Jan Gray
@jangray.bsky.social
Kilocore RISC-V FPGA accelerators; former Microsoft dev tools architect; Vice-chair RISC-V SoftCPU SIG & Composable Custom Extensions Task Group; blog: https://fpga.org. Cyclist. Let's try kindness. 🇨🇦-🇺🇸
Pinned
Jan Gray
@jangray.bsky.social
· Nov 12
2GRVI Phalanx at Hot Chips 31 (2019): The First Kilocore RISC-V RV64I with High Bandwidth Memory
This week at Hot Chips 31 (2019) I am presenting a status update poster on the work-in-progress GRVI Phalanx Accelerator Kit: 2GRVI Phalanx: Towards Kilocore RISC-V FPGA Accelerators with HBM2 DRAM…
fpga.org
#FPGA #RISCV
2GRVI Phalanx at Hot Chips 31: The First Kilocore RISC-V RV64I with HBM2 High Bandwidth Memory
fpga.org/2019/08/19/2...
2GRVI Phalanx at Hot Chips 31: The First Kilocore RISC-V RV64I with HBM2 High Bandwidth Memory
fpga.org/2019/08/19/2...
Reposted by Jan Gray
Initial conditions:
m1=21.4 m2=17.1 m3=52.9 (solar masses)
v1x=2.253 v1y=5.14 v2x=3.9 v2y=-1.408 v3x=-4.494 v3y=-6.724 (km/s)
x1=15.0 y1=1.0 x2=34.0 y2=6.0 x3=35.0 y3=-28.0 (AU from center)
Music: What Did You Do (The Expanse) – Shorter
m1=21.4 m2=17.1 m3=52.9 (solar masses)
v1x=2.253 v1y=5.14 v2x=3.9 v2y=-1.408 v3x=-4.494 v3y=-6.724 (km/s)
x1=15.0 y1=1.0 x2=34.0 y2=6.0 x3=35.0 y3=-28.0 (AU from center)
Music: What Did You Do (The Expanse) – Shorter
November 10, 2025 at 7:20 AM
Initial conditions:
m1=21.4 m2=17.1 m3=52.9 (solar masses)
v1x=2.253 v1y=5.14 v2x=3.9 v2y=-1.408 v3x=-4.494 v3y=-6.724 (km/s)
x1=15.0 y1=1.0 x2=34.0 y2=6.0 x3=35.0 y3=-28.0 (AU from center)
Music: What Did You Do (The Expanse) – Shorter
m1=21.4 m2=17.1 m3=52.9 (solar masses)
v1x=2.253 v1y=5.14 v2x=3.9 v2y=-1.408 v3x=-4.494 v3y=-6.724 (km/s)
x1=15.0 y1=1.0 x2=34.0 y2=6.0 x3=35.0 y3=-28.0 (AU from center)
Music: What Did You Do (The Expanse) – Shorter
Reposted by Jan Gray
Nearly two decades after my book on Cheney, I took a shot at assessing his legacy. It’s more complicated than you might think. www.theatlantic.com/ideas/2025/1...
What I Learned About Dick Cheney
The former vice president was certain he knew better than the citizens he served what was good for them.
www.theatlantic.com
November 9, 2025 at 6:40 PM
Nearly two decades after my book on Cheney, I took a shot at assessing his legacy. It’s more complicated than you might think. www.theatlantic.com/ideas/2025/1...
Reposted by Jan Gray
For onegreatidea @thestar.com
IDEA: reimagine and reopen Ontario Science Centre in Flemingdon Park as the Future of Canada Centre
CC: @mark-carney.bsky.social @onsciencecentre.bsky.social @saveosc.bsky.social @canadianarchitect.bsky.social @normsworld.bsky.social @dontrailtalker.bsky.social
IDEA: reimagine and reopen Ontario Science Centre in Flemingdon Park as the Future of Canada Centre
CC: @mark-carney.bsky.social @onsciencecentre.bsky.social @saveosc.bsky.social @canadianarchitect.bsky.social @normsworld.bsky.social @dontrailtalker.bsky.social
November 9, 2025 at 4:17 AM
For onegreatidea @thestar.com
IDEA: reimagine and reopen Ontario Science Centre in Flemingdon Park as the Future of Canada Centre
CC: @mark-carney.bsky.social @onsciencecentre.bsky.social @saveosc.bsky.social @canadianarchitect.bsky.social @normsworld.bsky.social @dontrailtalker.bsky.social
IDEA: reimagine and reopen Ontario Science Centre in Flemingdon Park as the Future of Canada Centre
CC: @mark-carney.bsky.social @onsciencecentre.bsky.social @saveosc.bsky.social @canadianarchitect.bsky.social @normsworld.bsky.social @dontrailtalker.bsky.social
In the present idiotic self-inflicted FAA brown-out, to reduce ATC workloads in busy airspaces, planes with the most pax (commercial aviation) should "trump" planes with the fewest pax (oligarch bizjets), am I right?
