Gelsinger told there are no delays on Arrow Lake and Intels TSMC 3nm program. I think they are aiming for a desktop launch in October-November 2024 if it's really on TSMC 3nm. Similar to ADL-S "k" timeframe, also they might use MTL-S (P) 6+8 for non K variants to lower the volume on TSMC 3nm. Mobile however is due on 2025 by the looks of it because it's on 20A which is later.
Well that might have very interesting implications. The EMIB latency is not just one or two cycles. And just an observation, but Intel used SNC4 mode for the benchmarks in this article. https://insidehpc.com/2022/11/recen...ake-cpus-the-desired-platform-for-ai-and-hpc/
Maybe ServeTheHome will eventually cover it.
Those are Xeon Max CPU with HBM2e(every quadrant gets full access to a HBM2e stack instead of having to jump to a different tile or worst to a different socket).
Sure, but the same principle applies. Even AMD's had a very similar functionality since Rome to partition the IO die, and that's monolithic. https://developer.amd.com/wp-content/resources/56338_1.00_pub.pdf
Last I heard, the default recommendation for many workloads was NPS4.
That's kinda outlandish imo. If RPL-R is on Intel 4 though, I would be very happy, not due to the competitiveness of Intel changing, but because it would be so interesting for a direct node to node comparison.
Update on Availability.
The Big Xeon 3400 series will be available by the end of April as per STH.
The Big Xeon 3400 series will be available by the end of April as per STH.
According to STH the Xeon W9 3495X will be available by the end of April. The monolithic 2400 series by the end of March.
If I'm reading this correctly......
May-June 2023 = Raptor Lake R
May-June 2024 = Zen 5
August-October 2024 = Arrow Lake S
????
I'm guessing the RPL-R means refresh and not some secret Intel suffix here. Has there been any confirmation through sources was the refresh entails? On my notes I've handwritten cache increase and speed increases. I did read one rumor of the igpu getting a small update but this seems very unimportant given the sweeping changes on the Arrowlake platform.
Okay. This makes more sense. I think this Intel delay rumor was spurred by the rumor spread around last and this week that Apple is purchasing all of N3 for the next 2 years. I don't remember if they're still on N3, but I was wondering why this would be so if others were going to N3E. I've read some conflicting information on MTL-S but I think we all know it's not coming to desktop at this point and that may have been the plan all along. I get Intel wanting to push Arrowlake S on mobile only in 2025. They've still got mobile on lock but are facing stiff competition from AMD's new hardware. AMD however cannot provide as much raw product to manufacturing partners as Intel can. They can take a potential small beating or market loss there only to come roaring back with Arrowlake S. Though mobile MTL-S should provide them enough breathing room. Gelsinger stated MTL was a game changer. Whether this was a white faced lie or he really meant it is to be seen.
Keep in mind this is just me shooting from the hip. Don't take it too seriously.
Yes, that's what I meant by RPL-R. And officially, there's been no word that such a refresh even exists, much less what it entails. No particular leaks either.
Then is this a fabrication? https://www.eteknix.com/intel-slide-confirms-raptor-lake-refresh-q3-2023/
Ah, thanks, missed that one. There's some wiggle room in that Intel can launch a part but not use it for certain markets (like enterprise), but at least that's a reference.
moinmoin
Diamond Member
- Jun 1, 2017
- 5,247
- 8,462
- 136
Semi off topic: You know, one thing that irks me is the lack of original sourcing of information. So this is eteknix (who?) referring to TechPowerUp referring to VideoCardz referring to HXL on Twitter... Phew!
coercitiv
Diamond Member
- Jan 24, 2014
- 7,362
- 17,453
- 136
Reminds me of one time when someone came up with such an article to support a claim made on Anandtech forums... only for us to realize the article was based on the Anandtech forum claim
Hulk
Diamond Member
- Oct 9, 1999
- 5,143
- 3,742
- 136
As many of you know I've been keeping quite detailed Intel records for quite some time. While I realize no chart will ever be absolutely correct and even posting them infuriates many people due to inaccuracies or simply the way the chart is set up, while we wait for the next cool thing to discuss I have created one based around Intel Core, only including the P cores, and organized by process first and release date second. A couple of things to consider.
I'm basing time from 14nm to 10nm on Ice Lake (10+) since Cannon Lake was obviously an attempt to get something out for stock reporting or something like that. It was not a viable product.
I've also included Intel promoted IPC increases just for fun.
Finally if you have data you think should be deleted or added let me know. Not saying I'll change it... but I might!
Bottom line, until Ice Lake at 10nm the longest time between nodes was 27 months. 10nm required 57 months to get a viable product.
We are at 42 months and counting for Intel 4. "Houston we have a problem."
I'm basing time from 14nm to 10nm on Ice Lake (10+) since Cannon Lake was obviously an attempt to get something out for stock reporting or something like that. It was not a viable product.
I've also included Intel promoted IPC increases just for fun.
Finally if you have data you think should be deleted or added let me know. Not saying I'll change it... but I might!
Bottom line, until Ice Lake at 10nm the longest time between nodes was 27 months. 10nm required 57 months to get a viable product.
We are at 42 months and counting for Intel 4. "Houston we have a problem."
Attachments
Problem... or expected difficulty. Scaling down past where they currently are is getting massively more difficult. Witness other leading foundaries having their own problems with achieving better total processes and many others just throwing in the towel along the way. As it stands, all of the foundaries on the leading edge are essentially peddling half and quarter steps between node variants to eek out even the tiniest of improvements to fill the long gaps between full nodes. I don't see Intel as being in trouble as much as I see the crowd catching up to them and occassionally gaining a small lead. We may bag on Intel's 10nm process, but it was often within spitting distance or even better than competing nodes for absolute circuit density. When fed reasonable power, it returned reasonable performance that improved with each revision.
If you think Intel is doing so much worse, look at their competition. Samsung has struggled to stay within a full node of TSMC and Intel, underperforming all along the way for years. TSMC is experiencing delays and having to issue less aggressive subnodes on N3 (N3e being one) to keep customers happy. Who else is anywhere near these three?
Intel's schedule is, as usual, overly optimistic. They aren't falling behind on process tech though.
If you think Intel is doing so much worse, look at their competition. Samsung has struggled to stay within a full node of TSMC and Intel, underperforming all along the way for years. TSMC is experiencing delays and having to issue less aggressive subnodes on N3 (N3e being one) to keep customers happy. Who else is anywhere near these three?
Intel's schedule is, as usual, overly optimistic. They aren't falling behind on process tech though.
Just a nitpick but Pentium M is more derived from Pentium III than 4.
coercitiv
Diamond Member
- Jan 24, 2014
- 7,362
- 17,453
- 136
I could pretend to find a reason this video is not off-topic in the Intel thread, but I won't even bother, this interview it's just that good and worth watching:
Hulk
Diamond Member
- Oct 9, 1999
- 5,143
- 3,742
- 136
Absolutely correct. Core was the next desktop architecture after P4, which is why I specified it as next in line after P4 chronologically, not architecturally as you correctly pointed out.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 22K
-
News NVIDIA and Intel to Develop AI Infrastructure and Personal Computing Products- Started by poke01
- Replies: 384
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
Facebook
Twitter