Sunday, March 22nd 2020

Intel Rocket Lake-S Platform Detailed, Features PCIe 4.0 and Xe Graphics

Intel's upcoming Rocket Lake-S desktop platform is expected to arrive sometime later this year, however, we didn't have any concrete details on what will it bring. Thanks to the exclusive information obtained by VideoCardz'es sources at Intel, there are some more details regarding the RKL-S platform. To start, the RKL-S platform is based on a 500-series chipset. This is an iteration of the upcoming 400-series chipset, and it features many platform improvements. The 500-series chipset based motherboards will supposedly have an LGA 1200 socket, which is an improvement in pin count compared to LGA 1151 socket found on 300 series chipset.

The main improvement is the CPU core itself, which is supposedly a 14 nm adaptation of Tiger Lake-U based on Willow Cove core. This design is representing a backport of IP to an older manufacturing node, which results in bigger die space due to larger node used. When it comes to the platform improvements, it will support the long-awaited PCIe 4.0 connection already present on competing platforms from AMD. It will enable much faster SSD speeds as there are already PCIe 4.0 NVMe devices that run at 7 GB/s speeds. With RKL-S, there will be 20 PCIe 4.0 lanes present, where four would go to the NVMe SSD and 16 would go to the PCIe slots from GPUs. Another interesting feature of the RKL-S is the addition of Xe graphics found on the CPU die, meant as iGPU. Supposedly based on Gen12 graphics, it will bring support for HDMI 2.0b and DisplayPort 1.4a connectors.
Intel Rocket Lake-S Platform
Some things like Direct Media Interface (DMI) will double the bandwidth and now there will be eight links present, compared to four of the previous platforms. Announced at CES 2020, ThunderBolt 4 will also be present along with USB 3.2 20G. Additionally, Intel Software Guard Extensions (SGX) have been removed to improve the security of the platform, as the SGX has proved to be quite vulnerable to many kinds of attacks and exploits. There are some updated media encoding standards as well, like 12-bit AV1/HEVC and E2E compression. Source: VideoCardz
Add your own comment

113 Comments on Intel Rocket Lake-S Platform Detailed, Features PCIe 4.0 and Xe Graphics

#76
notb
ARF
Both are with configurable up and down TDPs of 25-watt and 12-watt.

i7-1065G7 is a bad CPU. Period.
But why? Which part is so bad?

Features are quite even, IMO slightly better on the Intel side.
Performance... well, let's wait for the actual benchmarks.
Zen+ mobile APUs were also great on paper.

If you're amazed by the fact that AMD runs 2x more cores on 15W, don't be.
The real question is how they perform over longer period and how much energy they pull on the way.
All Renoir -U have 15W TDP - even the 4C/4T 4300U.

And of course by the time Renoir becomes available in mainstream products, it'll compete against 6 or 8-core Tiger Lake, not 4-core Ice Lake. :)
Posted on Reply
#77
Valantar
ARF
Notebookcheck and CPU-Monkey are independent enough.


https://www.cpu-monkey.com/en/cpu-amd_ryzen_7_4800u-1142


https://www.cpu-monkey.com/en/cpu-intel_core_i7_1065g7-933

Both are with configurable up and down TDPs of 25-watt and 12-watt.

i7-1065G7 is a bad CPU. Period.
...except that we can't actually read NBC's review, so we can't know the details of the configuration of the device. Hence the need to wait and see, right? I don't doubt their findings whatsoever, it's just that the numbers alone don't tell the whole story.

