % difference in the first place, likely completely imperceptible in real world. Hence should be discarded as meaningless, yet some folks blew it out of proportion.
 
Don't know why but Zen 3 seems to run best with 4 sticks of fast ram... I've played with 2 and 4 stick configuration after watching Gamers Nexus video and achieved 0-9% difference at 1080p low settings depending on a title. I compared 2x8gb and 4x8gb Patriot Viper Steel DDR4-3733 DIMM CL17 (both combos with manually fine tuned timings) on friend's 5600X & my Gigabyte B550 Aorus PRO MB. Worth considering 4x8 combo if you're buying 32 gigs of ram.
 
Literate people had no problems understanding the original review, this one is for noobs.
 
There is no Best Case Scenario. AMD's ZEN3 is the PC Gaming Winner hands down, no ifs or buts about it. ;):D:)

Don't know why but Zen 3 seems to run best with 4 sticks of fast ram... I've played with 2 and 4 stick configuration after watching Gamers Nexus video and achieved 0-9% difference at 1080p low settings depending on a title. I compared 2x8gb and 4x8gb Patriot Viper Steel DDR4-3733 DIMM CL17 (both combos with manually fine tuned timings) on friend's 5600X & my Gigabyte B550 Aorus PRO MB. Worth considering 4x8 combo if you're buying 32 gigs of ram.
They explained by in Hardware Unboxed on YouTube.
 
Great ideas, keep them coming, I'll test them all over the weekend

Keep in mind of single rank and dual rank memory sticks. GN just tested 2 vs 4 memory sticks and the results were off because of differences of memory rank during testing.

Edit: Keep in mind that most 8gb sticks are single rank as in most 16gb kits (8gbx2).
 
Thank you for the Follow up on this, and sorry for the fan boys sending nasty emails.
 
Back
Top