This policy would resolve the "crisis" most expeditiously.
This policy would resolve the "crisis" most expeditiously.
November 7, 2025 at 2:48 PM
In the present idiotic self-inflicted FAA brown-out, to reduce ATC workloads in busy airspaces, planes with the most pax (commercial aviation) should "trump" planes with the fewest pax (oligarch bizjets), am I right?
This policy would resolve the "crisis" most expeditiously.
This policy would resolve the "crisis" most expeditiously.
Reposted by Jan Gray
3 days left for our ChipFoundry sponsored TTSKY25b shuttle!
We still have 36 early bird dev kits available, so you can tapeout and receive a copy of the ASIC for just €185!
Plenty of interesting projects already submitted: app.tinytapeout.com/shuttles/tts...
We still have 36 early bird dev kits available, so you can tapeout and receive a copy of the ASIC for just €185!
Plenty of interesting projects already submitted: app.tinytapeout.com/shuttles/tts...
November 7, 2025 at 11:08 AM
3 days left for our ChipFoundry sponsored TTSKY25b shuttle!
We still have 36 early bird dev kits available, so you can tapeout and receive a copy of the ASIC for just €185!
Plenty of interesting projects already submitted: app.tinytapeout.com/shuttles/tts...
We still have 36 early bird dev kits available, so you can tapeout and receive a copy of the ASIC for just €185!
Plenty of interesting projects already submitted: app.tinytapeout.com/shuttles/tts...
Reposted by Jan Gray
Musk, X, Feb 2, 2025:
“We spent the weekend feeding USAID into the wood chipper.
Could gone to some great parties.
Did that instead.”
Please read this article 👇
“We spent the weekend feeding USAID into the wood chipper.
Could gone to some great parties.
Did that instead.”
Please read this article 👇
Elon Musk says that no one has died because he slashed humanitarian aid. I went to South Sudan to check if that's true. It's not. Meet Peter, Achol and Evan, ages 10, 8 and 5. The reckless actions of the world's richest men are killing the world's poorest kids www.nytimes.com/interactive/...
Opinion | Musk Said No One Has Died Since Aid Was Cut. That Isn’t True. (Gift Article)
A journey through the front lines of global poverty shows that when the world’s richest men slash aid for the world’s poorest children, the result is sickness, starvation and death.
www.nytimes.com
March 16, 2025 at 2:23 AM
Musk, X, Feb 2, 2025:
“We spent the weekend feeding USAID into the wood chipper.
Could gone to some great parties.
Did that instead.”
Please read this article 👇
“We spent the weekend feeding USAID into the wood chipper.
Could gone to some great parties.
Did that instead.”
Please read this article 👇
Reposted by Jan Gray
One analytical model shows that, as of November 5th, the dismantling of U.S.A.I.D. has already caused the deaths of 600,000 people, two-thirds of them children. https://newyorkermag.visitlink.me/jUzNSc
The Shutdown of U.S.A.I.D. Has Already Killed Hundreds of Thousands
The short documentary “Rovina’s Choice” tells the story of what goes when aid goes.
newyorkermag.visitlink.me
November 6, 2025 at 9:00 PM
One analytical model shows that, as of November 5th, the dismantling of U.S.A.I.D. has already caused the deaths of 600,000 people, two-thirds of them children. https://newyorkermag.visitlink.me/jUzNSc
Reposted by Jan Gray
Reposted by Jan Gray
If you're in line to bat at Game 7 of the World Series STAY IN LINE
November 2, 2025 at 3:59 AM
If you're in line to bat at Game 7 of the World Series STAY IN LINE
Reposted by Jan Gray
Yesterday I received an email informing me that my November 2025 ballot has been validated and counted. A ballot I received in the mail, completed at home at my own convenience, and sent back via mail (no postage required). #VoteWA
November 1, 2025 at 7:43 PM
Yesterday I received an email informing me that my November 2025 ballot has been validated and counted. A ballot I received in the mail, completed at home at my own convenience, and sent back via mail (no postage required). #VoteWA
Reposted by Jan Gray
A clock with two hour hands indicating daylight savings time and standard time for your timezone. You’d never have to adjust it, just remember to read the right hour hand for the season. And a constant reminder of pernicious sticky standards.
May 22, 2025 at 1:24 AM
A clock with two hour hands indicating daylight savings time and standard time for your timezone. You’d never have to adjust it, just remember to read the right hour hand for the season. And a constant reminder of pernicious sticky standards.
Surely the oldest person at Rocky Horror tonight at The Princess. It’s been 45+ years since that first showing that I tagged along with the MathSoc crew.
