AMD Ryzen 5 3600X vs Intel Core i5-9600K: Mid-Range Rumble

[Admin]

We put the Ryzen 5 3600X up against the Core i5-9600K in a seven-round faceoff.

AMD Ryzen 5 3600X vs Intel Core i5-9600K: Mid-Range Rumble : Read more

 

[King_V]

However, opting for a 3600X often equates to a savings if you aren't interested in that last bit of overclocking performance with the Ryzen 5 3600X, so keep that in mind at checkout.

Is that one supposed to be 3600 (non-X)? The wording here is a bit strange.


That said, I'll admit that being better overall in gaming and productivity, well, I did NOT see that coming. I suspect there will be protests from the Intel side about overclocking about this.

 

[PaulAlcorn]

Is that one supposed to be 3600 (non-X)? The wording here is a bit strange.


That said, I'll admit that being better overall in gaming and productivity, well, I did NOT see that coming. I suspect there will be protests from the Intel side about overclocking about this.

Thanks for the heads up, will fix

 

[sizzling]

Gaming benchmarks without 0.1% % 1.0% are only half the picture especially when comparing these 2 cpu's

 

[philged]

AMD has made a fantastic set of processors for a great value , but from a purely gaming perspective this article seems a little off.

Most of the reviews from Toms itself and other sites have put the i5-9600k clearly above AMD's set of offerings when you just look at gaming.

The article seems to try and skew some of those results incorrectly in favor of AMD.

The biggest issue, running a straight up average, or even a geometric average (as used by the article) Intel wins in the overall test suite. AMD's numbers are properly reported, but for some reason the Intel ones are lower than they should be. In the overall average, AMD is reported at 119.4 (actual is 119.4588) but Intel is reported at 118.3 (actual is 120.0273, meaning Intel wins, completely changing the result)

Looking at the individual test results, the article claims AMD wins 6/10 test suites (again, differing from previous results) but in reality, AMD wins 4 tests, and ties in 3 more where the results between the 99th and average are mixed.

Additionally, because the AoTS run is a CPU test, not a real-world gaming result, taking out that result lands Intel even higher on the overall test suite averages and makes it an even tie on individual game wins.

Again, AMD is great, but from a gaming perspective, something is wrong with your results.

 

[TJ Hooker]

Gaming benchmarks without 0.1% % 1.0% are only half the picture especially when comparing these 2 cpu's

They do have 1% (listed as 99%) FPS results.

 

[TJ Hooker]

In the overall average, AMD is reported at 119.4 (actual is 119.4588) but Intel is reported at 118.3 (actual is 120.0273, meaning Intel wins, completely changing the result)

I got 118.3 fps (118.2613) for the 9600K geomean average FPS. You probably just made a typo during your calculations.

Looking at the individual test results, the article claims AMD wins 6/10 test suites (again, differing from previous results)

If the 3600(X) performs a bit better than it did when it first came out that's not really surprising. There have a been a bunch of BIOS updates with performance/boost tweaks that have come out since then.

 

[iam2thecrowe]

Seems like the real winner is the 3600 non-X. Basically same performance for $50 less.

 

[PaulAlcorn]

AMD has made a fantastic set of processors for a great value , but from a purely gaming perspective this article seems a little off.

Most of the reviews from Toms itself and other sites have put the i5-9600k clearly above AMD's set of offerings when you just look at gaming.

The article seems to try and skew some of those results incorrectly in favor of AMD.

The biggest issue, running a straight up average, or even a geometric average (as used by the article) Intel wins in the overall test suite. AMD's numbers are properly reported, but for some reason the Intel ones are lower than they should be. In the overall average, AMD is reported at 119.4 (actual is 119.4588) but Intel is reported at 118.3 (actual is 120.0273, meaning Intel wins, completely changing the result)

Looking at the individual test results, the article claims AMD wins 6/10 test suites (again, differing from previous results) but in reality, AMD wins 4 tests, and ties in 3 more where the results between the 99th and average are mixed.

Additionally, because the AoTS run is a CPU test, not a real-world gaming result, taking out that result lands Intel even higher on the overall test suite averages and makes it an even tie on individual game wins.

Again, AMD is great, but from a gaming perspective, something is wrong with your results.

Thanks for the feedback. I went back and double-checked, and it appears our numbers are accurate. The problem partially stems from the fact that we have more detailed data that we round off for the charts for each title. For instance, the 3600X's average frame rate for Hitman 2 is 98.40134 (we calculate this from measurements of every single frame presented), but for obvious reasons we don't include the full number in the chart, instead rounding off. However, we calculate our overall geometric mean based off of the more granular data. In either case, I'm not able to replicate your math for the 9600k based off the data presented, so might be a miskey there.

