Intel Core i5-10400 vs. AMD Ryzen 5 3600

[Julio Franco]

Intel's new Core i5-10400 is a locked 6-core, 12-thread processor that operates at between 2.9 GHz and 4.3 GHz depending on the workload. At around $180, this chip seems to be going head to head against AMD's formidable Ryzen 5 3600. Can the i5 hold its own?

Read the full article here.

https://www.techspot.com/review/2032-intel-core-i5-10400/

 

[Vulcanproject]

It's not enough, simply due to the Ryzen platform's features. PCIe 4.0 on higher end boards, unlocked memory support on B450, forwards compatibility from said good B450 board and likely cheaper future CPU upgrades.

Many core Ryzens usually get cleared out for bargain basement prices when new generations are around. You'll be able to go to a 10700k with this later, but they will probably still end up more expensive than say a knock down 3900x in a year's time if you go with AMD's platform.

This has to be cheaper all in than AMD's alternatives for it to make sense for the average consumer.

 

[Irata]

Thanks for the review. But....

Seriously - why don't you test both CPU with the boxed cooler that comes with them ?

I can somewhat understand why you are using high end boards for both as those are part of your standard platform (or boards aren't available yet), but why use a high end cooling solution that basically costs as much as each of the tested CPU ?

Really, if I am buying in that price segment (and I was), one of the things I want to know is if a CPU performs well with the included HSF or if I need to spend another $30 for a decent HSF, money that may well be spent on better specced memory or a somewhat nicer mainboard.

I know that you mention that both aren't particularly impressive but tbh they don't need to be, just good enough to cool the CPU even at max load for extended periods without being painfully loud.

Not singling you out in particular. Many sites seem to have this test approach, sadly.

 

[krizby]

Thanks for the review. But....

Seriously - why don't you test both CPU with the boxed cooler that comes with them ?

I can somewhat understand why you are using high end boards for both as those are part of your standard platform (or boards aren't available yet), but why use a high end cooling solution that basically costs as much as each of the tested CPU ?

Really, if I am buying in that price segment (and I was), one of the things I want to know is if a CPU performs well with the included HSF or if I need to spend another $30 for a decent HSF, money that may well be spent on better specced memory or a somewhat nicer mainboard.

I know that you mention that both aren't particularly impressive but tbh they don't need to be, just good enough to cool the CPU even at max load for extended periods without being painfully loud.

Not singling you out in particular. Many sites seem to have this test approach, sadly.

To avoid any potential variation, I think using the same AIO across the entire test setup is a better idea. You can get an estimated temp for your own setup by adding 20C to the result if you are using the stock cooler and another 10C if your ambient temp is around 30C for example.

For anyone who value their hearing, throwing out the box cooler and paying 30usd for a good air cooler is a bigger priority than having some RGB motherboard or RGB RAM anyways.

If you want CPU to be tested with their box cooler, you have to prove that the majority of user would use it that way, and I seriously doubt that the majority of people do that...

 

[Peter Farkas]

I love my new 3300X!
However, I am concerned what will happen when Intel introduces their 7nm or 5nm chips. Currently AMD and Intel are kind of even in gaming in my opinion though AMD is doing it a better price and overall value.
Don't you guys find it weird that 7nm AMD equals 14nm Intel? I would be concerned if I were Lisa Su.

Any thoughts?

 

[Irata]

To avoid any potential variation, I think using the same AIO across the entire test setup is a better idea. You can get an estimated temp for your own setup by adding 20C to the result if you are using the stock cooler and another 10C if your ambient temp is around 30C for example.

For anyone who value their hearing, throwing out the box cooler and paying 30usd for a good air cooler is a bigger priority than having some RGB motherboard or RGB RAM anyways.

If you want CPU to be tested with their box cooler, you have to prove that the majority of user would use it that way, and I seriously doubt that the majority of people do that...

I do not know about the majority of users, but I personally do use the boxed cooler on my 2700x.
Was thinking about getting a nice be quiet HSF but there isn't really any reason for me to spend the extra money as it's good enough - temps are OK, it's not overly loud...

And I would bet that many people getting a sub 200 CPU will use the stock HSF.

Look at choices - I could either get a Ryzen 3100 plus a decent aftermarket HSF or get a 3300x and use the supplied HSF while spending slightly less. Or I get a 3100, use the stock cooler and get a second stick of 8GB RAM or better specced RAM. Alternatively, I could get a better mainboard (am not talking about RBG, btw) that has better connectivity or additional features. Who knows, but budgets are not unlimited - spending more on one item usually means spending less on another.

Now, the higher up you go CPU wise, the less difference getting a good aftermarket cooling solution makes wrt total cost.

 

[Achaios]

More of the same.

No savvy enthusiast will buy Comic Lake vs Ryzen.