As for the i7-1065G7 being bad? Nah. It has the best iGPU Intel has ever made - which is genuinely quite good - and perfectly adequate performance overall. It definitely shows the weaknesses of the 10nm node - I would expect higher clocks overall if that node wasn't still seriously flawed - but despite this it manages to be perfectly decent. And I personally value the boost in iGPU performance far more than any boost in CPU performance at 15-25W. Still, I wouldn't buy a laptop with one, mainly because I think Renoir will be better overall (particularly in iGPU performance), but if I had the choice between ICL and CML I would definitely go for the former.
Posted on Reply
#78
ARF
Valantar
As for the i7-1065G7 being bad? Nah. It has the best iGPU Intel has ever made - which is genuinely quite good - and perfectly adequate performance overall. It definitely shows the weaknesses of the 10nm node - I would expect higher clocks overall if that node wasn't still seriously flawed - but despite this it manages to be perfectly decent.
It competes with Ryzen 5 2500U, Ryzen 7 2700U, Ryzen 5 3500U and Ryzen 7 3700U. Or the previous generation AMD performance.
Posted on Reply
#79
notb
ARF
It competes with Ryzen 5 2500U, Ryzen 7 2700U, Ryzen 5 3500U and Ryzen 7 3700U. Or the previous generation AMD performance.
Based on AMD's own numbers (the ones that you're promoting so much), i7-1065G7 should be more than capable competitor for 6-core Renoir U.
In Blender, which is probably the best case scenario for Ryzen, 4800U is just 40% faster.

But of course this year Intel launches Tiger Lake, which is a different beast. Most likely 20%+ faster cores and improved iGPU.
So even if Tiger Lake stops at 4 cores, it won't be far behind the best from AMD.

I'm not sure why you're so inclined to persuade us that Intel is so far behind. Roughly equivalent node, same scientific level, similar know-how and workforce quality. There's no reason why one company would make a product much better than the other.
There will be some nuances and some implications of architectures, but we should expect both companies to turn 15W into similar performance.
Posted on Reply
#80
Parn
The feature set of the 500 series platform look very promising. But I'll stick to my 9700k for another year or two. Hopefully both AMD and Intel would be on 7nm by then.
Posted on Reply
#81
ARF
Parn
The feature set of the 500 series platform look very promising. But I'll stick to my 9700k for another year or two. Hopefully both AMD and Intel would be on 7nm by then.
AMD will 95% be on TSMC's N5 by then, for Intel we don't know - they can stay on their N14.
Posted on Reply
#82
notb
ARF
AMD will 95% be on TSMC's N5 by then, for Intel we don't know - they can stay on their N14.
There is no N14 either. Seriously, are you unable to learn this or what? :o
Parn
The feature set of the 500 series platform look very promising. But I'll stick to my 9700k for another year or two. Hopefully both AMD and Intel would be on 7nm by then.
Well, so I thought: next desktop upgrade with DDR5 and all that.

But frankly, this would be fine as well. WiFi 6, USB 4, PCIe4, 2.5GbE, Xe GPU with all it's goodies.
10nm desktops would only add performance or lower power draw - something I can live without shopping for a mid-range i5.
Posted on Reply
#83
ARF
notb
There is no N10 process.
notb
There is no N14 either. Seriously, are you unable to learn this or what? :eek:
Everyone knows that these are only marketing terms which have nothing to do with the reality.

Give me a single transistors dimension on N10 and N14 that is equal to ten (or fourteen, for that matter) nanometres in length, width or depth!
Posted on Reply
#84
Valantar
ARF
Everyone knows that these are only marketing terms which have nothing to do with the reality.

Give me a single transistors dimension on N10 and N14 that is equal to ten (or fourteen, for that matter) nanometres in length, width or depth!
Why are you making up your own node names?
Posted on Reply
#85
ARF
Valantar
Why are you making up your own node names?
TSMC doesn't use "nm" in its production processes designation.

Because 10 nm doesn't correspond to the reality, to the set rules, and because it's easier to standardise across the board using something like N in front.
TSMC has 10FF.


https://en.wikipedia.org/wiki/10_nm_process
Posted on Reply
#86
Valantar
ARF
TSMC doesn't use "nm" in its production processes designation.

