Thankfully those are made by many. Makes for a healthy competition. There were times when everyone was dependent on the MS/Intel liaison and that for sure was not OK.
Besides that for everything that disappears something new wil take its place. Having said that I do regret Intel selling its SSD business as these were plain good/reliable.
Last edited:
Trouble is, you had to do it both ways. There was a fair chance that Gemini would be plain wrong, it is so often. So you have to check... And that summary -- LLMs churn those out, but the content is bland and lacks nuance. Looks impressive but is not actually very good except at a superficial level.
On the other hand it amuses me all the coverage that said deepseek wouldn't answer questions about Taiwan, Tiananmen Square etc. I asked, and got fair summaries - same sort of quality as the intel answers for sure, but no attempt to dodge criticizing China....
It was a bland broad brush summary of Intel’s journey the lady 10 yrs. It basically says what we all know: Intel is in decline. No nuances needed there!
IMHO, the technical issue at Intel is that it's trying to be both a chip manufacturer AND designer. It's very hard to do nowadays.
I think they're the only one left.
Did Motorola ever do that too?
I think they're the only one left.
Did Motorola ever do that too?
https://www.reuters.com/technology/...-fumbled-revival-an-american-icon-2024-10-29/
tl;dr: Intel had a sweet deal with TSMC until Gelsinger p****d them off. He had a good idea with setting up Intel as a foundry, but it's not as simple as it looks so the process had problems.
Foundry process problems can be solved, but right in the middle of the transition the board canned the CEO and replaced him with a couple of finance people, just the recipe for success they had with Bob Swan (MBA) and Paul Otellini (MBA), who blew strategic decisions. What are the odds the two new beancounters will can the upcoming fabs, snatching defeat from the jaws of victory? Stay tuned.
MBAs have their place, perhaps, but running technical companies is not one of them. Look at MacDonnell-Douglas then Boeing for more examples.
tl;dr: Intel had a sweet deal with TSMC until Gelsinger p****d them off. He had a good idea with setting up Intel as a foundry, but it's not as simple as it looks so the process had problems.
Foundry process problems can be solved, but right in the middle of the transition the board canned the CEO and replaced him with a couple of finance people, just the recipe for success they had with Bob Swan (MBA) and Paul Otellini (MBA), who blew strategic decisions. What are the odds the two new beancounters will can the upcoming fabs, snatching defeat from the jaws of victory? Stay tuned.
MBAs have their place, perhaps, but running technical companies is not one of them. Look at MacDonnell-Douglas then Boeing for more examples.
Motorola designed and manufactured chips and discrete semiconductors (diodes and transistors) predating my career at Mot which began in 1972. I still have a digital music synthesizer that I built with Motorola RTL logic chips from the 1960's that got from boards obtained from the dumpster at Coulter diagnostics in Miami in the late 60's. I know that the 8 bit MC6800 and related families, the 16 and 32 bit MC68000 families, and the spinoff's for the automotive industry were all designed and produced at Motorola plants in Texas and Arizona. The Power PC stuff was a joint design between IBM and Motorola, with Motorola making the parts. When chips were still relatively simple, times were good.
Things got sticky as the economy faltered, so some unprofitable business operations were shut down or sold. The shut down of the discrete semiconductor business was planned, but a group of employees got funding to spin that group off as On Semiconductor. Motorola made some expensive "not ready for prime time" investments, most notably, Iridium. This made them ripe for the corporate raider Carl Icahn to take enough control to bust up the company and part it out like a junk car. The chip business became Freescale, which is now part of NXP and the cell phone group got sold to Google who took the patents and sold the rest to Lenovo, so my Motorola branded phone is actually a Lenovo product. The original two way radio division where I worked was pared down to less than half its former size and remains the only part of the original Chicago based Motorola left. I was "encouraged" to take a buyout in 2014, so I took the money and left.
The picture shows the old synthesizer boards that I built in 1971. The chips in the round metal cans are all Motorola RTL logic chips. The black ones are Fairchild equivalent chips. Fairchild is now part of On Semiconductor. Each 8 pin round can is the size of a TO-5 transistor and contains a single gate or Flip-Flop comprising of 5 to 20 transistors and resistors. Max clock speed, maybe 1 MHz on a good day with a little extra B+ voltage. 3.6 volts was the specified voltage. The chips locked up or died between 4 and 5 volts. The green board is a monophonic (one note at a time) synth that divides a master clock down by a "magic number" to produce 12 unique musical notes. One of the 12 is fed to another set of dividers to produce the lower octaves. The three brown boards are basically twelve copies without the note selection logic.
Attachments
Last edited:
Can we stop with the AI summaries they are mostly garbage
Whatever, I couldn't give a rats aerosol about Intel or any other uselefova companies.
