Discussion Intel current and future Lakes & Rapids thread

Page 845 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Khato

Golden Member
Jul 15, 2001
1,251
321
136
PL2 is likely to remain high as well, since even if it's not for this specific generation, it's possible future chips (cough 8+32) would be able to much better utilize the extra power.
Indeed. So many are incorrectly assuming that the PL1=PL2=250W means that's how much power the ARL SKU in question needs to achieve the specified performance. This is similar to the i7-13700k and i9-13900k both having a 253W max turbo power, but the i7 part both doesn't reach that max power consumption and offers notably lower performance. The 8+32 ARL part will have markedly higher MT performance with the same power budget, while the same 8+16 with non-k power levels (maybe still same 65W PL1, 219W PL2 as raptor lake) will be the true mark of how much efficiency has improved. Unfortunately those weren't the numbers that were leaked.
 
Jul 27, 2020
26,022
17,952
146
in your example intel may have handed over those processors for free as penalty. this tracks given their lateness and poor income.
Free processors maybe but a lot of those customers may still have to pay for the special accelerators they want through SDSi.
 

Dayman1225

Golden Member
Aug 14, 2017
1,160
996
146
While it is a relatively small upgrade it looks like Emerald Rapids has been pulled forward from Q4’23 to Q3 according to Pat


You know, data center, hey, we did a little bit better than we thought we would in Q2. The product roadmap Sapphire Rapids, Gen 4, Gen 5 with Emerald Rapids, looking to appear in Q3 a bit ahead of schedule and the roadmap for next year So I'll say we're still sort of at the green shoots level there.
 

ondma

Diamond Member
Mar 18, 2018
3,276
1,679
136
Indeed. So many are incorrectly assuming that the PL1=PL2=250W means that's how much power the ARL SKU in question needs to achieve the specified performance. This is similar to the i7-13700k and i9-13900k both having a 253W max turbo power, but the i7 part both doesn't reach that max power consumption and offers notably lower performance. The 8+32 ARL part will have markedly higher MT performance with the same power budget, while the same 8+16 with non-k power levels (maybe still same 65W PL1, 219W PL2 as raptor lake) will be the true mark of how much efficiency has improved. Unfortunately those weren't the numbers that were leaked.
8+32 ARL was cancelled, was it not? Too bad, because without hyperthreading, they really need additional e cores.
 

SiliconFly

Golden Member
Mar 10, 2023
1,925
1,284
106
8+32 ARL was cancelled, was it not? Too bad, because without hyperthreading, they really need additional e cores.
At this point, 8+32 ARL is just a rumor. Too early to say anything.

As a side note, those extra 16 E-cores (in 8+32 ARL), still require significant additional power. Can't run within the same power envelope without hampering the other cores.
 

SiliconFly

Golden Member
Mar 10, 2023
1,925
1,284
106
I keep seeing many posts on the internet saying Intel & mainly Pat Gelsinger doing a great job. And I'm sure he does. But what many casually overlook is the fact that the credit for most of the following products & technologies does not go to Pat. It was mostly initiated and/or designed by his predecessors:

- Intel 7, Intel 4/3
- Alder Lake, Raptor Lake (+refresh)
- SPR, EMR, etc.

Pat's contributions start with Meteor Lake, Arrow Lake, Granite Rapids, Sierra Forest, etc. Lunar Lake will be 100% Pat i think. Likewise, Intel 20A & 18A will be 100% Pat. But they're all still way off.

Think we shouldn't give him too much credit until he delivers. But he did save Intel from totally sinking as of now.
 
Last edited:
  • Like
Reactions: Henry swagger

Khato

Golden Member
Jul 15, 2001
1,251
321
136
At this point, 8+32 ARL is just a rumor. Too early to say anything.

As a side note, those extra 16 E-cores (in 8+32 ARL), still require significant additional power. Can't run within the same power envelope without hampering the other cores.
Exactly my point. The current generation i7-13700k and i9-13900k have the exact same PL1=PL2=253W specification, but the i7-13700k with 8 fewer E-cores both uses less power at full load and is notably lower performance. Why is the assumption that the leaked ARL performance/power numbers are analogous to the i9-13900k configuration rather than the i7-13700k?
 
