nina5d ago
oh, found a cute thing while cleaning up for moving

it's cpu silicon, from ibm power9 ​

i got it when i crashed an ibm conference in france some years back, they were giving them out as souvenirs
Show threadHaelwenn /элвэн/ 
@q66 I guess it's from a failed batch? Or like IBM is that made out of money.
Mar 25 at 9:51amWeb
Show threadnina5d ago
@lanodan no idea, i would assume so

everyone got one
Show threadAaron Sawdey, Ph.D.5d ago

@lanodan @q66 assuredly it’s a failed one, there are usually quite a few with any chip this size. It’s not usually that entire wafers don’t work, more like there are defects sprinkled across the wafer and some can’t be corrected either the redundancy in the design leading to some chips being unusable.

I’m jealous, I worked on POWER9 and I don’t have one of these.

Show threadnina5d ago
@acsawdey @lanodan yeah what can the yield be for this kind of complex chip and process, 90%? and power9 had a lot of different core count configurations to take care of defective cores, but i guess even then...
Show threadAaron Sawdey, Ph.D.5d ago
@q66 @lanodan cores can be gated off, caches have redundant cells that can be switched in .. this covers a lot of the area but not all. There are some places where defects are just going to make it unusable.
Show threadnina5d ago
@acsawdey @lanodan yup, i have a bunch of (working) configs (i have 3 working power9 machines right now? one is in a rack in a data center though)