Intel i9-9900K Review: The Gaming World's Most Powerful - MMORPG.com

SBFord

Intel i9-9900K Review: The Gaming World's Most Powerful - MMORPG.com

The day has finally come: The Intel i9-9900K is here. The new “world’s best gaming processor” brings the i9 line to mainstream consumers for the first time and packs a full 8 cores and 16 threads of performance, boosts 5GHz out of the box, and marks the return of a soldered heat spreader to keep it cool. Let’s see how it held up!

Read the full story here

Mowzer

This review confirmed what has been the status quo for so long now, and that is the VGA is where its all at, and that the cpu plays such a small part in gaming that you'd have to be stupidly wealthy or like stroking your epeen to bother with these overpriced offerings.

Sephrus

It just says volumes when Intel cuts 1/4 of the L3 cashe per core off the 9700K just so that it can still hold up the stupidly overpriced 9900K as the gaming king. And the 9700K STILL has a higher single core score on almost every chart because honestly with 8 physical cores you dont need to overburden your cpu by making it produce virtual cores for low threaded workloads.

Phry

Aori said:

Limiting tests to 1080p is a silly. Considering the gap shrinks dramatically at higher resolutions which more and more people are playing at these days anyways.

1k tests are good information to have, but they really should include 2k and 4k tests, as those are more likely to be used than 1k is and as such are likely far more relevant than any 1k test could be. 

Quizzical

Gorwe said:

Does this mean that the 7k series will get cheaper? I could get a new CPU, Black Friday...hm...nice!

No, or at least not unless you're planning on getting it used.  Intel doesn't drop prices on old generations when they launch a new one.  They discontinue old generations when they launch a new one.

Quizzical

Aori said:

Limiting tests to 1080p is a silly. Considering the gap shrinks dramatically at higher resolutions which more and more people are playing at these days anyways.

If this were a GPU review, you'd be right.  But it's not.  It's a CPU review.  And charts that say that all CPUs perform equally if you have a GPU bottleneck really aren't useful.  Higher resolutions add a ton of GPU load, but nothing to the CPU load, which makes a GPU bottleneck far more likely.  You could make a case that the review should have used resolutions lower than 1080p just to make even less of a GPU bottleneck.  But it doesn't look like any of the games reviewed have a GPU bottleneck, so that's not really necessary.

Quizzical

I think the critical takeaway from the gaming benchmark chart is that all of the CPUs are really fast.  The lowest frame rate on the entire chart is 96 frames per second, which is plenty good enough for just about anything.  Some reviews will try to pick games that will give lower frame rates, but that can easily lead to focusing on badly coded games, so that's a bad thing if taken too far.

Ozmodan

Still keeps AMD the king far as I am concerned.  Less expensive and gets the job done for the vast majority of games.  The nice thing about AMD cpus, if you desire a faster one, you don't have to buy a new motherboard to use it.  The new 7nm cpus coming next year from AMD will just plug into any current Ryzen motherboard.  AMD 7nm chips will probably compete equally with Intel's if not exceed them.  

Scolioz

you don't need 32 gb of ram for gaming. fuqing noob

sipu

parrotpholk said:

Why not benchmark against a threadripper from AMD. If this the top intel processor then it should be benched against AMD premium line and not the 2700x in my opinion. All this proves is it is still a better value to go with AMD or if you must have intel then an 8700K. Where is the incentive to buy this? Sure there are those who buy the biggest just for sake of having it but nothing here screams buy me now over a Ryzen chip. And again this was not benched vs a threadripper which I would be curious to see the results.

Threadripper isn't a gaming cpu.

Quizzical said:

I think the critical takeaway from the gaming benchmark chart is that all of the CPUs are really fast.  The lowest frame rate on the entire chart is 96 frames per second, which is plenty good enough for just about anything.  Some reviews will try to pick games that will give lower frame rates, but that can easily lead to focusing on badly coded games, so that's a bad thing if taken too far.

