• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Discussion Intel current and future Lakes & Rapids thread

Page 255 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
About Alder Lake:

Asked what enthusiasts should be excited about in the future with Intel, he replied:

"Alderlake – it was only a small part of our Architecture Day event, but it’s the biggest architecture movement since the Core architecture in 2006. It’s a huge leap and will be very exciting.

We haven’t always delivered content creator-level performance, or at the very least you would have to choose between one platform and another, but with Alder Lake and hybrid technology, you can get both great gaming performance and content creation performance at the same time.
 
Intel clearly told that Tigerlake supports DDR4 and LPDDR4 initially. A later version could support LPDDR5. The point is that the hardware itself supports LPDDR5 but Intel dropped it.

....

That directly contradicts what Anandtech said about it.

@jpiniero two years ago when I bought this AMD was a mobile nobody, they are only final becoming available in real business systems now. Maybe this quadro would be OK for Photoshop or simple CAD programs but it would still choke on complex models and heavy lifting... and Lenovo is still selling it with these thin and light workstations even today.

The fact was, though, was I was not going to invest in a $1,200 laptop and have it only have Intel HD graphics with performance straight out of the early two thousand teens. If Intel graphics can now competently drive several large resolution desktops AND run the occasional Borderlands 3 or lighter fair without switchable graphics that's still a big win.

About drivers... Their position being what it is, I would think they drivers are going to have to get up to speed in a hurry or even competent hardware is going to sit on the shelves. AMD gets raked over the coals for their uneven drivers with every new architecture launch (for good reasons, usually) but they have a dedicated following will to tolerate and work through that - normally for better performance for the dollar than nvidia is offering. Intel will have to do at least as well out of the gate. We'll see 🙂
 
About Alder Lake:



Golden Cove will have to convincingly beat Zen 4 on per-core performance to be competitive since Zen 4 will definitely have core count advantage. It will be interesting if that actually happens.
 
And when I actually see bench,arks, real, not Intel benchmarks, then I may be impressed.

There are numerous ones for GB5 and a few other benchmarks, sure it isn’t much, but one thing is for sure, Intel isn’t out of the race yet. Tiger Lake clearly walks all over Zen 2. I hope that Tiger Lake pushes AMD to release mobile parts faster, but I worry that with Intel beginning to spin up 10nm, AMD is going to get hit hard. Zen 3 should have launched a few months ago, but I understand why they couldn’t. 7nm supply is beating them up right now.

Luckily I don’t see Intel pulling another Core architecture on them. However, when Intel ends up moving to 10nm for most of it’s products, AMD will have a damn near vertical battle ahead.

EDIT: Part of the the issue is that mobile (15W) Tiger Lake appears to soundly beat Zen 2. While Zen 3 will improve things, that puts them neck-in-neck.
 
Last edited:
There are numerous ones for GB5 and a few other benchmarks, sure it isn’t much, but one thing is for sure, Intel isn’t out of the race yet. Tiger Lake clearly walks all over Zen 2. I hope that Tiger Lake pushes AMD to release mobile parts faster, but I worry that with Intel beginning to spin up 10nm, AMD is going to get hit hard. Zen 3 should have launched a few months ago, but I understand why they couldn’t. 7nm supply is beating them up right now.

Luckily I don’t see Intel pulling another Core architecture on them. However, when Intel ends up moving to 10nm for most of it’s products, AMD will have a damn near vertical battle ahead.

EDIT: Part of the the issue is that mobile (15W) Tiger Lake appears to soundly beat Zen 2. While Zen 3 will improve things, that puts them neck-in-neck.

Lets wait and see power consumption for those 15w 4.7ghz parts. normalised to clock Zen2 and tigerlake in the benchmark you linked for ST are pretty close per clock ( tigerlake wins ~7%). On my Lenovo Slim5 the 4700u on the single thread parts doesn't even touch 10watts package power peak. What would Renoir look like with the SMU allowing ~20 watt single core burst ~4.7ghz clock? probably pretty damn close to tigerlake wouldn't you say..............

