Falvyu
@falvyu.bsky.social
PhD | French | Hardware-aware Algorithm design | Image Processing | HPC
SIMD-friends: #SSE, #AVX512, #NEON, #RVV
(Opinions are my own)
SIMD-friends: #SSE, #AVX512, #NEON, #RVV
(Opinions are my own)
Reposted by Falvyu
Recent discussion about the perils of doors in gamedev reminded me of a bug caused by a door in a game you may have heard of called "Half Life 2".
I wrote it up over on Mastodon (I find it's better at long threads):
mastodon.gamedev.place/@TomF/115589...
I wrote it up over on Mastodon (I find it's better at long threads):
mastodon.gamedev.place/@TomF/115589...
Tom Forsyth (@TomF@mastodon.gamedev.place)
Attached: 1 image
Recent discussion about the perils of doors in gamedev reminded me of a bug caused by a door in a game you may heard of called "Half Life 2". Are you sitting comfortably? Then I sha...
mastodon.gamedev.place
November 21, 2025 at 10:00 PM
Recent discussion about the perils of doors in gamedev reminded me of a bug caused by a door in a game you may have heard of called "Half Life 2".
I wrote it up over on Mastodon (I find it's better at long threads):
mastodon.gamedev.place/@TomF/115589...
I wrote it up over on Mastodon (I find it's better at long threads):
mastodon.gamedev.place/@TomF/115589...
Reposted by Falvyu
Tenstorrent decided to publish the first benchmark data for Ascalon's RVV implementation using the instruction throughput benchmark of my rvv-bench benchmark suite. <3
camel-cdr.github.io/rvv-bench-re...
Overall, the results look really good so far:
camel-cdr.github.io/rvv-bench-re...
Overall, the results look really good so far:
September 25, 2025 at 5:49 PM
Tenstorrent decided to publish the first benchmark data for Ascalon's RVV implementation using the instruction throughput benchmark of my rvv-bench benchmark suite. <3
camel-cdr.github.io/rvv-bench-re...
Overall, the results look really good so far:
camel-cdr.github.io/rvv-bench-re...
Overall, the results look really good so far:
Reposted by Falvyu
The most underutilized #AVX512 instruction, VPSHUFBITQMB, is actually very useful for compactly generating a mask with arbitrary criteria based on b[5:0], e.g. to select special characters for parsing: if the second operand is a broadcasted 64-bit vector, where the ones represent the searched values
May 17, 2025 at 6:59 PM
The most underutilized #AVX512 instruction, VPSHUFBITQMB, is actually very useful for compactly generating a mask with arbitrary criteria based on b[5:0], e.g. to select special characters for parsing: if the second operand is a broadcasted 64-bit vector, where the ones represent the searched values
Reposted by Falvyu
Intel researchers have proposed Earth, an efficient architecture for RISC-V vector memory access patterns, enabling coalesced strided instruction memory access and buffer-free segment instruction processing.
arxiv.org/pdf/2504.08334
arxiv.org/pdf/2504.08334
April 15, 2025 at 7:39 PM
Intel researchers have proposed Earth, an efficient architecture for RISC-V vector memory access patterns, enabling coalesced strided instruction memory access and buffer-free segment instruction processing.
arxiv.org/pdf/2504.08334
arxiv.org/pdf/2504.08334
Reposted by Falvyu
Oh shit. I just realised no - it really is. I thought I was joking. This is going to require a long and complex backstory. Sorry absolutely not sorry it'll be funny maybe but also you'll learn something.
March 29, 2025 at 7:31 AM
Oh shit. I just realised no - it really is. I thought I was joking. This is going to require a long and complex backstory. Sorry absolutely not sorry it'll be funny maybe but also you'll learn something.
Reposted by Falvyu
Hey Folks! cool little FOSS community tool released for simulating the way @tenstorrent.bsky.social Hardware works!
Namely the way the tensix instruction stream operates: github.com/tenstorrent/...
Namely the way the tensix instruction stream operates: github.com/tenstorrent/...
GitHub - tenstorrent/tensix-isa-simulator
Contribute to tenstorrent/tensix-isa-simulator development by creating an account on GitHub.
github.com
March 27, 2025 at 5:32 PM
Hey Folks! cool little FOSS community tool released for simulating the way @tenstorrent.bsky.social Hardware works!
Namely the way the tensix instruction stream operates: github.com/tenstorrent/...
