RTX 4090 & 53 Games: Core i9-13900K vs Ryzen 7 5800X3D

[Tek-Check]

I also want to see the miracle of how a 7950X, reached 95 degrees (maximum allowed), "self-regulates" without losing performance.
It seems that you also missed the lesson on heat dissipation, the W/mm2 chapter.

Read, watch and learn dude, from reviewers and other members here. Don't embarrass yourself out of ignorance and/or arrogance. It's not worth it. There's a lot of heat from new CPUs and GPUs and each company deals with it in a slightly different way. Nvidia has even more heat to deal with on optimized 4 nm node, and so new GPUs are gigantic for this very purpose. As regards to CPUs, AMD has more elegant solution for thermal and power management this time around and Intel uses brute force on i9 to let us know that unrestricted power use beyond 253W will bring you a few petty percentages of performance, but also enormous amount of power and heat to deal with.

7950X hits 95 degrees immediately when MT workload starts and all-core frequency is very stable throughout tests. See GN below.

Watch GN review of 7950X
Both temperature and core behaviour are by design on Zen4 and choice of coolers will make much smaller difference for 7950X/7900X than for i9 and i7 CPUs that suck wapping ~300W in similar workloads. And so, GN found i9 to be the worst power efficiency CPU on the market for continuous MT workloads.
Watch GN review of 13900K
It is i9 that is hardest to cool precisely because it uses insane amount of power in those workloads. In HUB review, i9 was not able to finish a test without thermal throttling because cores hit 100 degrees soon, so soon that it throttles just after a few seconds into the test, even with 420 mm AIO cooler! It loses ~8% of performance in those workloads, which professionals who need strong MT systems for daily workloads will certainly think about very, very carefully, indeed.
This means that if 7950X can save them 20-30 minutes per 8 hours, day-by-day, to do the same job over 13900K, just imagine how many hours of work could be saved over several months of usage. Marginal gains are important in this field, as faster jobs done mean more jobs done over longer period of time. And time is money for them. Plus, it's easier to deal with 355W of system power consumption, than with almost 500W with 13900K. See HUB below.

Watch HUB review of 13900K
How much performance does 7950X lose, you asked? Here is another watch for you, to show that 360 mm AIO makes little difference to 120 mm or air cooler in most cases. If you are on the edge of saving time because you run multiple-hour workloads, 360 mm will save precious seconds and minutes. If not, good air cooler will do the job too. Other members here posted for you more explanations and links regaridng cooling of Zen4. Pretty simple stuff.
HC review of 7950X cooling requirements
Enjoy it and let us know what you think, after digesting relevant information.

I agree with all of your other points, but this one seems iffy. Based on pricing, the 13700k squares off against the 7700x, and the 13600K squares off against the 7600x. These aren't favorable match ups for AMD; the 7600x and 7700x compete well in gaming, but lose hugely in productivity workloads to Raptor Lake at analogous price points.

Price-wise, agreed. Intel wants us to think this way and pick up lower in stack Zen4 CPU to compare with higher in stack RPL. Of course 7700X looks worse than 13700K. No doubt. However, once prices change, 7700X will compare with 13600K, as it should.

In terms of power efficiency, core count and workload purpose, I disagree. It's amazing what Zen4 CPUs can do with less cores than Intel parts. 16-core CPU matches 24-core CPU in productivity workloads. The same for 12-core 7900X against 16-core i7. Those are high-end consumer CPUs, so they have broader application, far beyond gaming and casual MT. 7700X loses in productivity only ~5% to 13600K, which is negligible. That's why I suggested to compare the top three Zen4 with top three Raptor Lake, rather then falling into a price trap comparison. 7700X needs to be $320-330 to be very competitive with i5 on price.

If I was a professional who uses MT system daily, 8-10 hours a day over entire year, I'd buy 7900X over 13700K and 7950X over 13900K system, for simple reasons. Easier to cool, definitely more power efficient and future-proof platform. More expensive initially, true, but gains come over time and initial investment pays off. I only need to slot in a single CPU at Zen5 and/or Zen6 point and call it a day, rather than buying entire system again.

And if the low-to-mid range Zen 4 CPUs* are going to hang their hats on gaming perf, well, they also lose pretty convincingly to AMD's own 5800x3d, based on platform cost. AMD needs to release some non-X SKUs, or something. I'm sure AM5 will come into its own eventually, but it ain't there yet.