Exciting, illicit, and transgressive then, now? Well, quite lovely to see it again with this new generation.
Thank you bygone friends and times.
Exciting, illicit, and transgressive then, now? Well, quite lovely to see it again with this new generation.
Thank you bygone friends and times.
October 31, 2025 at 3:43 AM
Surely the oldest person at Rocky Horror tonight at The Princess. It’s been 45+ years since that first showing that I tagged along with the MathSoc crew.
Exciting, illicit, and transgressive then, now? Well, quite lovely to see it again with this new generation.
Thank you bygone friends and times.
Exciting, illicit, and transgressive then, now? Well, quite lovely to see it again with this new generation.
Thank you bygone friends and times.
Reposted by Jan Gray
Nvidia said Tuesday that it plans to integrate its Grace CPU and ConnectX-9 SuperNIC into the next-generation BlueField-4 DPU to bring 800 Gbps of network throughput to future AI data centers for high-performance inferencing. www.crn.com/news/compone...
Nvidia Reveals BlueField-4 DPU, Packed With 64-Core Grace CPU For AI Data Centers
Nvidia said Tuesday that it plans to integrate its Grace CPU and ConnectX-9 SuperNIC into the next-generation BlueField-4 DPU to bring 800 Gbps of networking throughput to future AI data centers for h...
www.crn.com
October 28, 2025 at 5:34 PM
Nvidia said Tuesday that it plans to integrate its Grace CPU and ConnectX-9 SuperNIC into the next-generation BlueField-4 DPU to bring 800 Gbps of network throughput to future AI data centers for high-performance inferencing. www.crn.com/news/compone...
Reposted by Jan Gray
Nvidia unveils BlueField-4 data processing unit, combining a 64-core Grace CPU and ConnectX-9 SuperNIC for 800 Gbps of network throughput, launching in 2026 (Dylan Martin/CRN)
Main Link | Techmeme Permalink
Main Link | Techmeme Permalink
October 28, 2025 at 5:50 PM
Nvidia unveils BlueField-4 data processing unit, combining a 64-core Grace CPU and ConnectX-9 SuperNIC for 800 Gbps of network throughput, launching in 2026 (Dylan Martin/CRN)
Main Link | Techmeme Permalink
Main Link | Techmeme Permalink
“… and do try and find time to get the moose’s head up!”
Breaking News: Prunella Scales, the British actress best known for her role as Sybil on “Fawlty Towers,” died at 93.
Prunella Scales, Sybil on ‘Fawlty Towers,’ Dies at 93
Best known “for playing unfortunate wives,” she had a decades-long career in the theater and on television.
nyti.ms
October 28, 2025 at 11:48 AM
“… and do try and find time to get the moose’s head up!”
Me, on low precision floating point LUT golf.
“Then multiplication is just a table lookup of the resulting 6b fraction after normalization and rounding. With 3b fraction inputs a.f and b.f you require 2^(3+3) = 64 entries, perfect for a 6 6-LUT 64x6b ROM. …”
From fpga.org/2023/11/27/r...
“Then multiplication is just a table lookup of the resulting 6b fraction after normalization and rounding. With 3b fraction inputs a.f and b.f you require 2^(3+3) = 64 entries, perfect for a 6 6-LUT 64x6b ROM. …”
From fpga.org/2023/11/27/r...
![From https://fpga.org/2023/11/27/risc-v-composable-extensions-for-microscaling-data-formats-for-ai-tensors/ :
“It is inexpensive to multiply such low precision floats in 6-LUT FPGAs. For example assume (flush-to-zero) normalized E4M3 × E4M3 products have a target precision of E4M6. The product of two E4M3 1.f 4b significands lie in [1.0000002 , 11.1000012]. If the product >= 2.0, normalize it by shifting the binary point to the left, i.e., 1.11000012 x 2-1, and round/truncate the (underlined) least significant bit. Then multiplication is just a table lookup of the resulting 6b fraction after normalization and rounding. With 3b fraction inputs a.f and b.f you require 23+3 = 64 entries, perfect for a 6 6-LUT 64x6b ROM. Add one LUT for the (>=2.0) shift signal and 0.5 of a 5,5-LUT for the XOR of the sign bit. Add 0.5-2 LUTs for zeroes. The (twice biased) product exponent is (a.e + b.e + shift), 4 LUTs. In all, perhaps 14 LUTs and two LUT delays. Cheap.
