Discussion Intel 9900KS review and availability thread

[Hitman928]

Since the other thread was started specifically for the Tom's preview article, here's a thread for actual review discussion. I'll add links to this post as more come out. If you see a review posted in the thread not posted in the OP just send me a PM and I will update.

I haven't seen the CPU in stock anywhere yet but will update this post when I do.

In stock at newegg for $570 and at amazon for $613.75

https://bit-tech.net/reviews/tech/cpus/intel-core-i9-9900ks-review/1/
https://www.computerbase.de/2019-10/intel-core-i9-9900ks-cpu-test/
https://www.techspot.com/review/1936-intel-core-i9-9900ks/
https://www.tomshardware.com/reviews/intel-core-i9-9900ks-special-edition-review/
https://www.phoronix.com/scan.php?page=article&item=intel-corei9-9900ks&num=1

 

[Kenmitch]

It has cc, but it goes by too quick, so what score did he get in cb15 ?

LN2 pot with dry ice and acetone.

2176 @5.6GHz
2250 @5.8GHz
2353 @6GHz

 

[Markfw]

LN2 pot with dry ice and acetone.

2176 @5.6GHz
2250 @5.8GHz
2353 @6GHz

So a 3700x beats it on air (maybe aio?) at 5.8 ? It takes the whole 6 ghz to beat it ? and dry ice ? Thats kind of pathetic.

And thanks for the summation @Kenmitch

 

[Arkaign]

Any 9900ks at 5.2 gaming reviews where the gpu is not bottle necking the scores. Something with a 2080ti at 1080p, mabe not using ultra settings?
I see the 9900ks wins in just about every gaming benchmark but I'm curious how well it will handle a 40% faster 3080ti next year at 1440p.
Techpowerup usually does a nice job with this but there review is not up yet.

I noticed that. What I also noted was some massive gaps in minimums, like from 80 (3900X OC) to 122 (9900KS OC). And that's with Ultra settings, which I'm only too happy to dial back in pursuit of 144hz and beyond. As a 2080ti high refresh gamer, this thing is about exactly what I need. 8086k/8700k/3700X/3900X are all slow enough to leave me with CPU bottlenecks. Even the 9900KS isn't perfect, I think it would have to be another 20-30% faster to completely eliminate CPU bottlenecks in situations where you're not GPU limited.

The unwritten rule where benchmarks are done with all ultra settings skews things somewhat. I've noticed it in how GPUs like Vega 56/64, GTX 1070ti/1080 don't look in reviews like you could use them at 4k for 60fps, and that IS true if you insist on 4K ULTRA lol. But dialing down AA, shadows, etc, basically a little bit of testing and optimization, and boom, no problem. The same thing is my experience at 1440p high refresh. Benchmarks might make it appear that you would be GPU limited, but with reasonable settings I actually changed to CPU bottlenecks in many situations even when I had 1080 Strix, 1080ti Aorus, let alone the 2080ti.

Hopefully something can deliver that major move forward, be it Zen3, Intel 10th gen, whatever.

 

[jpiniero]

Hopefully something can deliver that major move forward, be it Zen3, Intel 10th gen, whatever.

That would be the new consoles. You'll have to wait until then.

 

[TheGiant]

The Intel Core i9-9900KS Special Edition Review: 5.0 GHz on All the Cores, All the Time

Intel's Core i9-9900KS arrives with impressive gaming performance.

www.tomshardware.com

toms has a good test
based on the differences in power between the sites, I dont believe its just only the sample
but the "package power" drives me crazy, they should add the wall power
EDIT:

there is still a significant difference my old progaming friends are telling me 9900K@5+ is the best, they tried the ryzens 3k and its still not there
toms shows it in the frametimes charts
I think they removed the 2700X, cause it would look like crap

 

[itsmydamnation]

there is still a significant difference my old progaming friends are telling me 9900K@5+ is the best, they tried the ryzens 3k and its still not there
toms shows it in the frametimes charts
I think they removed the 2700X, cause it would look like crap

I dont know what data your looking at but in doesn't match your statement, there doesn't appear to be any behavior difference in frame times and actual performance is quite close as well.