Agreed.

Gaming enthusiasts who don't plan to buy an extremely expensive graphics card are almost always best off buying a CPU in the $150-200 range. Frankly, an R5 5600 or an i5-12400 still offers vastly better value for most gamers than any of the new shiny stuff from either AMD or Intel.)

Agreed.

 

[Gica]

Read, watch and learn dude

Agreed.

The same as you.
How to have the same performance in renders when the frequency decreases, only you can know.
Agreed?

 

[HTC]

The same as you.
How to have the same performance in renders when the frequency decreases, only you can know.
Agreed?

I've seen that graph before: it simulates "lower quality coolers" by having different fan speeds.

If only a similar test were to be done with the 13900K, just to see if the impact from "a lower quality cooler" was smaller, the same or higher: i'd bet it would be higher because the CPU already gets hotter to start with.

 

[Gica]

The gentleman claims that the 7950X maintains its performance when it reaches 95 degrees. I knew that it reduces its frequency (default performance) when it reaches this critical threshold.

Very fresh. My "hot Intel" 12500 beat by 10 W, average CPU+GPU. (printscreen)
Hero!

Seriously, it smells of frustration among red fans because the 13900K took the crown. If you don't like it, don't buy it. Simple! I don't think that in any competition, apart from the position occupied, anyone cares how much the winner consumes or what he eats.

 

[HTC]

The gentleman claims that the 7950X maintains its performance when it reaches 95 degrees.

In ST scenarios he's mostly right, but NOT in MT scenarios, like the graph you posted earlier clearly states.

The same should be right for the 13900K. In fact, it SHOULD be even worse because the "starting full load temp" is higher but, until a similar test is conducted on the 13900K, we won't know for sure.

@W1zzard proved that 95º for the 7000Zen 4 CPU series wasn't a problem and that, even with "weaker coolers", the CPU wouldn't overshoot it's temperature though IT WOULD lose performance. I wish he'd do a similar test for the 13900K so that we could compare the impact the "various coolers" he used with the 7950X have on 13900K's performance.

 

[Tek-Check]

A fair comparison would be the review with W11 because it does not disadvantage the platforms, but W10 can create problems for the Intel platform because it cannot efficiently manage P and E cores.

Hardware Unboxed has already compared Intel's performance in Win10 and Win11. Watch it.
Plus, it was Thread Director that had more problems with 12th gen rather than OS. Now, Thread Director 2.0 has improved in 13th gen.

The same as you.
How to have the same performance in renders when the frequency decreases, only you can know.
Agreed?

One, I have already posted performance differences with different coolers in #165
You replied to it in #169, but you completely ignored cooler and performance chart.

Two, I also gave you a link to the very video from which the chart was taken in #176
Have you seen it? I am asking you to engage with reference content and pay more attention to what is posted.

Both @W1zzard and Hardware Canucks found negligible differences in MT rendering performance.
In Blender, in 15 minute workload, you will save 14 seconds on 360 mm AIO versus good air cooler.
Would you like the same chart to be posted again or will you read #165 and watch the video from #176?

The TPU chart you posted is for academic purposes. No one sane will ever run in everyday life 32-threads in Blender with Noctua at 40%. It does not work like that. What is meaningful is to look at just how small frequency drop is from 420 mm AIO to Noctua single tower air cooler at 80-100%, only 70-80 MHz on all threads. This is where those additional seconds occur. The difference is still negligible and taps into W1zzard's conclusion, which you should read again. The CPU will attempt to run at 95 degrees with any cooler and if the cooler is not as performant, the CPU will drop the frequency a bit to stay within temperature envelope.

The same does not happen with 13900K. In Blender, it hits thermal throttling at 100 degrees just after a few seconds from the start of workload on 360 and 420 mm AIO. No matter how much additional power you throw at it, that's what you get - a power hungry beast that is extremely difficult to cool at those extended professional workloads, with performance drop up to 8% as it approaches 300W. You want more power and performance? Use LN cooler. Just don't burn your hands, please.