 

[MikitaM]

Could have been interesting if Intel unlocked RAM on all mobos, not just Z-series. As for now, doesn't make any sense at all: 1200 mobos are way overprices, "budget" ones with locked RAM, OC, and no MCE, useless trash cooler. B450 + 3600 build may appear almost 2x time cheaper weighing all the factors.

 

[dangh]

Thanks for the review. But....

Seriously - why don't you test both CPU with the boxed cooler that comes with them ?

I can somewhat understand why you are using high end boards for both as those are part of your standard platform (or boards aren't available yet), but why use a high end cooling solution that basically costs as much as each of the tested CPU ?

Really, if I am buying in that price segment (and I was), one of the things I want to know is if a CPU performs well with the included HSF or if I need to spend another $30 for a decent HSF, money that may well be spent on better specced memory or a somewhat nicer mainboard.

I know that you mention that both aren't particularly impressive but tbh they don't need to be, just good enough to cool the CPU even at max load for extended periods without being painfully loud.

Not singling you out in particular. Many sites seem to have this test approach, sadly.

This is cpu performance testing, to identify which cpu is faster, no a throttling or cheap circuit mb limitations testing. In any single component testing external limitations have to be either avoided or well defined and similar, that was memory settings. Test is done very well to match requirements, and it is easy to extrapolate the results.

 

[grumblguts]

3700k £150 from amazon bargain bucket due to damaged box.
Laughing at intel users who pay 4x that price for half the speed.

 

[Irata]

This is cpu performance testing, to identify which cpu is faster, no a throttling or cheap circuit mb limitations testing. In any single component testing external limitations have to be either avoided or well defined and similar, that was memory settings. Test is done very well to match requirements, and it is easy to extrapolate the results.

But what does that tell me as a potential customer ? Like I said, I understand the board choice, but I really want to see what I will get for my money.

It's a bit like testing budget cars and swapping the stock wheels for an expensive aftermarket wheel and tire combo, upgrading the suspension...yes, this will show what these vehicles are capable of under ideal conditions, but leaves many questions unanswered for the average customer.

Now, I am not saying that there is anything wrong with showing the best possible results even for cheaper CPU and I realize that takes extra time (and time = money), but at least I, personally, would like to see what I get for my money out of the Box in a typical scenario.

But that is just my personal preference, others' mileage may vary.

 

[Aryassen]

3700k £150 from amazon bargain bucket due to damaged box.
Laughing at intel users who pay 4x that price for half the speed.

If that it indeed the case and not a typo on the level I think (because 3700k doesn't exist, but I thought maybe you meant 3600/X), than good for you, fantastic buy, and enjoy! However, that is an very-very rare and extreme bargain...the 3700x is currently at the 280-ish bar here, with the occassional "box damaged' sample going for 250. To find one at 150, that must have been a one-in-a-million chance (and probably someone made a clerical error somewhere).

Still, at current prices here int he UK, the 3600 at Ł165 (with a cosmetic damage sample sitting at 140, but even new ones can be ordered for 151, included with prime) offers very good value, while the 10400 costs...oh wait, it is not even available yet. So while it is good to see some competition finally, and Intel getting down from the high horse (was about damn time), it will take some more time till the actual fight begins, and I reckon it will be a tough one.

 

[Eric Mozzone]

Even though these CPU's are so close to each other, my brand favoritism will dominate and I'll still be going with Intel.

 

[Aryassen]

+1 for those opinions to see how the stock coolers perform. I used the one for 2700 because I really liked the looks, even if I have changed it eventually to a monstrous Brocken 3 later (it was too good a deal to pass on it, OK?)...still, there is value to see how well the CPUs can run with their included cooling solution.
On another note, including the clock speeds also while running the test would be also beneficial (maybe not only for me - just food for thought)

 

[neeyik]

@Irata - you raise a good point but more often than not, time constraints greatly limit what can be achieved. This is especially true when its a new platform, as these invariably need additional testing to check for any stability and compatibility problems, before commencing data sampling. Every test is also run multiple times before calculating the relevant statistics. There comes a point where you have to strike off a number of things you’d like to examine, otherwise you’d never hit the deadline.

 

[Irata]

@Irata - you raise a good point but more often than not, time constraints greatly limit what can be achieved. This is especially true when its a new platform, as these invariably need additional testing to check for any stability and compatibility problems, before commencing data sampling. Every test is also run multiple times before calculating the relevant statistics. There comes a point where you have to strike off a number of things you’d like to examine, otherwise you’d never hit the deadline.

Thanks @neeyik - I am aware of time / resource constraints, but one can still wish. ?

