Exclusive: Testing Intel's Unreleased Core i9-9900KS

Aspiring techie

Reputable
Mar 24, 2015
824
8
5,365
118
There's almost zero reason to buy this CPU if the rumored price is accurate. It's only for those who want the absolute best gaming experience. For almost every other usage case in the $500 price bracket, the 3900X is the clear winner.

I still find it funny that AMD's 12 core CPU consumes less power than Intel's 8-core. My how the times have changed...
 

jimmysmitty

Polypheme
Moderator
There's almost zero reason to buy this CPU if the rumored price is accurate. It's only for those who want the absolute best gaming experience. For almost every other usage case in the $500 price bracket, the 3900X is the clear winner.

I still find it funny that AMD's 12 core CPU consumes less power than Intel's 8-core. My how the times have changed...
Considering the clock speed difference and that while its not quite 7nm it is still a smaller node than Intels 14nm its not surprising at all. If anything its to be expected. If Intel was using less power with a massive clock speed advantage on a larger process then something would be wrong.
 

Aspiring techie

Reputable
Mar 24, 2015
824
8
5,365
118
Considering the clock speed difference and that while its not quite 7nm it is still a smaller node than Intels 14nm its not surprising at all. If anything its to be expected. If Intel was using less power with a massive clock speed advantage on a larger process then something would be wrong.
I totally agree. 7nm especially has been a huge advantage on that front. It's just that AMD CPUs have traditionally been power hogs while Intel has been fast and power efficient. I find it funny that the roles are somewhat reversed now.
 

vinay2070

Distinguished
Nov 27, 2011
62
2
18,645
2
9900K @5Ghz performs better than 9900KS @5GHz and costs lesser ? OMG! I understand there are bug fixes, but the cost increase for very little performance increase, that too at 5.2GHz OC is not worth it IMHO. For the price of this CPU, I can get the 3700X/Mobo/16GB ram and it does not make a lot of difference at 1440P anyway.
 
Reactions: dorsai and unityole
Oct 18, 2019
1
0
10
0
AMD 3900X is not a 14nm CPU. Guess that is a typo, should be 7 nm.
Another typo is the 49.95 per thread price for the 3900X. 499/24=49.95?? The whole point of this was to try and act like Intel was releasing a competitive product, but they are just pulling the same stuff as when they did the 8086K. A special bin for <2% increase and more $$
 

PaulAlcorn

Senior Editor
Editor
Feb 24, 2015
704
56
5,060
0
Another typo is the 49.95 per thread price for the 3900X. 499/24=49.95?? The whole point of this was to try and act like Intel was releasing a competitive product, but they are just pulling the same stuff as when they did the 8086K. A special bin for <2% increase and more $$
Good eye, fixed!
 

jimmysmitty

Polypheme
Moderator
9900K @5Ghz performs better than 9900KS @5GHz and costs lesser ? OMG! I understand there are bug fixes, but the cost increase for very little performance increase, that too at 5.2GHz OC is not worth it IMHO. For the price of this CPU, I can get the 3700X/Mobo/16GB ram and it does not make a lot of difference at 1440P anyway.
Must be an extremely cheap board since the 3700X will take up over 50% of the cost.

Another typo is the 49.95 per thread price for the 3900X. 499/24=49.95?? The whole point of this was to try and act like Intel was releasing a competitive product, but they are just pulling the same stuff as when they did the 8086K. A special bin for <2% increase and more $$
The 8086K was a celebratory CPU and considering the original 8086 is why we are all here I see no reason for them not to celebrate history.

AMD has done the same, more recently they sold s 2700X 50th anniversary CPU that was signed by Lisa Su. Its just like buying anything labeled "Special Edition".
 

bit_user

Splendid
Herald
Yet another super underwhelming Intel release.
Um, it's not even pretending to be a new generation. How much improvement are you expecting, for simply adding a 'S' on the model number?

For me, the power graphs (page 2) say it all:


152 vs. 200 Watts? That's huge!

