Intel Core i9-10900K - 10 Core Is Coming

[Rvenger]

When it is released it should be faster (provided it does have the 5.X GHz clock) than the 3900X. Intel will have no answer for the 3950X. AMD knows that and has priced the 3950X as a result.

I have a feeling that $749 price will be another "jebait" situation and land at $599 or $649. I have a feeling AMD is waiting for them to release at $749 and drop the price on the 3950x. Hence the September release date.

 

[DuronBurgerMan]

I have a feeling that $749 price will be another "jebait" situation and land at $599 or $649. I have a feeling AMD is waiting for them to release at $749 and drop the price on the 3950x. Hence the September release date.

I think the September release date is just because the top-binned 8 core dies are going elsewhere right now.

 

[drescherjm]

I think the September release date is just because the top-binned 8 core dies are going elsewhere right now.

I think they are expecting TSMC to improve their 7nm process by then so that there will be more top binned 8 core dies that can hit the frequencies that were claimed for the 3950X.

 

[odditory]

If you want a pure gaming CPU, 10/20 won't be any better than what they have available (9900x).. if you need cores... well, the 3950x or TR is your go to... not really sure what this part is meant to satisfy.

People that want both
without compromise.

 

[Nolan7689]

People that want both
without compromise.

Except there is still a compromise on the number of cores/threads.

Then it just comes down to which is more important to a buyer, single thread or multi thread. Which is the same situation we have right now.

 

[Ready4Dis]

People that want both
without compromise.

If what you meant was people that want to compromise both... Then I agree. They will, in my best guess, end up somewhere near the 9900k in games and still slightly behind and in largely threaded loads. Aka, they are compromising, just compromising to somewhere in between.

 

[odditory]

Except there is still a compromise on the number of cores/threads.

Then it just comes down to which is more important to a buyer, single thread or multi thread. Which is the same situation we have right now.

Huh? Assuming a $499 10900k vs $499 3900x, 10 fast cores ALL running at 5Ghz+ isn't a compromise against 12 slower cores with only 2 at 4.xGhz and the rest 3.xGhz. 10 Cores@5Ghz+ will be faster and better at everything is the point.

Yes, 3950x (2 x 8-Cores glued together) at $749 will beat 10900k at Cinebench, as you'd expect for 50% more cost.

 

[N4CR]

Can barely scrape 5ghz on 30% of 9900k samples and huge thermal load, this ten core will be even hotter and more difficult to scale. This thread makes amd hype train threads look realistic.

 

[Dan_D]

Can barely scrape 5ghz on 30% of 9900k samples and huge thermal load, this ten core will be even hotter and more difficult to scale. This thread makes amd hype train threads look realistic.

So far I'm three for three for getting 5.0GHz on Core i9 9900K's.

 

[N4CR]

So far I'm three for three for getting 5.0GHz on Core i9 9900K's.

Interesting. Always thought the 30% number was a bit low but statistics and all.. like Kyle with his 0.02% space invaders triple win.

 

[Nightfire]

Huh? Assuming a $499 10900k vs $499 3900x, 10 fast cores ALL running at 5Ghz+ isn't a compromise against 12 slower cores with only 2 at 4.xGhz and the rest 3.xGhz. 10 Cores@5Ghz+ will be faster and better at everything is the point.

Yes, 3950x (2 x 8-Cores glued together) at $749 will beat 10900k at Cinebench, as you'd expect for 50% more cost.

Well, that is not true. The 9900k already runs at 4.7ghz in review tests so 5.0ghz is only a 6% o/c. Combined with a 25% core increase means that it will lose in any are where the 3900x is more than 33% faster which includes rendering, encryption, and some media encoding.

And yes I know 25+6=31 but 1.06×1.25=1.325 or 33% which is how you solve that. Most techies are really stupid at math so I figured I would spell that one out.

 

[N4CR]

Well, that is not true. The 9900k already runs at 4.7ghz in review tests so 5.0ghz is only a 6% o/c. Combined with a 25% core increase means that it will lose in any are where the 3900x is more than 33% faster which includes rendering, encryption, and some media encoding.

And yes I know 25+6=31 but 1.06×1.25=1.325 or 33% which is how you solve that. Most techies are really stupid at math so I figured I would spell that one out.

Very curious if they'll have to go to ring bus or not.

 

[Nightfire]

Very curious if they'll have to go to ring bus or not.

I believe staying ring bus is better for core to core latency which is preferred in gaming, but with that many cores, not going mesh would hurt it in overall latency. It's confusing in any case.

 

[Derangel]

AMD isn't going to get any faster in games with Zen 2, and Intel has shown >10% IPC increase with Ice Lake, so if desktop clocks are similar Intel will be leaping ahead again. Which really shouldn't be any sort of revelation.

Intel has CLAIMED a 10% IPC increase. Intel's own slides really don't prove anything other than "we claim this under circumstances which we will not reveal to the public." Until they're out and properly tested there is no "shown", only unsubstantiated claims. Intel has made BS claims of IPC improvement in the past which have turned out to be incredibly underwhelming on release. Namely, pretty much everything in the last decade. Even if we assume the chips do end up with 10% IPC gain across the board, clocks are going to be the big question. Clock-for-clock they'll have Zen 2 handily beat, but if the desktop chips aren't hitting near the 5GHz mark of 8th and 9th gen chips the total improvement might not end up being all that outstanding.