I have worked with intel, had friends let go in the last round of desperation. Intel aligned to close with wall street instead of the tech industry, paid more attention to dividends then innovation, turned down profitable project because they weren’t profitable enough and broke the golden rule which had kept the doors open , they lost the speed/ size/stability/marketing/manufacturing lead that they held (in public perception) for 30 years, they started to deliver slow defective produce then they thumbed their nose at their own tenant and turned to TSMC to manufacture lunar lake cause they couldnt do it or couldn’t be asked, all the while dumping money into ‘modernizing’ there our manufacturing so it can ‘compete’ with TMSC in a bid to spin it off and sell it…investors have squeezed all the dirty profit out of this bloated husk and left it to be cut up and dismantled.
No, they gave in to HP's proposal to use their nearly finished 432 processor where mostly the
the Si implementation & fabrication was left to do. Well b4 64 bit time btw.
Actually this was quite pretty a CPU with a weight on safety and object orientation but
with the memory bandwidth requirements for descriptor fetches alone it could never be
the speed king. Maybe it could be interesting with the cache technology as of today.
We even had one on a Multibus I card in our CPU zoo. (multi-chip CPU)
I'm still angry that I wasn't there when it was discarded. It would be a nice trophy in a picture frame.
The idea behind the zoo was that our OS prof. wanted an implementation of
Ardrew Tanenbaum's Experimental Machine on most cpus to serve as a foundation
for a portable operating system. I did the Z8000 implementation ( where AMD proved
to be an unreliable 2nd source, which did them in for me with the forever delayed DMA chip
and the adventures around the 1000 incompatible versions of the Am9511 floating point unit)
Alpha was history when the 200 MHz Pentium came out. No contest. Did I mention our zoo?
Other roadkills were TI99, National 32K, de facto SPARC, Inmos Transputer ( I liked it, I even had a cluster)
Moto threw the towel when they could not get double memory indirect deferred addressing right in a
CISC machine. It even was possible to construct instructions that would never end (multiple page faults
in conjunction with the fancy addressing etc).
Pic: one of my designs from roughly that time span. I had some poor soul to do the layout in Orcad...
Attachments
I was on the iAPX-432 team in Portland in 1980. Specifically the "432-02" floating point chip.
The choice of 32 bit datapaths using Nchannel MOS fabrication technology and 5 volt power supplies was stupid and fatal. Any clod can multiply PowerPerGate times GateCount for a 32 bit CPU. But Justin Rattner was a special and unique kind of clod.
The choice of 32 bit datapaths using Nchannel MOS fabrication technology and 5 volt power supplies was stupid and fatal. Any clod can multiply PowerPerGate times GateCount for a 32 bit CPU. But Justin Rattner was a special and unique kind of clod.
IMHO... have you seen a modern SoC? Tons of cores... ARM, Intel... and then you get the hybrids with cores and FPGAs...
The hardware guys spend their days, nights and weekends designing the switch fabric...
The field is moving faster today that it has EVER moved... it's incredible. As I noted, you can no longer be a fab and chip designer... choose one.
Even the hardware today is mostly firmware. Historically we had software and hardware.. then we came up with firmware... but now we need a ProgHWLayer.
The hardware guys spend their days, nights and weekends designing the switch fabric...
The field is moving faster today that it has EVER moved... it's incredible. As I noted, you can no longer be a fab and chip designer... choose one.
Even the hardware today is mostly firmware. Historically we had software and hardware.. then we came up with firmware... but now we need a ProgHWLayer.
He did stop shaking his hips.
intel architecture is outdated and and their lithography is not competitive
they are done
and i am not an AMD fanboy - all my DIY builds are Intel based
only my laptop is AMD which i chose for reasons unrelated to the CPU ...
they are done
and i am not an AMD fanboy - all my DIY builds are Intel based
only my laptop is AMD which i chose for reasons unrelated to the CPU ...
Pfft... you are measuring Intel by their PC processor chips...
I've programmed this family.. SoC, Xeon, Arria FPGA
https://www.nextplatform.com/2018/05/24/a-peek-inside-that-intel-xeon-fpga-hybrid-chip/
Just to show you I'm not stuck on Intel... I've also programmed stuff like this ( Xlinx SoC with ARM and FPGA )
https://www.iveia.com/atlas-som?gad...MIpKfi2o7siwMVUB5ECB1ChgD4EAAYASAAEgLRK_D_BwE
That's where the market is going... not PCs.
The growth in mobile devices is what's has been driving ARM... because of the lower power requirements... but remember, ARM only sells you the designs, the customer takes the design and stiches them into the resulting "chip". Most often an SoC design.
Check this out..
https://www.cadence.com/en_US/home/...tion/emulation-and-prototyping/palladium.html
Last edited: