Intel Core i5-10600K vs. AMD Ryzen 5 3600 vs. Ryzen 7 3700X

[Strawman]

It’s not my point. I keep telling you this! I’m just saying that another tech publisher has demonstrated that these chips differences are hidden when using average frame rates . And they have used graphs and examples to show this. They have also demonstrated stuttering happening and the frame time drops when it happened. So it does happen. Unless you think DF have made it all up!

I’m saying I think the DF is giving a clearer picture of the differences in gaming than Techspot and the picture is actually quite different. Of course DF also say that outside of gaming the Ryzen CPUs are generally better. Looking at the results and analysis from all the reviews I’ve seen im inclined to agree.

Techspot is testing more for the average joe situation. DF is taking the most demanding sections of the most demanding games. Pick your poison. Yes, the absolute maximum difference in framerates is around 20% between the top intel and the top ryzen.

I don't think DF is giving a clearer picture though. I do love DF and their benchmarks, but they are more for the 0.1% of the buyers rather than the average Joe. The average Joe doesn't really care if he gets 20% less fps in a particular scene of a particular game.

 

[Strawman]

I watched that video and DF highlights two 3 second sections in games where the Ryzen CPUs experience a large frame drop and 1 section where both Intel and AMD experience frame drops.

You want to know why TechSpot doesn't pick and choose specific sections of a benchmark to highlight? It's misleading plain and simple. Frame drops like that show up on the 1% and 0.1% low results, there is zero reason to pick out specific sections. GamersNexus does the same thing as TechSpot with 0.1% lows added in and IMO are far better then DF. By picking out specific data instead of plotting everything out in a graph (they could have even done a scatterplot) where everyone can draw their own conclusions from the data they decide to force a conclusion based on a small section of the data. DF looses major points in my book, if they have the data to show this occurs consistently only on AMD then they should have given us the whole data set like every other review on the internet does.

DF's method isn't better, it's worse by far.

No it's not. It's the DF's reviews that made me pick the R5 1600 over the 7600k (go watch that one). It's specifically because I care more about how my CPU performs in the demanding sections of the game rather than the generic scenes that every CPU performs decently. It's the worst case scenario if you will, and in those Intel has a clear ~20%+ advantage.

 

[Evernessince]

No it's not. It's the DF's reviews that made me pick the R5 1600 over the 7600k (go watch that one). It's specifically because I care more about how my CPU performs in the demanding sections of the game rather than the generic scenes that every CPU performs decently. It's the worst case scenario if you will, and in those Intel has a clear ~20%+ advantage.

As I pointed out earlier: 1% and 0.1% lows are EXACTLY that, worst case scenarios. They were added to benchmarks to precisely represent those dips DF are showing. Only on TechSpot and GamersNexus, they don't cherry pick specific lows.

At this point I don't think you understand what 1% and 0.1% lows so I'm going to recommend you watch the gamersnexus video on it.

Testing Methodology Explanations: 1% & 0.1% Lows, Delta T over Ambient | GamersNexus

stub Following several YouTube commenter questions on some of our testing methodology and presentation, we decided to put together a short guide to FPS and temperature measurements. Specifically speaking to framerate and frametime testing, we've spent a few years refining our collection and...

www.gamersnexus.net

 

[Steve]

Techspot is testing more for the average joe situation. DF is taking the most demanding sections of the most demanding games. Pick your poison. Yes, the absolute maximum difference in framerates is around 20% between the top intel and the top ryzen.

I don't think DF is giving a clearer picture though. I do love DF and their benchmarks, but they are more for the 0.1% of the buyers rather than the average Joe. The average Joe doesn't really care if he gets 20% less fps in a particular scene of a particular game.

I have no idea what you're basing this on, are you just making assumptions? FYI we also test demanding sections of the games. We also show the sections of the games used for testing in the videos.

No it's not. It's the DF's reviews that made me pick the R5 1600 over the 7600k (go watch that one). It's specifically because I care more about how my CPU performs in the demanding sections of the game rather than the generic scenes that every CPU performs decently. It's the worst case scenario if you will, and in those Intel has a clear ~20%+ advantage.

We've also shown where the R5 1600 beats the 7600K in more demanding titles: https://www.techspot.com/review/1859-two-years-later-ryzen-1600-vs-core-i5-7600k/

 

[PEnnn]

So, the successor to i7-8700K is barely faster than its predecessor??

Unless you call 5 - 8% real improvement!!

 

[BoboOOZ]

Nonsense, you won't see stuttering with Ryzen processors.

Looking forward to the comparison with the 3600XT and 3800XT

 

[Rafael Gehhlen]

HAHAHAHAHAHAHA Why do test using DDR4-3200?????? Memoria MAX SUPORTED DDR4-2666...
https://ark.intel.com/content/www/b...0600k-processor-12m-cache-up-to-4-80-ghz.html

 

[neeyik]

Why do test using DDR4-3200?

The RAM is run overclocked to minimise any performance limitations caused by hardware other than the CPU. In other words, when checking out the capability of a new CPU, you want to make sure nothing else affects the results.

This is why a GeForce RTX 2080 Ti and low latency, overclocked RAM are used.

 

[Rafael Gehhlen]

The RAM is run overclocked to minimise any performance limitations caused by hardware other than the CPU. In other words, when checking out the capability of a new CPU, you want to make sure nothing else affects the results.

This is why a GeForce RTX 2080 Ti and low latency, overclocked RAM are used.

Intel recommends not to use the memory above the indicated frequency as it can damage the memory controller thus causing the loss of its warranty, so you cannot use a technology created by Intel itself at the risk of losing your warranty, however you have an option , which is to PAY for "INSURANCE" against damage due to OC, it is necessary to follow the manufacturer's specifications when doing "TESTS", so that there is no more performance than possible.

 

[neeyik]

It's only a recommendation, not a requirement, and the risk of damage is due to excessive heat being retained in the CPU. This is why heavy duty cooling is always used on our test systems.

The majority of DRAM sold in online stores is speed rated higher than the default rating of any CPU, which clearly indicates that running memory clocked at, say, 3200 MHz on a CPU with a 2666 MHz controller, is perfectly safe.

 

[Rafael Gehhlen]

What I'm saying about the tests and why it is necessary to export what is recommended by the manufacturer, and what is not safe or not.

 

[BoboOOZ]

What I'm saying about the tests and why it is necessary to export what is recommended by the manufacturer, and what is not safe or not.

They are trying to conduct tests closer to what the majority of the users will use on their machines, the majority of users will not follow the manufacturer recommendation in this case.

However, it's quite difficult to understand your messages, it seems you disagree with something, but not clear what with, try google translate

 

[elmkay3]

So, the successor to i7-8700K is barely faster than its predecessor??

Unless you call 5 - 8% real improvement!!

At this point, it's clearly a pissing contest. Any sane person would just pick the cheap 6 or 8 core CPU which is best value, and save their time and energy for other things.

It's obviously not a popular opinion in the PC enthusiast community, but I find the idea that people piss away hundreds of dollars on these minor improvement is a joke. All the benchmarks are done with every setting maxed out, which is just ridiculous.

If you're thinking of building a new PC, get a decent B450 board with either a 1600AF, 3600 or 3700X depending on your budget and needs. Any of those will be future proof as games make use of more threads in the future. Get fast RAM and tune it properly. Save your money. DDR5 will change no-one's life when it comes to motherboards.

Gain some sanity and perspective. We're clearly at the point where you can get all the performance you need for whatever budget you have. You can't reach 144fps with everything maxed out? Try using the Graphics menu to change a few settings, and see how it looks exactly the same as it did before. You can't get afford a new RTX 2080? Get a GTX 1060 and see how it can still play any game you ask it to. Sorry, but so much of the back and forth commenting on here is just utter BS.

 

[Shadowboxer]

Nonsense, you won't see stuttering with Ryzen processors.

Well, I won’t see the stutters because I’m not benchmarking games and I don’t use a Ryzen CPU in my gaming rig. But digital foundry did See them and even went as far to show it in frame times and live gaming runs on their 10900k review to highlight that the real differences when actually gaming rather than just showing an average score for the whole run. So, il believe you, if you can demonstrate why that you don’t get stutters in the same scenarios they were using.

Of course, the stutters aren’t that big and to me personally wouldn’t break the experience. But on the other hand I am a PC gaming enthusiast and when buying hardware would like to know about these sort of differences when buying a cpu as I don’t really use my system for anything but gaming.

 

[MikitaM]

Of course, since only Intel's Z series mobos have ability to overclock the RAM, then it would be nice to perform all tests with both OCd and 2666 memory to represent what the users with all other boards will get (B/H/etc). Because this is extremely popular, watching results with OCd memory then gettting same cpu, but with cheap B365 board and getting disappointed.

 

[BoboOOZ]

Of course, since only Intel's Z series mobos have ability to overclock the RAM, then it would be nice to perform all tests with both OCd and 2666 memory to represent what the users with all other boards will get (B/H/etc). Because this is extremely popular, watching results with OCd memory then gettting same cpu, but with cheap B365 board and getting disappointed.

Steve did this comparison for the 10400

As expected, there's a difference, but it's not mind-boggling.