This is so untrue. 96 fps is way too low for someone who uses high frequency monitor such as a 144hz or even 120hz.

Phry

sipu said:

parrotpholk said:

Why not benchmark against a threadripper from AMD. If this the top intel processor then it should be benched against AMD premium line and not the 2700x in my opinion. All this proves is it is still a better value to go with AMD or if you must have intel then an 8700K. Where is the incentive to buy this? Sure there are those who buy the biggest just for sake of having it but nothing here screams buy me now over a Ryzen chip. And again this was not benched vs a threadripper which I would be curious to see the results.

Threadripper isn't a gaming cpu.

Quizzical said:

I think the critical takeaway from the gaming benchmark chart is that all of the CPUs are really fast.  The lowest frame rate on the entire chart is 96 frames per second, which is plenty good enough for just about anything.  Some reviews will try to pick games that will give lower frame rates, but that can easily lead to focusing on badly coded games, so that's a bad thing if taken too far.

This is so untrue. 96 fps is way too low for someone who uses high frequency monitor such as a 144hz or even 120hz.

That is a good point, as the only reason to have a 1080 monitor is to have one that is 144hz etc.  at least that is if your using it for gaming, which being on this website would be the norm.

Vrika

Aori said:

Quizzical said:

Aori said:

Limiting tests to 1080p is a silly. Considering the gap shrinks dramatically at higher resolutions which more and more people are playing at these days anyways.

If this were a GPU review, you'd be right.  But it's not.  It's a CPU review.  And charts that say that all CPUs perform equally if you have a GPU bottleneck really aren't useful.  Higher resolutions add a ton of GPU load, but nothing to the CPU load, which makes a GPU bottleneck far more likely.  You could make a case that the review should have used resolutions lower than 1080p just to make even less of a GPU bottleneck.  But it doesn't look like any of the games reviewed have a GPU bottleneck, so that's not really necessary.

If the GPU is the bottleneck at a higher resolution you're using then why buy the much more expensive though better CPU if its going to be a limited by the GPU? How can you say the games wouldn't have a GPU bottleneck at higher resolutions? This computer build to me is complete overkill for a 1080p setup, why have such an expensive system and then a simple 1080p display.

Who is this review for?

The review is for people who want to know about I9-9900K.

If you want to know about GTX 1080 Ti, then you need to look at GTX 1080 Ti reviews. For example here:
 https://www.guru3d.com/articles_pages/msi_geforce_gtx_1080_ti_gaming_x_trio_review,13.html

Reviewers try to avoid reviewing combinations, because if you reviewed how well I9-9900K and GTX 1080 Ti perform together that kind of review would be next to useless for someone who has I9-9900K and RTX 2080 Ti instead.

Quizzical

Aori said:

Quizzical said:

Aori said:

Limiting tests to 1080p is a silly. Considering the gap shrinks dramatically at higher resolutions which more and more people are playing at these days anyways.

If this were a GPU review, you'd be right.  But it's not.  It's a CPU review.  And charts that say that all CPUs perform equally if you have a GPU bottleneck really aren't useful.  Higher resolutions add a ton of GPU load, but nothing to the CPU load, which makes a GPU bottleneck far more likely.  You could make a case that the review should have used resolutions lower than 1080p just to make even less of a GPU bottleneck.  But it doesn't look like any of the games reviewed have a GPU bottleneck, so that's not really necessary.

If the GPU is the bottleneck at a higher resolution you're using then why buy the much more expensive though better CPU if its going to be a limited by the GPU? How can you say the games wouldn't have a GPU bottleneck at higher resolutions? This computer build to me is complete overkill for a 1080p setup, why have such an expensive system and then a simple 1080p display.

Who is this review for?

Why not let people see how well a CPU can perform if the GPU isn't the bottleneck, and then let the reader decide how high of frame rates are good enough?  That gives a lot more information than saying that all the CPUs give 40 frames per second when you have a GPU bottleneck there, and who knows what would happen if you didn't.

