Digital Design & Computer Arch. - Lecture 2b: Mysteries in Comp. Arch. (ETH Zürich, Spring 2020)

[@muhammadgamal9071]

I cannot find a word to describe my gratitude for your generosity that is more precise and definitive than thank you professor.

[@Mr.trending669]

Wow, I just admire how he exploit problems and uses it to generate new innovations. There is an opportunity for the next BIG THING.

[@runggp]

really can feel the energy of prof Onur Mutlu which makes the lectures so fun and motivational!

[@Redaxi]

Hocam bilkentte 2. sınıf cs öğrencisiyim cs223 dersim için bu serinizi izliyorum çok teşekkürler çok net ve güzel anlatmışsınız gerçekten.

[@feribabahir]

Dear Onur Mutlu thank you very much for the relevant, educational, and interesting Lectures. I am a IT-lawyer and I enjoy very much your lectures.

[@JonMasters]

Thanks for the shoutout to our video on Meltdown and Spectre! I gave a long form lecture on it for FOSDEM 2018 as the keynote also

[@Tracks777]

awesome content

[@Oingoboingo710]

We're waiting for the new lecture professor Onur! Thank you so much.

[@LuffyGearSecond2]

Thanks from Mexico, your lectures are great!

[@saikashyapkurella8299]

Great initiative to share the amazing knowledge you have. Thank you very much Professor!! :)

[@tiger4086]

interesting lecture！！！

[@faanross]

You are an incredible teacher, I love this so much ❤

[@dwein008]

Thanks from Colombia.

[@areejahmad3998]

where can i watch lab video?

[@ervaucher]

Dear Onur Mutlu, First of all, thank you for your amazing content. Just a speculative question about speculative execution and Spectre attacks. I couldn't find the answer anywhere in the documentation. Wouldn't it be possible to defend these kinds of attacks by measuring to an infinitely precise level the time needed to execute a particular program given a particular set of inputs and to consider any significant deviation from this timing as a potential sign of an attack ? I assume that a considerable Spectre attack would require a certain amount of conditional branches to be executed in order to fetch valuable information, and given this that the time deviation from the standard execution would be visible (as long as we work with extremely precise timings and that we control almost every parameter of the execution). If it is a current area of research, would you be so kind to provide links to sources describing these "timing" defence mechanisms ? Sincerely yours, Elliot.

[@isthereanyname]

very good on the attacks

[@JonMasters]

Rowhammer on CXL would be interesting - hopefully everyone has working ECC

[@YosseufHassan]

Amazing how much content you put here for free, but is there any prerequistes before seeing this course? like electrical knowledge?

[@TheFalsoblanco]

How do the labs look like ?

[@ramiadel5589]

جامد

[@lancelotrampaculan2075]

At what year level are his students? anyone knows?

[@minhtoannguyen1862]

1:04:30

[@TheFalsoblanco]

Why is not a solution for Row Hamering just go back and do what we did in 2008-2010 with seperated cell are the trade offs so terrible that this solution sounds so unlikeable?

[@giorgosK936]

speed

[@umutcagatay5597]

Because it requires much more space than it ever had so we wouldn't have devices which are as little-sized as mobile phones that fits in our pockets. Also it decreases the speed and performance which are the most significiant demands in technology science rn.