Anyway, the conclusion is clear for all these reviews, you buy one of these because you specifically want Intel, if you're after price-performance AMD is better placed each time.

 

[MikitaM]

Steve did this comparison for the 10400

Okay, need to check HU before I post next time

As expected, there's a difference, but it's not mind-boggling.

Anyway, the conclusion is clear for all these reviews, you buy one of these because you specifically want Intel, if you're after price-performance AMD is better placed each time.

The difference reaches 10% by lows in some cases - sensible IMHO. And completely agree that getting Intel platform in 2020 is just for a sake of getting Intel platform with no logic behind it. Actually, only Ryzen 3600 alone has blown away half of the Intel's current and coming line-ups. And 3300X finished the budget part of the list.

 

[Mithan]

The only thing these reviews are proving to me from a gaming perspective is the i7-7700k is still a capable gaming CPU and there is no reason to look at these new Intel CPUs if you already have an i7-8700/9700. On the AMD side I see no reason to go past the 3600 if all you want is a gaming CPU.

That is my issue. I have the 7700k. Great for gaming but the lack of cores is starting to cause me grief.

The problem is, Rocket Lake is right around the corner and will offer a much better CPU. The 8 core will out perform the 10900k for example (don't ask for a source, I wont provide it). I see no reason to jump on Series 10, when friends tell me to skip it. I guess I need to wait another 7 months.

Zen 3 might even be worth looking at, as I suspect they will close the gap in gaming performance, but my last experience with AMD (1700x) was bug filled, probably due to motherboard/bios maturity issues, but not something I want to repeat.

 

[pcnthuziast]

(don't ask for a source, I wont provide it)

I think you meant can't, not won't. Nobody can since the chips don't exist yet.

 

[Turkiyeeee]

3700x vs 10600k
WHY?

 

[Lionvibez]

I think you meant can't, not won't. Nobody can since the chips don't exist yet.

I found that pretty funny.

 

[Strawman]

I have no idea what you're basing this on, are you just making assumptions? FYI we also test demanding sections of the games. We also show the sections of the games used for testing in the videos.

We've also shown where the R5 1600 beats the 7600K in more demanding titles: https://www.techspot.com/review/1859-two-years-later-ryzen-1600-vs-core-i5-7600k/

First of all, thanks, I didn't know the sections on the video are where you actually test.

Yes, I do assume you are not running the most demanding places, judging by the framerates. A quick example, in the Zen 3 launch one of the games you tested was AC odyssey, and the 3600 (as well as the 8700k) were getting 100+ average framerates on your test.. The most demanding area of the game is the city of athens, where the 3600 cannot break the 60 fps barrier and 8700k barely does.

Anyways, I wasnt judging your methodology, it's neither better or worse than DF's, just different.

 

[mAdmAnDingo]

First of all, thanks, I didn't know the sections on the video are where you actually test.

Yes, I do assume you are not running the most demanding places, judging by the framerates. A quick example, in the Zen 3 launch one of the games you tested was AC odyssey, and the 3600 (as well as the 8700k) were getting 100+ average framerates on your test.. The most demanding area of the game is the city of athens, where the 3600 cannot break the 60 fps barrier and 8700k barely does.

Anyways, I wasnt judging your methodology, it's neither better or worse than DF's, just different.

Digital Foundrys Ryzen systems are poorly configured, I also believe it has something to do with the MSI X570 Carbon motherboard they use for the Ryzen CPUs (known to be a terrible MB).

In DF's own video they state the 3700x beats the 3900x in some tests. Ok cool, maybe the latency of the 2 chiplets of the 3900x causing these issues. But then in their 10600k video, they say the 3600 beats the 3700x in some tests. That should not be possible at all, the 3700x has 4 cores per CCX and higher clocks across the board. I have never seen the 3700x lose to the 3600 period. Something is very wrong with DF's Ryzen setup, no doubt about it.

And even if you look at the Cinebench test for the 3700x it is lower across the board to other sites including Techspot. DF gets 494 or 498 on the single core tests and lower multi core test as well. Even my 3700x gets 509-510 the same as Techspot.

DFs Ryzen systems are poorly configured on an poor MB platform. No doubt that is part of the problem, but DF also tests older DX11 games like Crysis 3 and also Kingdom come deliverance. I don't play either of those games and I don't find them reasonable as a benchmark for future games either. DX12 and Vulkan are the future. Even ACO is DX11. But I am also running a 2080s with my 3700x and I don't experience the issue that DF has in any of the games I have played over the past year now. I have played ACO from start to finish and have had an stutter free experience.

I get no stuttering or issues. I find DF to not represent an optimal Ryzen platform, even without tweaking any timings their systems are poorly configured. And as other posters have already pointed out, DF are also taking the smallest fraction of a specific game and reporting those figures. That is not representative of every game or future games either. I do find their testing flawed.

