Intel's 10th-gen processors haven't been announced, but they are for sale

[mongeese]

What just happened? Intel has sent out a wave of engineering sample processors, a few dozen of which are (allegedly) available for purchase on Taobao. If they actually shipped you one, then you’d be getting some very powerful hardware: those same engineering samples are beating their predecessors by 15-20% in a handful of online benchmarks.

Intel’s tenth generation processors aren’t confirmed, yet there’s little mystery surrounding them. In late December, Informatica Cero obtained slides detailing the entire lineup, and their information has been almost confirmed by about a dozen online sightings.

The flagship is the decacore i9-10900K, with a reported 5.1 GHz single-core boost and a 5.3 GHz single-core velocity boost. Two of its Fire Strike Physics scores have been uncovered, and at 28,462 and 28,988, they’re 15% and 17% higher than the 9900K’s average score of 24,741.

Its Time Spy CPU result of 13,142 is 19.5% higher than the 9900K’s average score of 10,997 as well. Both the 10900K and 10900 have been listed on Taobao, with photographs.

The table below has been put together based on details provided by Informatica Cero and Videocardz.

Model	Cores/ Threads	Base (GHz)	All/Single Core Boost (GHz)	Single Core Turbo 3.0 (GHz)	All/Single Core Velocity (GHz)	TDP (Watts)
i9-10900K	10/20	3.7	4.8/5.1	5.2	4.9/5.3	125W
i9-10900	10/20	2.8	4.5/5.0	5.1	4.6/5.1	65W
i7-10700K	8/16	3.8	4.7/5.0	5.1	-	125W
i7-10700	8/16	2.9	4.6/4.7	4.8	-	65W
i5-10600K	6/12	4.1	4.5/4.8	-	-	125W
i5-10600	6/12	3.3	4.4/4.8	-	-	65W

In second place is the Core i7-10700K and i7-10700, which aren’t for sale – what a shame.

The 10700K has been spotted with a 5.3 GHz boost which is higher than the 5.0 GHz Informatica Cero’s info suggested, while the 10700 itself is falling a bit short at only 4.6 - 4.7 GHz. What’s more interesting is a Fire Strike Extreme Physics score for the 10700, which is 23.7% higher than the 9700’s average score at 23,326.

Taobao is also listing a Core i5-10600K and non-K models. Informatica Cero’s info, 3DMark entries, and CPU-Z screenshots all suggest these processors will be Intel’s first multi-threaded mid-range parts, with six cores and twelve threads.

The other mid-range and budget parts haven’t appeared much outside of Informatica Cero’s slides, so we’ll direct you back to that article if you’re curious about those as well.

Xfastest apparently got their hands on this 10900 engineering sample.

Seeing that Intel is shipping engineering samples in a fairly large volume, they must be getting close to release. And with such enticing leaks, that’s exciting.

However, Intel has spent the past five years plucking performance out of an architecture and node they introduced in 2015. Their then-flagship 6700k quad-core gaming piece had a 91W TDP and a $339 price tag. The present day’s 9900K is a $488 octa-core furnace that regularly consumes in excess of 150W (but is marketed with a misleading 95W TDP, mind you…). Consequently, we can only expect that they’ve continued wringing more performance out of something that should’ve been put down years ago and have produced yet another literal burning hole in your pockets and CPU sockets.

And thus, we’d like to remind readers that the sanguinity of leaks should invoke some suspicion. The 10900K and 10700K will probably both turbo over 5 GHz and outdo their predecessors by anywhere from 15-20% in certain (ideal) workloads.

But like the 9900K, which requires specific and pricey motherboards and coolers to sustain peak performance, these new processors might conceal hidden compromise. Please wait for reviews before purchasing these chips, even if Taobao’s listing makes for an enticing offer.

Permalink to story.

https://www.techspot.com/news/84033-intel-10th-gen-processors-havent-announced-but-they.html

 

[Lounds]

15-17% increase in performance for a node shrink isn't anything to brag about.

 

[krizby]

Yah, 20% in "synthetic benchmark" such as cinebench, for games maybe <5% due to 100-200mhz core clock bump vs gen9. Really we need more raw CPU horsepower than this to fully utilize Ampere/RDNA 2 at 1080p/1440p.

 

[VitalyT]

And they are all still supporting only PCIe 3.0. That's definitely not worth the premium.

 

[Lew Zealand]

15-17% increase in performance for a node shrink isn't anything to brag about.

