Monday, July 30th 2018

Intel Core i9-9900K 3DMark Numbers Emerge: Beats Ryzen 7 2700X

Some of the first benchmark numbers of Intel's upcoming 8-core/16-thread socket LGA1151 processor, the Core i9-9900K, surfaced, from Thai professional overclocker TUM APISAK. A 3DMark database submission sees the processor score 10,719 points in the CPU tests, with an overall score of 9,862 points, when paired with a GeForce GTX 1080 Ti graphics card. According to WCCFTech, the CPU score is about 2,500 points higher than the 6-core/12-thread Core i7-8700K, and about 1,500 points higher than the 8-core/16-thread AMD Ryzen 7 2700X. The tested processor features 8 cores, 16 threads, a nominal clock of 3.10 GHz, and boost frequency of 5.00 GHz, as measured by 3DMark's internal SysInfo module. Intel is expected to launch the Core i9-9900K on 1st August, 2018.
Sources: TUM APISAK (Twitter), WCCFTech
Add your own comment

65 Comments on Intel Core i9-9900K 3DMark Numbers Emerge: Beats Ryzen 7 2700X

#1
londiste
1500 points is 16% which is notably higher.
We do not seem to know what clocks that CPU was running at. Difficult to draw any conclusions.

Base clock is listed at 3.1 GHz (compared to 3.6 GHz from most other rumors/listings) and I saw comments in several places claiming the CPU was running at 4.1 GHz on all cores which may or may not be true. It would match the Coffee Lake behaviour though so it would not be unexpected. However, the results would match better with around 4.5 GHz if not higher.
Posted on Reply
#2
Arjai
Yawn, whatev's.
Posted on Reply
#3
techy1
well I did expect nothing less: intels i9 the 9-th gen (whatever it means these days) 8c/16t vs AMDs 300-ish $ 8c/16t
Posted on Reply
#4
Vayra86
Matthew Linke said:
I wonder if AMD was waiting for intel to come out with its i9-9900k before it drops its ryzen 2800x if it has one.
I don't think any Ryzen 2 will close this gap though.

Still, we haven't seen temps, actual clocks, volts, etc. 3DMark is not the best testbed for a CPU test. All it has is a physics bench and it scales almost linearly with HT/SMT which is not the real world. It barely sweats the CPU in terms of temperature or actual stress.
Posted on Reply
#5
nemesis.ie
Indeed, as we all know it's "easy" to have a machine bench stable but that would crash in a couple of minutes running something continuously.

We also don't know if it was stock or OCed.

If the launch is this week we will hopefully see some proper reviews pop up. :)
Posted on Reply
#6
btarunr
Editor & Senior Moderator
Matthew Linke said:
I wonder if AMD was waiting for intel to come out with its i9-9900k before it drops its ryzen 2800x if it has one.
There was a rumor from a semi-reliable source which later turned out to be unreliable, and I decided against running the story. The rumor spoke of a new 10-core/20MBL3$ die by AMD based on 12 nm Zen+. AMD seems to be focused on late-2018 rollout of "Rome," which is rumored to be based on 16-core/7nm/Zen2 dies. This die will debut in EPYC MCMs, and Ryzens based on it will follow sometime in 2019.

Till then, I guess we'll just see higher-clocked "Pinnacle Ridge" parts which could still fall short of the 9900K but be attractively priced.
Posted on Reply
#7
Vayra86
I just realized Intel went over 9000. At last.
Posted on Reply
#8
gmn 17
Took them long enough
Posted on Reply
#9
HTC
For AMD to beat this chip, it will require more then 8 cores because the clocks gap is simply too wide, and Intel also OCs more, which makes it even worse. They will only beat it in price and perhaps power consumption: pretty much all other metrics will be ahead on the I9-9900K.

AMD most likely saw this coming, which is why rumors of more then 8 core chips for AM4 platform began. I think Intel responded a bit quick and AMD are being forced to up their schedule: pretty sure they didn't expect to need to launch more then 8 cores on AM4 platform so soon.

That said, this will only work because desktop has "few" cores: for the server platform, i don't see this approach working due to the high core count requirements.

