4GHz CPU Battle: Ryzen 3900X vs. 3700X vs. Core i9-9900K

[grvalderrama]

How is that IDLE?? How is 5+ GHz at below 32 degrees not impressive? I've been reading your posts above.... What are you an AMD employee or something??
I like AMD, I never use other than AMD Graphics cards, I'm just relaying what I'm seeing, If that confuses you, you got much bigger problems!!!

Must agree (again) with Puiu there's no way to get that temperature with a stock cooler (which that cpu doesn't have) and hardly even possible to achive with an after market one AT IDLE let alone at heavy cpu usage.

From what I read, you are running non intensive tasks and I'm guessing you must have a pretty decent cooling solution so that should be the reason why temps won't escalate to "normal" chip temps.

But please, do tell us what is your cooling solution so we understand how you get those thermal figures...

 

[hk2000]

Because it isn't impressive. It's just 1 or 2 cores being used in very short workloads.

You can't convince anyone who knows a thing or two about computers that your CPU works at 32 degrees, it's impossible. You are either misreading the temps or you have a sensor issue. Most web browsing involves some form of GPU acceleration so does video playback.

Run any modern game on it and without a good cooler you are looking at some pretty "hot temps". The 9900k is know to run hot, especially when not capped at 95W by cheap motherboards.

FYI you CAN insert an image here.
FYI2 here are some real results that everybody in the world can replicate: https://www.techspot.com/review/1744-core-i9-9900k-round-two/

You can insert an image if you upload it to a URL. I'm talking about an image on my PC- You cannot upload an image from your computer !
I told you I'm giving you what the ASUS utility is reporting. But that same utility is reporting the motherboard temp to be around 38 degrees, and my own hemostat is measuring about the same. Regardless, it is at 5GHz without any overclocking on my part. I couldn't care less what it does at 4GHz!!! and I understand the concept of the experiment- I just think it's an irrelevant experiment to owners of the i9 9900K.
Looking at the <Help> section, I now see there is a way to attach an image, but I still can't figure out how.

 

[Puiu]

You can insert an image if you upload it to a URL. I'm talking about an image on my PC- You cannot upload an image from your computer !
I told you I'm giving you what the ASUS utility is reporting. But that same utility is reporting the motherboard temp to be around 38 degrees, and my own hemostat is measuring about the same. Regardless, it is at 5GHz without any overclocking on my part. I couldn't care less what it does at 4GHz!!! and I understand the concept of the experiment- I just think it's an irrelevant experiment to owners of the i9 9900K.
Looking at the <Help> section, I now see there is a way to attach an image, but I still can't figure out how.

I'm sorry, but if you can't do something as simple as attacking a picture then how am I going to believe that you are someone tech-savvy? And I also have no idea how you are using a surgical tool (hemostat) to measure temps



And just to make sure I sound like a broken record: your CPU is not operating at 38 degrees during even half (50%) workloads, let alone high workloads. It makes zero sense. Just open a game, run it for 30 minutes and look at the temps shown by something like hwmonitor (from cpuid).

Most people who are using good expensive coolers are reporting sub 40 degrees during full idle.

 

[Ludak021]

single core performance = IPC x Clock speed
I can only count the years since they re-entered the market. yes development took longer.

Knowing the difference in IPC at the same clock speeds is very important because you can infer performance across the entire product stack without having to look up benchmarks.

Cheaper Intel CPUs have closer clock speeds to what AMD is offering. For example, knowing what Intel and AMD can achieve at 4GHz you can easily calculate what the 9400F is capable of at 4.1GHz boost clocks vs the 3600 at 4.2GHz (with very small real world differences).

I seriously don't understand why people are so confused about what was tested in this article and why.

I am not confused, other people that measure IPC of AMD vs Intel and downclock intel acting as if AMD CAN reach 5+ GHZ are confused. They think IPC matters in those cases. It doesn't. Comparing $180 intel to $500 AMD in IPC is also misguided. 9400F uses the same tech 8700K uses, and that one does 5 GHz too.

 

[Puiu]

I am not confused, other people that measure IPC of AMD vs Intel and downclock intel acting as if AMD CAN reach 5+ GHZ are confused. They think IPC matters in those cases. It doesn't. Comparing $180 intel to $500 AMD in IPC is also misguided. 9400F uses the same tech 8700K uses, and that one does 5 GHz too.

Again, this is not the point of the article. If you want a comparison of retail products then go read the reviews (there are plenty). You didn't understand the point of the article and also misread my comment (or misunderstood it too).

I can't explain in simpler terms than this, nor do I have the patience to do so for someone who doesn't want to understand. Such articles have always been made with new CPU generations to see how much they evolved over time.

 

[Ludak021]

I agree. I don't want to explain things to you either since you don't understand that comparing apples and organizes is pointless.

edit: as for the article, it's point is to praise AMD. They like them apples and I get that. Author could have told you something in the lines of IF AMD COULD reach Intel Speeds, the IPC would matter, but since it can't, intel is faster per core with worse IPC because iit reaches higher speed to make up the difference, and AMD gives you more cores and/or threads to make up the difference in speeds that it cannot reach.

 

[cliffordcooley]

I agree. I don't want to explain things to you either since you don't understand that comparing apples and organizes is pointless.