 

[Gica]

OK. We pay $1000+ for the cpu+mobo+RAM, we pay another $1000+ for the video card... at least $3000 for the system, but we buy a cooler under $100 and complain about the consumption.
Which part will add a comma, because it sounds like hell to me? Those who want the best do not look at costs or consumption.
What would be the difference between 7950X and 13900K in the normal use of a home user? A kilowatt a week? Two a month? You talk as if the owner of an AMD has no expenses, his processor produces energy and sells it on the free market.
Do you understand that the reviews test the extremes and you will not meet such consumption in real life? It seems not. If the 13900K consumes significantly less than the 7950X in single-threaded mode (automatically more economical in idle), the average consumption depends only on the user and with the watt performance ratio even as you speak, you may have a question, because it is sensitively equal: the 13900K can win some applications , may lose in others.

Below you have a 12500, used for an hour only for internet. 3.9 W/h average, CPU+igp. Show me a Ryzen that consumes that much or less, please.

 

[Tek-Check]

OK. We pay $1000+ for the cpu+mobo+RAM, we pay another $1000+ for the video card... at least $3000 for the system, but we buy a cooler under $100 and complain about the consumption. Which part will add a comma, because it sounds like hell to me? Those who want the best do not look at costs or consumption.

I have tried to explain this in the post #176 above. Have you missed it? Working people who want the best DO look at costs, consumption, efficiency and time-saving. It is a prudent thing to look into for your business. They are not impressed by the brute force of CPU. From #176:

"It is i9 that is hardest to cool precisely because it uses insane amount of power in those MT workloads. In HUB review, i9 was not able to finish a test without thermal throttling because cores hit 100 degrees soon, so soon that it throttles just after a few seconds into the test, even with 420 mm AIO cooler! It loses ~8% of performance in those workloads, which professionals who need strong MT systems for daily workloads will certainly think about very, very carefully, indeed. This means that if 7950X can save them 20-30 minutes per 8 hours, day-by-day, to do the same job instead of using 13900K, just imagine how many hours of work could be saved over several months of usage. Marginal gains are important in this field, as faster jobs done mean more jobs done over longer period of time. And time is money for them. Plus, it's easier to deal with 355W of system power consumption than with almost 500W with 13900K."

What would be the difference between 7950X and 13900K in the normal use of a home user? A kilowatt a week? Two a month? You talk as if the owner of an AMD has no expenses, his processor produces energy and sells it on the free market.

You are putting words into my mouth. I did not say that. I say that "normal" home user does not need any of these two CPUs. It's a huge overkill in both cases.

Do you understand that the reviews test the extremes and you will not meet such consumption in real life? It seems not. If the 13900K consumes significantly less than the 7950X in single-threaded mode (automatically more economical in idle), the average consumption depends only on the user and with the watt performance ratio even as you speak, you may have a question, because it is sensitively equal: the 13900K can win some applications , may lose in others.

Extreme testing cases are an indicator for how those CPUs would behave in daily usage in different fields, businesses and industries.
And I did say that users need to buy a CPU that would best fit their case use and workloads. Read my posts again.
When the two top CPUs perform roughly equally in extended MT workloads, 7950X is the system to go to, for the reasons I posted above. It is that simple.

If I was a professional who uses MT system daily, 8-10 hours a day over entire year, I'd buy 7900X over 13700K or 7950X over 13900K system, for simple reasons. Easier to cool, definitely more power efficient, no throttling and AM5 is future-proof platform. More expensive initially, true, but gains come over time and initial investment pays off. I only need to slot in a single CPU at Zen5 and/or Zen6 point and call it a day, rather than buying entire system again for my business or home use.

Below you have a 12500, used for an hour only for internet. 3.9 W/h average, CPU+igp. Show me a Ryzen that consumes that much or less, please.

There are many efficient Intel CPUs. 12400 is even better. But 13900K is not one of them and it was never designed to be in those extended MT workloads. Man, they threw 24 cores (!) with power usage never seen before, to be competitive against 16-core 7950X in those workloads.

