Initially, FPGA's were not a serious threat to LSI's gate-arrays/standard-cell designs because the per-unit cost of the FPGA devices were very high by comparison, and the turnaround time for gate arrays was very fast. When I was an LSI customer in the early 1990s (at Intel), I got silicon back in 8 days (gate array), and 17 days (standard cell with 3 metal layers). ASIC costs were quite low, too. Gate-array NRE charges were around $10K for the LMA100K designs I did in 1 micron, and I think the LCB007 standard cell which was around 0.7 micron were around $35K. From there, you went into production at costs well below $20 per chip and that was way less than FPGAs. Now, as process technology advanced, the number of masks increased and the costs exploded. NRE costs for 0.18 microns (LSI's G11 process) was $250K and up. At the same time, FPGA costs were coming down as the number of available gates continued to increase, so LSI got squeezed at both ends: Better-faster-cheaper FPGAs luring customers away from the low-end, and the high-cost of taping-out designs at the high end. I just retired from the chip industry last year. When I started in 1985, I literally did the entire design myself (architecture, design, simulation, debug, layout, timing analysis, test vectors, etc). Today, you have an army of engineers working on chips, and we only get to work on small bits of the design.
Now that you told us what you did, we are hoping for a nice video about those projects you were working on back in time, nicelly ilustrated showing us the good, the bad and the ugly. Thank you for all the good work
These days FPGAs seem like an obviously great product, but surprisingly even some of Xilinx founders doubted that it was so. Bill Carter remembers that when Ross Freeman told him about his idea for an FPGA, he thought that it was "stupid" -- "the least efficient use of silicon possible. And it was going to be expensive and slow, and nobody would buy it." Not half as obvious as it seems now in retrospect.
I just started in the chip industry 2 years ago. I am designing analog components of mixed signal ASICs, and that's the only thing I do. Our main advantage is that we need very precise devices and that we can integrate multiple specific functions on a single chip, which would require multiple chips otherwise, as we're in quite a niche market. It's the only way we can justify custom silicon to our customers.
I would imagine it took quite a while before one could do complex designs on a single FPGA that were possible on ASIC's at the time. Didn't they have much lower density?
FPGA's and CPLDs will never have the same logic density as a custom IC, then, now, or in the future because there is a lot of overhead for the routing of signals and configuration of the logic. ASICs became popular in the 1980's because they were able to replace a lot of discrete ICs for a lower cost and less PCB area. I think the tipping point for FPGAs was around 100K usable gates; the per-unit cost of FPGA's is higher than an ASIC, but there is no NRE charge for FPGAs so they are common in lower-volume designs. At higher volumes, ASICs become cheaper and there were even conversion paths from FPGA to ASIC. Speed is another issue; you will never as high of a speed on an FPGA that you can get with an ASIC/SoC. Another factor is development time; today it takes almost 3 months from tapeout to silicon for an SoC in a 10nm or smaller process, not to mention many millions of $ for a set of masks. FPGAs can be recompiled in minutes, basically for free.
I couldn't count the number of systems and expansion cards I've encountered over the years with an LSI logic chip on them. They were everywhere, especially in storage.
@@gregebert5544 Hey Greg! I definitely remember visiting the Roseville DC, and meeting you, when I was working on the HP account and before the many rounds of DC consolidation.
I definitely remember seeing LSI Logic chips on all sorts of cards wondering what they did in the early 90s. I was perplexed because when I got into the industry in the mid 90s they were known for storage solutions. It's pretty clear now.
I still have two LSI-made HBAs in my home file server. They’re still the gold standard for home NAS use. It helps that bulk storage for home use is still entirely SATA, a standard that has not been meaningfully updated in almost two decades.
@@KibitoAkuya The answer is murky. First of all, gate arrays did not allow you to pick different transistors to get around defects on a die-by-die basis. In other words, gate arrays did not have fuses to select redundant cells the way on-die memory does today. However, not all defects are visible to a given customer because their design might not use certain available gates, whereas another customer design would.
I did a project with them back in the late 1980's. Flew off to their site for a few days and cranked out a custom design we then used in our products at National Instruments.