The greater cost of the dot product’s multiply-accumulate is the adder / adder tree. The sum of (e.g.) four element-wise products 1×2-6 + 1×2-4 + 1×2-2 + 1×20 = 1.0101012 x 20. Naively each add requires a pre-addition shifter to align mantissas, an add/sub, and then a CLZ circuit and another shifter to renormalize the sum/difference. How wide those shifters are, how wide the sum reduction adder precision is, ideally, depends on the adder tree requirements. But compared to the two LUT delays of of the multiplier, this FP reduction adds more LUT delays / pipeline stages as well as area. “It all adds up.](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:egm6nnyusca4icjiqlb7pyiw/bafkreia77yooyztxgq7an6ebzzuxg3jzigmtgmpupjge24c7kyxfxtabce@jpeg)
October 27, 2025 at 11:41 AM
Me, on low precision floating point LUT golf.
“Then multiplication is just a table lookup of the resulting 6b fraction after normalization and rounding. With 3b fraction inputs a.f and b.f you require 2^(3+3) = 64 entries, perfect for a 6 6-LUT 64x6b ROM. …”
From fpga.org/2023/11/27/r...
“Then multiplication is just a table lookup of the resulting 6b fraction after normalization and rounding. With 3b fraction inputs a.f and b.f you require 2^(3+3) = 64 entries, perfect for a 6 6-LUT 64x6b ROM. …”
From fpga.org/2023/11/27/r...
Reposted by Jan Gray
Initial conditions:
m1=112.9 m2=119.6 m3=18.3 (solar masses)
v1x=2.146 v1y=1.763 v2x=2.576 v2y=-1.125 v3x=-4.415 v3y=0.737 (km/s)
x1=-35.0 y1=-21.0 x2=6.0 y2=33.0 x3=12.0 y3=-26.0 (AU from center)
Music: Rhapsody in Blue – Gershwin
m1=112.9 m2=119.6 m3=18.3 (solar masses)
v1x=2.146 v1y=1.763 v2x=2.576 v2y=-1.125 v3x=-4.415 v3y=0.737 (km/s)
x1=-35.0 y1=-21.0 x2=6.0 y2=33.0 x3=12.0 y3=-26.0 (AU from center)
Music: Rhapsody in Blue – Gershwin
October 27, 2025 at 6:23 AM
Initial conditions:
m1=112.9 m2=119.6 m3=18.3 (solar masses)
v1x=2.146 v1y=1.763 v2x=2.576 v2y=-1.125 v3x=-4.415 v3y=0.737 (km/s)
x1=-35.0 y1=-21.0 x2=6.0 y2=33.0 x3=12.0 y3=-26.0 (AU from center)
Music: Rhapsody in Blue – Gershwin
m1=112.9 m2=119.6 m3=18.3 (solar masses)
v1x=2.146 v1y=1.763 v2x=2.576 v2y=-1.125 v3x=-4.415 v3y=0.737 (km/s)
x1=-35.0 y1=-21.0 x2=6.0 y2=33.0 x3=12.0 y3=-26.0 (AU from center)
Music: Rhapsody in Blue – Gershwin
LUT golf after my own heart.
Misaki Kida, Shimpei Sato: Hardware-Efficient Accurate 4-bit Multiplier for Xilinx 7 Series FPGAs https://arxiv.org/abs/2510.21533 https://arxiv.org/pdf/2510.21533 https://arxiv.org/html/2510.21533
October 27, 2025 at 11:08 AM
LUT golf after my own heart.
Completely fix baseball with five words.
October 26, 2025 at 6:19 PM
What a processor’s front end sees
October 26, 2025 at 3:45 PM
What a processor’s front end sees
Reposted by Jan Gray
If you loved the Living Computer Museum in Seattle as much as I did, you should know that SO MANY of those systems are still alive and running at icm.museum?donate under the same stewardship of Stephen Jones who was curator at LCM. You can visit OR connect remotely! DONATE AND DOUBLE YOUR DONATION!
The Interim Computer Museum
icm.museum
October 23, 2025 at 10:35 PM
If you loved the Living Computer Museum in Seattle as much as I did, you should know that SO MANY of those systems are still alive and running at icm.museum?donate under the same stewardship of Stephen Jones who was curator at LCM. You can visit OR connect remotely! DONATE AND DOUBLE YOUR DONATION!
Hey @waterloomath.bsky.social in the “Drive of 85” we (CSC and MathSoc kids) ran WATBALL’85 — bought a couple dozen of then inexpensive bleachers seats, rented a bus, and Oct 4 we saw the Jays fall to the Yankees 4-3 — “Lloyd Moseby dropped the ball”.
They won the division the next day. Good times.
They won the division the next day. Good times.
October 25, 2025 at 1:13 PM
Hey @waterloomath.bsky.social in the “Drive of 85” we (CSC and MathSoc kids) ran WATBALL’85 — bought a couple dozen of then inexpensive bleachers seats, rented a bus, and Oct 4 we saw the Jays fall to the Yankees 4-3 — “Lloyd Moseby dropped the ball”.
They won the division the next day. Good times.
They won the division the next day. Good times.