Intel Rocket Lake-S CPU Rumors: Core-i9 chips can turbo all the way up to 5.4-5.5 GHz | Engineering Samples already reaching 5.3GHz

[schwellmo92]

3 hours ago, porina said:

The cores were originally designed to go on their 10nm process, so changes may be required to make it work on 14nm. One recent rumour suggested it was going to be more Sunny Cove like in cache size, so wont have the full IPC uplift expected from Willow Cove. That's not to say there wont be an IPC uplift, just not as much as if it was the whole experience. Sunny Cove is still a good uplift from Skylake.

Willow Cove is only marginally better than Sunny Cove in IPC (low single digits). Regardless of whether this desktop CPU uses Sunny Cove or Willow Cove, if it can get 5GHz+ it will be a monster.

[Stahlmann]

Idk why people think this will bring Intel back to the top. Clocks are only 200MHz higher than their current top CPU and that's probably only the TVB again, so no realistic boost clock. Also almost any comet-lake CPU was able to hit 5GHz all-core when overclocked, so that's not really an argument either. Anyone under air cooling won't reach the thermal requirements for TVB to kick in, so the boost clock will likely be at 5.3-5.4GHz. Even if there is a IPS increase, i doubt on it will be the same increase on 14nm compared to the native core architecture on 10nm. Don't get me wrong, i don't prefer AMD or Intel over one or the other but i don't quite think they will catch up to Zen3 in the near future. Next problem is, they need so much more power (and heat) to reach these levels of performance. I think people are just blindly taking the 5.5GHz marketing as a bait again, even if raw clock speed is not nearly the only performance indicator.

[leadeater]

2 hours ago, Fatih19 said:

@RejZoR@leadeaterThat's crazy skipping a whole generation of memory lmao.

Well Intel only have themselves to blame for that one. I'm in the group of people that purchase HEDT for the PCIe lanes, not the core counts. I've always run dual GPU (thought that might be impossible now instead of impractical), 10Gb/25Gb NICs and RAID cards. When Intel cut down the number of PCIe lanes in the lower core count HEDT parts the entire product stack became useless to me and is the direct reason I did not upgrade.

 

They started that on X99 with the 6800K limiting it to 28 lanes from 40 and then did it to even more on X299, and with how generally bad a platform X299 was/is there was no way I was going to buy in to that. So I just gave up caring until my PC is literally unable to play games properly, which it very much still can, a testament to how little progress there has actually been until now (CPU and GPU).

 

I'm not the only person either that felt Intel completely lost sight of the point of HEDT platform either, simply repurposing Xeon architecture to run up the core count to only chase the CB/3DMark benchmark scores is not the point of that platform, but that's all Intel was focusing on. Meanwhile on the desktop platform side Intel was introducing new features and technologies, I mean sure HEDT usually lags consumer desktop by a bit but nothing was added at all. So you had a platform with none of the last 2/3 generations of features/technologies as the consumer desktop platform and none of the Xeon features/technologies either. Yea real good deal there, not. Oh also they jacked up the prices to insanity and beyond.

 

Spoiler

 

[porina]

44 minutes ago, Stahlmann said:

Idk why people think this will bring Intel back to the top.

Regardless if it does or not, to get to the top you have to improve regardless, and that's what they're trying to do.

 

1 hour ago, schwellmo92 said:

Willow Cove is only marginally better than Sunny Cove in IPC (low single digits). Regardless of whether this desktop CPU uses Sunny Cove or Willow Cove, if it can get 5GHz+ it will be a monster.

The claims were bigger than that, from memory, but I don't recall the exact number. I think most people kinda ignored it around here since it wasn't going on desktop any time soon.

[schwellmo92]

3 minutes ago, porina said:

The claims were bigger than that, from memory, but I don't recall the exact number. I think most people kinda ignored it around here since it wasn't going on desktop any time soon.

The claim was 20% performance uplift, but that was almost entirely from the significantly increased clock speeds.

[porina]

4 minutes ago, schwellmo92 said:

The claim was 20% performance uplift, but that was almost entirely from the significantly increased clock speeds.