 

[Zucker2k]

The Intel Core i9-9900KS Special Edition Review: 5.0 GHz on All the Cores, All the Time

Intel's Core i9-9900KS arrives with impressive gaming performance.

www.tomshardware.com

toms has a good test
based on the differences in power between the sites, I dont believe its just only the sample
but the "package power" drives me crazy, they should add the wall power

The OP only has three links... one to a German website. He curiously references the Tom's preview but makes no effort to link them, or Phoronix, etc, where the power numbers don't fit the popular narrative too well about how the 9900KS is just an overclocked 9900K.

On the phoronix Linux test, the first thing that jumps at the reader is context switching.



In Gaming the 3900x system curiously guzzles power while being slower. The 9900KS is the dominant and more efficient gaming processor platform. Here's what the reviewer says:

Even with the Core i9 9900KS having a 127 Watt TDP, the overall AC system power consumption with the i9-9900KS was still better off than the particular Ryzen 9 3900X comparison system.

Power Consumption for 116 Non-Gaming workloads:



And the geometric mean of all Non-Gaming tests:



Remarks:

Across dozens of Linux benchmarks, it was certainly an interesting showdown between the Core i9 9900KS and Ryzen 9 3900X trading blows. For the workloads that didn't scale as well or favored higher clock frequencies, the Core i9 9900KS stole the lead, but for those workloads scaling well the Ryzen 9 3900X easily took the crown.

Conclusion:

If taking the geometric mean of all those (non-gaming) results, the Core i9 9900KS was 14% higher over the Core i9 9900K. Unfortunately though that came up short still of the AMD Ryzen 9 3900X performance with its geometric mean being just under 10% faster. Then again, Linux has more multi-threaded workloads typically speaking than Microsoft Windows, so it will be interesting to see how the Core i9 9900KS vs. Ryzen 9 3900X performance pans out there. Of course, the one area where the Core i9 9900KS did come out hands-down stronger on Linux was the gaming performance. So for gamers out there with current-generation Linux games at least, the i9-9900KS does deliver better performance while being more power efficient. The other area where the Core i9 9900KS now tended to win most often was in the Chrome/Firefox web browser benchmarks, but for there you generally don't need a Core i9 / Ryzen 9 if all you are doing is browsing the web. The other good news with the Core i9 9900KS besides the raw performance is improved hardware mitigations for Spectre V2 and MDS.

Here's Tom's power consumption numbers:











Remarks:

We began our power testing with the AIDA 64 stress test with AVX enabled. The stock -9900KS draws 7W of additional power over the Ryzen 9 3900X. After overclocking, that gap widens to 17W, but it is important to remember that AMD's auto-overlocking precision boost overdrive feature (PBO) doesn't impart very large performance gains in most workloads. However, the feature does get us within range of the 3900X's maximum overclock ceiling.


The -9900KS only draws 12W more than the stock -9900K during the AVX test, but runs at 5.0 GHz on all cores. Meanwhile, the -9900K downshifts to 4.7 GHz under the AVX load. The KS also draws 20W less than the vanilla model after overclocking but gains an extra 200 MHz of clock speed. We see much of the same trend during the AIDA 64 non-AVX test.

Peak stock power draw for the -9900KS weighs in at 158W during the y-cruncher benchmark, which computes pi using a heavily-threaded AVX workload. That falls neatly in line with Intel's PL2 value (although the chip certainly boosts longer than the 28-second Tau guideline). The stock Core i9-9900KS stays under 150W during our AVX-enabled HandBrake x264 and x265 tests, that latter of which uses a heavier distribution of the densely-packed SIMD instructions.

The -9900KS consumes far less power at its stock 5.0 GHz than we need to push the -9900K to similar levels of performance, a delta that meets or exceeds 50W in some workloads. That equates to less waste heat dumping into your case and also helps explain the additional overclocking headroom.