Or, to look at it another way, 4% more overclocking for 10% less power. Definitely a worthwhile update!
 

jimmysmitty

Polypheme
Moderator
Um, it's not even pretending to be a new generation. How much improvement are you expecting, for simply adding a 'S' on the model number?

For me, the power graphs (page 2) say it all:


152 vs. 200 Watts? That's huge!

Or, to look at it another way, 4% more overclocking for 10% less power. Definitely a worthwhile update!
Thats actually pretty impressive TBH.
 
Reactions: Phaaze88

bit_user

Splendid
Herald
9900K @5Ghz performs better than 9900KS @5GHz and costs lesser ? OMG!
Not if you compare them both running @ the same memory speed.

From the top of page 3:
We present the 'Core i9-9900K @ 5.0 GHz DDR4-2666' entry so you can compare an overclocked -9900K to the stock KS model at 5.0 GHz with the same memory settings.

I might've missed clicking through a couple of the graphs, but I didn't see a case where the K @ 5.0 GHz with DDR4-2666 significantly outperformed the KS (stock) with same.
 
Last edited:

bit_user

Splendid
Herald
PaulAlcorn said:
As we can see, several workloads show some IPC regression, albeit mostly minor ones. It's important to remember that IPC can vary by workload, so dissimilar tasks may yield different outcomes.
Thanks for the tests, but FWIW the bigger impacts are going to be seen in more I/O-intensive and message-passing workloads. For instance, Phonix' recent testing found an average (GeoMean) slowdown of 11.1% for the i9 9900K and only 4.3% for Ryzen 9 3900X:

https://www.phoronix.com/scan.php?page=article&item=3900x-9900k-mitigations

It seems the i9 9900K's Achilles heel is context switching. In Stress-NG, it suffered a 67.0% slowdown in context switches/sec. In a different test, context switch time was measured to increase by over 3.6x, with mitigations enabled.

Obviously, your testing should be biased towards how this site's readers are likely to use the product being reviewed, but I'd suggest an I/O heavy app be added to the mix. Perhaps you could measure the time needed to do an incremental backup, where very little had changed on the client PC, as that should be dominated by the time needed to crawl through the filesystem. Or, maybe something like a chkdisk, on a NVMe storage volume with lots of files.
 

sreams

Distinguished
Jul 15, 2009
6
0
18,510
0
It's amazing to me just how many benchmarks completely ignore the extra cores/threads on the 3900x. Several of them have the 3700x in a tie with the 3900x. It seems to be a lack of optimization for more than 8 cores that is keeping the 3900x (and soon 3950x) from dominating many tests.
 

bit_user

Splendid
Herald
It's amazing to me just how many benchmarks completely ignore the extra cores/threads on the 3900x. Several of them have the 3700x in a tie with the 3900x. It seems to be a lack of optimization for more than 8 cores that is keeping the 3900x (and soon 3950x) from dominating many tests.
Um, so like which benchmarks? I'd look for myself, but there's some buggy full-screen ad that keeps popping up which I cannot close.

Anyway, possible reasons for lack of scaling include:
  • GPU-bottleneck - common for games.
  • I/O-bottleneck - common for compression or productivity benchmarks.
  • Memory bottlenecks - there's only so far you can push a dual-channel memory subsystem.
  • Poorly-scaling algorithms - threading model scales poorly or use of algorithms that do not scale well.
  • Poor load-balancing - algorithms might scale, but load is not well-distributed.
  • Implementation issues - lock contention, cache contention, heavy context-switching.
In the first 3 cases, there's not necessarily a lot to be done about it. In the latter 3, there are usually solutions, but you have to weigh the implementation complexity (and potentially more overhead that could weigh down fewer-core CPUs) against the benefits of the improvements. With most users having <= 8-core CPUs, it's going to be an uphill battle to improve scaling much.
 

mitch074

Distinguished
Must be an extremely cheap board since the 3700X will take up over 50% of the cost.
Thing is, with the 3700X's TDP of 65W, you can use it fully up to spec even with a basic B450-based $60 board; falling RAM prices make a 16 Gb 2666 RAM kit fit into that too.
So yeah, for the price of a single Intel CPU (without cooler) you get an almost complete AMD system that performs 5-10% worse on average - that's quite a premium for so little benefit. Now, AMD systems don't really overclock, and you sure can push some extra performance out of these i9s, but consider that to get it you need a $170+ motherboard, a $100+ AIO and a beefy power supply...
On the AMD side, getting your hands on a 3900X does not require you to beef up the rest of your system; a $100 B450 motherboard with better power delivery is the only thing you'd really need.
 