We use a geometric mean to reduce the impact of any massive deltas in a given title that aren't represented elsewhere. That smoothing effect is widely used for these types of cumulative measurements, and it does reduce the impact of the Ashes test you mention. Removing ashes from the test roster takes the 9600K from a -1.1834 fps loss to a 0.7856 fps lead, an overall swing of 1.969 fps. However, replacing that test with another would cause a comparable swing based upon title chosen. At the end of the day, it is all about the mix, and merely swapping one out for another impacts cumulative results as well. The Ashes test is CPU-centric, but this is a CPU review, after all, and we're measuring the CPUs ability to issue draw calls while under heavy load.

Please remember that we're dealing with small variances here when discussing 3600X to 9600K, and the overall impression we have is that there is little differentiation between the two, especially with lesser cards/higher resolutions. Cumulative results make the overall picture easier to digest for casual readers, but they certainly aren't the end-all-be-all for performance. We provide the detailed per-game charts to allow readers to get a gauge of performance in games that best represent what they're playing, so keep that in mind.

AMD does pull off the slightest of wins here, so we have to call that out lest we be accused of not calling out a win. I will say it is a bit refreshing that someone thinks we're skewing in favor of AMD, as most of the time people suspect we do that for Intel

 

[RodroX]

At 1080p Intel core i5 9600K still give you higher AVG FPS in some older games and even in some newer ones when OC. But once you take a close look to the 1% and 0.1%low in games like RDR 2, you understand why a R5 3600 (the non-X one) is a better idea than a i5 9600K or 9700K even when you overclock them.

At 1440p, the story change a bit, but in a way that wont help intel too much either, since the numbers tend to be more close to each other cause of the bigger GPU load.

 

[TJ Hooker]

But once you take a close look to the 1% and 0.1%low in games like RDR 2, you understand why a R5 3600 (the non-X one) is a better idea than a i5 9600K or 9700K even when you overclock them.

In the case of RDR2 there's clearly some issue affecting non-multithreaded CPUs, that goes beyond simply not having enough cores/threads. In Gamernexus' benchmarks the 7700K has massively better 1% and 0.1% FPS than a 9700K. That makes no sense, a 4C/8T CPU should obviously not be outperforming a 8C/8T CPU.

I guess you could still argue that it's a 'win' for AMD as all their (non-APU) chips are multithreaded, but I don't know if I'd use it as evidence that 6 threads are no longer enough for gaming.

View: https://youtu.be/z_ty-gajwoA?t=422

 

[sizzling]

They do have 1% (listed as 99%) FPS results.

Completely missed that

 

[st379]

AMD does pull off the slightest of wins here, so we have to call that out lest we be accused of not calling out a win. I will say it is a bit refreshing that someone thinks we're skewing in favor of AMD, as most of the time people suspect we do that for Intel

I guess an apology is in order. The truth is that i thought that you are biased toward Intel because of the athlon 200ge review....
But maybe I came to a quick conclusion. There were some articles that I did not like but I guess that could happen from time to time.
Anyway I usually enjoy the articles on this site so I wanted to apologize if I came out to harsh in the comments on the Intel new vulnerability.

 

[RodroX]

In the case of RDR2 there's clearly some issue affecting non-multithreaded CPUs, that goes beyond simply not having enough cores/threads. In Gamernexus' benchmarks the 7700K has massively better 1% and 0.1% FPS than a 9700K. That makes no sense, a 4C/8T CPU should obviously not be outperforming a 8C/8T CPU.

I guess you could still argue that it's a 'win' for AMD as all their (non-APU) chips are multithreaded, but I don't know if I'd use it as evidence that 6 threads are no longer enough for gaming.

View: https://youtu.be/z_ty-gajwoA?t=422

True, thats some heavy crap on software level going on there. But yeah, RDR 2 then is out of equation.

Its seems even today the same crap still going on with RDR 2..... and Core i5 9600K and i7 9700K.

 

[philged]

Thanks for the feedback. I went back and double-checked, and it appears our numbers are accurate. The problem partially stems from the fact that we have more detailed data that we round off for the charts for each title. For instance, the 3600X's average frame rate for Hitman 2 is 98.40134 (we calculate this from measurements of every single frame presented), but for obvious reasons we don't include the full number in the chart, instead rounding off. However, we calculate our overall geometric mean based off of the more granular data. In either case, I'm not able to replicate your math for the 9600k based off the data presented, so might be a miskey there.

I apologize, looks like I jumped the gun there before checking my own numbers! Thanks for the detailed response. Please feel free to use that in any fights against Intel bias

 

[mdd1963]

We put the Ryzen 5 3600X up against the Core i5-9600K in a seven-round faceoff.

AMD Ryzen 5 3600X vs Intel Core i5-9600K: Mid-Range Rumble : Read more

It is inconceivable that the 9600K could actually be selling for $260+.....; IMO, it's price should be less than that of the R5-3600, currently going for $180-ish...