Even though both processors are operating at 5.0 GHz on all cores, the stock Core i9-9900KS is much more efficient than the overclocked Core i9-9900K during our round of performance benchmarks. However, the Ryzen 9 3900X's hefty allotment of 12 cores and 24 threads easily provides more performance than Intel's eight-core -9900 models.

 

[TheELF]

Anybody watch the video from der8auer? His scores in CB 15 looks pretty unimpressive for the super high clocks he's achieved. I guess it explains why Intel now frowns on Cinebench as a benchmark.

For comparison I think the highest I've gotten with my 3700x is 2282 at a measly 4.325 GHz all cores.

You use CB to check the stability of overclocking because it uses a lot of the CPU as you see at the end he runs validation on cpu-z just for fun just to see how high he get it just for validation because that only uses a very small amount of the CPU.
Yeah Cinebench scales with the amount of instructions the cores has which was why FX having only two per half core was so bad at it.
intel added instructions to 10nm and immediately gained about 20% on CB.
Intel doesn't dig it because FOR MAINSTREAM there is no correlation towards anything that a mainstream user would run,nothing runs in the same way and nothing gaines the same level of performance,again for mainstream uses.
https://www.anandtech.com/show/14664/testing-intel-ice-lake-10nm/6

 

[DrMrLordX]

Current estimate is April/May, although it could be earlier. It's definitely not worth waiting for, unless it somehow overclocks well beyond 5 Ghz. As you can see the 5 Ghz 9700K delivers basically the same gaming performance as the 9900KS. You can see why Intel decided against releasing it early for K because of the shortages.

I think it's the opposite. Anyone seriously thinking about a 9900KS would be better off waiting for 10c Comet Lake-S.

None of the reviews I have seen have gotten it to 5.2 stable, just 5.1

That one German review had it at 5.3 GHz, I think.

Power numbers for this chip are all over the map. Seems odd.

 

[Hitman928]

The OP only has three links... one to a German website. He curiously references the Tom's preview but makes no effort to link them, or Phoronix, etc, where the power numbers don't fit the popular narrative too well about how the 9900KS is just an overclocked 9900K.

On the phoronix Linux test, the first thing that jumps at the reader is context switching.

I like the insinuation but maintaining this thread isn't my job and sometimes I get caught up doing, you know, my real job, which is why I asked people to PM me if there were reviews they thought I should add to the OP. . . checks inbox. . . nope, no PMs. So if you want to be upset about a review not appearing in the OP, maybe you should be upset at yourself for not reading the OP?

The context switching improvement happens because of the added hardware mitigations which tamz_msc already posted a link to.

Phoronix already mentioned in a previous review that their Zen2 systems for some reason use a significantly higher amount of power when they run Linux vs. Windows and they weren't sure why. It looks like it's still an issue under Linux for them but I haven't had a chance to read the review to see if they talk about it.

As far as power goes overall, GamersNexus has done the best job I've seen of going in depth about power use and how the 9900K(S) handle turbo frequencies and power limits. They also measure power at the CPU 12V rail so they get actual CPU draw without the added system draw so it's a more accurate reading than power at the wall. I didn't link to it yet in the OP because I was waiting for their print version to come out but I posted a link to their video on the first page of this thread.

 

[jpiniero]

I think it's the opposite. Anyone seriously thinking about a 9900KS would be better off waiting for 10c Comet Lake-S.

It's not really going to be any faster in games, unless it clocks higher.

 

[Hitman928]

Power numbers for this chip are all over the map. Seems odd.

It was the same with the 9900K. Basically intel has a set stock configuration but then almost none of the MB manufacturers are following actual stock parameters (even without enabling MCE) so you get these wild fluctuations in power between motherboards used at "stock". GamersNexus talked pretty in depth about it, its worth a watch if you can. They should have the print version out in a couple of days if you want to wait for that, but it might not be as detailed as their video.

 

[jpiniero]

Hmm, in AT's review they mention that there was talk of the 9900KS only being available for the "holiday season". That would only make sense if Comet Lake-S was being released at CES, which seems unlikely but that might have been the plan at one time.

 

[DrMrLordX]

It's not really going to be any faster in games, unless it clocks higher.