Because 10 nm doesn't correspond to the reality, to the set rules, and because it's easier to standardise across the board using something like N in front.
TSMC has 10FF.


https://en.wikipedia.org/wiki/10_nm_process
... so will you also start naming AMD's CPUs and APUs i3, i5, i7 and i9? Because that's what you're doing here, applying the marketing names of one company to another's products. Please stop, you're only sowing confusion. These nodes aren't directly comparable, so naming them by their marketing names (Intel 14nm (+/++/+++/++++/+++++/+++++++++++++), TSMC N7/N7P, etc) makes it easier to understand what you are talking about, not the other way around.
Posted on Reply
#87
ARF
Valantar
... so will you also start naming AMD's CPUs and APUs i3, i5, i7 and i9? Because that's what you're doing here, applying the marketing names of one company to another's products. Please stop, you're only sowing confusion. These nodes aren't directly comparable, so naming them by their marketing names (Intel 14nm (+/++/+++/++++/+++++/+++++++++++++), TSMC N7/N7P, etc) makes it easier to understand what you are talking about, not the other way around.
It's true that Intel's processes are more dense but to use nm in the name is misleading as well.

Everyone calls it 7 nm, I will call it "N10" or "10 nm".

What do you mean "These nodes aren't directly comparable"?
Posted on Reply
#88
Berfs1
I thought PCIe 4.0 was going to be present but disabled since there were so many issues with it?

notb
Based on AMD's own numbers (the ones that you're promoting so much), i7-1065G7 should be more than capable competitor for 6-core Renoir U.
In Blender, which is probably the best case scenario for Ryzen, 4800U is just 40% faster.

But of course this year Intel launches Tiger Lake, which is a different beast. Most likely 20%+ faster cores and improved iGPU.
So even if Tiger Lake stops at 4 cores, it won't be far behind the best from AMD.

I'm not sure why you're so inclined to persuade us that Intel is so far behind. Roughly equivalent node, same scientific level, similar know-how and workforce quality. There's no reason why one company would make a product much better than the other.
There will be some nuances and some implications of architectures, but we should expect both companies to turn 15W into similar performance.
yeah except Ryzen has more than 2x the performance/watt. So yes, Ryzen will be SIGNIFICANTLY better than Intel in laptops since perf/watt matters. It doesn't matter in desktops, that's why Intel CPUs pull ahead because they are drawing much more power. Performance matters in desktops, Effficiency (perf/watt) matters in laptops.
Posted on Reply
#89
notb
ARF
TSMC doesn't use "nm" in its production processes designation.
So? :o
Because 10 nm doesn't correspond to the reality, to the set rules, and because it's easier to standardise across the board using something like N in front.
TSMC has 10FF.
Why would "10nm" correspond to anything real? And if yes, to which property of node? You've shown a few.

N6 is a marketing name, like Ford Focus. So you're now calling a competing car VW Focus, because you like that naming better and the car absolutely doesn't look like a piece of clothing.
And your wicked idea of "standarizing" is that we should call all compact cars "Focus", because you can't handle multiple naming schemes.
Berfs1
I thought PCIe 4.0 was going to be present but disabled since there were so many issues with it?
What's wrong with PCIe 4.0?
yeah except Ryzen has more than 2x the performance/watt. So yes, Ryzen will be SIGNIFICANTLY better than Intel in laptops since perf/watt matters. It doesn't matter in desktops, that's why Intel CPUs pull ahead because they are drawing much more power. Performance matters in desktops, Effficiency (perf/watt) matters in laptops.
Ryzen will be significantly better in laptops than Intel's architecture that won't be used in laptops? I don't understand this comment. :)
Rocket Lake S is a desktop platform. Intel will make mobile platforms using more dense nodes.
Posted on Reply
#90
Valantar
ARF
It's true that Intel's processes are more dense but to use nm in the name is misleading as well.

Everyone calls it 7 nm, I will call it "N10" or "10 nm".

What do you mean "These nodes aren't directly comparable"?
Not directly comparable means that one fab's 7nm/M7/whatever they choose to call it node might be similar to one fab's 10nm/N10/whatever node, etc.

As for everyone calling it 7nm: well, no, TSMC doesn't, at least not in their node names.

And you fail to respond to the main point here, which is reiterated and exemplified beautifully by @notb above. Why do you insist on using one brand's marketing names on other brands' products?
Posted on Reply
#91
Berfs1
notb
So? :eek:

Why would "10nm" correspond to anything real? And if yes, to which property of node? You've shown a few.

N6 is a marketing name, like Ford Focus. So you're now calling a competing car VW Focus, because you like that naming better and the car absolutely doesn't look like a piece of clothing.
And your wicked idea of "standarizing" is that we should call all compact cars "Focus", because you can't handle multiple naming schemes.