(Concur with other poster that suggested inclusion of .1% and 1% lows....; average FPS often is less important than getting rid of the sub-60 fps dips incurred by some CPUs with lower thread count. (normally the 4c/4t CPUs are decimated here)

 

[cryoburner]

Seems like the real winner is the 3600 non-X. Basically same performance for $50 less.

Well, it's more like a $25 difference now, at least going by current US pricing. The 3600X is now around $200 at most major online retailers, while the 3600 is around $175. That's still about 14% more for a marginal performance difference of just a couple percent or so, but you do get a better stock cooler out of it, along with slightly better-binned silicon. And if you consider the full cost of a $1000 system, the higher price becomes nearly in-line with the level of CPU performance gained, so either could be considered reasonable options.

The 9600K can be had for as little as $220 now, though it's currently a bit more at Amazon and Newegg, around $240. It might still be a decent option if one wants to overclock for a bit higher lightly-threaded performance for things like esports titles, though with Intel adding Hyperthreading to their i5 range in the coming months, it would probably be better to wait a bit for that unless one is really in need of a new system right away.

 

[PaulAlcorn]

I guess an apology is in order. The truth is that i thought that you are biased toward Intel because of the athlon 200ge review....
But maybe I came to a quick conclusion. There were some articles that I did not like but I guess that could happen from time to time.
Anyway I usually enjoy the articles on this site so I wanted to apologize if I came out to harsh in the comments on the Intel new vulnerability.

No worries. Passionate readers keep us afloat! We've got our share of scars from things, no doubt, but the community assures we stay on an even keel.

 

[alextheblue]

Well, it's more like a $25 difference now, at least going by current US pricing. The 3600X is now around $200 at most major online retailers, while the 3600 is around $175. That's still about 14% more for a marginal performance difference of just a couple percent or so, but you do get a better stock cooler out of it, along with slightly better-binned silicon. And if you consider the full cost of a $1000 system, the higher price becomes nearly in-line with the level of CPU performance gained, so either could be considered reasonable options.

Yeah they're both good options. If you use the stock cooler, that might tip things in favor of the 3600X. I personally would favor the vanilla 3600, since I always use aftermarket cooling.

 

[hotaru251]

thought every1 knew for straight gaming intel has higher OC potential but if u do anythign other than gaming amd is hands down winner?


only thing intel has going for it is gaming (and only for those diehard who want that extra few fps)


oh and 3600 w/ proper cooler will basically beat x version. (as some channels have said .....ur paying more just for x when performence is miniscule in differences)

 

[LordVile]

thought every1 knew for straight gaming intel has higher OC potential but if u do anythign other than gaming amd is hands down winner?


only thing intel has going for it is gaming (and only for those diehard who want that extra few fps)


oh and 3600 w/ proper cooler will basically beat x version. (as some channels have said .....ur paying more just for x when performence is miniscule in differences)

Depends on the games though, if they prefer core count vs raw frequency then Intel gets demolished. Plus AMD actually has higher IPC now it think? Just into has the like 0.6GHz advantage, though for that you need to by the already more expensive intel chip, a more expensive Z370/90 board and a beefy cooler. vs a B450 Max and just running stock AMD

 

[boju]

Would it be possible to measure cpu usage while doing these tests? Afaik, 1% and 0.1% is suppose to kind of reflect how a cpu is fairing but it would be good to actually see numbers how close a cpu is to being maxed out in given tasks.

Also popularity in high frame rate gaming is ever increasing and i'd say a good chunk of people would like to know which processor is best for that purpose without hitting a high usage wall causing potential stutters. I know the 9600k is a bull and can dish out the frames but even at 5GHz, games that are heavy on the cpu, having to also pre-render frames for the gpu, this cpu is known to high usage stutter. I wouldn't think the 3600 would be in the same boat, or even 2600 for that matter. Even though frame rates aren't as high with Ryzen's, at least cpu usages wont be maxed out.

An example with this video, there'd be other games as well. 8600k @ 5GHz (pretty similar performance to a 9600k) would struggle at high frame rates if the player were to go into the thick of it.

Cpu usage pegged on a 100 isn't fun and i don't really believe 1% and 0.1% entirely reflects how close a cpu is to being maxed out.

 

[calken]

Gaming benchmarks without 0.1% % 1.0% are only half the picture especially when comparing these 2 cpu's

So is benchmarking $200 CPUs with a $1000 GPU.

 

[sizzling]

So is benchmarking $200 CPUs with a $1000 GPU.

Who said anything about a $1000 cpu, what’s the point you are making?

 

[King_V]

oh and 3600 w/ proper cooler will basically beat x version. (as some channels have said .....ur paying more just for x when performence is miniscule in differences)

Well, depending on how close the prices are, sometimes paying a little extra for the X version is worth it even to just get the Wraith Spire over the Wraith Stealth.