PCWarrior

Honorable
May 20, 2013
16
5
10,515
0
The in-silicon mitigations are not meant to hurt performance when compared against a software-mitigated system. If anything, they are meant to claw lost performance back. So, was this comparison made against a 9900K with software mitigations that take care of the exact same bugs? I don't think so. In any case I am willing to bet that this slight 1-2% IPC performance regression has nothing to do with the mitigation and instead is due to being a pre-release sample and due to using an early bios. With a proper bios release it will claw back the lost performance and probably even improve it. I remember there was the exact same “issue” with the 8086K upon release versus the 8700K when both were clocked at 5GHz (all cores fixed) and the 8700K was consistently scoring higher in benchmarks. It turned out this was due to early bios and that a subsequent bios optimisation restored parity and even improved the performance of the 8086K over its predecessor at equal speeds.

On a different note, this is the third time (in the mainstream platform) in recent years that Intel has re-released the exact same processor (4770K -> 4790K, 8700K -> 8086K, 9900K -> 9900KS) with improved frequencies achieved by process maturity alone (i.e. as the process matures, what was once considered to be a golden sample is now the norm so they can increase stock frequencies). And the fifth time they have done it if you also count an actual process modification (6700K->7700K) or a move to an entirely new process node (2600K->3770K), but as these moves also included added features (PCIE3, boost algorithm improvements, 4K video encoding) I don’t personally count them.

As for the price, I think it will be $545, not $559, we will see. But the better power efficiency (152W versus 200W, a 24% power reduction @5GHZ) plus the higher OCing potential are really worthwhile improvements. The only issue is that, like it happened with the hexacore 8086K with the 5GHz single core turbo boost (but 4.3GHz all core), it looked good at the time but a few months later Intel released a cpu with even better clock speeds (5GHz dual core turbo and 4.7GHz all core), better efficiency and more cores - the octacore 9900K. And that is exactly what will happen again. Comet lake is expected to be available for purchase by February 2020 i.e. in 4 months and that on a new platform, and is poised to have the same or even higher stock speeds and that with 10 cores.
 
Reactions: bit_user
Thing is, with the 3700X's TDP of 65W, you can use it fully up to spec even with a basic B450-based $60 board; falling RAM prices make a 16 Gb 2666 RAM kit fit into that too.
So yeah, for the price of a single Intel CPU (without cooler) you get an almost complete AMD system that performs 5-10% worse on average - that's quite a premium for so little benefit. Now, AMD systems don't really overclock, and you sure can push some extra performance out of these i9s, but consider that to get it you need a $170+ motherboard, a $100+ AIO and a beefy power supply...
On the AMD side, getting your hands on a 3900X does not require you to beef up the rest of your system; a $100 B450 motherboard with better power delivery is the only thing you'd really need.
You can't just mix any random info you want to get the outcome you like.
At 65W the 3700x will run an all core of 3.6 maybe 3.7 that means that anything that is not GPU limited will run at least 35% faster on the 5Ghz all core 9900ks,higher if you push it even more.
You only get a 5-10% slower Ryzen system if you push your 3700x quite hard to reach that 4.4Ghz all core which isn't even possible on most CPUs,you surely won't be able to do this at 65W TDP and probably also won't be able to do this on a cheap mobo.
 
Sep 29, 2019
21
3
15
0
This CPU has no place , For Gamingi n which you need more frequency , 16 threads are not fully utilized. hence loss of money.

and for productivity , more cores at the same price with Ryzen 12/24 cores/threads is better at the same price.

Nothing to look at here.
 
Last edited:

ASK THE COMMUNITY

TRENDING THREADS