That is the "Golden Question". But if you are interested, they gave a breathtaking talk at Hot Chips a couple of years ago about the first version of the chip and they explain how they deal with the defects on the wafer. The lecture is available on RUclips and I really highly recommend watching it.
Hi Dr. Sampatrao, Indeed, the faculty website is down due to security concerns. If you need to access the slides in the meantime, please email me at adam.teman@biu.ac.il and tell me which slide decks you need. I also may move them to a different server if it is not resolved soon. Thank for your interest, Adi
Do you know that from 2008 prefixes for binary multiples are in ISO 80 000 standard? It is incredibile that a university lession does not contemplate it
Indeed, I should adhere to the standard. However, like it or not, this notation has hardly been adopted in practice, to the point where many (probably most...) engineers would not know what the difference between KB and KiB is and I'm afraid that many (maybe most?) would not know what a MiB is and this could lead to more confusion than benefit. I, for example, have never heard anyone use the term "mebibyte" or "gibibyte" and I believe that if they were to be used, a bit of laughter may arise in the presence of other engineers. That being said, it would be better practice to use the correct notation. Thank you for the comment.
Hi Isshhh, I'm not sure what you mean by "available". However, probably 90% or more of the embedded memories you will find on commercial ASICs are based on 6T SRAM. There are compilers that produce two-port 8T or dual-port 8T SRAM, but they are less commonly used. There are also register file (RF) compilers that may use one of the above or may use something closer to a flip flop. There are also content-addressable memories (CAMs), which I do not discuss in this course. But really, the vast majority is 6T SRAM, unless you go for something very specific, proprietary, or custom designed.
Thanks for your appreciation. As an educator, I long ago took the approach to spread knowledge and contribute to research, and therefore, I choose to share this content in the best way I can, which is making it accessible on RUclips.
Thank you so much Proffessor Adi Teman for this GOLD content.
My pleasure!
Great video ! I wonder what the yield is for the Cerebras Wafer Scale Engine 2 @ 2:00 !? Thank you very much for this lecture series :)
That is the "Golden Question". But if you are interested, they gave a breathtaking talk at Hot Chips a couple of years ago about the first version of the chip and they explain how they deal with the defects on the wafer. The lecture is available on RUclips and I really highly recommend watching it.
Dear Dr Adam Teman , Your videos are really great, A result of serious research. I tried opening the lecture
ppt link but its not opening.
Hi Dr. Sampatrao,
Indeed, the faculty website is down due to security concerns. If you need to access the slides in the meantime, please email me at adam.teman@biu.ac.il and tell me which slide decks you need. I also may move them to a different server if it is not resolved soon.
Thank for your interest,
Adi
My faculty website is back online www.eng.biu.ac.il/temanad/teaching/
Great lecture sir
Thanks and welcome
Do you know that from 2008 prefixes for binary multiples are in ISO 80 000 standard? It is incredibile that a university lession does not contemplate it
Indeed, I should adhere to the standard.
However, like it or not, this notation has hardly been adopted in practice, to the point where many (probably most...) engineers would not know what the difference between KB and KiB is and I'm afraid that many (maybe most?) would not know what a MiB is and this could lead to more confusion than benefit.
I, for example, have never heard anyone use the term "mebibyte" or "gibibyte" and I believe that if they were to be used, a bit of laughter may arise in the presence of other engineers.
That being said, it would be better practice to use the correct notation. Thank you for the comment.
Nice Lectures
Thanks for liking
What all type of Bitcell are available?
Hi Isshhh,
I'm not sure what you mean by "available". However, probably 90% or more of the embedded memories you will find on commercial ASICs are based on 6T SRAM.
There are compilers that produce two-port 8T or dual-port 8T SRAM, but they are less commonly used. There are also register file (RF) compilers that may use one of the above or may use something closer to a flip flop. There are also content-addressable memories (CAMs), which I do not discuss in this course.
But really, the vast majority is 6T SRAM, unless you go for something very specific, proprietary, or custom designed.
@@AdiTeman i mean to say that how many types of bitcell
@@isshhh...740 Are you asking about HDSP bitcell(High density single port), HCSP(High current single port).
OR 6T SRAM, 12T SRAM
Hi sir
Hello
lol how can this be free.
Thanks for your appreciation.
As an educator, I long ago took the approach to spread knowledge and contribute to research, and therefore, I choose to share this content in the best way I can, which is making it accessible on RUclips.