Namely the way the tensix instruction stream operates: github.com/tenstorrent/...
Reposted by Falvyu
I've now published my blog post, "I want a good parallel computer." raphlinus.github.io/gpu/2025/03/... . Thanks much to all the feedback on the draft, I'd like to think I've clarified some things that might have been confusing.
I want a good parallel computer
The GPU in your computer is about 10 to 100 times more powerful than the CPU, depending on workload. For real-time graphics rendering and machine learning, you are enjoying that power, and doing those...
raphlinus.github.io
March 21, 2025 at 7:58 PM
I've now published my blog post, "I want a good parallel computer." raphlinus.github.io/gpu/2025/03/... . Thanks much to all the feedback on the draft, I'd like to think I've clarified some things that might have been confusing.
Reposted by Falvyu
March 19, 2025 at 3:30 PM
Reposted by Falvyu
Intel AVX10 Drops Optional 512-bit: No AVX10 256-bit Only E-Cores In The Future - https://www.phoronix.com/news/Intel-AVX10-Drops-256-Bit
Intel AVX10 Drops Optional 512-bit: No AVX10 256-bit Only E-Cores In The Future
Intel updated their AVX10 whitepaper and associated open-source compiler patches around this next Advanced Vector Extensions standard... While AVX10 had intended to allow either 256-bit or 612-bit modes depending upon processor capabilities, Intel has dropped the 256-bit-only approach and going for 512-bit everywhere. Thus it would seem to indicate that Intel E cores of the future will properly support AVX 512-bit operation!..
www.phoronix.com
March 19, 2025 at 10:57 AM
Intel AVX10 Drops Optional 512-bit: No AVX10 256-bit Only E-Cores In The Future - https://www.phoronix.com/news/Intel-AVX10-Drops-256-Bit
Reposted by Falvyu
Lights, camera, action!
The world's largest digital camera has been installed at NSF–DOE Rubin Observatory! 🤩
The LSST Camera was the final major component of the observatory. With it in place, Rubin officially enters its final phase of testing!🔭
🔗: rubinobservatory.org/news/lsst-camera-installed
The world's largest digital camera has been installed at NSF–DOE Rubin Observatory! 🤩
The LSST Camera was the final major component of the observatory. With it in place, Rubin officially enters its final phase of testing!🔭
🔗: rubinobservatory.org/news/lsst-camera-installed
March 12, 2025 at 3:00 PM
Lights, camera, action!
The world's largest digital camera has been installed at NSF–DOE Rubin Observatory! 🤩
The LSST Camera was the final major component of the observatory. With it in place, Rubin officially enters its final phase of testing!🔭
🔗: rubinobservatory.org/news/lsst-camera-installed
The world's largest digital camera has been installed at NSF–DOE Rubin Observatory! 🤩
The LSST Camera was the final major component of the observatory. With it in place, Rubin officially enters its final phase of testing!🔭
🔗: rubinobservatory.org/news/lsst-camera-installed
Reposted by Falvyu
"All lessons in tech will be relearned every 25 years"
-Me, looking at us relearning the same thing again
-Me, looking at us relearning the same thing again
February 24, 2025 at 2:14 PM
"All lessons in tech will be relearned every 25 years"
-Me, looking at us relearning the same thing again
-Me, looking at us relearning the same thing again
Reposted by Falvyu
I gotta see so many posts saying "Rust is slower" and like I think these people just evaluate languages and designs based on vibes and not based on like the material output from the compilers.
February 7, 2025 at 4:44 PM
I gotta see so many posts saying "Rust is slower" and like I think these people just evaluate languages and designs based on vibes and not based on like the material output from the compilers.
Reposted by Falvyu
The chance of asteroid 2024 YR4 hitting out planet in 2032 is now 1.5%, or 1 in 67.
The OVERWHELMING likelihood is that the asteroid will miss Earth.
But for the first time ever, we might have to seriously consider a deflection mission soon.
Let me explain. 🧵 (1/x)
The OVERWHELMING likelihood is that the asteroid will miss Earth.
But for the first time ever, we might have to seriously consider a deflection mission soon.
Let me explain. 🧵 (1/x)
a picture of the earth and an asteroid with the national geographic logo
Alt: a picture of the earth and an asteroid heading towards it
media.tenor.com
February 4, 2025 at 11:53 AM
The chance of asteroid 2024 YR4 hitting out planet in 2032 is now 1.5%, or 1 in 67.