AMD has so much room to move with Zen2 let alone Zen3 in terms of mobile SKU configuration if they choose to match the way intel 10/14nm currently consumes power.
 
Lets wait and see power consumption for those 15w 4.7ghz parts. normalised to clock Zen2 and tigerlake in the benchmark you linked for ST are pretty close per clock ( tigerlake wins ~7%). On my Lenovo Slim5 the 4700u on the single thread parts doesn't even touch 10watts package power peak. What would Renoir look like with the SMU allowing ~20 watt single core burst ~4.7ghz clock? probably pretty damn close to tigerlake wouldn't you say..............

AMD has so much room to move with Zen2 let alone Zen3 in terms of mobile SKU configuration if they choose to match the way intel 10/14nm currently consumes power.
He linked a Windows result for TGL but a Linux result for Renoir - the two aren't comparable.
Compare with a Windows result for the 4800U and TGL is ~15% faster clock-for-clock, not 7%. Besides clock-for-clock performance is largely an academic discussion, the reality is that the fastest TGL-U will be some ~1.3x faster than Renoir-U in ST workloads with ~0.75x the multicore performance, which is not too shabby IMO.
 
Intel said that with Tigerlake, they did not focus on IPC improvements (i.e. larger ROB etc.) . Instead I think they focused on making it more efficient (even without better transistors) by overhauling everything. (i.e. what is normally called "CDyn optimization" by the designers). I.e. the circuit blocks get optimized so the core runs cooler at the same voltage.
 
Lets wait and see power consumption for those 15w 4.7ghz parts.
Rumor is that it guzzles power. The burst ST performance is impressive and so is the vast improvement over ICL in general, but I want to see sustained clocks in MT workloads at reasonable wattages. These GB results are nice and all, but there's more to a laptop CPU.
Golden Cove will have to convincingly beat Zen 4 on per-core performance to be competitive since Zen 4 will definitely have core count advantage. It will be interesting if that actually happens.
It won't, not at all.
I hope that Tiger Lake pushes AMD to release mobile parts faster
It has, Cezanne is coming relatively soon after TGL.
 
These are tactics to bring doubt to peoples minds so not to buy AMD. Very good PR. I'll wait for the real benchmarks and see what to buy next year.
Extensive, independent benchmarks are the way to go. Marketing benchmarks are pretty unreliable.

That said, I think a Tiger Lake / Ryzen 4000 face-off would be pretty interesting. My money is on Ryzen 4000 edging ahead.
 
There are numerous ones for GB5 and a few other benchmarks, sure it isn’t much, but one thing is for sure, Intel isn’t out of the race yet. Tiger Lake clearly walks all over Zen 2.

He asked about Alder Lake, there are no GB5 or other relevant benchmarks available at the moment. I'm curious how it will perform, Raja sounds very excited about ADL. Regarding Tigerlake, for me this is Intels best mobile ULV product they have ever released. The package in overall is just the best, mainly because they have finally a really good GPU under the hood.
 
Extensive, independent benchmarks are the way to go. Marketing benchmarks are pretty unreliable.

That said, I think a Tiger Lake / Ryzen 4000 face-off would be pretty interesting. My money is on Ryzen 4000 edging ahead.
TGL is definitely the better laptop CPU for most people.
 
Intel said that with Tigerlake, they did not focus on IPC improvements (i.e. larger ROB etc.) . Instead I think they focused on making it more efficient (even without better transistors) by overhauling everything. (i.e. what is normally called "CDyn optimization" by the designers). I.e. the circuit blocks get optimized so the core runs cooler at the same voltage.

Ice Lake brought the IPC improvements. Everyone ignored it because of the lower clockspeeds. The uplift from Skylake to Tiger Lake is significant.

Rumor is that it guzzles power. The burst ST performance is impressive and so is the vast improvement over ICL in general, but I want to see sustained clocks in MT workloads at reasonable wattages. These GB results are nice and all, but there's more to a laptop CPU.

It won't, not at all.

It has, Cezanne is coming relatively soon after TGL.