What's wrong with PCIe 4.0?

Ryzen will be significantly better in laptops than Intel's architecture that won't be used in laptops? I don't understand this comment. :)
Rocket Lake S is a desktop platform. Intel will make mobile platforms using more dense nodes.
Intel was having issues implementing PCIe 4.0 in their next gen chips, that's what I meant. PCIe 4.0 as a whole is nice, but Intel reportedly was going to disable the feature because there were too many issues with the chipsets.

notb
Ryzen will be significantly better in laptops than Intel's architecture that won't be used in laptops? I don't understand this comment. :)
Rocket Lake S is a desktop platform. Intel will make mobile platforms using more dense nodes.
Clearly you have never heard of an i7-9750H or i9-9980HK or the upcoming 10th gen Comet Lake? Pretty sad.
Posted on Reply
#92
Valantar
Berfs1
Intel was having issues implementing PCIe 4.0 in their next gen chips, that's what I meant. PCIe 4.0 as a whole is nice, but Intel reportedly was going to disable the feature because there were too many issues with the chipsets.


Clearly you have never heard of an i7-9750H or i9-9980HK or the upcoming 10th gen Comet Lake? Pretty sad.
You guys are just talking past each other. @notb is talking about Rocket Lake-S, which is what this thread is (supposedly) about, and which is a desktop platform like all other -S Intel Platforms. It'll likely be closely related to some form of Rocket Lake-H for high-end mobile, but again, not the subject of this thread. The i7-9750H and i9-9980HK are Coffee Lake-H, not Rocket Lake-S. If we are going to compare across generations or platforms let's at least ensure that we're all talking about the same things, eh?
Posted on Reply
#93
Harry Lloyd
Can someone explain all those Intel Lakes? Because I have lost count and have no idea what is what. We have like 1500 different Lakes, but every one seems exactly the same.
Posted on Reply
#94
Valantar
Harry Lloyd
Can someone explain all those Intel Lakes? Because I have lost count and have no idea what is what. We have like 1500 different Lakes, but every one seems exactly the same.
That's because the majority of them are; just variations of the Skylake arch on variations of the 14nm node, with each one clocking slightly higher than the previous, some adding cores, and some adding hardware mitigations to security issues. Ice Lake (currently on the market, 10nm, mobile), Tiger Lake (upcoming, 10nm, mobile) and Rocket Lake (upcoming, ??nm desktop, possibly) are based on an updated architecture with better IPC. Though Rocket Lake is entirely unknown beyond vague rumors like this one.
Posted on Reply
#95
efikkan
I wouldn't put too much faith in any of the specifics of "Rocket Lake" yet. Whenever rumors are pointing in all kinds of directions it's usually a sign of people just speculating.

I find it a little odd that there is so little substantive about Rocket Lake, if it's supposed to be launched in a few months. I haven't found any references to it in the drivers either, while the Tiger Lake support was just updated a few days ago.
Posted on Reply
#98
bug
Harry Lloyd
Can someone explain all those Intel Lakes? Because I have lost count and have no idea what is what. We have like 1500 different Lakes, but every one seems exactly the same.
I've been with Intel since Sandy Bridge and i couldn't tell you that. I've lost count and the differences between them seem so minute, it's not worth keep an eye on them. I imagine it's a bit different on the mobile front, but that interests me even less.
Posted on Reply
#100
ARF
Valantar
Someone random posting stuff on a random BBS is not a "leak", it's a rumor. For it to qualify as a leak there needs to be some level of trust backing up the information, which can be either directly attached to the source (has previously given accurate information ahead of time etc.) or through the leak being verified by well-connected journalists. Until then, it's a rumor and nothing more.
I disagree. We don't know actually if it's a rumour or a leak because there is no way to verify the information, obviously because of lack of CPUs in the wild.
But it more closely resembles to a leak by an internal for the matters person..

Actually, every single employee who has access to this data, can leak it anonymously and you will never know who exactly were they.

Rumour is information by a person who doesn't understand.
Leak is information by a person who has access to the proper data.
Posted on Reply
Add your own comment