These are still 14nm processors, with 25% more cores (10 vs. 8). Hence only 15-20% improvement.

 

[Paul Dougherty]

And they are all still support only PCIe 3.0. That's definitely not worth the premium.

PCIe 4.0 is a big deal for... sequential file reads? Don't get me wrong, I'd prefer having it over but having it, but a 100 MHz clock increase is probably a bigger deal, and that doesn't matter much either.

 

[Shadowboxer]

15-17% puts them above equivalently priced Ryzen 3xxx performance. And with a single core speed improvement it will take its gaming prowess even higher.

This is awesome, one manufacturer outdoes the other in turn. This is what we want. Stuff the fanboys, I can’t wait to see Ryzen 3xxx overturned. And il be anticipating Ryzen 4xxx outdoing this in 6 months time too!

 

[yRaz]

PCIe 4.0 is a big deal for... sequential file reads?

It doesn't sound like a big deal until you realise all your computer does is read and process files

 

[CharmsD]

Blowing that dog whistle, cmer cmere everyone throw rocks at intel, amd is bleeding in the bad drivers bad video card thread.

 

[CharmsD]

It doesn't sound like a big deal until you realise all your computer does is read and process files

Do you personally recommend pcie v. 4? Have you bought it?

 

[noel24]

If I were to buy one of those CPU, any good motherboard You could recommend off the Alibaba? ;-)

 

[Paul Dougherty]

15-17% puts them above equivalently priced Ryzen 3xxx performance. And with a single core speed improvement it will take its gaming prowess even higher.

This is awesome, one manufacturer outdoes the other in turn. This is what we want. Stuff the fanboys, I can’t wait to see Ryzen 3xxx overturned. And il be anticipating Ryzen 4xxx outdoing this in 6 months time too!

It's 3DMark physics. That translates over basically not at all in any actual games because no game uses CPU crushing physics. This isn't good competition from Intel, it's them selling a ~$500 CPU + 200 motherboard (or whatever) that's maybe 2% faster than the 9900K that's been out for more than a year.

 

[hahahanoobs]

If performance holds up I'm going with Comet Lake over Ryzen.

3970X on LN2 can pull 1100w alone @ 4.9GHz....

I couldn't care less about a 150W power draw from CPU when actually putting it to work versus what a chip pulls when browsing and watching YouTube. Aren't we all using 500W+ PSU's? I imagine it's sub 100W while doing that?

 

[Paul Dougherty]

It doesn't sound like a big deal until you realise all your computer does is read and process files

I'm using an Optane 900p because my computer my computer reads and processes tiny, random files all day. If I spent more time allocating 160GB Steam game installs, then a large jump in sequential transfers would be great. But after a day of use HWinfo64 generally shows only a few hundred megabytes per second peak transfer speed. Run a disk transfer speed test, and it's up at a couple gigabytes.

 

[Shadowboxer]

It's 3DMark physics. That translates over basically not at all in any actual games because no game uses CPU crushing physics. This isn't good competition from Intel, it's them selling a ~$500 CPU + 200 motherboard (or whatever) that's maybe 2% faster than the 9900K that's been out for more than a year.

Oh I was referring to general performance. I’m hoping it can be out the 3xxx series of Ryzen in terms of overall compute. As I said we all benefit from competition. I’m curious to what the reviews will say. At the end of the day Ryzen 3xxx didn’t beat Intel by a huge amount so it’s not out of the realm of possibility that Intel could come back with a part like this.

Of course everyone can see that games are affected by single core performance more than anything else. And if these chips can boost to 5.3 out of the box then they will be an improvement over what’s out there. And we haven’t had much single core improvement over the last decade or so. So whilst it’s small, it’s still rare.

I think what’s shocking me is that this is still on 14nm. What a node that turned out to be for Intel, especially if it fends off the 7nm competition.

 

[yRaz]

Do you personally recommend pcie v. 4? Have you bought it?

The biggest thing for me is having 10gigabit while using less PCIe Lanes. A network card can get everything it needs while running a GPU without messing up performance. A Ryzen 3600 only has 24 PCIe Lanes. In 3.0, that's a graphics card, network card and an NVMe then youve reach your limit. I, potentially, could run 10gigabit only using 1 Lane and a graphics won't be throttled by taking it down to just 8x. If I throw in an NVMe drive that still leaves me with 11 more Lanes on a Ryzen 3600 without any performance loss

 

[poohbear]

Yah, 20% in "synthetic benchmark" such as cinebench, for games maybe <5% due to 100-200mhz core clock bump vs gen9. Really we need more raw CPU horsepower than this to fully utilize Ampere/RDNA 2 at 1080p/1440p.