 

[wheresmycar]

I have tried to explain this in the post #176 above. Have you missed it? Working people who want the best DO look at costs, consumption, efficiency and time-saving. It is a prudent thing to look into for your business. They are not impressed by the brute force of CPU. From #176:

"It is i9 that is hardest to cool precisely because it uses insane amount of power in those MT workloads. In HUB review, i9 was not able to finish a test without thermal throttling because cores hit 100 degrees soon, so soon that it throttles just after a few seconds into the test, even with 420 mm AIO cooler! It loses ~8% of performance in those workloads, which professionals who need strong MT systems for daily workloads will certainly think about very, very carefully, indeed. This means that if 7950X can save them 20-30 minutes per 8 hours, day-by-day, to do the same job instead of using 13900K, just imagine how many hours of work could be saved over several months of usage. Marginal gains are important in this field, as faster jobs done mean more jobs done over longer period of time. And time is money for them. Plus, it's easier to deal with 355W of system power consumption than with almost 500W with 13900K."

You are putting words into my mouth. I did not say that. I say that "normal" home user does not need any of these two CPUs. It's a huge overkill in both cases.

Extreme testing cases are an indicator for how those CPUs would behave in daily usage in different fields, businesses and industries.
And I did say that users need to buy a CPU that would best fit their case use and workloads. Read my posts again.
When the two top CPUs perform roughly equally in extended MT workloads, 7950X is the system to go to, for the reasons I posted above. It is that simple.

If I was a professional who uses MT system daily, 8-10 hours a day over entire year, I'd buy 7900X over 13700K or 7950X over 13900K system, for simple reasons. Easier to cool, definitely more power efficient, no throttling and AM5 is future-proof platform. More expensive initially, true, but gains come over time and initial investment pays off. I only need to slot in a single CPU at Zen5 and/or Zen6 point and call it a day, rather than buying entire system again for my business or home use.

There are many efficient Intel CPUs. 12400 is even better. But 13900K is not one of them and it was never designed to be in those extended MT workloads. Man, they threw 24 cores (!) with power usage never seen before, to be competitive against 16-core 7950X in those workloads.

I haven't read this but good luck in knocking some common sense into GICA.... he's one of the strangest characters on here, pays no attention to member input, derails the topic in hand to conceal his lack of understanding on any given situation and will repeat everything again no matter how well you present FACTs/etc.

We couldn't convince Gica that its not the "actual market" which determines the MSRP of GPUs but the manufacturers themselves. If that doesn't sink home, NOTHING WILL!!

 

[Tek-Check]

Well, back to the topic. 5800X 3D is an amazing CPU for gamers and far, far better value for buck than 13900K will ever be. Nothing more to add.

 

[wheresmycar]

Well, back to the topic. 5800X 3D is na amazing CPU for gamers and far, far better value for buck than 13900K will ever be. Nothing more to add.

I hope he's not suggesting otherwise lol. Wouldn't touch a 13900K for gaming not even with a 10-foot telescopic pole which stretches to 20-foot hehe.

Can't blame the ordinary folk though... the amount of times i've read "X900/XX900 BEST GAMING CPU" certainly doesn't help people who don't know any better. The 13600K or 13700K is where gaming options should draw the line.

 

[Tek-Check]

The 13600K or 13700K is where gaming options should draw the line.

Even 12400 is great in gaming and great value.

 

[wheresmycar]

Even 12400 is great in gaming and great value.

can't argue with that! TBH i'm all for previous Gen CPUs at this point of time. AM5 went AWOL and RPL is juicing it too hard. Neither tickle my fancy! The 12400/12600K/5600/5600X/5700X is plenty for any gamers dream PC... although ive got my eyes set on the little more pricier 5800X3D.

 

[Nopa]

@W1zzard Is this 13900K's power consumption data from HWUB included MB, RAM, Fans, and even GPU too?
I mean 493W seems a little off somehow...

 

[Hervon]

Read the title : SYSTEM power consumption.

 

[Tek-Check]

View attachment 269359
@W1zzard Is this 13900K's power consumption data from HWUB included MB, RAM, Fans, and even GPU too?
I mean 493W seems a little off somehow...

It's without GPU.
It's bonkers with 13900K! So bad for i9 that DerBauer spent three days tweaking the CPU, e.g undervolting etc. to try to tame it into meaningful usage.

 

[Gica]

I haven't read this but good luck in knocking some common sense into GICA.... he's one of the strangest characters on here, pays no attention to member input, derails the topic in hand to conceal his lack of understanding on any given situation and will repeat everything again no matter how well you present FACTs/etc.

We couldn't convince Gica that its not the "actual market" which determines the MSRP of GPUs but the manufacturers themselves. If that doesn't sink home, NOTHING WILL!!