  • Like
Reactions: SiliconFly

A///

Diamond Member
Feb 24, 2017
4,351
3,160
136
Exactly my point. The current generation i7-13700k and i9-13900k have the exact same PL1=PL2=253W specification, but the i7-13700k with 8 fewer E-cores both uses less power at full load and is notably lower performance. Why is the assumption that the leaked ARL performance/power numbers are analogous to the i9-13900k configuration rather than the i7-13700k?
he's a psychic mate. he knows all.
 

Abwx

Lifer
Apr 2, 2011
11,837
4,790
136
Why is the assumption that the leaked ARL performance/power numbers are analogous to the i9-13900k configuration rather than the i7-13700k?

Because that s a 8 + 16.? why would it be analogous to a 8 + 8 ?

The 8 + 8 ARL will use less power than the 8 + 16 one in the same extent than RPL 8 + 8 in respect of the 8 + 16...

You can count on Intel trying to extract as much perf as possible for the flagship and for this they ll undoubtly exhaust the full 250W, otherwise why would they let perfs on the table given that the competition will surely also get a perf boost with their new design.?
 

coercitiv

Diamond Member
Jan 24, 2014
7,225
16,982
136
Exactly my point. The current generation i7-13700k and i9-13900k have the exact same PL1=PL2=253W specification, but the i7-13700k with 8 fewer E-cores both uses less power at full load and is notably lower performance. Why is the assumption that the leaked ARL performance/power numbers are analogous to the i9-13900k configuration rather than the i7-13700k?
The 13700K uses just a little less power than the 13900K when configured per Intel's spec.
1691149331541.png

The reason you probably thought that's not the case is that some reviews measured "motherboard stock" configuration instead, in which case some mobos run above Intel spec. In this case, the 13900K ends up in a power league of it's own. Notice the massive 100W+ delta between the 13900K and 12900K, even though their official PL2 values are ~10W apart.
1691149545790.png
 

lightisgood

Senior member
May 27, 2022
242
119
86
It seems that Sierra Forest samples had achieved product level frequency.
I guess that SRF is going to arrive on Jan24.

> Several Sierra Forest customers have already turned on the boards and the silicon is meeting all performance and energy targets.
 

Hitman928

Diamond Member
Apr 15, 2012
6,642
12,245
136
It seems that Sierra Forest samples had achieved product level frequency.
I guess that SRF is going to arrive on Jan24.

> Several Sierra Forest customers have already turned on the boards and the silicon is meeting all performance and energy targets.

A January launch doesn't seem likely to me, but we'll see.

Also, from the article:

Pat Gelsinger said:
Emerald Rapids, our 5th Gen Xeon Scalable set to launch in Q4 of ‘23.

Seems strange that Pat says in their earnings call that EMR is launching Q4 and then just a couple of days later tells Yahoo we'll see it in Q3. . .

Edit:

Checked the earnings call, seems the Intel CFO also confirms EMR won't launch in Q3 by mentioning it will be pre-PRQ in Q3:

David Zinsner said:
We will have pre-PRQ reserves in the third quarter. . . we have Emerald Rapids. . .

It's hard to imagine plans changed within a couple of days after the earnings call.
 
Last edited:

moinmoin

Diamond Member
Jun 1, 2017
5,234
8,442
136
I think it just means they are in the early innings of the growth or recovery phase.
I assume it means nascent.
I'm sorry that I wasn't clear with my question. I know it's literally a plant starting to grow and figuratively something nascent.

What I don't understand is why Pat used that term there. To me using it in a business or financial context sounds very negative, akin to saying it didn't really start yet or we haven't really done much work on it yet etc.

So in the context of "Emerald Rapids looking to appear in Q3 a bit ahead of schedule" and using above term, is he saying the schedule is still in flux? EMR being ahead of schedule can still change? What's the exact purpose of using this term here? Thanks.
 

Hitman928

Diamond Member
Apr 15, 2012
6,642
12,245
136
I'm sorry that I wasn't clear with my question. I know it's literally a plant starting to grow and figuratively something nascent.