A basic principle of benchmarking hardware is that you want for whatever it is that you're trying to benchmark to be your bottleneck.  Sometimes that means having some atypical workloads.  It definitely means that you don't want to use exactly the same benchmark for a CPU as you would for a GPU review, memory review, SSD review, hard drive review (who cares about that anymore, anyway?), motherboard review, case review, or power supply review.

Vrika

sipu said:

parrotpholk said:

Why not benchmark against a threadripper from AMD. If this the top intel processor then it should be benched against AMD premium line and not the 2700x in my opinion. All this proves is it is still a better value to go with AMD or if you must have intel then an 8700K. Where is the incentive to buy this? Sure there are those who buy the biggest just for sake of having it but nothing here screams buy me now over a Ryzen chip. And again this was not benched vs a threadripper which I would be curious to see the results.

Threadripper isn't a gaming cpu.

I'd argue that I9-9900K isn't really gaming CPU either. If all you want to do is game then you're likely to be more interested in I5-9600K.

I guess the reason they didn't include Threadrippers in this review is simply because they haven't any Threadripper reviews in their database. 

sipu

Vrika said:

sipu said:

parrotpholk said:

Why not benchmark against a threadripper from AMD. If this the top intel processor then it should be benched against AMD premium line and not the 2700x in my opinion. All this proves is it is still a better value to go with AMD or if you must have intel then an 8700K. Where is the incentive to buy this? Sure there are those who buy the biggest just for sake of having it but nothing here screams buy me now over a Ryzen chip. And again this was not benched vs a threadripper which I would be curious to see the results.

Threadripper isn't a gaming cpu.

I'd argue that I9-9900K isn't really gaming CPU either. If all you want to do is game then you're likely to be more interested in I5-9600K.

I guess the reason they didn't include Threadrippers in this review is simply because they haven't any Threadripper reviews in their database. 

Everything that can hit nearly 5Ghz clock (with or without slight OC) is a gaming CPU. Period. I5 is a budget choice in this case.

Quizzical

Ozmodan said:

Still keeps AMD the king far as I am concerned.  Less expensive and gets the job done for the vast majority of games.  The nice thing about AMD cpus, if you desire a faster one, you don't have to buy a new motherboard to use it.  The new 7nm cpus coming next year from AMD will just plug into any current Ryzen motherboard.  AMD 7nm chips will probably compete equally with Intel's if not exceed them.  

It's not that likely that the 7 nm AMD CPUs will be faster than this at gaming.  Part of what allows Sky Lake Refresh Refresh Refresh (or whatever they're officially calling it these days) to clock so high is that it's build on a very old, very mature, very well understood process node.  That hardly ever happens, as you generally want to move from one process node to the next.  Intel's 14 nm++*^! (or whatever they're calling the latest iteration) is far, far more mature than TSMC's 7 nm is, or even will be next year.  But the same lack of a new process node that allows this chip to clock so high for gaming today means that Intel's server division is going to be in a world of hurt next year.

What we can pretty much guarantee, however, is that AMD's 7 nm, 8 core CPUs will use vastly less power than a Core i9-9900K.  Die shrinks like that save you a ton of power, and part of how Intel got so much performance out of this part is that they were willing to blow out the power budget and just expect you to put a great big cooler on it.  That you can get more performance if you're willing to burn more power isn't news.  Anandtech found a Core i9-9900K burning more than 50 W more than a Ryzen 7 2700X:

https://www.anandtech.com/show/13400/intel-9th-gen-core-i9-9900k-i7-9700k-i5-9600k-review/21

And yes, that does let you get more performance.  You can argue that it's nice to at least have the option to burn a ton of power and get high performance that way.  But it's also important to realize that there are few workloads where that's an interesting thing to do.  It's certainly not something you'd want to do in a laptop, let alone a mobile device.  Nor are servers likely to be interested in runaway power consumption.  Even most desktop uses would prefer to get all the performance you need while using little power rather than a little more performance at the cost of a lot of power.

