How is Intel Beating AMD Zen 3 Ryzen in Gaming?

[Super XP]

At most , I can tell you what to do to have an Intel CPU better in games than AMD Ryzen Zen 3 :
- for 8 cores battle : Intel I7-10700k clocked at 5.1-5.2Ghz all cores, at least 4.5Ghz cash speed, RAM with at least 4133Mhz speed , tight adjusted timings (16 CAS for e.g)
- for higher cores : Intel i9-10900k clocked at 5.3-5.4Ghz all cores , Cash speed at least 4.8Ghz, RAM at least 4266Mhz , tight timings.

Also important : 4 ranks of the RAM . (2x16Gb dual rank or 4 x 8 Gb single rank , for e.g)

Nowadays Intel runs too hot and continues to suffer from security vulnerabilities. They've been slowly patching what they can but it is what it is.

 

[Sisyfos]

I absolutely agree that we need these "normal gaming" benchmarks. If I have a good GPU and a great CPU, I am not going to game at 720p low settings. I save those settings for the occasions when my GPU is broken or I am on the road with an ultraportable. When I have my good gear available I use my 1440p 144Hz large monitor and Ultra settings and enjoy the game play and the visuals. If there is no difference between CPUs on those scenarios, I want to know it before I spend unnecessary Euros on that front

 

[Super XP]

I absolutely agree that we need these "normal gaming" benchmarks. If I have a good GPU and a great CPU, I am not going to game at 720p low settings. I save those settings for the occasions when my GPU is broken or I am on the road with an ultraportable. When I have my good gear available I use my 1440p 144Hz large monitor and Ultra settings and enjoy the game play and the visuals. If there is no difference between CPUs on those scenarios, I want to know it before I spend unnecessary Euros on that front

Makes total sense.
BENCHMARKS should be done in 1440p and 4K. To test the entire test system on real performance figures.

 

[tirasoft]

Nowadays Intel runs too hot and continues to suffer from security vulnerabilities.

That is true . Intel definitively lags in terms of power efficiency. And to achieve results you have to supply lot of power compared to AMD.
BUT, due to the Intel architecture, Intel CPU's are getting better if you overclock them , the ONLY limit is the heat ... Which you cannot say if you refer to AMD.

AND, there are practical ways for better heat handling : liquid metal apply between die and IHS , copper IHS, direct die cooling ....

And, the latest , the Intel/Cooler master thing: ML360 SUB-ZERO POWERED INTEL® CRYO COOLING TECHNOLOGY . (That's 5.6 - 5.8 Ghz CPU speed )
Which AMD Fan boys mock as it's a desperate Intel thing to keep up with Ryzen , lol

 

[Mtom]

"I'm also relieved that I found out what was "wrong" with my Zen 3 gaming performance results—nothing. I simply tested in a scenario that's not the most favorable for AMD. "

LOL testing the CPUs in a GPU limited environment...wow...

 

[Super XP]

That is true . Intel definitively lags in terms of power efficiency. And to achieve results you have to supply lot of power compared to AMD.
BUT, due to the Intel architecture, Intel CPU's are getting better if you overclock them , the ONLY limit is the heat ... Which you cannot say if you refer to AMD.

AND, there are practical ways for better heat handling : liquid metal apply between die and IHS , copper IHS, direct die cooling ....

And, the latest , the Intel/Cooler master thing: ML360 SUB-ZERO POWERED INTEL® CRYO COOLING TECHNOLOGY . (That's 5.6 - 5.8 Ghz CPU speed )
Which AMD Fan boys mock as it's a desperate Intel thing to keep up with Ryzen , lol

I love competition. I just believe the gaming performance crown now belongs to AMD. It's going to go back and forth as it should, that's what keeps the industry healthy and honest.

 

[tirasoft]

I just believe the gaming performance crown now belongs to AMD

I've just said few posts above it is not the case.

For e.g. an overclocked I9-10900k with the right tuning effectively destroys AMD Ryzen 5900x or 5950x in some games in terms of FPS . Like for example Shadows of the Tomb Raider internal Benchmark ... (this is benchmark known as CPU intensive, the most of the OC-ers take it as the best comparison)
And that's true whatever you do to the AMD in therms of overclocking ...

 

[W1zzard]

Can someone explain how we get over 100% load on the x-axis? I can not really understand the part: towards to the right GPU limited and left CPU limied. It is relative to what?

