Intel Core i9-9900K 9th Gen Review

[kyotokid]

...wait, an i7 without hyperthreading? Something is seriously wrong there. I could understand that on an i5 or i3, but an i7? You're talking Core Duo Quad L775 there.

 

[hessfy]

save your time.
Article's summery:
if there is Intel's money: Core i9-9900k is the best CPU in the universe + 5 star further.
if not: read the comments for real review.

 

[papasha408]

Intel does make an excellent CPU. Unfortunately, their pricing is a good example of Usury.

 

[papasha408]

Intel does make an excellent CPU. Unfortunately, their pricing is usually an exercise in usury.

 

[papasha408]

save your time.
Article's summery:
if there is Intel's money: Core i9-9900k is the best CPU in the universe + 5 star further.
if not: read the comments for real review.

 

[papasha408]

Intel has a bad habit of overcharging. The 9900 is maybe 10% faster than Ryzen's 7-2700 but, over twice the price. Only fools or people with too much money will buy Intel's product.

 

[mlee 2500]

I would like to see the i9-9900 bench-marked with Hyperthreading turned off in the firmware (but otherwise at stock speeds).

Not only to see the performance relative to the 9700K in games which suffer a performance penalty from threads, but especially to see what impact disabling threads has on power and temperature.

 

[TCA_ChinChin]

There was nothing in the background. I've ran the benchmark about 15 times on two different fresh Windows installs and it's pretty much the same everytime. I was wondering if the game was an anomaly. I'll have to install more games to test. But i gave my i7 to a friend because I also bought a z390 Mobo, just have to buy a CPU. I'm kind of leaving towards an i5-9600K since my internet is too slow to stream. I wanted the 9700K but not for over $400.

Might be an oddity. I've been seeing really sketchy results from benchmarks of Odyssey where the results for r7-2700x vs i9-9900k are wonky too (somehow r7 has higher fps than i9). Obviously that doesn't really make any sense. Do agree that i5-9600k is Intel's best overall chip this refresh. How much is it compared to i5-8600k?

 

[redgarl]

Wait, so it performs within a spitting distance of the 2700x

I wish I could wear those AMD goggles everyone else is wearing. I have the R7 2700X and an i7-7700K and I just don't see it. For the most part they are similar, but when things get rough in a game my 2700X turns belly up while my i7 chugs along with 20-30% higher frames. Maybe that's an anomaly, but Ryzen's architecture just doesn't seem like it's designed for high-performance gaming.

See
https://www.youtube.com/watch?v=eIsK58nUcE0
https://www.youtube.com/watch?v=eS37fjP_FWw

*Update: To clarify, from where I'm sitting the 6-core i5 is possibly the best gaming CPU price/performance.

Remove your CPU OC and test it with CPU boost only. Anyway, you are talking about a single game. Might be your motherboard to. It could be driver related to because other reviewers report good performances on Ryzen at 1440p.

https://static.techspot.com/articles-info/1730/bench/ACO.png

I am playing at 2160p, I am GPU bound all the time, however the multi-threaded performances over Intel when the 1700x was released were just way better than anything from Intel.

Everybody is GPU bound unless you play at 1080p with a 1080 GTX and above, and even there, you cannot see the difference if you are above 144Hz.

Basically, a gaming CPU doesn't exist anymore. The GPU is all that matter. If you are playing at 1440p, it is good if you are going to get 10% more performances.

Finally, most of the people own something like a RX 580 - 1060 GTX. With any mainstream cards, this nice picture of CPU bottleneck doesn't exist. SO these benches are useless for 99% of the Use cases out there.

 

[redgarl]

There you go Toms... your value proposition.

https://youtu.be/_I--zROoRws?t=1241

2700x with 1080TI still 100$ cheaper than 9900k with a 1070 GTX...

