Core i9-10900K up to 30% Faster than i9-9900K: Intel

[btarunr]

Intel's upcoming Core i9-10900K desktop processor is up to 30 percent faster than the Core i9-9900K according to the company, which put out a performance guidance slide that got leaked to the web. Based on the 14 nm "Comet Lake-S" silicon and built for the new LGA1200 platform (Intel 400-series chipset motherboards); the i9-10900K is a 10-core/20-thread processor that leverages increased TDP headroom of 125 W to sustain higher clock-speeds than 9th generation "Coffee Lake Refresh," while also offering a 25% increase in processing muscle over the i9-9900K, thanks to the two additional CPU cores.

In its performance guidance slide, Intel shows the i9-10900K scoring 30% more than the i9-9900K in SPECint_rate_base2006_IC16.0. There's also a 25% boost in floating-point performance, in SPECfp_rate_base2006_IC16.0, which roughly aligns with the additional core count, as both these tests are multi-threaded. Other noteworthy results include a 26% gain in Cinebench R15, and 10% in SYSMark 2014 SE. In tests that don't scale with cores, Intel appears to rely entirely on the increased clock-speeds and improved boosting algorithm to eke out performance gains in the low-to-mid single-digit percentages. Intel is introducing a new clock-speed boosting technology called Thermal Velocity Boost, which can dial up clock-speeds of the i9-10900K up to 5.30 GHz.



View at TechPowerUp Main Site

 

[chodaboy19]

Reeks of desperation...

 

[fancucker]

Cooper-Lake will arrive MCM with 48/56 cores H2 2020. As for this, Skylake at two nodes behind still manages to have greater ST than AMD's tweaked Ryzen. Zen 3 will arrive to find backported Willow Cove on 14nm++ or 10nm/7nm, Zen 4 will probably be the first to equal WC clock-for-clock in IPC. So Intel has winning cards in all battles here.

 

[HTC]

26% gain in Cinebench R15

Intel says Cinebench isn't a "real world" benchmark.

 

[chaosmassive]

Intel : its not valid benchmark because its doesnt reflect to real-word usage.
Also Intel : include a 26% gain in Cinebench R15

 

[biffzinker]

Intel should have added on four more cores for at least twelve cores instead of a measly two core bump.

Intel already looks be in second place compared to AMD's twelve, and sixteen core chips.

 

[GoldenX]

Intel should added on four more cores for at least twelve cores instead of a measly two core bump.

Intel already looks be in second place compared to AMD's twelve, and sixteen core chips.

I bet they can't push the Skylake arch that much.

 

[Nkd]

Cooper-Lake will arrive MCM with 48/56 cores H2 2020. As for this, Skylake at two nodes behind still manages to have greater ST than AMD's tweaked Ryzen. Zen 3 will arrive to find backported Willow Cove on 14nm++ or 10nm/7nm, Zen 4 will probably be the first to equal WC clock-for-clock in IPC. So Intel has winning cards in all battles here.

lol. Did you just say AMD won’t match intel IPC until zen 4? Really? They are already trading blows with them with zen 2.

 

[cucker tarlson]

lol. Did you just say AMD won’t match intel IPC until zen 4? Really? They are already trading blows with them with zen 2.

I think they're slightly ahead,though for single thread gaming intel still beats them due to lower latency of the ring design and clocks.

This looks like same old same old core,only 10th gen locked skus will match 9th gen k-series on stock clocks and there's HT on every cpu + a 10 core.

will probably end up really competitive against ryzen 3000/4000,imagine stock 9900k rivalling 3700x/3800x not 3900x,but intel has no new core design still.

 

[Hyderz]

the 30% performance is a welcome indeed and lets see how intel price it
i think the 10900k will be priced at $549, but would rather see it at $499-$525
this is good for consumers as more options available from both amd and intel.
now what do u guys think the z490 chipset will it include pci-e 4.0 or not?

 

[R-T-B]

As for this, Skylake at two nodes behind still manages to have greater ST than AMD's tweaked Ryzen.

No, it doesn't.

It has less latency which is better for gaming yes. But it is actaully worse in ST IPC.

 

[DeathtoGnomes]

Reeks of desperation...

its proves Intel isnt standing around playing switch.

 

[cucker tarlson]

I bet they can't push the Skylake arch that much.

Not only they cant,they absolutely should stay away from the core race if they dont want to embarass themselves.
Keep the ring design and give it all it has,that is the only way they can still stay ahead.

 

[Noztra]

Cooper-Lake will arrive MCM with 48/56 cores H2 2020. As for this, Skylake at two nodes behind still manages to have greater ST than AMD's tweaked Ryzen. Zen 3 will arrive to find backported Willow Cove on 14nm++ or 10nm/7nm, Zen 4 will probably be the first to equal WC clock-for-clock in IPC. So Intel has winning cards in all battles here.

Not even proper MCM. Its just 2x 28C clued together. And you know that it will be a 400/500W part that require watercooling. No enterprise is gonna buy or use it. TCO is gonna be 4 times higher than AMD’s best server CPU, while still being slower.

 

[XL-R8R]

Maybe someone else has already posted this... but:

So... we should take from this, that the 10900K is actually only 5% faster... the sad part is, the 4.8GHz clock speed of the newer part vs 4.6GHz for the 9900k..... a reduction of 5% from 4.8GHz puts you at 4.56GHz..... so with 2 extra cores, 4 extra threads and a bump in effect base clock... I would have expected more than 5%(30%??) improvements overall.




