@erikcorry

Software Engineer. V8 JS pioneer. Anglo-dane. Curious person at large.

 Denmark
medium.com/@erik_68861
Joined September 2008
  • Tweets13.3K
  • Following648
  • Followers1.1K
1.1KMedia
Windy day in February #Aarhus
3
1
12
After the little setback, did some work on stack allocating instead of heap allocating frames <2M!. And some improvements to gcc to handle far branches better. Love being able to fix things in any layer: The compiler, the language runtime, gcc, the CPU design or even in the ISA.
Erik@erikcorryOct 24
Oh no I ran out of space on the ice40up5k. It's a fairly small FPGA. A tiny computer (even a virtual one) isn't much fun without USB and video output. Had to remove the BTB again. Still, the other progress has been pretty good.
1
1
1
200
more replies
The sort of thing that would be a _lot_ of work to do by hand, and of course it also requires a forked binutils and gcc. But I already forked those to add Fructus support github.com/erikcorry/fructus

GitHub - erikcorry/fructus: Free 16 bit RISC-inspired ISA with 8 registers, optimizing gcc toolch...

Free 16 bit RISC-inspired ISA with 8 registers, optimizing gcc toolchain, minimal libc, beginnings of an FPGA implementation. Optimized for code size & an 8-bit data bus. - erikcorry/fructus

github.com
1
47
This last step is just caused by Richards needing quite a lot f globals (class object mostly) and looking them up with a linear search of a globals table. So it's kinda cheating. On embedded SOM++ new globals can't be created so we just embed a direct pointer in the bytecodes.
15
Hunter Biden's grift looking decidedly half-assed right now. $1m annual salary for pretending to use his Dad's influence when actually you are too busy with hookers and blow to do anything for your paymasters. Pfft. How about 1000x that while actively working in diplomacy & Govt?
Mike Bird
@BirdywordOct 24
Finally got round to reading this. In any other administration it's something that would be a top 5 history-making story, published shortly before an election!
117
My circuit for Fructus VGA output is designed to support many formats including Teletext. The rendered data just has to fit in the 32k frame buffer I have available on the ICE40. Luckily the Teefax project is still producing pages for this retro medium.
1
1
141
more replies
Rendering a page from the 40x25 byte input takes about 10ms according to the simulator. The CPU is underclocked to the 25.175MHz VGA pixel clock. Without VGA the CPU goes about 31MHz according to the simulator.
1
66
The font is about 20k and the renderer in fructus machine code is <2k. Written in C++.
22
The UK should standardize (sic) on Oxford spelling, which has -ize. In return Americans should think long and hard about how many syllables aluminium has. en.wikipedia.org/wiki/Oxford…