All in all DF's has sub par Ryzen systems with flawed benchmark testing. And I can assure you that if DF had a properly configured Ryzen system they would have much less or no issues. And if they tweaked their RAM timings as shown by Steve in his Ryzen RAM tweaking guide they would raise their 0.1% and 1% lows hugely with that 2080ti they are using. But first they need to sort out their Ryzen platform issues, they do have a major problem with their current setup.

And DF also need to revise their benchmark lineup after sorting out their platform issues. Some of their games are so old that they came out years before Ryzen (hence no optimizations for Ryzen) and really not a representation for what to expect from future games at all. I already get enough FPS in older games like the Witcher 3, but it is future DX12/Vulkan games and upcoming next gen console games that will be what to expect in the future. Games will be using the newer APIs and getting more multi-threaded, not less. DF's testing is short sighted and inaccurate. We will absolutely be seeing more multi-core and Ryzen aware games coming out, not less.

And here, just type "intel cpu stutter in games reddit" into google and look at all the Intel owners complaining about stutter. It happens to CPUs from both companies, either setup incorrectly or issues with specific games.

9900k stuttering - Google Search

intel cpu stutter in games reddit - Google Search

 

[Shadowboxer]

Digital Foundrys Ryzen systems are poorly configured, I also believe it has something to do with the MSI X570 Carbon motherboard they use for the Ryzen CPUs (known to be a terrible MB).

In DF's own video they state the 3700x beats the 3900x in some tests. Ok cool, maybe the latency of the 2 chiplets of the 3900x causing these issues. But then in their 10600k video, they say the 3600 beats the 3700x in some tests. That should not be possible at all, the 3700x has 4 cores per CCX and higher clocks across the board. I have never seen the 3700x lose to the 3600 period. Something is very wrong with DF's Ryzen setup, no doubt about it.

And even if you look at the Cinebench test for the 3700x it is lower across the board to other sites including Techspot. DF gets 494 or 498 on the single core tests and lower multi core test as well. Even my 3700x gets 509-510 the same as Techspot.

DFs Ryzen systems are poorly configured on an poor MB platform. No doubt that is part of the problem, but DF also tests older DX11 games like Crysis 3 and also Kingdom come deliverance. I don't play either of those games and I don't find them reasonable as a benchmark for future games either. DX12 and Vulkan are the future. Even ACO is DX11. But I am also running a 2080s with my 3700x and I don't experience the issue that DF has in any of the games I have played over the past year now. I have played ACO from start to finish and have had an stutter free experience.

I get no stuttering or issues. I find DF to not represent an optimal Ryzen platform, even without tweaking any timings their systems are poorly configured. And as other posters have already pointed out, DF are also taking the smallest fraction of a specific game and reporting those figures. That is not representative of every game or future games either. I do find their testing flawed.

All in all DF's has sub par Ryzen systems with flawed benchmark testing. And I can assure you that if DF had a properly configured Ryzen system they would have much less or no issues. And if they tweaked their RAM timings as shown by Steve in his Ryzen RAM tweaking guide they would raise their 0.1% and 1% lows hugely with that 2080ti they are using. But first they need to sort out their Ryzen platform issues, they do have a major problem with their current setup.

And DF also need to revise their benchmark lineup after sorting out their platform issues. Some of their games are so old that they came out years before Ryzen (hence no optimizations for Ryzen) and really not a representation for what to expect from future games at all. I already get enough FPS in older games like the Witcher 3, but it is future DX12/Vulkan games and upcoming next gen console games that will be what to expect in the future. Games will be using the newer APIs and getting more multi-threaded, not less. DF's testing is short sighted and inaccurate. We will absolutely be seeing more multi-core and Ryzen aware games coming out, not less.

And here, just type "intel cpu stutter in games reddit" into google and look at all the Intel owners complaining about stutter. It happens to CPUs from both companies, either setup incorrectly or issues with specific games.

9900k stuttering - Google Search

intel cpu stutter in games reddit - Google Search

Tosh! DF do the best testing in the business.

 

[mAdmAnDingo]

Tosh! DF do the best testing in the business.

Sure, they are truly the best. That explains why the 3700x beats the 3900x in some of their tests and the 3600 beats the 3700x in some of their 10600k tests. Those words came out of Richards own mouth.

DF's Ryzen tests are a train wreck. and no doubt it is because of their MSI X570 carbon MB that they have been using since Ryzen 3rd gen launch. Every publication has slammed MSI for that MB being terrible.

And if that is not the case, then DF is just plain incompetent. You choose.