 

[volkgren]

Wait, so it performs within a spitting distance of the 2700x

I wish I could wear those AMD goggles everyone else is wearing. I have the R7 2700X and an i7-7700K and I just don't see it. For the most part they are similar, but when things get rough in a game my 2700X turns belly up while my i7 chugs along with 20-30% higher frames. Maybe that's an anomaly, but Ryzen's architecture just doesn't seem like it's designed for high-performance gaming.

See
https://www.youtube.com/watch?v=eIsK58nUcE0
https://www.youtube.com/watch?v=eS37fjP_FWw

*Update: To clarify, from where I'm sitting the 6-core i5 is possibly the best gaming CPU price/performance.

Remove your CPU OC and test it with CPU boost only. Anyway, you are talking about a single game. Might be your motherboard to. It could be driver related to because other reviewers report good performances on Ryzen at 1440p.

https://static.techspot.com/articles-info/1730/bench/ACO.png

I am playing at 2160p, I am GPU bound all the time, however the multi-threaded performances over Intel when the 1700x was released were just way better than anything from Intel.

Everybody is GPU bound unless you play at 1080p with a 1080 GTX and above, and even there, you cannot see the difference if you are above 144Hz.

Basically, a gaming CPU doesn't exist anymore. The GPU is all that matter. If you are playing at 1440p, it is good if you are going to get 10% more performances.

Finally, most of the people own something like a RX 580 - 1060 GTX. With any mainstream cards, this nice picture of CPU bottleneck doesn't exist. SO these benches are useless for 99% of the Use cases out there.

Even at 1080p it drops below 60fps at very high settings OC or no OC. 1080p High settings it doesn't drop below 60fps; but at 1440p it does.

Well, AC Odyssey 1440p Very High settings Intel beats AMD. I see Joker comparing High settings at 1080p and somehow the i9 performed worse than the 2700X.

It's weird..

 

[SonSon1]

How much did Intel pay for this article?

 

[soski.krota]

I think we can't trust Tom's anymore. Intel clearly paying them.
10% more performance for 2x more money = editors choice best buy?
How you dare?

 

[cryoburner]

Although AMD's second-gen Ryzen processors narrowed the gap with Intel's Coffee Lake-based line-up, these ninth-generation Core chips redefine the playing field. The $263 Core i5-9600K at stock settings regularly beat an overclocked $378 Ryzen 7 2700X in games, and we expect even more performance from the Core i5 once we overclock it. Ryzen 7 2700X does come with a capable cooler, but the Core i5’s lower price diminishes AMD’s value proposition.

This comparison seems kind of bizarre. Why are you comparing the price of an 8-core, 16-thread processor against a 6-core, 6-thread processor, particularly when an overclocked Ryzen 2600 or a 2600X should perform relatively similar to a 2700X in today's games, at a far lower price than either of those CPUs?

And the pricing isn't even right. The Ryzen launched for just $329, so I'm not sure where you get $378 from. Plus, the pricing for these Ryzen CPUs has been quite a bit lower lately, with the Ryzen 2700X currently costing just $305 at most major online stores, and it's gone on sale for even less in recent weeks, and that's including a capable cooler in the box. The i5-9600K is currently $280 everywhere, and doesn't even include a cooler. Once one is figured into the price, it actually costs more than a 2700X. And much like you said about the i9-9900K, unless you regularly use heavily-threaded applications, there are options that offer better value than a Ryzen 2700X as well.

The 6-core, 12-thread Ryzen 2600 is currently just $150 at a couple of the biggest US online tech retailers. After adding a $30 aftermarket cooler for overclocking, it should be able to manage gaming performance close to that of the 2700X, at a price far lower than what an i5-9600K costs once a similar cooler is figured in. Sure, the 9600K clocks higher, and will obviously be a bit faster in today's games, provided it's paired with a high-end graphics card and a high refresh rate monitor that's at a low enough resolution where the graphics card isn't what's limiting performance. It could be argued that this extra per-core performance will become more relevant as games become more demanding in the years to come, but the 2600 has SMT, allowing it to more efficiently handle more than 6 threads, which may also become more relevant as it becomes common for games to better utilize more cores. You can already see a hint of this in the Hitman benchmark, where all of the 6-thread i5 processors appear to take a disproportionately large drop in 1% lows compared to their average frame rates during the benchmark run, even compared to the 4-core, 8-thread i7-7700K, so the extra threads seem to be making a difference there.

