Discussion Intel current and future Lakes & Rapids thread

Page 699 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Jul 27, 2020
16,155
10,234
106
I don't buy that. If they had them working other than any 'accelerators' they would already be selling them minus those that feature, and could start selling the ones with the accelerators later.
I think a big feature with SPR is their software licensing feature to unlock accelerators for later use. They don't want to admit defeat by launching a range of SKUs without those accelerators and licensing options. That's supposed to be their big ticket to make them "great again" in the enterprise world.
 
  • Like
Reactions: ftt
Jul 27, 2020
16,155
10,234
106
I hope you are right about the big IPC increase. Do you have a source for the four way hyperthreading? The rumors in this forum is the first I have heard of it being used.
No official source so far. You would have to get into an NDA with MLID for that revelation.
 

nicalandia

Diamond Member
Jan 10, 2019
3,330
5,281
136
Certain AI/HPC Apps that combine AMX/AVX512 that benefit from High Bandwidth memory will see a boost in performance in the Magnitude of Order when compared to standard settings. Thats where Intel will claim victory over any new AMD offerings. But on general task workload apps, Sapphire Rapids is 2 years late.
 
  • Like
Reactions: ftt

yuri69

Senior member
Jul 16, 2013
387
616
136
Royal core aims for double single core performance of golden cove and will have four way hyperthreading.. so royal core cinebench prediction 4000 single core performance.. just guessing
During that Keller's talk at Berkeley, he sounded like Intel was truly aiming for a something different. Was it a simple PR or is Intel really going to experiment? Who knows. There are always wild R&D projects with a long way to the market.

Ocean Cove is gone. Royal core on the radar? Who knows
 

inf64

Diamond Member
Mar 11, 2011
3,697
4,015
136

Doug S

Platinum Member
Feb 8, 2020
2,251
3,481
136
Intel's management is faced with many big issues from process problems they've battled for years, a resurgent AMD, and the COVID PC buying spike ending, and this is what they deem important to spend their time on?

I guess from the school of "we have nothing positive to market, so let's distract people with pointless name changes!"
 

nicalandia

Diamond Member
Jan 10, 2019
3,330
5,281
136
Intel Sapphire Rapids 300mm Si Wafer.

Total Complete Dies: 148

With 50 killer defects per wafer they could be getting 98 fully functional dies. Out of that how many 60C/120T Xeon Platinum 8490 are they getting?

1663371324042.png


This was a very good yield calculator but it's been on "Maintenance" for a few months now.
 

Ranulf

Platinum Member
Jul 18, 2001
2,348
1,165
136
It's the dumbest...

Until they do something even dumber.

Hey, if anything it got me thinking about the Pentium 2 ads. The department of redundancy department is apparently out of ideas at Intel.

RIP Intel Spaceman

 

DrMrLordX

Lifer
Apr 27, 2000
21,617
10,826
136
Intel's management is faced with many big issues from process problems they've battled for years, a resurgent AMD, and the COVID PC buying spike ending, and this is what they deem important to spend their time on?

Marketing's gotta do something. Plus I think the idea was that the Pentium and Celeron product names were confusing people, so they figured they could up sales somehow if they just stuck to a series of numbers and letters and did away with the Pentium/Celeron titles.
 

BorisTheBlade82

Senior member
May 1, 2020
663
1,014
106
@nicalandia
If find this calculator to be pretty decent: http://cloud.mooreelite.com/tools/die-yield-calculator/index.html
Without knowing Intel's defect density we cannot calculate the percentage of good dies. But instead we can do a comparison:
  • Let us assume a defect density of 0,1 defects/cm²
  • For the Alder Lake 8+8 die with a size of 215mm² (10.5 * 20.5) you will get 207 good dies out of 256 in total - around 81% yield
  • To get 148 complete dies from 300mm with a squared die it needs to be 19mmx19mm = 361mm² (there is margin for error because of scribe lines)
  • Everything else being equal you get 104 good dies and a yield of 70.5%
 

poke01

Senior member
Mar 8, 2022
724
697
106
The dumbest naming scheme in the history of Intel. Everyone involved with that decision should be fired IMO.
no, not just Intel but in the history of any company.

Convo:

Person 1: I got the new Intel Processor!
Person 2: Which Intel processor?
Person 1: You know, the one thats called Processor.
Person 2: ahh, I see. (wtf is this person talking about...)