It is all well and good to be a fanboy, but spreading misinformation is dishonest.

Tiger Lake U has a TDP of 15W and a base clock of 2.7 Ghz. That is much higher than Renoir, so it is clearly a more efficient chip. This isn’t 14nm.
 
Tigerlake and 10nm SF greatly improves V/F curves:

-Tigerlake's Willow Cove can operate at 2.6GHz in the same voltage as Sunny Cove at 1.6GHz.
-3.6GHz in the same voltage as SNC at 2.6GHz.
-4.6GHz in the same voltage as SNC at 3.9GHz.
-Finally, it can go even higher with higher voltages.

Sunny Cove at 15W runs at 1.8-2GHz with MT workloads and 2.4-2.6GHz at 25W.
 
That is much higher than Renoir, so it is clearly a more efficient chip. This isn’t 14nm.
4C/8T vs 8C/16T?
Well, modern power gating is pretty advanced, but is it really fair to compare a quad core and octo core die?
In an ideal world, AMD would have to budget and volumes to have actually made quad core Renoir if for no other reason than to keep Shivansps happy on the Renoir thread, but we have what we have.
However, in terms of marketing 8C vs 4C would be hard sell even if the quad has a bit more ST performance. It is no longer Construction Core Vs Intel Core after all.
 
4C/8T vs 8C/16T?
Well, modern power gating is pretty advanced, but is it really fair to compare a quad core and octo core die?
In an ideal world, AMD would have to budget and volumes to have actually made quad core Renoir if for no other reason than to keep Shivansps happy on the Renoir thread, but we have what we have.
However, in terms of marketing 8C vs 4C would be hard sell even if the quad has a bit more ST performance. It is no longer Construction Core Vs Intel Core after all.

Actually 4300U Renoir(4c) has exactly same 2.7Ghz base clock as Tiger Lake.......
 
Tigerlake and 10nm SF greatly improves V/F curves:

-Tigerlake's Willow Cove can operate at 2.6GHz in the same voltage as Sunny Cove at 1.6GHz.
-3.6GHz in the same voltage as SNC at 2.6GHz.
-4.6GHz in the same voltage as SNC at 3.9GHz.
-Finally, it can go even higher with higher voltages.

Sunny Cove at 15W runs at 1.8-2GHz with MT workloads and 2.4-2.6GHz at 25W.

Well, it isn't a new node but it sure performs like one: about 40% from clocks alone at 28W if that voltage graph holds true.

4C/8T vs 8C/16T?
Well, modern power gating is pretty advanced, but is it really fair to compare a quad core and octo core die?
In an ideal world, AMD would have to budget and volumes to have actually made quad core Renoir if for no other reason than to keep Shivansps happy on the Renoir thread, but we have what we have.
However, in terms of marketing 8C vs 4C would be hard sell even if the quad has a bit more ST performance. It is no longer Construction Core Vs Intel Core after all.

45-50% increase in MT scores according to clocks above, ST gains... not as large but still decent at 20-25%

It's yet another 4 core, but who cares when it performs like 6?
 
Do you have a source? Not that it matters much, however AnandTech says: “The current processor we know about today is a four core processor at 15 watts.”

Forget where I saw it, but the default TDP is 25 I believe (or 28). The OEM can of course change the PL1 and PL2 to whatever they want.

There are also U SKUs that are meant to replace Y, I don't know what the 'official' TDP is for those but it's probably higher than the Icelake Y's (9).
 
Last edited:
Do you have a source? Not that it matters much, however AnandTech says: “The current processor we know about today is a four core processor at 15 watts.”

The base frequency has little relevance nowadays anyways. The 1065G7 with "1.3GHz" clocks never reach that point. I don't know why its set that low. Probably in some very limited super intensive AVX-512 code that's a special weakness for Icelake.

Also, the 1068NG7 with 2.3GHz base clocks perform essentially identically to majority of the 1065G7 devices.

The only thing the 1068NG7 guarantees is that the device you buy doesn't have a wildly varying TDP level, that can be set by the manufacturer.
 
Back
Top