The i5-9600K certainly appears to be a fine processor, with higher clocks than its 8000-series counterpart, but I would hardly say that it "redefines the playing field" or that it significantly "diminishes AMD’s value proposition". It's ultimately just an i5-8600K with higher boost clocks and better thermal material. Additionally, unless you already have an 8000-series CPU on hand to perform a BIOS update, you'll likely need to buy a Z390 chipset motherboard for it, which adds another $25-$30 over even Z370 at the lower end. On AMD's side, it's possible to even get a decent amount of overclocking out of a B450 board. So again, for someone looking to maximize "value", a Ryzen 2600 with a B450 motherboard currently costs around $180-$190 less than an i5-9600K with a Z390 board. That i5 will be faster in today's games, particularly with an overclock in a high refresh rate setup, but most would likely get better gaming performance by putting that money toward their graphics card. That's nearly the difference in price between a 1050 Ti and a 1070, after all.

Even at 1080p it drops below 60fps at very high settings OC or no OC. 1080p High settings it doesn't drop below 60fps; but at 1440p it does.

Well, AC Odyssey 1440p Very High settings Intel beats AMD. I see Joker comparing High settings at 1080p and somehow the i9 performed worse than the 2700X.

It's weird..

From what I've heard, AC: Odyssey is kind of poorly optimized, and certain graphics settings in particular cause a big hit to performance while offering little improvement to visuals. In particular, the volumetric cloud setting can massively affect performance. Likewise, some settings make a notable difference to visuals, but not much of a difference to performance. So it can definitely help to go in and make some modifications to settings beyond the presets. This video from Hardware Unboxed went over the impact of the various settings on their test system, though of course, other hardware might potentially behave a bit differently...

https://www.youtube.com/watch?v=chqQanHcvHk

 

[mac_angel]

I was wondering if anyone could give me any information on the release date of the Skylake-X refresh and/or performance/hardware info. I was initially planning on getting the Core i9 9900K, but I'm still pissed about the 16x PCIe lane limit. Sadly it's still an on going argument, but I have a Core i7 6850k with 40 PCIe lanes, and two Asus Strix GTX 1080ti OC cards in SLI. There's a very noticeable difference going from 16x/8x over 8x/8x. My Asus X99-A Deluxe II was badly designed and you can't use 16x/16x and the m.2 slot at the same time, so I don't know how much of a difference that makes.
Anyway, time for an upgrade, but sticking with the HID models. They last longer, and if a pair of 1080tis can see the difference going from 8x/8x to 16x/8x, then future GPUs definitely will even more.

 

[mac_angel]

also, I understand that all game testing is done at 1920x1080 to take more of the GPU out of the equation, but I'm curious that if you put the same GPU in all systems anyway, how do the CPUs take handling the workload of something like 4k? I may be naive and not fully understand exactly how much the CPU handles in game architecture, but I'd think that 4 times the resolution and stressing a system will also add more to the CPU and whole platform. I may be wrong, but I believe most people that go out and pick up a Core i9 9900k isn't going to be gaming at 1080p.

 

[ah]

At this rate of the Intel's asking price, my next built will definitely be on TR4.

 

[volkgren]

I may be naive and not fully understand exactly how much the CPU handles in game architecture, but I'd think that 4 times the resolution and stressing a system will also add more to the CPU and whole platform.

Yes, you are wrong (*about the CPU at least). Low settings 1080p (or lower resolutions) will stress the CPU more than 4K because the GPU is pushing out higher frame rates at lower resolutions and settings and the CPU has to work harder to keep up with those higher frame rates. 4K puts more load on the GPU and system RAM but not the CPU.

I may be wrong, but I believe most people that go out and pick up a Core i9 9900k isn't going to be gaming at 1080p.