The X axis does not represent "GPU load". It represents how complex the frame is, it's an arbitrary choice, I just had to set the values on the scale to something, didn't want to remove the X axis labels completely

"I'm also relieved that I found out what was "wrong" with my Zen 3 gaming performance results—nothing. I simply tested in a scenario that's not the most favorable for AMD. "

LOL testing the CPUs in a GPU limited environment...wow...

Are you saying AMD CPUs only show favorable results when CPU limited? The reality of AAA PC gaming is that you are GPU limited. Yes I know 345498564586 people play MOBAs, which runs on a potato

 

[Vayra86]

Very nice analysis. Well done

 

[RandallFlagg]

While I think this article is great for its technical content, it really didn't address the contention in the forums which had nothing to do with 10900K vs 5900X at all, at least from any posts I ever saw. I have not seen a single person advocate the 10900K as a good deal or a better performer since Zen 3 came out (see below post), and certainly no one has said that Intel Gen 10 was as good as Zen 3 in a 1:1 core to core comparison.

In other words, a straw man seems to have been made.

Also of concern, since this is a synthetic measure for rendering using an engine, It's quite likely that you wound find a 3900X outclasses a 10900K on those same tests - by virtue of the 3900X having more cores.

See post below, and I quote :

"..the 5800X / 5900X are clear winners in overall performance at the high end. If one is looking for a $500ish CPU right now for a new build, these two seem to be the clear front runners. The 10850K and 10900K just can't compete at their current prices."

The contention that I saw was not a 10C / 20T Gen 10 vs a 12C / 24T Zen 3. It was the $300 5600X vs a $300 10700 or $350 10700K.

I mean really, this article is not very enlightening except to tell us a 5900X beats a 10900K in games when paired with a 3090 and DDR4-3600. We pretty much already knew that from other game sites.

The problem of course is going to be comparable performance vs already benched chips like the Intel 10th Gen and Zen 2, so it would be a lot of work. Basically have to re-bench most of what was released in the last 18 months, so like 5x more work than just benching the Zen 3 using the standard test bed. I do agree it would be more relevant.

Having said that, the 720P benchmarks with a 2080Ti are probably representative of max performance with little or no GPU limiting factors.

Compared to the 10700K - the results aren't much different with only 1% fps gain from a 5800X or 5900X and essentially a tie with a 5600X in games. The 10700K actually beats the 5600X on CPU tests (moar cores, higher freq).
I thought the 5600X would be a winner here, but IMO the 10700K is a much better deal as it generally outperforms it and has much more OC headroom (you can easily get another 5% out of the 10700K, widening that gap). It can also benefit from higher speed memory, as most Intel chips can.

That said, the 5800X / 5900X are clear winners in overall performance at the high end. If one is looking for a $500ish CPU right now for a new build, these two seem to be the clear front runners. The 10850K and 10900K just can't compete at their current prices.

Final statement: Rocket Lake is probably going to destroy this, Zen 3 is too close to Gen 10 performance.

 

[R0H1T]

And, the latest , the Intel/Cooler master thing: ML360 SUB-ZERO POWERED INTEL® CRYO COOLING TECHNOLOGY . (That's 5.6 - 5.8 Ghz CPU speed )
Which AMD Fan boys mock as it's a desperate Intel thing to keep up with Ryzen , lol

Well it's not Cryo anything, it's TEC. Either Intel marketing got to you or you're obviously ignoring how inefficient & ineffective that is compared to top end AIO or custom loops.

 

[Sisyfos]

"I'm also relieved that I found out what was "wrong" with my Zen 3 gaming performance results—nothing. I simply tested in a scenario that's not the most favorable for AMD. "

LOL testing the CPUs in a GPU limited environment...wow...

Testing CPUs in REAL environment. I have heard so much talk lately how we would be getting "better fps" with all these CPU upgrades. Yeah, not going to happen in any realistic gaming scenario.

 

[Searing]

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.

He writes this entire article showing AMD up to 25 percent ahead, and you get from that "single digit"...

Anandtech's 43 percent or whatever Civilization turn time result was the highest difference.

 

[tirasoft]

It doesn't matter how you try to skin it the article reads Why and not if.

I've just read the article, sorry , just missed the link previoulsy ...

