Discussion Intel current and future Lakes & Rapids thread

[DrMrLordX]

Please show that "supporting information" that placed meteor lake "solidly as a mobile product".

It's been littered through this thread and elsewhere in the forum. The same one we read. How long you got? I don't currently have the time to search that up but if you really need reminders I can hunt it all down. Meanwhile:

https://www.tomshardware.com/news/intel-meteor-lake-pictured

That's one of the mockups that Intel has been showing off. Not looking like a desktop chip. The core count looks too low.

 

[dullard]

It's been littered through this thread and elsewhere in the forum. The same one we read. How long you got? I don't currently have the time to search that up but if you really need reminders I can hunt it all down. Meanwhile:

https://www.tomshardware.com/news/intel-meteor-lake-pictured

That's one of the mockups that Intel has been showing off. Not looking like a desktop chip. The core count looks too low.

No, I'm looking for something official. Something at Intel that definitively states that Meteor Lake would, in fact, not be a desktop product. Do you have any? I'm not really sure how a photo of a chip that could be a mobile chip definitively means that a desktop chip is not in the plans.

 

[IntelUser2000]

So davidbepo at Twitter seems to have some knowledge and he thinks due to Intel 4(and thus 3) taking out cobalt, it'll impact frequency, nevermind having to come straight from Raptorlake clocking insane high.

4.7GHz is not bad, but they won't be able to claim leadership desktop with this. Some BGA desktop like Tigerlake might be the case with Meteorlake. On the plus side, that opens up timing to release an Arrowlake desktop chip as fast as they can get it ready without worrying about market timing.

 

[Exist50]

So davidbepo at Twitter seems to have some knowledge and he thinks due to Intel 4(and thus 3) taking out cobalt, it'll impact frequency, nevermind having to come straight from Raptorlake clocking insane high.

4.7GHz is not bad, but they won't be able to claim leadership desktop with this. Some BGA desktop like Tigerlake might be the case with Meteorlake. On the plus side, that opens up timing to release an Arrowlake desktop chip as fast as they can get it ready without worrying about market timing.

It's utter nonsense. TSMC 5nm doesn't have cobalt, and it clocks great. Frequency should improve across the board, not decrease. Much less by a GHz.

 

[biostud]

Is anyone excited about Raptor Lake or is it just a usual Intel 9900k - >10900k upgrade?

 

[ondma]

Is anyone excited about Raptor Lake or is it just a usual Intel 9900k - >10900k upgrade?

Not particularly excited. Will be an improvement for multithreaded due to doubling of the E cores, but probably not much in gaming. I really would like to see more P cores, but I dont think power and die space allow it. May be some gaming uplift due to revised cache.

 

[coercitiv]

Will be an improvement for multithreaded due to doubling of the E cores, but probably not much in gaming.

I'm cautiously optimistic about Raptor's gaming performance:

• cache is getting pumped up at both L2 and L3 level
• memory controller will probably get optimized

I was watching a video on DDR5 vs. DDR4 technical differences from Buildzoid, and during his usual rants he made a few mentions about Intel having previously used a mem controller topology that behaved more like a single 128bit channel vs 2x64bit. The controller obviously worked in dual-channel mode, but my understanding is that memory operations were conducted in such a way as to prioritize latency at the expense of performance in complex workloads. According to Buildzoid this changed with Alder Lake, the new memory controller topology is modern, with better performance in complex workloads, but there was a latency price to pay.

The information above I took for granted, nowadays I lack the time to do proper reading on the subject, but it does fit nicely with what we've seen and measured with Alder Lake: latency at ISO timings is up but other memory metrics are doing great (or at least ok). This would suggest we're likely to see further refinement and probably latency improvements with Raptor. One thing we know for sure is memory subsystem is "holding back" the Cove cores in gaming, or to say it in a more balanced fashion: Coves have room to scale with memory system improvements.

That being said, I'm not excited about Raptor Lake, as I'm not a fan of the hybrid design on the desktop. That's not to say I won't necessarily buy one if Intel ends up selling the E cores for nothing like they did with Alder Lake. I'm very happy with my current DDR4 setup, and I may simply pick up a Raptor Cove on discount if opportunity presents itself.

 

[dullard]

Is anyone excited about Raptor Lake or is it just a usual Intel 9900k - >10900k upgrade?

I wouldn't say I'm excited by it since I have no plans for a computer late this year/early next year. But, Raptor Lake should fix the main problem with Alder Lake: 4 E cores is the worst possible hybrid situation if you multitask at all (which makes the top i5 and all i7 Alder Lakes non-advisable to purchase). That is, unless they just dump the 4 E core problem onto the i3 line and ruin it.

I think the other rumored improvements are minor in comparison:

• More cache will help in synthetic benchmarks, not sure if it will help much in real life. Maybe a bit of a boost for gaming.
• Faster official DDR5 support doesn't do much unless you have a bottom of the barrel motherboard that doesn't let you use faster memory.
• Clock speed bumps are nice but nothing spectacular.
• Power usage is a wash: higher PL2 levels but lower PL4 levels for Raptor Lake. That is, unless you are on a laptop, then power usage with dual voltage might be a game changer.

 

[biostud]

I wouldn't say I'm excited by it since I have no plans for a computer late this year/early next year. But, Raptor Lake should fix the main problem with Alder Lake: 4 E cores is the worst possible hybrid situation if you multitask at all (which makes the top i5 and all i7 Alder Lakes non-advisable to purchase). That is, unless they just dump the 4 E core problem onto the i3 line and ruin it.

I think the other rumored improvements are minor in comparison:

• More cache will help in synthetic benchmarks, not sure if it will help much in real life. Maybe a bit of a boost for gaming.
• Faster official DDR5 support doesn't do much unless you have a bottom of the barrel motherboard that doesn't let you use faster memory.
• Clock speed bumps are nice but nothing spectacular.
• Power usage is a wash: higher PL2 levels but lower PL4 levels for Raptor Lake. That is, unless you are on a laptop, then power usage with dual voltage might be a game changer.

So, some extra cores and minor changes like 9900k -> 10900k?

 

[deasd]

The most important thing I care about is if there's a chance that sub i5 RPL model would have native 8 cores die, we know ADL have 6+0 die to diferentiate the lineup of 8+8 die, since RPL can have up to 16 E-cores, the RPL 8+0 die could also be well diffentiate to the 8+16 option as well.

 

[jpiniero]

The most important thing I care about is if there's a chance that sub i5 RPL model would have native 8 cores die, we know ADL have 6+0 die to diferentiate the lineup of 8+8 die, since RPL can have up to 16 E-cores, the RPL 8+0 die could also be well diffentiate to the 8+16 option as well.

If there's a change to the core count with the small Raptor Lake, it's extremely likely to be 6+8. But I would bet on 6+0.

 

[igor_kavinski]

ADL-N, where art thou? I want to buy thee, my first Pentium laptop ever.

 

[jpiniero]

ADL-N, where art thou? I want to buy thee, my first Pentium laptop ever.

Might be two years before you can buy it, even after launch. Also it doesn't seem like the Tremont Atom is that much cheaper than the dual core Tiger Lake i3.

 

[dullard]

So, some extra cores and minor changes like 9900k -> 10900k?

I'd say it is more like the 9700K -> 10700K. That transition had better multithreaded performance (added hyperthreading), more cache, and faster clock speeds but no more P-cores. Just like Rocket Lake is rumored to do (although Rocket Lake will have better multithreaded performance via more E cores instead of adding hyperthreading).

 

[AtenRa]

If there's a change to the core count with the small Raptor Lake, it's extremely likely to be 6+8. But I would bet on 6+0.

I believe Core i5 13400 will be 6+4 and Core i5 13600K 6+8

 

[igor_kavinski]

NVIDIA rips Intel Alder Lake Architect out from under the company's eye to assist with future Arm technology

NVIDIA poaches Intel Alder Lake architect Rafi Marom from Team Blue to help spearhead new Arm-based technologies with their designs.

wccftech.com

Going from 5 GHz clocks to max 3 GHz may take him some getting used to.

 

[mikk]

It's utter nonsense. TSMC 5nm doesn't have cobalt, and it clocks great. Frequency should improve across the board, not decrease. Much less by a GHz.

Also some say Cobalt was one of the big culprits for Intels big 10nm struggle, as a result they removed it. They are using "enhanced copper" instead.

T1-1 Fig. 9.jpg on Egnyte

File T1-1 Fig. 9.jpg shared using Egnyte

btbmarketing.egnyte.com

 

[biostud]

NVIDIA rips Intel Alder Lake Architect out from under the company's eye to assist with future Arm technology

NVIDIA poaches Intel Alder Lake architect Rafi Marom from Team Blue to help spearhead new Arm-based technologies with their designs.

wccftech.com

Going from 5 GHz clocks to max 3 GHz may take him some getting used to.

Like going from pentium 4 3.06 Ghz, to core 2.33 GHz.

 

[Carfax83]

Is anyone excited about Raptor Lake or is it just a usual Intel 9900k - >10900k upgrade?

I think a lot of people are underestimating Raptor Lake, especially in regard to Zen 4. Raptor Lake will be an enhancement of Alder Lake, an already strong architecture and I suspect it will have double digit IPC gains from microarchitecture enhancements plus cache upgrades in addition to clock frequency enhancements and a more robust and performant memory controller, so it should significantly outperform Alder Lake in every metric; especially in multithreaded apps due to having twice as many Gracemont cores.