You are actually right here. I've thought the same thing. Who buy a $500+ CPU to use on a 1080p monitor?? But that just allows us to see it's capability as mentioned above. So, who needs the i9 when the CPU is less stressed at higher res? People who can afford the CPU want the absolute best and have room to spare for whatever might fancy their interest. It's a niche, kind of like (almost exactly like) the High-End Desktop platform; Intel just now allow it on the mainstream platform.

 

[cryoburner]

also, I understand that all game testing is done at 1920x1080 to take more of the GPU out of the equation, but I'm curious that if you put the same GPU in all systems anyway, how do the CPUs take handling the workload of something like 4k? I may be naive and not fully understand exactly how much the CPU handles in game architecture, but I'd think that 4 times the resolution and stressing a system will also add more to the CPU and whole platform. I may be wrong, but I believe most people that go out and pick up a Core i9 9900k isn't going to be gaming at 1080p.

In games, typically increasing resolution won't increase load on the CPU. The CPU is mainly handling other parts of the game, such as AI, physics, netcode and general game logic, and while it handles providing instructions to the graphics card as well, those calls should generally be more or less the same no matter the resolution. The graphics card is what draws the game to your screen, and its the graphics card that needs to work harder to render more pixels at higher resolutions. There have been some exceptions to this in certain games, but those tend to be uncommon. In general, at a given frame rate, the CPU's workload won't be significantly impacted by resolution. Of course, like Volkgren said, at lower resolutions a graphics card can push higher frame rates, making it more likely to reach the point where the CPU can't keep up, at which point it has to wait for the CPU to finish its processing each frame.

Conversely, at high resolutions like 4K, typically the graphics card won't be able to push high enough frame rates to where the CPU makes much of a performance difference, unless it's really underpowered relative to the graphics hardware. Plus, most current 4K screens can only take a 60Hz input anyway, so frame rates above 60fps won't tend to matter that much. So really, most people gaming at 4K don't actually need a particularly fast processor, and could arguably get away with a mid-range one if they don't need the extra CPU performance for other tasks.

If you are interested in how different CPUs compare at high resolutions, some other sites cover that in their reviews. For example, here's the summary page of TechPowerUp's 9900K review, showing averaged performance across their benchmarked games using a 1080 Ti, with charts for various resolutions ranging from 720p to 4K...

https://www.techpowerup.com/reviews/Intel/Core_i9_9900K/19.html

As you can see, the average gaming performance difference between a mid-range processor priced around $200 and an overclocked i9-9900K only amounts to about one or two percent at 4K. I'm not super-familiar with the SLI-scaling of two 1080 Tis in current games, but I suspect that the CPU performance difference at 4K might be closer to the 1440p results there if a game offers good SLI scaling, but even then we're only looking at around a five percent performance difference between these mid to high-end processors. So results tend to be very similar at high resolutions, and would effectively be even more similar on a display that can only update 60 times per second.

 

[mdd1963]

...wait, an i7 without hyperthreading? Something is seriously wrong there. I could understand that on an i5 or i3, but an i7? You're talking Core Duo Quad L775 there.

Int he i7-9700K's defense, it did defeat the 8700K in pretty much all games....which is saying something....; so clearly having 'just 8 cores and 8 threads' is not quite a crippling condition....

 

[mitch074]

[...]

But they went with that, since in the worst case they are just competing with themselves, and at best they can claim that they have the fastest processor at SOMETHING, even if its advantage over the nearest, much cheaper competition is almost irrelevant. For "money is no object" gamers, this will be the go-to processor, no matter how wasteful of a purchase that is, but considering the tiny market for >$500 processors in general, I guess that's a better fit than any other.

I agree - yet you forgot one point : considering how much cherry-picking they probably have to do to manufacture this chip in any reliable way with how much power it guzzles, they'd rather not sell too many of them.
Thus, they can claim the crown ("best gamin' CPU evah") and not deliver ("it's too much for too little, I'm not buying that!").

