𝐷𝑟. 𝐼𝑎𝑛 𝐶𝑢𝑡𝑟𝑒𝑠𝑠Account verificato

Similarly, AMD/Intel reserve the right to adjust the bins on their products without saying anything. Adjusting the 3.0 GHz point up or down 0.025 mV for yield could happen, and you wouldn't even know.

Seriously though, I asked SSD vendors back then if they would commit that they will never change controllers or NAND on a product currently in market. None of them would outright. They reserve the right to replace like-for-like mid-cycle. Some do change product rev, others do not

Re: the news that some SSD vendors change controllers and/or NAND in a product mid-cycle. My take: By the time we wrote about it in 2014 *!*, it was already old hat, and we were criticized for being late to report. So why is it suddenly news in 2020? https://www.anandtech.com/show/7763/an-update-to-kingston-ssdnow-v300-a-switch-to-slower-micron-nand …

Remember a time when Intel made Arm CPUs for smartphones? @Dellchannel21 has the info. This was way ahead of its time https://youtu.be/4TvlpMK35Sk 

My goal was to transform the serial compute into something MT, so that the individual workload was faster. I then ported a form of it to GPUs. That was basically my PhD thesis.

This is coming from a guy who worked in a research group where every simulation was ST. The way they took advantage of 4C machines was to run 4 threads. They didn't look at turning the code MT, so the limit of their perf is strictly a serial instruction path.

ST tests are testing the efficacy of that single serial stream of workload. If you can double your workload across two threads, then good for you - make it a million threads across a million cores, increase your throughput. You have TLP. Most workloads do not.

To add to this, saying that x86 vs arm ST benchmarks are wrong because you're not using the whole core is SMT2 is still a very bad argument. If a workload is serially limited, you can't multi-thread it. *Most* of workloads are serially limited. You don't just add MT.https://twitter.com/anandtech/status/1334512726928814080 …

I went into this testing with an open mind. I'm still undecided, and it's a topic we've been discussing internally a lot, especially with super wide M1 cores now available. What are your opinions?https://twitter.com/anandtech/status/1334512726928814080 …

The Crown Idle I'm pretty sure that's the game Prince Charles is playing anywayhttps://twitter.com/CamBunton/status/1334490039758299136 …

This is DCG, not CCG. Not consumer focused.

Atom based Xeons coming to TSMC.

'You will play a key role in the development and integration of QAT into Atom ; Xeon based SOC on Intel and TSMC process' So, Atom-based Xeon? Like Snow Ridge? or Atom+Xeon, like a hybrid?