Intel do have some funny press announcements recently.

 

[biffzinker]

Not only they cant,they absolutely should stay away from the core race if they dont want to embarass themselves.
Keep the ring design and give it all it has,that is the only way they can still stay ahead.

Intel at some point will need to move on from the ring bus to their mesh interconnect architecture in the mobile, and desktop space.

Mesh Interconnect Architecture - Intel - WikiChip

Intel's mesh interconnect architecture is a multi-core system interconnect architecture that implements a synchronous, high-bandwidth, and scalable 2-dimensional array of half rings. Their mesh architecture has replaced the ring interconnect architecture in the server and HPC markets.

en.wikichip.org

 

[R0H1T]

Meanwhile at CES ~ AMD CEO To Unveil "Zen 3" Microarchitecture at CES 2020

 

[R-T-B]

Not even proper MCM. Its just 2x 28C clued together.

And AMD is using better glue? What is your point here? This is just as valid a criticism as Intel calling Ryzen "glued together."

 

[R0H1T]

Intel at some point will need to move on from the ring bus to their mesh interconnect architecture in the mobile, and desktop space.

Mesh Interconnect Architecture - Intel - WikiChip

Intel's mesh interconnect architecture is a multi-core system interconnect architecture that implements a synchronous, high-bandwidth, and scalable 2-dimensional array of half rings. Their mesh architecture has replaced the ring interconnect architecture in the server and HPC markets.

en.wikichip.org

They're likely to move on from both AFAIK. Remember the glue

XeMF: The Scalable Memory Fabric, with RAMBO CACHE

For all the talk about Intel being a leader, they sure do follow AMD a lot not to mention badmouth & then copy them.

 

[londiste]

Meanwhile at CES ~ AMD CEO To Unveil "Zen 3" Microarchitecture at CES 2020

A preemptive move to Intel's 10-series launch?
From what AMD has said, the CPUs will be out in H2 2020.

 

[R0H1T]

Hardly, except for gaming & applications which rely heavily on ST performance, I can't see any xx900k coming close to 3950x let alone beat it. And if you really want ST performance then buying a 10c isn't that wise of a choice, same goes for 8c or 6c arguably. Zen3 seems like yet another major push for servers, especially as 10nm Intel server chips are nowhere in sight & 7nm is possibly ~2 years away.

 

[londiste]

3950X is an 800€ CPU. Even if 10900K follows the i9-9900K pricing, it should be a 500€ CPU and go against 3900X.
Keep in mind that 10-series according to what we know will have HT. There will not be that huge cap in productivity performance any more.

While much of the hoopla is around i9-10900K the actual Intel 10-series CPU to watch is i5-10400(F).
Similarly, AMD's Renoir APUs should be very excciting.

 

[R0H1T]

Even if 10900K follows the 9900K pricing

I'll eat my non existent hat if 10900k is priced (too) close to 3900x or 9900k.

Keep in mind that 10-series according to what we know will have HT

Yes & AMD will still have better SMT, 20~60% more cores & higher IPC. The gap can go down by 20~30% as compared to 9900k but then you're also paying a higher electricity bill & will likely need a new MB.

Purely in terms of value, this offers more than most Intel chips in the last decade though you will have to question why doesn't Intel use *cove at 14nm++++ instead of adding more (weak) cores. Surely their volumes will justify back-porting the design & IIRC the coves are node agnostic?

Don't focus too much on 10900K/9900K.

Anything lower than i9 currently does not have HT. This is the main concern in every review - Intel may do OK in gaming but there is a large automatic gap in productivity/threaded performance. HT will address most of that.

Yes that's why I said better value in a really long time though I'll add Intel's HT isn't as good as AMD's SMT implementation.

 

[londiste]

Don't focus too much on 10900K/9900K.

Anything lower than i9 currently does not have HT. This is the main concern in every review - Intel may do OK in gaming but there is a large automatic gap in productivity/threaded performance. HT will address most of that.

Purely in terms of value, this offers more than most Intel chips in the last decade though you will have to question why doesn't Intel use *cove at 14nm++++ instead of adding more (weak) cores. Surely their volumes will justify back-porting the design & IIRC the coves are node agnostic?

Intel's cores were not node agnostic. There are rumors (and I think Intel's people have mentioned that in several talks) about a project to make their cores node agnostic. This project should be going on for a little over a year now. Too early to actually use the cores.

When it comes to Coffee Lake vs Ice Lake, the claimed transistor counts are 217 million vs 300 million in a core. This is a considerable 38% difference.

 

[john_]

So, the next high end CPU will have NO IPC improvement over the current one. That 2-4% comes from the higher maximum Turbo speed.

Also NO extra performance from architectural improvements in multithreding, just what we get from those extra 2 cores / 4 threads and those higher Turbo frequencies.

That probably also means NO hardware security fixes. If there where any security fixes we could have seen some extra performance differences.

That 95W 9900K is in fact a 210W chip, if you want to get 100% of what that chip can offer. The new 10900K is in fact a 250W chip. 125W is at base frequency.

Great. Even Pentium 4 was looking better against Athlon64.