I think the 8 cores & 16 threads at 4.5ghz on all (even if with overclocking) are more than enough horsepower @ for Ampere/RDNA2 and probably the next 4 generations of graphics cards @ 1440p+. Once you pass 1440p, its mostly the GPU anyways.

 

[krizby]

I think the 8 cores & 16 threads at 4.5ghz on all (even if with overclocking) are more than enough horsepower @ for Ampere/RDNA2 and probably the next 4 generations of graphics cards @ 1440p+. Once you pass 1440p, its mostly the GPU anyways.

3900X vs 9900K
Check this article and you can see 9900K is still faster at 1440p. Now with the projected performance of Ampere being >50% faster than 2080 Ti, some games will definitely be CPU bottlenecked for 3080Ti at 1440p.

 

[pcnthuziast]

Prices won't be competitive, that's for sure.

 

[jbc029]

Good lord people, read the article. The 10700K *is* a 9900KS. Yeah, a 9900K is ~20% faster than a 9700K in the physics score.

The new 10900 is literally 15-20% faster because it has two more cores. 25% more cores = 15-20% faster. That's it.

The big win of intel 10th gen is that i3 and i5 are worth looking at again. 10300K = 7700K. Literally.
10500K = 8700K. Literally. That is the big news of 10th gen.

 

[Evernessince]

Yah, 20% in "synthetic benchmark" such as cinebench, for games maybe <5% due to 100-200mhz core clock bump vs gen9. Really we need more raw CPU horsepower than this to fully utilize Ampere/RDNA 2 at 1080p/1440p.

Unfortunately I don't think there is much Intel can do until their next big architecture.

If performance holds up I'm going with Comet Lake over Ryzen.

3970X on LN2 can pull 1100w alone @ 4.9GHz....

I couldn't care less about a 150W power draw from CPU when actually putting it to work versus what a chip pulls when browsing and watching YouTube. Aren't we all using 500W+ PSU's? I imagine it's sub 100W while doing that?

Given that the 9900K draws over 280w during full load, your 150w figure for a 10 core Intel CPU is likely less then half what the CPU will draw.

If all you need the system for is basic browsing you are fine with pretty much any CPU. If you are gaming you might as well go 9900K. There's really no point to spend more for the 2 extra cores. If it's cores you need, the 3900X and 3950X very likely already outperform Intel 10th gen parts. Not counting Zen 3.

 

[Lounds]

Good lord people, read the article. The 10700K *is* a 9900KS. Yeah, a 9900K is ~20% faster than a 9700K in the physics score.

The new 10900 is literally 15-20% faster because it has two more cores. 25% more cores = 15-20% faster. That's it.

The big win of intel 10th gen is that i3 and i5 are worth looking at again. 10300K = 7700K. Literally.
10500K = 8700K. Literally. That is the big news of 10th gen.

So no innovation just finally intel trying to compete on core count.

 

[poohbear]

3900X vs 9900K
Check this article and you can see 9900K is still faster at 1440p. Now with the projected performance of Ampere being >50% faster than 2080 Ti, some games will definitely be CPU bottlenecked for 3080Ti at 1440p.

Did you read the conclusion? It said the only time to consider a premium CPU is if buying a $1000+ card, otherwise all the other graphics cards perform generally the same with the CPUs. As such, I was referring to the majority of people that will be buying midrange /upper midrange cards like the Ampere 3060-3080.

 

[urbanman2004]

15-17% increase in performance for a node shrink isn't anything to brag about.

What node shrink are you talking a/b since the article specifically states:

Intel has spent the past five years plucking performance out of an architecture and node they introduced in 2015.

 

[Puiu]

PCIe 4.0 is a big deal for... sequential file reads? Don't get me wrong, I'd prefer having it over but having it, but a 100 MHz clock increase is probably a bigger deal, and that doesn't matter much either.

It depends a lot on what you use it. PCIe 4.0 will definitely allow you to get good speeds in mainstream mobos if you plan on using 2 or more high end NVMEs. For many I/O performance is very important and you'll notice it when using RAID.

For regular use, even PCIe 3.0 8x is enough