I do recall the clock improvements also. In a quick search I've not managed to find a credible source for what that IPC increase is now. I might look again later. I do recall that there wasn't an architecture change between Sunny and Willow Cove, the improvements in IPC come down to the cache upgrade. Hence if Rocket Lake doesn't get those larger caches, it is effectively like Sunny Cove.

[Bombastinator]

1 hour ago, Stahlmann said:

Idk why people think this will bring Intel back to the top. Clocks are only 200MHz higher than their current top CPU and that's probably only the TVB again, so no realistic boost clock. Also almost any comet-lake CPU was able to hit 5GHz all-core when overclocked, so that's not really an argument either. Anyone under air cooling won't reach the thermal requirements for TVB to kick in, so the boost clock will likely be at 5.3-5.4GHz. Even if there is a IPS increase, i doubt on it will be the same increase on 14nm compared to the native core architecture on 10nm. Don't get me wrong, i don't prefer AMD or Intel over one or the other but i don't quite think they will catch up to Zen3 in the near future. Next problem is, they need so much more power (and heat) to reach these levels of performance. I think people are just blindly taking the 5.5GHz marketing as a bait again, even if raw clock speed is not nearly the only performance indicator.

Do you mean by type or by individual chip?  We’re measuring manufacturers claim against manufacturers claim currently and garbage times garbage is still garbage.  Manufacturers seem to be making faster and faster chips, and the difference between the top chip of one and the top chip of the other is tiny.  Improvements are getting smaller and smaller.   Either manufacturer can generally beat the previous chip of the other, but not by a whole lot.   I think this is why price/performance is becoming a bigger and bigger deal.  

[jagdtigger]

5 hours ago, leadeater said:

4930K is pretty damn slow now, it's not a problem but like I mentioned any current CPU is significantly faster

I still have a 4670k in my gaming rig but it aint that bad:
https://www.dropbox.com/s/zcogebr9v00z2hq/CBR20.PNG?dl=0

[HenrySalayne]

Good job, Intel! You are proudly following your old Netburst tradition and turning power consumption to 11 to compete with AMD (pun intended). I'm just a little bit sad the name "Meltdown" is already associated with another "feature" of your processors.

[thechinchinsong]

Very interesting product from Intel if they can get both the rumored/claimed? 20% ipc increase from Skylake AND the 5.5 ghz clocks. Things are definitely spicing up in the 8 core and below segment, especially if Intel can make sure prices are competitive. Wonder if Intel is just going to ignore the 8+ core prosumer/enthusiast space though.

[BlueChinchillaEatingDorito]

10 hours ago, RejZoR said:

 Intel's only saving grace is their stupid high clocks, really. They just feel like they are in absolute panic mode and every meeting ends with "lets just raise the clock some more". 

Don't forget AMD was basically in the same position back when they released the FX-9590. 220W TDP for god's sake and it wasn't even a powerhouse because the FX Architecture wasn't much better than the Phenom II. In fact, the Phenom II surpasses FX it in several applications mainly in number crunching if I remember correctly. It was mainly the additional instruction sets that the FX had over the Phenom II that made it slightly better for gaming. For any other application, they were on par or Phenom II squeezed ahead. The fact that most reviews today of the FX series chips compared the FX-8350 to a Phenom II X4 965 is rather sad. The former is a Octa-core and the latter is a quad core. How terrible does an Octa-core chip have to be to make a comparison to a quad-core be relevant? 

 

At the same time their APUs were not taking the market by storm as they hyped it to be. In fact, they were rather uninspiring and it was basically an Athlon chip with almost Intel HD level of graphics slapped onto it. Sure, you can play some games but it was really at low resolutions and low details that you got any decent frame rates. So the value proposition wasn't there compared to the Core chips of that era. They were more like Pentium tier chips really. 

[aisle9]

18 hours ago, Fasauceome said:

Not sure how Intel can pull off a 10 core i9 to compete with the 5800X though, price wise. unless all of their yield problems are solved, and their IPC increase is actually noticable. otherwise, we're looking at a mere 5% or so increase in performance.

Even when they're trying really hard, it's still an incremental upgrade smh

 

*ducks to avoid flames*

[Vishera]

19 hours ago, BiG StroOnZ said:

the Core i9 variant of Rocket Lake-S will hit a turbo clock of 5.4GHz

Single core only

19 hours ago, BiG StroOnZ said:

The leaker claims Rocket Lake-S will handle an all-core overclock to 5GHz without any trouble.