No, it'll be about the same. But:

1). It'll be cheaper than the 9900KS. Probably
2). It'll bring two more cores and maintain high game performance
3). It will be in a platform that may have more life left in it than dead-end Z390/Z370.

I see no reason to pay $570+ for a CPU that will be rendered mostly-obsolete by Intel in maybe sixth months.

It was the same with the 9900K. Basically intel has a set stock configuration but then almost none of the MB manufacturers are following actual stock parameters (even without enabling MCE) so you get these wild fluctuations in power between motherboards used at "stock". GamersNexus talked pretty in depth about it, its worth a watch if you can. They should have the print version out in a couple of days if you want to wait for that, but it might not be as detailed as their video.

Yeah I hadn't thought about the influence of mobo OEMs in all this. What a mess.

 

[happy medium]

I think the Anandtech review said it best.
"Ultimately the Core i9-9900KS is going to end up in the hands of enthusiasts who want nothing more than the best, but don’t want to jump to the high-end desktop platform"

The Intel Core i9-9900KS Review: The 5 GHz Consumer Special

www.anandtech.com

 

[Markfw]

I think the Anandtech review said it best.
"Ultimately the Core i9-9900KS is going to end up in the hands of enthusiasts who want nothing more than the best, but don’t want to jump to the high-end desktop platform"

The Intel Core i9-9900KS Review: The 5 GHz Consumer Special

www.anandtech.com

And the sentence after that:
"Despite the Intel chipsets for consumers, it’s still a shame that these processors only have 16 PCIe 3.0 lanes, given the desire for direct attached PCIe storage in this market. "

And this one farther up
"One question that does remain however, is which set of results should we keep? The 255W results are what we get out of the box, and the 159W results are only 'Intel guidelines that Intel expects none of the board manufacturers to keep to'. Ideally we keep both, but that's a mess in its own right. "

The bottom line is, its expensive, limited edition, very limited quantity, HOT and "usually" power hungry. (that all depends on motherboard settings, as they said, but the lower power, reduces performance). Also, it looses in MT uses to the 3900x almost all the time, at a greater cost. And no heatsink is included, most all reviews went with a 360 AIO to keep it sort of cool.

 

[DigDog]

currently being sold for 540 GBP here ($700)

Intel Core i9-9900KS 4 GHz 8-Core Processor

uk.pcpartpicker.com

 

[Arkaign]

I think the Anandtech review said it best.
"Ultimately the Core i9-9900KS is going to end up in the hands of enthusiasts who want nothing more than the best, but don’t want to jump to the high-end desktop platform"

The Intel Core i9-9900KS Review: The 5 GHz Consumer Special

www.anandtech.com

Yeah, it's rather definitely the best high refresh gaming CPU, but a horrible value for productivity compared to HEDT or the bigger Zen2 SKUs.

Definitely a niche inside a niche, even for gaming.

60fps gaming? Ryzen 3600/9400f

Mid-range GPU? Ryzen 3600/9400f

4k Ultra Gaming? Ryzen 3600/9400f

144hz w/1080ti or better AND you run settings that reveal CPU bottlenecks? THAT is the rare case use where it pays off, and when you want a 4.8Ghz or better Coffee Lake 6C/8C.

Even as someone who wants one and will see a benefit, I think it's a poor value for most users.

Still, it's a definite truth that even the 9900KS@5.2 is not enough to get some titles to 144fps even at 1080p. It's in those cases where I'd rather be at 120-130fps vs 90-100 though. Which is the situation I'm many games today when comparing an OC 3700X/3900X, or non-K stock i5/i7.

I'm not sure it's going to get better any time soon either. I have my doubts that Zen3 or Intel 10th gen will notch past a 5+GHz 9900ks in gaming.

 

[happy medium]

Yeah, it's rather definitely the best high refresh gaming CPU, but a horrible value for productivity compared to HEDT or the bigger Zen2 SKUs.

Definitely a niche inside a niche, even for gaming.