Pointless is bringing your point to this article. That is the point you apparently refuse to grasp. That is OK though, because there is always someone that does. I'm starting to understand why authors don't spend time in the comments.

 

[msroadkill612]

They already did that last week (linked in the very first phrase of the very first sentence of this article).

The reason they needed to do this article, however, is because people consistently misinterpret IPC when trying to compare AMD & Intel CPUs anymore. The original (& standard) reviews for new CPUs, to use the car analogy, are testing to see whether a brand-new Ford Mustang (4-cylinder, V6 & both V8 models), Chevrolet Camaro (V6 & V8 models), Chevrolet Corvette (turbocharged & non-turbocharged models), Dodge Challenger (V6 & V8 models), & Dodge Charger (V6 & V8 models) is not only the absolute fastest car but also the fastest to go from 0-60MPH. This review, on the other hand, says, "OK, we're going to put a horsepower limiter on each of these cars so that they have the same maximum horsepower, & see which one has the best performance across a fairly even playing field" -- & note, while I'm not a huge gearhead, even I know there's a significant difference in max horsepower on those different engine options for those cars.

Since apparently people have trouble reading & interpreting the charts as well, here's a normalized (Normalized results means the i9-9900K's performance = 100%, written as 100.00, looking at the 1% Low/Average FPS for all CPUs, all at 1080p resolution):

Battlefield V
-- 1700X: 83.04/84.375
-- 2700X: 83.93/86.25 [1%/2% improvement over 1700X]
-- 3700X: 91.96/93.125 [10%/8% improvement over 2700X, 11%/10% improvement over 1700X]
-- 3900X: 91.96/97.5 [10%/13% improvement over 2700X, 11%/16% improvement over 1700X]
-- 9900K: 100/100

Far Cry New Dawn
-- 1700X: 77.17/75.42
-- 2700X: 80.43/78.81 [4%/4% improvement over 1700X]
-- 3700X: 91.30/91.53 [13%/16% improvement over 2700X, 18%/21% improvement over 1700X]
-- 3900X: 91.30/94.92 [13%/20% improvement over 2700X, 18%/25% improvement over 1700X]
-- 9900K: 100/100

Total War: Three Kingdoms
-- 1700X: 88.07/94.53
-- 2700X: 89.91/95.31 [2%/1% improvement over 1700X]
-- 3700X: 94.50/96.09 [5%/1% improvement over 2700X, 7%/2% improvement over 1700X]
-- 3900X: 97.25/97.66 [8%/3% improvement over 2700X, 10%/3% improvement over 1700X]
-- 9900K: 100/100

World War Z
-- 1700X: 89.61/84.36
-- 2700X: 91.56/86.73 [2%/3% improvement over 1700X]
-- 3700X: 96.10/94.79 [5%/9% improvement over 2700X, 7%/12% improvement over 1700X]
-- 3900X: 100/98.58 [9%/14% improvement over 2700X, 12%/17% improvement over 1700X]
-- 9900K: 100/100

World of Tanks
-- 1700X: 90.40/86.12
-- 2700X: 92.93/87.90 [3%/2% improvement over 1700X]
-- 3700X: 100.00/96.09 [%/% improvement over 2700X, 11%/12% improvement over 1700X]
-- 3900X: 103.54/99.64 [%/% improvement over 2700X, 14%/16% improvement over 1700X]
-- 9900K: 100/100

Rainbow Six Siege
-- 1700X: 93.26/88.45
-- 2700X: 95.34/90.44 [2%/2% improvement over 1700X]
-- 3700X: 97.41/92.83 [8%/9% improvement over 2700X, 4%/5% improvement over 1700X]
-- 3900X: 98.96/94.82 [11%/13% improvement over 2700X, 6%/7% improvement over 1700X]
-- 9900K: 100/100

And although they didn't include one, here's the overall average
6-game Average
-- 1700X: 88.23/85.90
-- 2700X: 90.33/87.90 [2%/2% improvement over 1700X]
-- 3700X: 96.04/94.26 [6%/7% improvement over 2700X, 9%/10% improvement over 1700X]
-- 3900X: 98.25/97.39 [9%/11% improvement over 2700X, 11%/13% improvement over 1700X]
-- 9900K: 100/100

Yes, you're reading that right: using the 6 games they tested here, 3rd-gen Ryzen made double-digit percentage improvements over the original Ryzen CPUs, to the point that they pull within a few percentage points (roughly 5% or less) of the 9900K's performance, when everyone is operating at the same frequency (3% or less in the case of the 3900X). When was the last time you heard of an AMD chip having that close of an IPC result to an Intel CPU of the same generation? As for overclocking...again, unless it's going to be "everyone is overclocked to the same frequency", that's not what this article is looking at.

The TLR results: 1st-gen Ryzen was 12-15% slower than the i9-9900K, 2nd-gen improved slightly but was still 10-13% slower; 3rd-gen Ryzen, however, has pulled within 5% or less of the i9-9900K's performance. And for 95% of the gamers out there, that's going to be "more than good enough".

What a great post. It would be perfect if the popular 3600 were included.
Its cores dont match your sample, but I suspect it would show the savings are worth the downsides for most.