Or I ignore some pro AMD characters and look for information from several channels.

Intel Core i9 13900K: Impact of MultiCore Enhancement (MCE) and Long Power Duration Limits on Thermals and Content Creation Performance​

What is the difference in performance between AIO and $100 air cooler, even with MCE ON (overclocking)??? Zero or ...zero????

P.S. The source of the review is of high quality and covers an important area of the reason for buying a processor.
Note: let's accept +/- 2% tolerance margin, however.

 

[Tek-Check]

Or I ignore some pro AMD characters and look for information from several channels.

Intel Core i9 13900K: Impact of MultiCore Enhancement (MCE) and Long Power Duration Limits on Thermals and Content Creation Performance​

What is the difference in performance between AIO and $100 air cooler, even with MCE ON (overclocking)??? Zero or ...zero????
P.S. The source of the review is of high quality and covers an important area of the reason for buying a processor.
Note: let's accept +/- 2% tolerance margin, however.

Dude, did you read their closing words?
"We are going to continue to disable MCE and enforce the P1/P2 power limits in our testing and workstations, but we want to make it clear that there is no right or wrong answer here. It is just a matter of tradeoffs, and the 30-40C temperature increase simply does not constitute an acceptable tradeoff for us as a workstation system integrator." Therefore, PudgetSystems are NOT happy to run 13900K in a hotter mode, regardless of cooler used. They are happy with performance drop 10-20% in specific workloads in order to run the CPU at acceptable temperature and power use. Simple.

End of story. Focus on the thread topic. 5800X 3D is far better value CPU for gaming than 13900K will ever be. That is the topic of this thread.

 

[Gica]

"In most of the workloads we tested, it is very clear that using Precision Boost Overdrive (PBO) and Core Performance Boost (CPB) is not worth it. Photoshop, Lightroom Classic, Premiere Pro, After Effects, and DaVinci Resolve all showed no difference in performance when we disabled these settings from their "Auto" default, yet the CPU temperature dropped as much as 30C, or even 40C!"
Review for 7950X, red boy.

5800X3D costs as much as 13600KF (7600X). If you opt for it, you invest in DDR4 memories in 2022, red boy. About the same performance in gaming, effectively destroyed by the new processors in the other applications.
13900K is clearly on topic. I do both, red boy.

If the 13900K reaches the maximum temperature in only two heavy loads, the 7950X reaches it in all. With the same cooler!
I think it was also discussed here about watt/mm2 and that ryzen needs a more efficient cooler than 13th because the dissipation surface is smaller. Elementary physics.

 

[Tek-Check]

"In most of the workloads we tested, it is very clear that using Precision Boost Overdrive (PBO) and Core Performance Boost (CPB) is not worth it. Photoshop, Lightroom Classic, Premiere Pro, After Effects, and DaVinci Resolve all showed no difference in performance when we disabled these settings from their "Auto" default, yet the CPU temperature dropped as much as 30C, or even 40C!"
Review for 7950X, red boy.

I have never suggested that 7950X should be overclocked with PBO or other tool to improve performance. Those tools are OFF by default on Zen4 systems. I'd never do it, as the CPU is already maxed out enough out-of-the-box, and I agree with what Pudget System said about those tools. Totally not worth it.

However, Intel allows motherboard partners to enable MCE or whatever name they use, so on several MSI and Gigabyte boards you can get unrestricted power use as "auto", out-of-the-box feature, without even knowing. And then reviewers benchmark with such motherboards, often without checking this.

13900K should come with motherboard set by deafult to Intel's limit of 253W, aka PL1=PL2. Anything beyond this in default settings is pure cheating by motherboard vendors for benchmark wins and sale purposes. If any vendor does the same for any AMD system, they should also be condemned immediately. I believe that AMD is pretty strict about those tools being OFF by default.

It is not surprising that 13900K attracts bad press for power usage and thermal throttling when motherboard vendors cheat the logic of PL2, in reality offering MCE by default, and allowing CPU to run at 300W out-of-the-box. Hence, by default on many motherboads, users are forced to buy high end cooler to avoid thermal throttling in heavy and continuous MT tasks, which is not the case with 7950X. Intel has only themselves to blame for this lax and careless power policy.