60fps gaming? Ryzen 3600/9400f

Mid-range GPU? Ryzen 3600/9400f

4k Ultra Gaming? Ryzen 3600/9400f

144hz w/1080ti or better AND you run settings that reveal CPU bottlenecks? THAT is the rare case use where it pays off, and when you want a 4.8Ghz or better Coffee Lake 6C/8C.

Even as someone who wants one and will see a benefit, I think it's a poor value for most users.

Still, it's a definite truth that even the 9900KS@5.2 is not enough to get some titles to 144fps even at 1080p. It's in those cases where I'd rather be at 120-130fps vs 90-100 though. Which is the situation I'm many games today when comparing an OC 3700X/3900X, or non-K stock i5/i7.

I'm not sure it's going to get better any time soon either. I have my doubts that Zen3 or Intel 10th gen will notch past a 5+GHz 9900ks in gaming.

I just read over some 1080p med setting and some 720p gaming scores.
On average the 9900ks without overclocking is 14% faster than a AMD 3900x.
17% faster when overclocked.

One thing I did notice is AMD cpu have horrible game minimums.

Fast foward 10 months from now, Nvidia releases its 3080ti which is 40% faster than the 2080ti at 4k,1440p and 1080p.
1440p is no longer gpu bottlenecked ,that's when the intel Intel 9900ks starts to shine.
I told this same story once before back when the AMD 1700 vs the 8700k debate was hot.
What's happen since then?
Those who bought a 8700k are now enjoying top speed gaming from a 2080 or 2080ti gpu, where the 1700 buyers ,have since bought a 2700 and now a 3700 and still can't match a overclocked 8700k.
Meanwhile the argument back then was but AMD motherboards and cpu's combos are $165 cheaper.
Fast foward 2 year to now, That same AMD buyer that saved $165 has bought 2 more cpu's and another motherboard for about $500 ,while the 8700k guy still has the fastest gaming system.
Who made the better choice?

33% of the United states population plays PC games, that's about 120 million .
I wonder how many need a 14 core cpu to do some video encoding or other tasks that would require a 12 or 16 core CPU?
If you own the cpu gaming market, you own the highest consumer based market.

 

[thesmokingman]

It has cc, but it goes by too quick, so what score did he get in cb15 ?

He got 2087, yikes. A 3900x should be around 3000 and some. Oh duh, its right there in his CD windows, 3900x = 3108.

And cue the jacked price whiners... incoming.

 

[krumme]

Yeah, it's rather definitely the best high refresh gaming CPU, but a horrible value for productivity compared to HEDT or the bigger Zen2 SKUs.

Definitely a niche inside a niche, even for gaming.

60fps gaming? Ryzen 3600/9400f

Mid-range GPU? Ryzen 3600/9400f

4k Ultra Gaming? Ryzen 3600/9400f

144hz w/1080ti or better AND you run settings that reveal CPU bottlenecks? THAT is the rare case use where it pays off, and when you want a 4.8Ghz or better Coffee Lake 6C/8C.

Even as someone who wants one and will see a benefit, I think it's a poor value for most users.

Still, it's a definite truth that even the 9900KS@5.2 is not enough to get some titles to 144fps even at 1080p. It's in those cases where I'd rather be at 120-130fps vs 90-100 though. Which is the situation I'm many games today when comparing an OC 3700X/3900X, or non-K stock i5/i7.

I'm not sure it's going to get better any time soon either. I have my doubts that Zen3 or Intel 10th gen will notch past a 5+GHz 9900ks in gaming.

I think what you really want is good coding and an engine made from scratch for dx12.

I game a bit 60hz with a 8700k and one of my kids game a lot of high res 144 on a 3600. He does the same stuff af you. Set most low and optimizes the entire settings for high res. We havnt bothered to change cpu. Sure there is some old games that is more like a memory bm, but they get played more and more seldom. A 3600 pulls mins in more throughout limited games like bf5 or even lighter games like overwatch pretty excellent. In one year you absolutely want 8c for those new heavy games anyway. The new consoles will make most high end high res gaming cpu absolete.

For high res gaming my advice would be to take a cheap 450 mb, plug in a cheap 3600 and replace it in 1-2 years with a cheap 4000 series 8c.