 

[Ready4Dis]

Huh? Assuming a $499 10900k vs $499 3900x, 10 fast cores ALL running at 5Ghz+ isn't a compromise against 12 slower cores with only 2 at 4.xGhz and the rest 3.xGhz. 10 Cores@5Ghz+ will be faster and better at everything is the point.

Yes, 3950x (2 x 8-Cores glued together) at $749 will beat 10900k at Cinebench, as you'd expect for 50% more cost.

Man, not only is everyone assuming this new wonder chip can hit 5ghz, has a low wattage use, it is not too g to be offered at the same price as the 9900k? I agree, these latest Intel leaks and slide leaks make the AMD fan boys look nuetral. Don't get me wrong, I hope they release 65w 10 core part that hits 5ghz on all cores without needing liquid nitrogen, and Intel doesn't raise prices.... Lol, just saying that all in one sentence makes me laugh.

 

[techguymaxc]

Man, not only is everyone assuming this new wonder chip can hit 5ghz, has a low wattage use, it is not too g to be offered at the same price as the 9900k? I agree, these latest Intel leaks and slide leaks make the AMD fan boys look nuetral. Don't get me wrong, I hope they release 65w 10 core part that hits 5ghz on all cores without needing liquid nitrogen, and Intel doesn't raise prices.... Lol, just saying that all in one sentence makes me laugh.

This is a strawman argument. No one needs to defend these obviously fake specs. Realistic 10-core chips from Intel will have TDP ratings of 95W (or higher) at base clock on 14nm, less if they are 10nm. We still don't know what the desktop parts from Intel will look like, it's all speculative at this point.

 

[odditory]

Man, not only is everyone assuming this new wonder chip can hit 5ghz, has a low wattage use, it is not too g to be offered at the same price as the 9900k? I agree, these latest Intel leaks and slide leaks make the AMD fan boys look nuetral. Don't get me wrong, I hope they release 65w 10 core part that hits 5ghz on all cores without needing liquid nitrogen, and Intel doesn't raise prices.... Lol, just saying that all in one sentence makes me laugh.

It's not an assumption or presumption -- its a hypothetical for the sake of discussion.

That said, if it can't 5Ghz and costs more than $499, Intel may as well not even show up - and I think they know that. And the credit for that goes squarely to AMD.

 

[Ready4Dis]

This is a strawman argument. No one needs to defend these obviously fake specs. Realistic 10-core chips from Intel will have TDP ratings of 95W (or higher) at base clock on 14nm, less if they are 10nm. We still don't know what the desktop parts from Intel will look like, it's all speculative at this point.

It's not strawman anything, I'm just a believe it when I see it kind of guy... How many AMD fans were expecting ryzen to hit 5ghz... And they believed it and the more level headed ones said it would be unrealistic and take it with a grain of salt (note, most *wanted* it to be true, but thought it sounded to good). Same thing except it's now Intel, so if u don't want to listen to logic and past actions, good on you and enjoy. Don't get me wrong, I would love competition to heat up and Intel (and AMD) to hit 5ghz all core on 10/20 and 16/32 chips... But I'm not going to talk about it like that's likely either.

 

[Red Falcon]

Oh good, we're going to need those additional cores and clocks to make up for the performance hits from a few more security exploit patches.
So, nearly 2 years later, Intel's processors will be almost back to Kaby Lake performance after this?

You know, I hate to sound like I'm shitting all over Intel about those security exploits coming to light all of the time, but it was bullshit that they hid it for so long from loyal customers, partners, etc. for the sake of their shareholders.
However, they are a technology-megacorp, so their actions are really nothing surprising (AMD would do the same if they could), and despite that, the rest of what Intel produces actually are surprisingly high quality, and near top-shelf, products.

Their SSDs (especially enterprise), wired and Wi-Fi NICs, and AI (NPU, VPU, etc.) are all very high performing and high quality, and I've had nothing but good, stable, and consistent experiences with the rest of their products.
Even their motherboard chipsets are high quality and offer a lot of options per market segment.

They burned the bridge with customers on their processors and trying to hide decades of hardware exploits, though.
Use what works and vote with your wallet - just be sure to keep what they have done as a company in mind before buying another CPU/SOC from them again.

 

[4saken]

Oh good, we're going to need those additional cores and clocks to make up for the performance hits from a few more security exploit patches.
So, nearly 2 years later, Intel's processors will be almost back to Kaby Lake performance after this?

You know, I hate to sound like I'm shitting all over Intel about those security exploits coming to light all of the time, but it was bullshit that they hid it for so long from loyal customers, partners, etc. for the sake of their shareholders.
However, they are a technology-megacorp, so their actions are really nothing surprising (AMD would do the same if they could), and despite that, the rest of what Intel produces actually are surprisingly high quality, and near top-shelf, products.

Their SSDs (especially enterprise), wired and Wi-Fi NICs, and AI (NPU, VPU, etc.) are all very high performing and high quality, and I've had nothing but good, stable, and consistent experiences with the rest of their products.
Even their motherboard chipsets are high quality and offer a lot of options per market segment.

They burned the bridge with customers on their processors and trying to hide decades of hardware exploits, though.
Use what works and vote with your wallet - just be sure to keep what they have done as a company in mind before buying another CPU/SOC from them again.

k.

 

[Red Falcon]

k.

 

[Ready4Dis]

It's not an assumption or presumption -- its a hypothetical for the sake of discussion.

That said, if it can't 5Ghz and costs more than $499, Intel may as well not even show up - and I think they know that. And the credit for that goes squarely to AMD.

I understand, but seems pretty far fetched. I hope it's true, just don't think it will be.