Marketing dept. at Intel is the one defining what product Intel sell. You'll tell me, that's an improvement from the P4 days where they actually designed them ("highest clock'd CPU evah" losing to chips with half the clock speed).

I can't help but wonder, would AMD cherry-pick one of their Zen+ core and sell it with a 200W power envelope - what clock speed and actual performance would it get? I bet that they would still lose in some games (which are compiled for Intel chips only) but absolutely decimate them everywhere else. But that would mean more R&D and potentially cannibalizing their Threadripper market for a game of "mine's bigger!" - not exactly smart.

 

[volkgren]

...wait, an i7 without hyperthreading? Something is seriously wrong there. I could understand that on an i5 or i3, but an i7? You're talking Core Duo Quad L775 there.

Int he i7-9700K's defense, it did defeat the 8700K in pretty much all games....which is saying something....; so clearly having 'just 8 cores and 8 threads' is not quite a crippling condition....

That was surprising until you think about the clock speed difference. All-core turbo for the 8700K is 4.3GHz while the 9700K is 4.7GHz. But, the 9700K still might have the edge of the 8700K. I went ahead and ordered the 9700K since the 9900K isn't available. 8C/8T is perfect right now.

Anyone know if consoles will be moving to 8C/16T CPUs in the next-gen?

 

[logainofhades]

I see Tom's Hardware is continuing their bullshit shilling for corporations. First the "NVIDIA Just Buy It" article, now they are spreading false information "Fastest Gaming Processor Ever" and shilling for Intel
Stop discrediting yourself even TOM

It is the fastest gaming processor ever. It just has a very horrible price/performance ratio. For the price of a 9900k, You can get a 2700x, x470 motherboard, and 16gb of 3600 ram.

PCPartPicker part list / Price breakdown by merchant

CPU: AMD - Ryzen 7 2700X 3.7GHz 8-Core Processor ($304.89 @ OutletPC)
Motherboard: MSI - X470 GAMING PLUS ATX AM4 Motherboard ($116.98 @ Amazon)
Memory: G.Skill - Ripjaws V 16GB (2 x 8GB) DDR4-3600 Memory ($144.99 @ Newegg)
Total: $566.86
Prices include shipping, taxes, and discounts when available
Generated by PCPartPicker 2018-10-22 09:19 EDT-0400

or for a bit more than the 9900k's newegg price, something like this.

PCPartPicker part list / Price breakdown by merchant

CPU: AMD - Ryzen 7 1700 3GHz 8-Core Processor ($179.99 @ Newegg)
Motherboard: ASRock - X370 KILLER SLI/ac ATX AM4 Motherboard ($133.17 @ Amazon)
Memory: G.Skill - Ripjaws V Series 16GB (2 x 8GB) DDR4-3200 Memory ($139.99 @ Newegg)
Storage: Intel - 660p Series 512GB M.2-2280 Solid State Drive ($89.99 @ Newegg)
Storage: Seagate - BarraCuda 1TB 3.5" 7200RPM Internal Hard Drive ($39.99 @ Amazon)
Total: $583.13
Prices include shipping, taxes, and discounts when available
Generated by PCPartPicker 2018-10-22 09:39 EDT-0400

 

[vapour]

thanks for the "real" benchmark comparison than the bogus principled tech...
Nothing surprising here, 9900k is better in many occasions, but mostly not 50% as claimed before. Not worth the cost if you want better bang for the money. However, if money is not an issue, 9900k is best choice all around so far. I am looking forward to 2800X's beating 9900k in productivity work soon.

 

[SkyBill40]

I am looking forward to 2800X's beating 9900k in productivity work soon.

While AMD has said they've got something in the wings that they've not seen the need in releasing, they've made no other indications beyond that on if that'll be a binned 2700X (and called 2800X) or something completely different. Will we see it now that the 99xx series are out or will we just have to wait until Zen 2 drops next year?

I personally doubt they drop anything else this year but that's just me. I suppose time will tell.