If the 13900K reaches the maximum temperature in only two heavy loads, the 7950X reaches it in all. With the same cooler! I think it was also discussed here about watt/mm2 and that ryzen needs a more efficient cooler than 13th because the dissipation surface is smaller. Elementary physics.

I have Intel, AMD and ARM systems at home and at work and happily use all of them. I don't get angry.

5800X3D costs as much as 13600KF (7600X). If you opt for it, you invest in DDR4 memories in 2022

That's exactly why this CPU is the most popular to buy right now. I don't buy it, as I don't need it, but thousands upon thousands of AM4 owners or builders can make a simple upgrade and extend the longevity of their systems for another couple of years, or build an affordable system they will enjoy. Pretty simple.

Yet, again, stick to the topic. 5800X 3D is both very competitive and more affordable than 13900K for gaming. No brainer.

 

[HTC]

That's exactly why this CPU is the most popular to buy right now. I don't buy it, as I don't need it, but thousands upon thousands of AM4 owners or builders can make a simple upgrade and extend the longevity of their systems for another couple of years, or build an affordable system they will enjoy. Pretty simple.

I'm using a 2600X and a 5800X3d is going to be my next upgrade.

Unfortunately, it will have to wait because the cheapest i can find it here in Portugal is 394€ ... since i'm not in any rush, i don't mind the wait ...

 

[Solaris17]

Really would've appreciated tests for Final Fantasy XIV, Valorant and Apex. I know you mentioned in a previous CPU specific review it's hard to bench games like that for comparison because the games or on different patch versions at different CPU release dates. But in a straight X vs Y I think it's a great opportunity to do it.

I pull between 100-190 FPS on a 13900k and 4090 at 4k settings cranked depending on zone in XIV. That is with my machine driving x3 4k displays though. so you might get more.

 

[Gica]

Bla bla bla

Look for the fine print in the reviews and you will see that those who want a processor for specific tasks can correctly opt for only one of them. For example, if you want a processor for video editing and encoding, you are definitely wrong if you choose AMD over Intel 13th (hint or hint and read the conclusions in Premiere Pro)). AMD also has advantages in other applications and then you are wrong if you choose Intel.

In short and on topic:
1. Whoever buys Intel is not crazy, nor stupid. The one who criticizes his choice is stupid. The same rule applies to AMD.
2. 5800X3D is an excellent AM4 upgrade for gaming only, not to invest in a "new" platform. It's ok only if you change the processor and only have gaming as your target.
It costs as much as the 5900X, I don't think you will notice differences in gaming, but the 5900X destroys the 5800X3D in the rest.

 

[Tek-Check]

Once again, temper your language and stop going personal with members. If not, your posts will be reported and removed.
You will never successfully argue your case if you post personal comments. You have to learn to refrain from such comments and disagree with respect.

Secondly, I have never said that 13900K is not suitable for video encoding and editing. You are again putting words into my mouth. I was specific about which workloads are less suitable, as shown in testing by others posted before. GN and HUB tested default configurations sold to end user, as many motherboards come unlocked to allow 300W usage. As I said, you only have Intel to blame for this reckless power policy, and not us members who observe the results of such policy. Do not blame the messengers.

If you use 13900K, enjoy it, but have decency to allow voices of criticism towards Intel and their partners for shoving the hot cake 300W CPU down the throat of end user from the get go. If you are ok with that policy, that's fine, but you are not in a position to go personal with members who point out this reckless power policy. On a tech forum, you are only allowed to argue your case with respect, without labelling anyone with insults or similar. If you are unable to argue your case respectfully without calling people names, this is not a good place for you. Move on.

Again, 5800X 3D is ~15% faster in gaming than 5900X, so it's a great option both as single component upgrade and CPU for someone who wants to build AM4 gaming system. No matter what you say about it, it's not going to fly, especially when you use emotive language such as "5900X destroys 5800X 3D in the rest" to make irrelevant MT comparison for gaming topic. Ridiculous. Someone who mainly games does not care about 15% less performance on average in productivity applications. Finally, 5800X 3D is, as said many times, undeniably a better value for buck in gaming than slightly faster 13900K, as the results in this thread show. There is nothing more to add to it.

 

[95Viper]

Stay on the topic.
Stop the insults.
Report problems... do not become a problem.
Read the Guidelines and follow them.

Any more of the <arguing/name calling/guideline violations> will incur warnings and thread bans.