The OVERWHELMING likelihood is that the asteroid will miss Earth.
But for the first time ever, we might have to seriously consider a deflection mission soon.
Let me explain. 🧵 (1/x)
The OVERWHELMING likelihood is that the asteroid will miss Earth.
But for the first time ever, we might have to seriously consider a deflection mission soon.
Let me explain. 🧵 (1/x)
Reposted by Falvyu
Hey folks! Will be talking about the wonderful, wacky world of floating point Sunday afternoon in the HPC, Big data, & Data science dev room at 15:30!
#fosdem
#fosdem
February 1, 2025 at 5:10 PM
Hey folks! Will be talking about the wonderful, wacky world of floating point Sunday afternoon in the HPC, Big data, & Data science dev room at 15:30!
#fosdem
#fosdem
Reposted by Falvyu
February 1, 2025 at 8:33 AM
Reposted by Falvyu
Mmmmmnnnn, data.... thephd.dev/the-big-arra...
Results! - The Big Array Size Survey for C
Happy New Year! It is time report the results of the Array Size Operator survey and answer some comments people have been asking for!
thephd.dev
January 24, 2025 at 4:05 AM
Mmmmmnnnn, data.... thephd.dev/the-big-arra...
Reposted by Falvyu
SDL 3 Official Release Now Available With New APIs - https://www.phoronix.com/news/SDL3-Official-Release
SDL 3 Officially Released With New APIs, Better HiDPI & Improved Audio Handling
In addition to the Wine 10.0 stable release today, making the day very exciting as well for Linux gamers is the first official SDL 3.0 release!..
www.phoronix.com
January 21, 2025 at 10:12 PM
SDL 3 Official Release Now Available With New APIs - https://www.phoronix.com/news/SDL3-Official-Release
Reposted by Falvyu
Linux 6.14 To Add Support For SpacemiT's "Energy Efficient AI" RISC-V CPUs - https://www.phoronix.com/news/Linux-6.14-SpaceMiT-RISC-V-CPUs
Linux 6.14 To Add Support For SpacemiT's "Energy Efficient AI" RISC-V CPUs
The upcoming Linux 6.14 kernel is poised to introduce initial support for SpacemiT platforms, the Chinese computing chip company developing "next-generation RISC-V high-performance CPUs." For this next Linux kernel release the SpaceMiT Key Stone K1 octa-core RISC-V AI CPU with SpacemiT X60 cores will see support...
www.phoronix.com
January 17, 2025 at 3:12 PM
Linux 6.14 To Add Support For SpacemiT's "Energy Efficient AI" RISC-V CPUs - https://www.phoronix.com/news/Linux-6.14-SpaceMiT-RISC-V-CPUs
Reposted by Falvyu
So long and thanks for all the fish! I just observed my last star this morning... #GaiaDPAC and @esa.int Gaia teams will do their best to make some great data releases from the data I gathered! #GaiaDR4 in 2026, #GaiaDR5 around the end of the decade. Keep posted for our news today at 10:00 CET.
rotating european space agency GIF - Find & Share on GIPHY
Discover & share this rotating european space agency GIF with everyone you know. GIPHY is how you search, share, discover, and create GIFs.
giphy.com
January 15, 2025 at 8:12 AM
Reposted by Falvyu
When you turn it on and it works spectacularly!
The Rubin team has successfully completed the first on-sky engineering tests using the 144-Mpx engineering camera. Up next: installing the largest digital camera in the world, LSST Camera! 🔭🧪
rubinobservatory.org/news/rubin-completes-comcam-tests
The Rubin team has successfully completed the first on-sky engineering tests using the 144-Mpx engineering camera. Up next: installing the largest digital camera in the world, LSST Camera! 🔭🧪
rubinobservatory.org/news/rubin-completes-comcam-tests
January 14, 2025 at 6:13 PM
When you turn it on and it works spectacularly!
The Rubin team has successfully completed the first on-sky engineering tests using the 144-Mpx engineering camera. Up next: installing the largest digital camera in the world, LSST Camera! 🔭🧪
rubinobservatory.org/news/rubin-completes-comcam-tests
The Rubin team has successfully completed the first on-sky engineering tests using the 144-Mpx engineering camera. Up next: installing the largest digital camera in the world, LSST Camera! 🔭🧪
rubinobservatory.org/news/rubin-completes-comcam-tests