Oxford spelling - Wikipedia

en.wikipedia.org
Allison Floyd@AllisonRFloydOct 23
Serious question for people in commonwealth counties the use British-based English spelling: Why even have the letter z if y’all won’t use it if the word doesn’t start with z? (It’s not a serious question, I’m just curious)
2
121
Once I lengthened the pipeline I started wanting some branch prediction (of conditional branches) and some branch target prediction (for virtual C++ calls and interpreter loops). I actually managed to fit both in without breaking the LUT limit on this small FPGA.
Erik@erikcorryOct 13
Replying to @erikcorry
Is it cheating to lengthen the pipeline of the CPU and make loads and stores faster, just so your Smalltalk VM goes faster? No, it isn't!
1
1
1
5
538
I wonder if I can get under 3 million cycles with FPGA changes before I start working on SOM++ again.
1
63
Answer was yes. Got close to 2.5 million by improving load and store cycle times. xitter.cf/erikcorry/status/21139…
Erik@erikcorryOct 24
Oh no I ran out of space on the ice40up5k. It's a fairly small FPGA. A tiny computer (even a virtual one) isn't much fun without USB and video output. Had to remove the BTB again. Still, the other progress has been pretty good.
1
25
Oh no I ran out of space on the ice40up5k. It's a fairly small FPGA. A tiny computer (even a virtual one) isn't much fun without USB and video output. Had to remove the BTB again. Still, the other progress has been pretty good.
Erik@erikcorryOct 19
Once I lengthened the pipeline I started wanting some branch prediction (of conditional branches) and some branch target prediction (for virtual C++ calls and interpreter loops). I actually managed to fit both in without breaking the LUT limit on this small FPGA.
1
1
2
400
Russia requires a security guarantee that Ukraine will not defend itself the next time we attack it.
62
2,078
63
15,311
267,716
I recommend everyone upgrade Ruby immediately. This is a DoS and RCE vector if you accept user input into Regexp.new() ruby-lang.org/en/news/2026/1…
19
117
39
507
129,031
Aside from the shocking substance of this story, what's amazing is how it was unearthed by Cole. The short version: The Russians were able to penetrate Jared Kushner's iPhone and extract screenshots and data. Then, a Russian intel official was captured by Dutch intelligence (!) and flipped. The Dutch believed Kushner may have been acting on behalf of Russia, so pursued him as an intel target and were able to get into his phone. They did not find any evidence of Russian collusion, instead they watched him moving hundreds of millions of dollars from slush funds created by Saudi Arabia and the UAE, one of them housed in Israel's central bank. Incredible reporting, and if Democrats have any sense, they will pull on these threads and learn a lot more about this money.
Ryan Grim
@ryangrimOct 22
Explosive story many years in the making, now out in Vanity Fair by Matthew Cole (who will be on Breaking Points next week, and publishing soon in @DropSiteNews): Saudi Arabia and the UAE set up multi-billion slush funds for Jared Kushner to use. Iran-Contra 2 here (except with Iran-Contra, at least the US got some hostages released. Not clear what we got out of this one.)
398
9,334
350
33,833
1,751,548
good thing that apt security doesn't rely on https
1
1
136
PHP sites behind Cloudflare already have an initial set of WAF rules deployed for the known exploit vectors in these vulnerabilities. The rules are available to all customers, including free. We are seeing early attacks emerging and are updating mitigations as new vectors show up. This is a STOPGAP only. Patch as soon as @ThePHPF releases the update at 14:00 UTC today.
The PHP Foundation@ThePHPFOct 20
The PHP security team is giving advance notice: Upcoming security release Wed, Oct 21st 2026, 14:00UTC, two critical vulnerabilities, both rated CVSSv4 9.2. PHP users are encouraged to prepare to update. Earlier release if details become public or active exploit evidence.
87
876
193
5,392
754,091
I was hopeful for a moment :-)
Matteo Collina
@matteocollinaOct 19
RAM is the new bottleneck. Every Node.js process is fighting for memory: containers, serverless, cloud bills, k8s limits. And here's the maddening part: V8 can use half the memory with one flag, and almost nobody turns it on.
124
So cool to have KillSwitch featured on this podcast! Some really interesting discussions too (and not just about KillSwitch 😃)
ChaosAgentsAI
@ChaosAgentsAIOct 19
Welcome to our live show! xitter.cf/i/broadcasts/1vJpPNvOl…
0:00

ChaosAgentsAI

Welcome to our live show!

1
3
7
468
It was really fun chatting with Brandon about KillSwitch today. I didn’t expect this project to get such a big reception but it’s fun to see everyone’s thoughts on it.
The Register
@TheRegisterOct 19
Lies, anagrams, and instructions designed to confuse AI. Meet KillSwitch, an experimental programming language that makes life difficult for LLM coding agents. More here: theregister.com/ai-and-ml/20… theregister.com/ai-and-ml/20…

Write bot-resistant code with KillSwitch, a programming language designed to confuse LLMs

Lies, lots of adverbs, and prosaic stories don't make for very machine-readable code, it turns out

theregister.com
1
1
132
Load more