 

[Nobina]

To avoid any potential variation, I think using the same AIO across the entire test setup is a better idea. You can get an estimated temp for your own setup by adding 20C to the result if you are using the stock cooler and another 10C if your ambient temp is around 30C for example.

For anyone who value their hearing, throwing out the box cooler and paying 30usd for a good air cooler is a bigger priority than having some RGB motherboard or RGB RAM anyways.

If you want CPU to be tested with their box cooler, you have to prove that the majority of user would use it that way, and I seriously doubt that the majority of people do that...

I don't think it's a stretch to say most people use boxed coolers for anything that's not a high-end CPU. If I was buying a R5 3600 which I will soon I'm using a boxed cooler. It should be sufficient and while I might think about going aftermarket later I will probably forget about it if there are no issues with the stock one.

Even though these CPU's are so close to each other, my brand favoritism will dominate and I'll still be going with Intel.

I absolutely respect the honesty.

 

[Alfatawi Mendel]

I love my new 3300X!
However, I am concerned what will happen when Intel introduces their 7nm or 5nm chips. Currently AMD and Intel are kind of even in gaming in my opinion though AMD is doing it a better price and overall value.
Don't you guys find it weird that 7nm AMD equals 14nm Intel? I would be concerned if I were Lisa Su.

Any thoughts?

There is that. But due to the difference in architectures, AMD gets much better yields, and the new 10+nm from Intel aren't exactly setting the heather on fire.
We'll see a couple of years down the road whether Intels new uarch @7nm lords it over AMD's @5nm.
Just be glad there is some competition now - otherwise we'd all still be on 4-core consumer desktop chips.

 

[Alfatawi Mendel]

Even though these CPU's are so close to each other, my brand favoritism will dominate and I'll still be going with Intel.

I'll be sticking with the better value AMD. Brand names mean nothing to me.

 

[Lew Zealand]

Intel needs to enable all possible memory speeds on B series MoBos. The numbers here prove that as the i5 doesn't compete nearly as well at 2666 as it does at 3200, and few people are going to spring for a ~$200 overclocking MoBo for a budget, locked CPU.

I've been using a Core i5-8400, the direct ancestor of the i5-10400, with a B360 Mobo and I have a single complaint: I'm stuck at 2666 memory. Sure, I have 3200 CL16 running at 2666 CL13 which helps but then my kids have R5 2600 and R5 1600AFs running on B350 and B460, both with 3200 memory because AMD supports it!

Here's your chance, Intel. AMD has F'd up with the B550 delay and PR flubs, you can reclaim some lost ground by opening memory overclocking on B460. Sadly I have little confidence that Intel will seize this opportunity.

 

[meric]

Many people will not be happy to be forced to buy a z490 board with a locked CPU. I remember the same situation with i5 8400, if I'm not mistaken. As far as I know, there's no indication as to whether or not intel will offer better memory support with their b460. That wouldn't be a consumer friendly approach. But yeah, people are people and they'll buy it no matter what.

 

[Lew Zealand]

Many people will not be happy to be forced to buy a z490 board with a locked CPU. I remember the same situation with i5 8400, if I'm not mistaken. As far as I know, there's no indication as to whether or not intel will offer better memory support with their b460. That wouldn't be a consumer friendly approach. But yeah, people are people and they'll buy it no matter what.

You are correct, I specifically waited for the B360 boards (and reviews!) to come out as it's just stupid to buy the budget locked chip (8400) and pair it with an expensive overclocking MoBo. You buy the 8600K for that.

 

[mrvco]

There is that. But due to the difference in architectures, AMD gets much better yields, and the new 10+nm from Intel aren't exactly setting the heather on fire.
We'll see a couple of years down the road whether Intels new uarch @7nm lords it over AMD's @5nm.
Just be glad there is some competition now - otherwise we'd all still be on 4-core consumer desktop chips.

Yep, it's not like AMD will be idling until Intel can launch 7nm. It's pretty remarkable how much pressure AMD has been able to apply to Intel on desktop, mobile and server. Too bad AMD's Radeon group hasn't been able to push Nvidia as hard.

 

[Evernessince]

If you want CPU to be tested with their box cooler, you have to prove that the majority of user would use it that way, and I seriously doubt that the majority of people do that...

You test with an AIO to ensure the cooler isn't a bottleneck and you test with the stock cooler to see out of the box performance. At this price point it's reasonable to assume a significant portion of people will be using the stock cooler. You don't have to prove anything, this is about giving customers a full picture of the performance they should expect, both out of the box and will an aftermarket cooler.

Nothing is stopping a publication from doing a test or two from their suite with both the stock and after-market coolers. As noted, the omission is likely due to time constraints rather then simply thinking it won't be beneficial.

 

[pcnthuziast]

The cinebench r20 single core score caught my eye, seems off. My 6700k with no OC scores better at lower clocks?