FlyByKnight

Watching MMORPG players talk about how most games are GPU bound is kinda funny considering the badly optimized early access/alpha Unreal Engine POS that WE like to play hog the the CPU.

Kajidourden

Quizzical said:

Aori said:

Quizzical said:

Aori said:

Limiting tests to 1080p is a silly. Considering the gap shrinks dramatically at higher resolutions which more and more people are playing at these days anyways.

If this were a GPU review, you'd be right.  But it's not.  It's a CPU review.  And charts that say that all CPUs perform equally if you have a GPU bottleneck really aren't useful.  Higher resolutions add a ton of GPU load, but nothing to the CPU load, which makes a GPU bottleneck far more likely.  You could make a case that the review should have used resolutions lower than 1080p just to make even less of a GPU bottleneck.  But it doesn't look like any of the games reviewed have a GPU bottleneck, so that's not really necessary.

If the GPU is the bottleneck at a higher resolution you're using then why buy the much more expensive though better CPU if its going to be a limited by the GPU? How can you say the games wouldn't have a GPU bottleneck at higher resolutions? This computer build to me is complete overkill for a 1080p setup, why have such an expensive system and then a simple 1080p display.

Who is this review for?

Why not let people see how well a CPU can perform if the GPU isn't the bottleneck, and then let the reader decide how high of frame rates are good enough?  That gives a lot more information than saying that all the CPUs give 40 frames per second when you have a GPU bottleneck there, and who knows what would happen if you didn't.

A basic principle of benchmarking hardware is that you want for whatever it is that you're trying to benchmark to be your bottleneck.  Sometimes that means having some atypical workloads.  It definitely means that you don't want to use exactly the same benchmark for a CPU as you would for a GPU review, memory review, SSD review, hard drive review (who cares about that anymore, anyway?), motherboard review, case review, or power supply review.

Most people fail to understand how testing actually works.  It doesn't matter what the more practical build is....the point is to gather as much data as possible about the CPU, nothing else. 

Quizzical

parrotpholk said:

sipu said:

Threadripper isn't a gaming cpu.

Gaming CPU??  Love that marketing.........

Threadripper isn't primarily targeted at gamers, or at least not people for whom gaming is their most intense workload.  It's not that it's bad at gaming; it's that it's not really better at it than a Ryzen 7 2700X, in spite of being a lot more expensive.  And that's even if we ignore the competition from Intel.

Traditionally, the choices were that you could get a lot of cores clocked low or a few cores clocked high.  For example, a Xeon Platinum 8176 has 28 cores, but a base clock speed of only 2.1 GHz, and even a max turbo speed of only 3.8 GHz.  If you have some workloads that need many cores and others that need very fast cores, you had to either accept that you'd be bad at one of them or have two different computers.  For example, someone who does a lot of rendering that needs as many cores as he can get, and also likes to play games.  Threadripper makes it possible to get both at once, though it comes at the expense of burning a ton of power.

Threadripper also manages to greatly undercut Intel's prices.  AMD will sell you 16 cores for $900 or 32 cores for $1750, as available on New Egg right now.  That 28 core Xeon cited above costs over $9000, and Intel charges $1800 for their own 16-core CPU.  If you want as much CPU performance as you can possibly get in programs that scale well to many cores, then some Threadripper CPU will win by a mile on any budget between about $400 and several thousand dollars.

Threadripper is also for hardware enthusiasts who think it's really cool that now they can get a 16-core CPU that is clocked high for under $1000.  For example, Hard OCP has been running a bunch of articles examining this or that about Threadripper, mostly because they think it's a cool product.  If AMD hadn't bothered to build Threadripper, it might not have been possible to get 16 cores for under $1000 several years from now, even.

The market for Threadripper isn't really that big.  But when you're selling CPUs for $1000 each, you don't need to sell a million of them to justify having the product.  Part of the reason why Threadripper even exists is that AMD didn't even have to make a new die for it.  Just take the same die that they had designed for both relatively high end consumer desktops and for servers and put two of them in a package, or four for the high end.  There's still a lot of non-trivial engineering necessary to make that work, but it's a lot cheaper than designing a huge die on a cutting edge process node.