I think Raptor Lake on desktop will have a good chance of competing favorably with Zen 4 in single threaded workloads, and edging it in gaming. 5800X3D basically ties with Alder Lake using high speed DDR5, and the extra cache provided the Zen 3 part with a massive uplift in performance that likely won't be replicated by normal Zen 4 parts. Raptor Lake's IMC should have lower latency than Zen 4's due to having a more integrated design and the ability to operate at higher speeds.

Personally, I will likely be doing a full upgrade to a Raptor Lake based rig later this year unless Zen 4 is significantly better.

 

[DrMrLordX]

No, I'm looking for something official.

What I linked you is as close as it's going to get then, since that is the only Meteor Lake Intel has shown, and it allegedly is a mockup of a 6+8 compute tile. If they are going to try making that their desktop flagship in 2023 then we're all in trouble. Everything else is rumour or speculation or educated guesses or people pointing to "low power" on slides next to Meteor Lake or people making references to (recently changed) roadmaps showing huge overlap between Meteor Lake and Arrow Lake.

Anyone claiming Meteor Lake has "always" been targeted at desktop have zero supporting evidence and have been coming up gooseeggs, until maybe now, and I do not think that Intel ever really meant for a 6+8 compute tile to be their flagship. Maybe they can cobble together a larger package with two such tiles, but we will see how they cover themselves what with Arrow Lake being moved to 2024 or . . . whenever.

Btw I would also like to point out this awesome tweet from Gregory M. Bryant:

https://twitter.com/x/status/1396860052145770497

Yay it taped in! Which means jack since it's delayed to 2023. Aww. "Taped in" means "expect delays" apparently. Yes I know the tweet is from 2021,but that's kind of my point: if Intel celebrates a "tape in", expect nothing good.

 

[dullard]

What I linked you is as close as it's going to get then, since that is the only Meteor Lake Intel has shown, and it allegedly is a mockup of a 6+8 compute tile

So, you have nothing but your own hopes to base your posts on. I see.

 

[mikk]

Yay it taped in! Which means jack since it's delayed to 2023. Aww. "Taped in" means "expect delays" apparently. Yes I know the tweet is from 2021,but that's kind of my point: if Intel celebrates a "tape in", expect nothing good.

You are going crazy about a tape-in while it taped out in the meantime lol.

 

[igor_kavinski]

You are going crazy about a tape-in while it taped out in the meantime lol.

https://www.reddit.com/r/AMD_Stock/comments/q0o7oy

Tape out means they have finished the design. It doesn't even mean they can manufacture test chips yet. This is like me announcing I'm going to get married in 2023, but I don't even have a girl friend atm.

How many test chips do they usually go through before it's ready for production?

 

[Ajay]

So, you have nothing but your own hopes to base your posts on. I see.

Well, it’s a new world compared to a decade ago or more. We, the enthusiast community, used to get plenty of advance info on upcoming products. Now it’s all about Parsing leaks and patents and deciphering quarterly reports and FADs. Release dates, expected features and performance are now actively hidden from consumers till their impeding release with 'product previews' and paper releases.

There are reasons that quality websites like AT are struggling to keep up. They post articles that contain only verifiable data, with maybe a tad bit of speculation bases on industry trends. Then the are other sites that have learned to live with posting stories on every major leak, so that they get more page hits - so that they can manage to keep a couple of good tech journalists on the payroll. Sadly, the best tech journalists can make better money elsewhere.

Now it's all YouTube, Twitter, Reddit and the rest of the mad SM world pumping out 'leaks' that are often just an extrapolation of a rumor that could be true, but are't 90%+ of the time. We have companies like NV, Apple and AMD have put out false rumors to find out where the real leaks are coming from and plugging them. It’s a strange time for some of us who have been interested in CPUs and GPUs for 20+ years.

TLDR; Rumors, leaks and reading tea leaves are no substitute for verifiable data. Modern tech companies are very secretive. It's a sucky time to be an enthusiast.

 

[dullard]

TLDR; Rumors, leaks and reading tea leaves are no substitute for verifiable data. Modern tech companies are very secretive. It's a sucky time to be an enthusiast.

I agree. But, I do still find it quite enjoyable to cobble together the different pieces of verifiable data and wild rumors to try to get the best estimate of the future. No estimate is infallible though.

I find DrMrLord's conjecture laughable. He is essentially claiming that since one die might be 6+8 therefore desktop Meteor Lake does not exist. As a thought comparison, how many Alder Lake dies are there? At least 3 (desktop, mobile, and ultramobile). Plus a rumored Alder Lake-X which might make 4 dies if that rumor turns out to be true. So, how on earth does the presence of one Meteor Lake die prevent other Meteor Lake dies from existing?