Here is my opinion:
The conditions CPU's were put in to do the benchmark are biased, clearly put Intel in disavantage in a lot of scenarios on internet / Youtube/ etc.
Le me tell you why :
As I said, at STOCK level, AMD is better than Intel.
BUT, don't you agree on this: if you do some performance comparison, you have top bring the CPU's at the most they are capable of ....
In this respect, there are some differences between Intel and AMD :
1. performance scales with the cash speed on Intel . Not the case with Infinity Fabric of AMD which is limited . Therefore, why do you leave the cash speed at stock for Intel . Set it to 5Ghz (perfectly possible ) and you will have a sizable performance boost for Intel.
2. Again, memory . Of course, it is convenient for AMD to stop the comparison at 3800Mhz RAM .... they cannot go too high . On the other hand, you can use on Intel platform a 4400 Mhz speed, with somewhere around 37nm latency with ease. Why don't you give the Intel the best memory it can handle when do the comparison ? (biased ...maybe?) . And more.... if you use 4 ranks of memory compared to 2 or whatever , there is a 5% -10% more or less performance bump ... Again, why wouldn't you give the best scenario Intel is capable in terms of pairing with RAM and setting ??
3. Of course, overclocking AMD is not the same as overclocking Intel . And I mean Intel has a better overclockability. Therefore do the comparison on, let's say 5.3Ghz all cores INtel i9-10900k (perfectly possible)

 

[windwhirl]

"I'm also relieved that I found out what was "wrong" with my Zen 3 gaming performance results—nothing. I simply tested in a scenario that's not the most favorable for AMD. "

LOL testing the CPUs in a GPU limited environment...wow...

Be that as it may, right now there are only two GPUs above the 2080 Ti, the 3080 and the 3090. And if all the complaints are true, then very few people have actually been able to acquire one of them. So it's still a rather realistic scenario testing with a 2080 Ti.

 

[hurakura]

Intel fanboys

 

[RandallFlagg]

@tirasoft post is very germane to this.

These are basically overclock tests that stop at the limits of the AMD platform, or completely ignore OC items on Intel because they aren't possible or provide no benefit on AMD.

I've just read the article, sorry , just missed the link previoulsy ...

Here is my opinion:
The conditions CPU's were put in to do the benchmark are biased, clearly put Intel in disavantage in a lot of scenarios on internet / Youtube/ etc.
Le me tell you why :
As I said, at STOCK level, AMD is better than Intel.
BUT, don't you agree on this: if you do some performance comparison, you have top bring the CPU's at the most they are capable of ....
In this respect, there are some differences between Intel and AMD :
1. performance scales with the cash speed on Intel . Not the case with Infinity Fabric of AMD which is limited . Therefore, why do you leave the cash speed at stock for Intel . Set it to 5Ghz (perfectly possible ) and you will have a sizable performance boost for Intel.
2. Again, memory . Of course, it is convenient for AMD to stop the comparison at 3800Mhz RAM .... thay cannot go too high . On the other hand, you can use on Intel platform a 4400 Mhz speed, with somewhere around 37nm latency with ease. Why don't you give the Intel the best memory it can handle when do the comparison ? (biased ...maybe?) . And more.... if you use 4 ranks of memory compared to 2 or whatever , there is a 5% -10% more or less performance bump ... Again, why wouldn't you give the best scenario Intel is capable in terms of pairing with RAM and setting ??
3. Of course, overclocking AMD is not the same as overclocking Intel . And I mean Intel has a better overclockability. Therefore do the comparison on, let's say 5.3Ghz all cores INtel i9-10900k (perfectly possible)

 

[tirasoft]

Well it's not Cryo anything, it's TEC. Either Intel marketing got to you or you're obviously ignoring how inefficient & ineffective that is compared to top end AIO or custom loops.

I am perfectly aware of the mismach in terms and of what a TEC is.
If you say I am ignoring facts , please tell me an example where a custom loop liquid cooling can do a 5.6-5.8Ghz on a i9-10900k ..... as the aboved mentioned TEC AIO can do ....

 

[Steevo]

I've just read the article, sorry , just missed the link previoulsy ...

