High Yield

@Reptalicant @YBgnn I’m like 99% sure both core clusters have a dedicated AMX unit. And that’s a lot more certainty than I’m on most other IP blocks ;)

@highyieldYT @YBgnn this is something I had in mind when I did M1, too, confused me so hard, LOL

Apple M1 dieshot annotated shot by : b23.tv/H2drEFh

@YBgnn @Reptalicant x.com/highyieldYT/st…

@highyieldYT @Reptalicant I think A13, A14 and M1, M1 Pro, M1 Max do not have AMX on E-CPU. dougallj.github.io/applecpu/icest… : No AMX for Icestorm, contrary to Firestorm.

@Reptalicant @YBgnn P-core vs E-core AMX units on the M1. I'd say they look too close to not be the same. Ofc a slightly different layout and the E-core AMX unit is smaller. AMX is an Apple specific ISA extension. If the E-cores didn't have a AMX unit, they wouldn't support the entire Apple ISA.

@highyieldYT @YBgnn god, f** google!

@Reptalicant @YBgnn Are we sure? I think they have.

@YBgnn ouch, just checked yeah 🥲 no wonder why I didnt saw any unit there, but since my memory insist there was e-amx, i labeled it anyway

@HiEq_2005 Nice, but I always wonder how precise the DSP/ISP/decode labeling is. Media encode I get, very visible. But the other stuff? Unless you are an Apple engineer, it's a lot of guessing.

🍎

@x86deadandback Thanks

Fantastic news that the RZL UL will be using 14A for only 2C. Smart move, high yield! 14A launches RibbonFET 2.0 & PowerDIRECT (BSPDN 2nd gen). Start by producing small dies on cutting-edge nodes and use mature nodes for high-core configurations.

@Taog_1575 With RZL-UL, shouldn't we see 14A late 2027 or early 2028?

Due to 14A HVM slip into 2029, so we can precisely compare the node with same HVM year. 18A cell level density 33~34% less. 14A cell level density 32~33% less. 18A to 14A reduce ~0.7% density gap! (If 14A HVM at 2028, then 14A will reduce ~8.5% density gap)

@jukan05 As long as we are compute constrained, the only thing that matters is actually having compute.

Seeing the news that Anthropic is going to use Maia made me think… At this point, does performance even matter as long as the chip works decently well? How can they use Maia? Maybe you can just go to TSMC, ask them to make a reasonably functional chip, and sell it to Anthropic… lol What exactly is the moat for ASICs?

@Darth_Goldsmith @FritzchensFritz @arenamagdotcom I haven’t seen any

@highyieldYT @FritzchensFritz @arenamagdotcom Any images of the Zen5c 16C CCD in there? :D I've seen the renders, but actual die shot seems scarce. Server parts are spendy

I basically paid for page 278 & 279 🫣 (yes, that’s the 9600X that was used by @FritzchensFritz for the die shots)

@Darth_Goldsmith @FritzchensFritz Silicon coffee table book by @arenamagdotcom

@highyieldYT @FritzchensFritz Which book?

@jwt0625 I'm planning to do this with iPhone logic boards.

3d gaussian splatting everyone, now do this for chips and advanced packaging stuff

@zelifxeh @FritzchensFritz They credit every image

@highyieldYT @FritzchensFritz No credit?

@Reptalicant @FritzchensFritz It seems to be like 50% Fritzchens Fritz die shots

@highyieldYT @FritzchensFritz love it

Also found the first mistake @arenamagdotcom Left is correct (Hawaii aka R9 290), but right is Polaris (RX 470).

@x86deadandback What's with the white bars?

"in full production" 😎

It’s finally here, the @arenamagdotcom Silicon coffee table book! Lots of amazing siliconpr0n inside 🫪

@VadimYuryev Mine should arrive tomorrow!

Just received the Silicon Arena coffee table book! Absolutely beautiful die shots and very informational so far.

@Luisrebollo1 @dylan522p It's a copypasta based on Leopolds last filings.

@dylan522p Is this satire? I can't tell anymore. I didn't know he was a multi-billionaire nor that he was into trading.

Breaking: Andrej Karpathy just filed his Q1 2026 13F. Here's everything you need to know about his recent 13F Top 10 positions: 1. VanEck Semiconductor ETF - $SMH - [Put] — $2.04B 2. Nvidia - $NVDA - [Put] — $1.57B 3. Oracle - $ORCL - [Put] — $1.07B 4. Broadcom - $AVGO - [Put] — $1.01B 5. Advanced Micro Devices - $AMD - [Put] — $969M 6. Bloom Energy - $BE — $879M 7. SanDisk - $SNDK — $724M 8. Micron - $MU - [Put] — $584M 9. CoreWeave - $CRWV — $556M 10. Taiwan Semiconductor - $TSM - [Put] — $535M New positions: • $SMH, $NVDA, $ORCL, $AVGO, $AMD, $MU, $TSM, $ASML, $INTC, $GLW — all puts • $MU [Call] — $422M • $TSM [Call] — $355M • $SNDK [Call] — $389M Summary: He kept his AI infrastructure longs and opened $8.45B in new puts against tech and semiconductor