The 2600K already does that.

 

Just empty statements that mean nothing.

[thechinchinsong]

2 minutes ago, BlueChinchillaEatingDorito said:

Don't forget AMD was basically in the same position back when they released the FX-9590. 220W TDP for god's sake and it wasn't even a powerhouse because the FX Architecture wasn't much better than the Phenom II. In fact, the Phenom II surpasses FX it in several applications mainly in number crunching if I remember correctly. It was mainly the additional instruction sets that the FX had over the Phenom II that made it slightly better for gaming. For any other application, they were on par or Phenom II squeezed ahead. The fact that most reviews today of the FX series chips compared the FX-8350 to a Phenom II X4 965 is rather sad. The former is a Octa-core and the latter is a quad core. How terrible does an Octa-core chip have to be to make a comparison to a quad-core be relevant? 

 

At the same time their APUs were not taking the market by storm as they hyped it to be. In fact, they were rather uninspiring and it was basically an Athlon chip with almost Intel HD level of graphics slapped onto it. Sure, you can play some games but it was really at low resolutions and low details that you got any decent frame rates. So the value proposition wasn't really there compared to the Core chips of that era. 

Intel isn't as far behind as AMD were in the FX era. You can deride Intel all day about how they don't have the best value at every price segment currently, but in the FX days, there was literally NO price point at which AMD was competitive. At least nowadays, you can still make an argument for Intel in several use cases and perhaps many more once rocket lake comes out. Of course we'll have to wait and see how Zen 3 and Rocket Lake match up, but its looking to be a good competition. Hopefully this time there isn't any shady backroom deals from Intel.

[Bombastinator]

27 minutes ago, thechinchinsong said:

Intel isn't as far behind as AMD were in the FX era. You can deride Intel all day about how they don't have the best value at every price segment currently, but in the FX days, there was literally NO price point at which AMD was competitive. At least nowadays, you can still make an argument for Intel in several use cases and perhaps many more once rocket lake comes out. Of course we'll have to wait and see how Zen 3 and Rocket Lake match up, but its looking to be a good competition. Hopefully this time there isn't any shady backroom deals from Intel.

One of the more interesting ones lately is apparently the 10100f which is more or less a 3100/3300 for less money.

[tim0901]

10 hours ago, porina said:

If someone already owns an Intel (or even AMD) 8 core CPU, probably not, but not everyone upgrades every generation. It will have to be considered amongst other offerings when people do decide to buy.

You misunderstand my question. People who already have an 8 core CPU probably aren't interested in a Rocket Lake i7 either, so that's kind of a moot point. Both would be a massive improvement over my 4790K, but that doesn't mean I don't care about the value proposition of each CPU within it's own generation.

 

What I'm curious about is: if both their $500 i9 and <$400 i7 are 8C/16T parts, what does Intel plan on doing to try and persuade people who are interested in an i7 to buy an i9 instead? How do they plan on differentiating these two products, when it seems like they've already played all their cards with the i7?

 

In the past Intel (and AMD) have primarily used thread count to differentiate their products, but if they both have the same thread count then will it just come down to clocks?

And if so, if the i9 is only a small clock bump over the i7, yet costs $100 more, then why would anyone buy it, except to say they own the halo product? The value proposition will suck balls, even by Intel's standards! 25% price increase for a 5-10% bump in clock speed? (which can likely be partially mitigated with an overclock) - no thanks!

[RotoCoreOne]

MOAR POWAH!!!!!

[Vishera]

43 minutes ago, RotoCoreOne said:

MOAR POWAH!!!!!

Whoooo,Yaahhhh!!!!!!,Takemikazuchi-no-kami!!!!!!!!!!!!!!

For POWAH you need adequate cooling and VRMs,for VRMs you need moneys.

In that pace Intel will reach 300W at no time,it has already surpassed the 220W of the FX 9590.

[porina]

2 hours ago, tim0901 said:

What I'm curious about is: if both their $500 i9 and <$400 i7 are 8C/16T parts, what does Intel plan on doing to try and persuade people who are interested in an i7 to buy an i9 instead? How do they plan on differentiating these two products, when it seems like they've already played all their cards with the i7?

