Daniel Lemire's blog
Peer review is not the gold standard in science
Peer review as we know it today was introduced very late, over a century after the scientific revolution. It happened after Einstein’s time… arguably the most productive era in science. Current scientists often equate a success with the publication in a selective peer-reviewed venue. But that was never the scientific paradigm. In fact, it is pre-scientific thinking. Back in Einstein’s time, many scientists believed in the ether. It would have been difficult to dismiss the ether as a concept. The prudent approach would have been to pay lip service to the ether. Similarly, most scientists believed in eugenics. They believed in forced sterilization for the greater good. Many of the racist laws in the US followed straight from progressive science. Opposing eugenics would have been difficult in the context of peer review. It would have been difficult to challenge eugenics openly as a scientists. Recently, people like Matt Ridley challenged the idea that the SARS-Cov2 virus originated from nature. Back when he published his book on the topic, it would have been difficult to pass peer review.
You may not remember, but early on, it would widely accepted that the lab origin of SARS-Cov2 was only for far-right conspiracy theorists. The Canadian State broadcaster (CBC) told us, in its ‘science’ section:
source
Peer review is not the gold standard in science
Peer review as we know it today was introduced very late, over a century after the scientific revolution. It happened after Einstein’s time… arguably the most productive era in science. Current scientists often equate a success with the publication in a selective peer-reviewed venue. But that was never the scientific paradigm. In fact, it is pre-scientific thinking. Back in Einstein’s time, many scientists believed in the ether. It would have been difficult to dismiss the ether as a concept. The prudent approach would have been to pay lip service to the ether. Similarly, most scientists believed in eugenics. They believed in forced sterilization for the greater good. Many of the racist laws in the US followed straight from progressive science. Opposing eugenics would have been difficult in the context of peer review. It would have been difficult to challenge eugenics openly as a scientists. Recently, people like Matt Ridley challenged the idea that the SARS-Cov2 virus originated from nature. Back when he published his book on the topic, it would have been difficult to pass peer review.
You may not remember, but early on, it would widely accepted that the lab origin of SARS-Cov2 was only for far-right conspiracy theorists. The Canadian State broadcaster (CBC) told us, in its ‘science’ section:
One of the most persistent and widespread pieces of disinformation during the COVID-19 pandemic has been the conspiracy theory that the novel coronavirus that causes the disease was created in a lab — and was let loose either by accident or on purpose by some nefarious actor.Remember: ‘The reasonable man adapts himself to the world; the unreasonable man persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man.’
source
属于CYY自己的世界
Spacemit X60 (K1) SPECINT 2006 Benchmark
I ordered a BananaPi F3 last week and it arrived on May 6th. Using the opensbi and kernel compiled from armbian , replace the rootfs with archlinux.
Surprisely, I found this core has Zicond extension which was ratified in late 2023, then I also add this extension to compiler. As RISC-V vector support in gcc13.2 is not good, I used llvm 17 insted.
The result is shown below:
As an in-order micro-architecture, the result is normal, as I expect. It’s not so good compared to some other cores, such as T-Head C910, but the ISA extension is good for developers. It is an excellent device to come in 2024 for the RISC-V ecosystem, especially for RISC-V Vector software development. I have used it to benchmark my chacha20 RVV patch for openssl, and it has been merged today.
The raw result is at https://blog.cyyself.name/static/CINT2006.057.ref.html
source
(author: 陈泱宇)
Spacemit X60 (K1) SPECINT 2006 Benchmark
I ordered a BananaPi F3 last week and it arrived on May 6th. Using the opensbi and kernel compiled from armbian , replace the rootfs with archlinux.
Surprisely, I found this core has Zicond extension which was ratified in late 2023, then I also add this extension to compiler. As RISC-V vector support in gcc13.2 is not good, I used llvm 17 insted.
The result is shown below:
As an in-order micro-architecture, the result is normal, as I expect. It’s not so good compared to some other cores, such as T-Head C910, but the ISA extension is good for developers. It is an excellent device to come in 2024 for the RISC-V ecosystem, especially for RISC-V Vector software development. I have used it to benchmark my chacha20 RVV patch for openssl, and it has been merged today.
The raw result is at https://blog.cyyself.name/static/CINT2006.057.ref.html
source
(author: 陈泱宇)
Chips and Cheese
Inside the Snapdragon 855’s iGPU
#ChipAndCheese
Telegraph | source
(author: clamchowder)
Inside the Snapdragon 855’s iGPU
#ChipAndCheese
Telegraph | source
(author: clamchowder)
Chips and Cheese
Can China’s Loongson Catch Western Designs? Probably Not.
#ChipAndCheese
Telegraph | source
(author: clamchowder)
Can China’s Loongson Catch Western Designs? Probably Not.
#ChipAndCheese
Telegraph | source
(author: clamchowder)
有一些测试数据值得商榷，我觉得还是JamesAslan的靠谱一点。比如极客湾的8MB LLC容量，应该是没考虑temporal prefetcher。