Quizzical

t0nyd said:

Wouldn't it make the most sense to show FPS at common resolutions. 1080p, 1200p, and 1440p are common. Ultra-wide resolutions are becoming common. Personally I ignore 1080p because it doesn't pertain to me. Why do I care if any processor gets 240 frames per second in 1080p when I'm at 3440 X 1440. 

If it's common to be GPU bottlenecked, then I would prefer a real world performance than a hypothetical. I'm mostly ignore synthetic benchmarks. I also ignore ashes of the benchmark, who cares what the FPS are, it isn't a game. 

The average person builds a computer based upon a budget. If you're building a computer for gaming, as long as you're 6 cores or higher, CPU choice doesn't really matter. So why spend more on a CPU when that money could be spent on a better GPU. I see no point in the 9900 k at this price point. I'd rather pick up a cheaper ryzen and put that extra money into the graphics card.

Why is it interesting to show, this is how a CPU performs under various circumstances where we change the GPU load?  Isn't it sufficient to say, this is how the CPU performs if there isn't a GPU bottleneck, and if the GPU caps you at something less than this, you get whatever the GPU can handle?

Quizzical

Vrika said:

sipu said:

parrotpholk said:

Why not benchmark against a threadripper from AMD. If this the top intel processor then it should be benched against AMD premium line and not the 2700x in my opinion. All this proves is it is still a better value to go with AMD or if you must have intel then an 8700K. Where is the incentive to buy this? Sure there are those who buy the biggest just for sake of having it but nothing here screams buy me now over a Ryzen chip. And again this was not benched vs a threadripper which I would be curious to see the results.

Threadripper isn't a gaming cpu.

I'd argue that I9-9900K isn't really gaming CPU either. If all you want to do is game then you're likely to be more interested in I5-9600K.

I guess the reason they didn't include Threadrippers in this review is simply because they haven't any Threadripper reviews in their database. 

I would argue that gaming is the primary, intended market for this CPU.  It's perhaps overkill, and not something that all that many people should buy, but it's generally going to be the fastest gaming CPU that you can buy.  It's not terribly appropriate for workloads that scale well to arbitrarily many cores, as it would likely lose to a much cheaper 12-core Threadripper 1920X, let alone the CPUs that actually have really a lot of cores.  And it's wildly inappropriate for workloads where you need a ton of memory bandwidth, memory capacity, or PCI Express connectivity, as it will get completely crushed by the HEDT platforms there.

Groqstrong

Will it help in 4k gaming? Im planning my next build around 4k gaming. Goal is to sustain 100+ fps at 4k.

Cleffy

I am not expecting 7nm AMD CPU parts anytime soon. They are competing for fab resources. For instance Vega was nearly a year after everyone was hoping it would release.
I think the jump to EUV will bring a couple more benefits. CPU prices will drop and less space will be required as the waste should be less doing 1 pass over 4 or more passes.
On this CPU. I think you would have to be dense to buy it. It's way too expensive for a consumer processor that offers a marginal performance benefit verse it's equivalent competitor while using more energy and doesn't even ship with a cooler. Don't buy hardware that's NDA lifts on release day and has a pre-order.

Renoaku

Someone told me the new AMD CPU's will be better than this new AMD I kinda double that not talking just basic specs like GHZ, or cores / threads, but CPU instructions, ability to run 3+ different games at one time like on my i7, and CPU Intesnive tasks / VM Ware while gaming.

I hear intel just does it better?

vegetableoil

I'm still using gen 2 quad core i5 3.2ghz with 16 G ram, I can pretty much run any game with ultra setting in 1070 ti lol. may be the newer one will help the loading times?

gervaise1

Its a processor and? 

Its faster; nothing wrong with that - we benefit when there is competition. Means we have more options (price, power etc.) as and when we decide to upgrade.