In the past they also differentiated on cache size. I don't recall seeing if that was a thing in rumours. For example, for most CPUs in the quad core era, the i7's got 2MB/core and i5's got 1.5MB/core. Broadwell was an odd ball, but I think that applied for Kaby Lake, Skylake, Haswell, and probably the two Bridges before that also.

[Godlygamer23]

12 hours ago, leadeater said:

Still running a 4930K, almost any CPU is an upgrade for me now lol

It's not THAT bad. I was still using a 3570K up until the beginning of last year when I was required(effectively) to upgrade. The CPU would choke on Forza Horizon 4 with my targeted frame rates lol. 

[thechinchinsong]

3 hours ago, Bombastinator said:

One of the more interesting ones lately is apparently the 10100f which is more or less a 3100/3300 for less money.

Yeah I found that one super interesting. Really I think once Rocket Lake comes out, the only big gripe I'll have with Intel is how much they lock up their non z-series motherboards and prevent overclocking on all sku's except K. 

 

Can you imagine if Intel made their B-series boards as attractive and affordable as AMD's B-series?

[Fatih19]

10 hours ago, leadeater said:

Well Intel only have themselves to blame for that one. I'm in the group of people that purchase HEDT for the PCIe lanes, not the core counts. I've always run dual GPU (thought that might be impossible now instead of impractical), 10Gb/25Gb NICs and RAID cards. When Intel cut down the number of PCIe lanes in the lower core count HEDT parts the entire product stack became useless to me and is the direct reason I did not upgrade.

 

They started that on X99 with the 6800K limiting it to 28 lanes from 40 and then did it to even more on X299, and with how generally bad a platform X299 was/is there was no way I was going to buy in to that. So I just gave up caring until my PC is literally unable to play games properly, which it very much still can, a testament to how little progress there has actually been until now (CPU and GPU).

 

I'm not the only person either that felt Intel completely lost sight of the point of HEDT platform either, simply repurposing Xeon architecture to run up the core count to only chase the CB/3DMark benchmark scores is not the point of that platform, but that's all Intel was focusing on. Meanwhile on the desktop platform side Intel was introducing new features and technologies, I mean sure HEDT usually lags consumer desktop by a bit but nothing was added at all. So you had a platform with none of the last 2/3 generations of features/technologies as the consumer desktop platform and none of the Xeon features/technologies either. Yea real good deal there, not. Oh also they jacked up the prices to insanity and beyond.

 

  Reveal hidden contents

 

Damn, why not consider Threadripper for your next upgrade then?

[leadeater]

8 minutes ago, Fatih19 said:

Damn, why not consider Threadripper for your next upgrade then?

Because I just gave up and stopped caring 

[Brooksie359]

21 hours ago, schwellmo92 said:

With Willow Cove cores that are a 15-20% IPC increase it is actually substantial.

I will wait and see if that is actually the case the same as I will do with the ipc claims from AMD. If they do have a significant ipc increase that would be great but part of me doubts that with them being stuck on 14nm. 

[Brooksie359]

7 hours ago, tim0901 said:

You misunderstand my question. People who already have an 8 core CPU probably aren't interested in a Rocket Lake i7 either, so that's kind of a moot point. Both would be a massive improvement over my 4790K, but that doesn't mean I don't care about the value proposition of each CPU within it's own generation.

 

What I'm curious about is: if both their $500 i9 and <$400 i7 are 8C/16T parts, what does Intel plan on doing to try and persuade people who are interested in an i7 to buy an i9 instead? How do they plan on differentiating these two products, when it seems like they've already played all their cards with the i7?

 

In the past Intel (and AMD) have primarily used thread count to differentiate their products, but if they both have the same thread count then will it just come down to clocks?

And if so, if the i9 is only a small clock bump over the i7, yet costs $100 more, then why would anyone buy it, except to say they own the halo product? The value proposition will suck balls, even by Intel's standards! 25% price increase for a 5-10% bump in clock speed? (which can likely be partially mitigated with an overclock) - no thanks!

Its basically the same as the 3800x vs the 3700x. Most people won't bother with the new i9 just like most people didn't bother with the 3800x