Sure you can get a 9900ks now and a 4000/i9 whatever each year forward but man you really really pay insane amounts for that strategy. Better pay for some game coaching. Seriously. Better investment for ranking.

 

[Arkaign]

I think what you really want is good coding and an engine made from scratch for dx12.

I game a bit 60hz with a 8700k and one of my kids game a lot of high res 144 on a 3600. He does the same stuff af you. Set most low and optimizes the entire settings for high res. We havnt bothered to change cpu. Sure there is some old games that is more like a memory bm, but they get played more and more seldom. A 3600 pulls mins in more throughout limited games like bf5 or even lighter games like overwatch pretty excellent. In one year you absolutely want 8c for those new heavy games anyway. The new consoles will make most high end high res gaming cpu absolete.

For high res gaming my advice would be to take a cheap 450 mb, plug in a cheap 3600 and replace it in 1-2 years with a cheap 4000 series 8c.

Sure you can get a 9900ks now and a 4000/i9 whatever each year forward but man you really really pay insane amounts for that strategy. Better pay for some game coaching. Seriously. Better investment for ranking.

You'd be better off swapping those CPUs. It's not just old games. It's basically ALL of them. Like Outer Worlds, caps at mid 80s with 4.4Ghz 3900X, but over 115 with 9700k. Same with GTAV, Far Cry 5, COD Modern Warfare (new engine 2019 version from last week), AC Odyssey, FFIV, Shadow of the Tomb Raider, Metro Exodus, etc. Always with a solid gap in minimums and especially average FPS.

Now ultra settings, 4k gamers, and 60fps gamers will not benefit whatsoever, and would be better off with Ryzen or cheaper Intel builds. But 144hz gamers who optimize for a mix of medium/high (especially if you cut down AA and shadows, two GPU bottlenecks), you see a huge benefit from the faster option.

It's the same thing in reverse for people who need heavy encode/render workhorses. Eg an 8700k is a complete waste of money next to a 3700x, or the 9900k is a waste of money next to a 3900x, it just depends what you want to do. It's why I have AMD Zen2 for my general purpose PC, and 8086k (until 9900ks) for the gaming rig. With a Threadripper 2920 for DC and general tinkering.

 

[Abwx]

One thing I did notice is AMD cpu have horrible game minimums.

To the contrary, they are very good, perfs tank less than Intel CPUs in respect of average/max FPS..

Intel Core i9-9900KS mit 5,0 GHz im Test: Benchmarks in Games und Anwendungen sowie Preis/Leistung

Intel Core i9-9900KS im Test: Benchmarks in Games und Anwendungen sowie Preis/Leistung / Benchmarks in Spielen (Full HD und UHD)

www.computerbase.de

 

[Arkaign]

To the contrary, they are very good, perfs tank less than Intel CPUs in respect of average/max FPS..

Intel Core i9-9900KS mit 5,0 GHz im Test: Benchmarks in Games und Anwendungen sowie Preis/Leistung

Intel Core i9-9900KS im Test: Benchmarks in Games und Anwendungen sowie Preis/Leistung / Benchmarks in Spielen (Full HD und UHD)

www.computerbase.de

It's the polar opposite of what Gamers Nexus gets.

As someone with a 2700X and 3700X and an 8086K, the AMD numbers there look fine, but their 9900KS runs worse than my 8086K lol. Mins (especially 1%) definitely hurt more on the 3700X for me. To be fair I'm running 3400 Ram with the 3700 w/Taichi, and 4000 on the 8086K w/Aorus 5. I moved the 2080ti to the gamer box, and swapped the 1080ti to the heavy lifting Ryzen.

 

[Arkaign]

Wait a minute, 2666 Ram? LOL what kind of absolute tool would pair a 9900KS with 2666. That's an abomination. And 'spec' is no excuse. They're enthusiast models. We live in an era where Gamers Nexus, Hardware Unboxed, Linus, etc videos get millions and millions of views. It's not a mystery to get some competent parts picked out. And ANY decent 370/390 board has the high end memory speeds plastered all over the packaging and marketing.