Discussion Intel current and future Lakes & Rapids thread

Page 702 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

LightningZ71

Golden Member
Mar 10, 2017
1,391
1,536
136
And, on a very fundamental level, we STILL don't have a good handle on how Intel 7 is actually yielding on the various fin combinations. We can externally see that Intel has been improving yields enough to make volume on progressively larger and larger chips (ice lake on volume 10nm, tiger lake on superfin, larger tigerlake 8 core on superfin [ice lake server at low speeds with big die], still larger alder lake 8+8 on enhanced superfin, finally rocket lake 8+16 on esf/intel 7. We don't know if Intel is still having yield issues on "7" that are greater than their hedging on an extra core and redundant pathways are able to make up for. In addition, saphire rapids is arguably one of their more complex packages that they've attempted in volume. There's still a lot that can go very wrong.

Simulation only takes you so far. Eventually, you put it all together and watch the train derail in new and exciting ways, until it doesn't.
 

igor_kavinski

Diamond Member
Jul 27, 2020
4,722
2,968
106
I believe that the issues/bugs are not at the Core/Logic/SDRAM parts of the CPU, but at the whol SOC system(Compute Tiles, Mesh Interconnect, HBM, UPI Links). I am not a CPU engineer but trying to simulate such complex SOC could be more complex than simulating a simple X86_64 CPU.
I believe they hired some ex-AMD engineer from the bulldozer days...
 

NostaSeronx

Diamond Member
Sep 18, 2011
3,513
1,060
136
I believe they hired some ex-AMD engineer from the bulldozer days...
SoftMachines/AMD-related (Bulldozer-related - 2001-2013) engineers were dropped from Core cores/SoCs and shifted to Atom cores/SoCs. Sapphire Rapids is on the standard team stack, not the refreshed American E-core/E-SoC team stack.

They did one P-core project and it never came out. Specifically, it never got into the roadmap.
 
Last edited:

name99

Senior member
Sep 11, 2010
402
302
136
I actually wish we could do just that, fly on the wall style! Not in a malicious way, I just think post mortems of such developments anywhere are very interesting and insightful.
You can for Wolfram (company that makes Mathematica). Their design meetings are open and available on YouTube and have been for at least five years or more.
Mathematica probably has the best bugs to complexity ratio of any software on earth (at least that I know of; maybe avionics is better – but avionics may also not be *that* complex a problem, and doesn't have as long a tail of backward compatibility?)
 

ASK THE COMMUNITY