What I don't understand is why Pat used that term there. To me using it in a business or financial context sounds very negative, akin to saying it didn't really start yet or we haven't really done much work on it yet etc.

So in the context of "Emerald Rapids looking to appear in Q3 a bit ahead of schedule" and using above term, is he saying the schedule is still in flux? EMR being ahead of schedule can still change? What's the exact purpose of using this term here? Thanks.

The green shoots comment was in regards to Intel’s overall recovery and expansion progress. So, EMR being on time or even a bit early is a good sign of their recovery in the data center segment, but they still have some work to do to fully recover is what he is saying.
 
  • Like
Reactions: Tlh97 and moinmoin

Abwx

Lifer
Apr 2, 2011
11,837
4,790
136
The green shoots comment was in regards to Intel’s overall recovery and expansion progress. So, EMR being on time or even a bit early is a good sign of their recovery in the data center segment, but they still have some work to do to fully recover is what he is saying.
They are just buying time by selling hopes to gullible customers and keep them from switching plateform brand.

EMR will still be 2x less efficent than the competition, and if said customers are dragged buying gear in 2024 then their set ups will be 3x less effcient than the competing Zen 5 updated plateform, both in perf/watt and in throughput, so they ll need to buy 3 EMR servers and consume 3x the power to compete with a single AMD one, in a way Gelsinger is selling some sort of bankruptcy model...
 

Khato

Golden Member
Jul 15, 2001
1,251
321
136
Because that s a 8 + 16.? why would it be analogous to a 8 + 8 ?

The 8 + 8 ARL will use less power than the 8 + 16 one in the same extent than RPL 8 + 8 in respect of the 8 + 16...

You can count on Intel trying to extract as much perf as possible for the flagship and for this they ll undoubtly exhaust the full 250W, otherwise why would they let perfs on the table given that the competition will surely also get a perf boost with their new design.?
Because even if the rumors of it being cancelled are true, the ARL platform power rating was still designed for 8+32 being the top end SKU, not 8+16. Therefore the 8+16 SKU would be analogous to the i7-13700k in the current generation - notably lower performance and some amount less power than the top end SKU. (Note that I never said that the i7-13700k uses a lot less power, just less - I don't believe I've seen any stock benchmarks where the i7-13700k at full load uses as much power as an i9-13900k?) It's the 8+32 SKU that would use the full 250W the platform is rated for to full effect.

Now I will agree that it likely won't be exactly the same case as the current i7-13700k vs i9-13900k. RPL process/design is adequately tweaked that they can push the frequency/power well into the depths of inefficiency which likely won't be the case on ARL. And instead of going from 10->12 P-core equivalent area (counting 4E=1P for rough estimate of area/power equivalence) it's 12->16.
 

SiliconFly

Golden Member
Mar 10, 2023
1,925
1,284
106
Because even if the rumors of it being cancelled are true, the ARL platform power rating was still designed for 8+32 being the top end SKU, not 8+16. Therefore the 8+16 SKU would be analogous to the i7-13700k in the current generation - notably lower performance and some amount less power than the top end SKU. (Note that I never said that the i7-13700k uses a lot less power, just less - I don't believe I've seen any stock benchmarks where the i7-13700k at full load uses as much power as an i9-13900k?) It's the 8+32 SKU that would use the full 250W the platform is rated for to full effect.

Now I will agree that it likely won't be exactly the same case as the current i7-13700k vs i9-13900k. RPL process/design is adequately tweaked that they can push the frequency/power well into the depths of inefficiency which likely won't be the case on ARL. And instead of going from 10->12 P-core equivalent area (counting 4E=1P for rough estimate of area/power equivalence) it's 12->16.
Well, it looks like ARL is actually going to push the frequency hard at the expense of power. Desktop ARL might not be that efficient compared to competition i believe (not the laptop parts though).
 

Khato

Golden Member
Jul 15, 2001
1,251
321
136
Well, it looks like ARL is actually going to push the frequency hard at the expense of power. Desktop ARL might not be that efficient compared to competition i believe (not the laptop parts though).
According to what exactly?