A question: has the exact architecture of I9-9900K been confirmed yet? I'm expecting it to be coffeelake, but this could potentially be skylake on a more refined 14nm process.
Posted on Reply
#11
dj-electric
Capitan Harlock said:
I wanna see the temps at 5ghz .
9000°C?
Not if its properly soldered.
Posted on Reply
#12
Vya Domus
HTC said:

A question: has the exact architecture of I9-9900K been confirmed yet? I'm expecting it to be coffeelake, but this could potentially be skylake on a more refined 14nm process.
Skylake , Kaby Lake , Coffee Lake are all the same uarch. Only difference is core count , cache and iGPU I think.
Posted on Reply
#13
chaosmassive
Ryzen inherently slightly slower in single core performance, lag behind 1-2 gen intel generation
coupled with ryzen chip maxed out to around 4.2 ghz,
it is not surprising ryzen 7 slower than intel counterpart with equal number of core/thread with superior clock speed
Posted on Reply
#14
bug
Capitan Harlock said:
I wanna see the temps at 5ghz .
9000°C?
It's still a 95W part, temps can't be too bad. The real question is for how long can it sustain 5GHz with a 95W power envelope.
Posted on Reply
#15
HTC
Vya Domus said:
Skylake , Kaby Lake , Coffee Lake are all based on the same uarch but built in different 14 nm process versions. Only difference is core count , cache and iGPU I think.
Corrected that for you.
Posted on Reply
#16
Vya Domus
HTC said:
Corrected that for you.
Well the architecture is independent from the node though Intel is still very keen on pointing out they use an optimized node.
Posted on Reply
#17
HTC
Vya Domus said:
Well the architecture is independent from the node though Intel is still very keen on pointing out they use an optimized node.
Then it's Haswell arch, because Broadwell (which is the 1st 14nm arch) is "just" a shrink of Haswell's 22nm process.
Posted on Reply
#18
Vya Domus
HTC said:
Then it's Haswell arch, because Broadwell (which is the 1st 14nm arch) is "just" a shrink of Haswell's 22nm process.
Skylake did bring a few actual changes to the architecture , as insignificant as they may be I wouldn't go as far as to say they are the same.
Posted on Reply
#19
dj-electric
bug said:
It's still a 95W part, temps can't be too bad. The real question is for how long can it sustain 5GHz with a 95W power envelope.
Some "95W parts" consume twice as much as others. TDP as we know it is long... long gone. Today its used to describe power consumption for base frequency, but even that is incorrect.
At his maximum potential, said 9900K would be able to cross the 200W line with ease.
Posted on Reply
#20
robert3892
I'll not make a judgment until the final pricing is announced.
Posted on Reply
#21
R0H1T
bug said:
It's still a 95W part, temps can't be too bad. The real question is for how long can it sustain 5GHz with a 95W power envelope.
I'm yet to see any place, on the web, where 95W is the official TDP. Where's this number coming from, apart from possibly Intel PR speak o_O
Posted on Reply
#22
Vayra86
R0H1T said:
I'm yet to see any place, on the web, where 95W is the official TDP. Where's this number coming from, apart from possibly Intel PR speak o_O
Every spec sheet Intel puts out?
Posted on Reply
#23
londiste
There are no official details about the 9000 series CPUs. Everything is technically a rumor at this point. However, the leaks we have come from manufacturer/retailer listings and seem to be very consistent so far. That makes them likely to be accurate.

bug said:
It's still a 95W part, temps can't be too bad. The real question is for how long can it sustain 5GHz with a 95W power envelope.
Indefinitely. On one core that the specs will say it can boost to 5.0 GHz. :D
Posted on Reply
#24
Vya Domus
Vayra86 said:
Every spec sheet Intel puts out?
Except this is a yet to be announced product with no spec sheet ?
Posted on Reply
#25
londiste
The leaked details sound very suspicious to me. 3.6 GHz base clock and 4.7 GHz all-core boost? 8700K has 3.7 GHz base on 6 cores. If 9900K can do 3.6 GHz on 8 cores within the same 95W power envelope as well as boost all cores to 4.7 GHz out-of-box, that last + in 14++ process would really be deserved with the efficiency boost.
Posted on Reply
Add your own comment