Here is my opinion:
The conditions CPU's were put in to do the benchmark are biased, clearly put Intel in disavantage in a lot of scenarios on internet / Youtube/ etc.
Le me tell you why :
As I said, at STOCK level, AMD is better than Intel.
BUT, don't you agree on this: if you do some performance comparison, you have top bring the CPU's at the most they are capable of ....
In this respect, there are some differences between Intel and AMD :
1. performance scales with the cash speed on Intel . Not the case with Infinity Fabric of AMD which is limited . Therefore, why do you leave the cash speed at stock for Intel . Set it to 5Ghz (perfectly possible ) and you will have a sizable performance boost for Intel.
2. Again, memory . Of course, it is convenient for AMD to stop the comparison at 3800Mhz RAM .... they cannot go too high . On the other hand, you can use on Intel platform a 4400 Mhz speed, with somewhere around 37nm latency with ease. Why don't you give the Intel the best memory it can handle when do the comparison ? (biased ...maybe?) . And more.... if you use 4 ranks of memory compared to 2 or whatever , there is a 5% -10% more or less performance bump ... Again, why wouldn't you give the best scenario Intel is capable in terms of pairing with RAM and setting ??
3. Of course, overclocking AMD is not the same as overclocking Intel . And I mean Intel has a better overclockability. Therefore do the comparison on, let's say 5.3Ghz all cores INtel i9-10900k (perfectly possible)

Are you willing to buy back the CPUs that don't clock well enough? Obviously if Intel could make all their CPUs do it don't you think they would have?

Stock VS Stock which shows the guaranteed performance of each AMD wins, on TDP, on IPC, and overall performance out of the box. Intel still beats AMD in some titles and saw a performance uplift with faster RAM, so really its down to how much money do you want to throw at it. That isnt what 98% of users are going to do, they will buy hardware that matches a budget, and some will tune their timings a little or depending on workload may go for an all core OC vs the dynamic and faster single threaded boost clocks. For those people unless Intel has a specific lead AMD is the better option overall.

 

[Sisyfos]

Are you willing to buy back the CPUs that don't clock well enough? Obviously if Intel could make all their CPUs do it don't you think they would have?

Stock VS Stock which shows the guaranteed performance of each AMD wins, on TDP, on IPC, and overall performance out of the box. Intel still beats AMD in some titles and saw a performance uplift with faster RAM, so really its down to how much money do you want to throw at it. That isnt what 98% of users are going to do, they will buy hardware that matches a budget, and some will tune their timings a little or depending on workload may go for an all core OC vs the dynamic and faster single threaded boost clocks. For those people unless Intel has a specific lead AMD is the better option overall.

I know this "stock vs stock" argument too, it is weird though that if people decide to put Intel at stock why are they picking the "K" parts for the testing and then complain about the price on the other hand :-D In Intel platform you pay a premium to get an overclockable part. If you do not need overclocking there are cheaper alternatives for almost every Intel chip.

 

[RandallFlagg]

I am perfectly aware of the mismach in terms and of what a TEC is.
If you say I am ignoring facts , please tell me an example where a custom loop liquid cooling can do a 5.6-5.8Ghz on a i9-10900k ..... as the aboved mentioned TEC AIO can do ....

You don't even need that.

These tests don't explain how a the #1 5800X on 3dMark TimeSpy charts loses to #31 10700K that is OC to 5.2Ghz, both running 3090s and the graphics card clock/mem near the same. You can do that with a real good air cooler or a sloppy $100 AIO 240mm.

This same pattern is for every one of their graphics benchmarks, with the one exception of FireStrike (normal).

It's probably the cache OC on the Intel + memory, just a guess. And these tests at TPU have AMDs Infinity Fabric overclocked, but I'm sure nothing is done on the Intel rigs.

 

[tirasoft]

Intel fanboys

 

[ExcuseMeWtf]

He writes this entire article showing AMD up to 25 percent ahead, and you get from that "single digit"...

Anandtech's 43 percent or whatever Civilization turn time result was the highest difference.

Article was about disproving claim of Intel supposedly "beating AMD in games" lmao...

And that difference was indeed single digit %...

 

[R0H1T]

please tell me an example where a custom loop liquid cooling can do a 5.6-5.8Ghz on a i9-10900k ..... as the aboved mentioned TEC AIO can do ....

Did you miss the part where it was a golden sample, literally, sent by Intel themselves & plucked especially for LTX? Outside of Silicon lottery, no one comes close to getting the kind of samples Intel would need to burn through to get to such high clocks not to mention even Silicon lottery would only have probably 1/10000 as many chips as Intel actually has access to!

 

[TumbleGeorge]

direct die delided cooling

I see your signature. Yes every user keep delided Intel in home.