“Now forgotten” - mate, as a sysadmin, LSI Logic is still the defacto name today for SAS controllers and storage HBAs. If you make a VM, it’s probably emulating an LSI storage controller even today
@@Skukkix23 It's because of its nearly universal compatibility out of the box. As someone with a Proxmox homelab myself, I have encountered a strange behavior where emulating the storage controller is multiple times faster than VirtIO in Windows VMs. Seeing 11GB/s read speed in the days of PCIe 3 is a beautiful sight.
@Asianometry Schlumberger is not pronounced like hamburger. It is pronounced SchlumberJAY, as it is a family name in French. Source, I used to work for Schlumberger. Strangely, the French do not pronounce hamburger, hamburJAY. Source, I live in France.
I was scrolling to see if someone else picked that up. I work for an electricity/water/gas metering company and upon starting there, I saw Schlumberger is thought "burger" due to the last two syllables. I quickly heard people refer to it by its French pronunciation. Found that interesting.
I used some kind of ASIC design tool on a computer to design LS IC back in 1990 or 91. We wrote the rules, ran simulations, and it supposedly laid out the IC layers. Mine never worked properly but I still got an A, somehow, as neither the professor or I could figure out what was wrong. Supposedly, the best designed would be built into a chip, but that was probably an exaggeration. Obviously, it would have taken months to fab, and we o my had 12 week courses. I never knew if LSI was a partner for the software or just short for Large Scale Integration.
The E-Mu Emulator and Emulator II (the Ferris Bueller synthesizer) were big hits in the early '80s, but by the time E-Mu was making the Emulator III (which was about $10K) they found themselves competing with new cheaper synths. Dave Rossum at E-Mu designed the G-chip - an LSI Logic chip that was basically a single-chip DSP synth/sampler. In 1988 it was completely revolutionary. It also used DMA to play samples so it didn't need extra RAM to have 32 voices. They put it in the Proteus/1 rack synth and sold it for $995 with the best samples they could fit into 4MB. The next year they sold an orchestral version, the Proteus/2. A huge number of film/TV composers bought both. If you watched any TV in the early '90s you've heard the Proteus/1 and Proteus/2. The N64 GoldenEye pause screen is nothing but a trap beat with the Proteus Infinite One sound. The X-Files theme is the Proteus Whistlin' Joe patch. All the Thomas the Tank Engine music was done on a Proteus. I just thought maybe some of the nerds watching this would find that interesting! I don't know any other synth manufacturers using custom chips before the Xylinx era. Roland made a lot of synths to compete with E-Mu, but they used Hitachi RISC chips.
LSI Logic techs were not enthusiastic about Daisy Systems hardware in the mid-80s, when I taped out a graphics chip with them. They really believed the Sun workstation was much better. However they ported my netlist and test vectors without too much trouble. I have the graphics card from that project. One minor event not mentioned is LSI Logics joint venture with Video Seven, which they called G2. The graphics card business quickly became a money loser and some executives had to wear ankle bracelets. But That's a story for another video.
Ferranti seems to have been a significant player, at least in Europe, for quite some time in the 1970s and early 1980s, and as you note, their ULA products got into some high-volume consumer products. There are various well-rehearsed stories about Ferranti's ULAs in the context of Acorn Computers and, I believe, Sinclair Research, with contentious accusations about Ferranti not knowing their own technology well enough and such. Acorn went with VLSI Technology for at least one component, and other gate array vendors seem to have benefited from the reputation that Ferranti got, perhaps unfairly, in the press. But Ferranti did seem to have an ongoing product line of design tools that may not have been used by these relatively small companies, or the evidence for their use is thin on the ground. I think the culture in companies like Acorn and Sinclair was to roll your own solutions, as opposed to pay serious money for the vendor's tools. Later on, Acorn did buy in solutions and hire domain experts to help their ARM design work along. Related to Ferranti, there was also a company called Qudos that did electron beam lithography in conjunction with Ferranti's ULAs, but I think it proved uneconomical. That would be an interesting footnote in the story, I suppose.
FPGA and custom logic was the first anti-copy protection used on a lot of older arcade game boards. The downside to using all off the shelf components at the time was it was really easy for someone to take your design. This was usually augmented with some microcontroller with fused lock preventing readout alongside the general CPU such as a 68000 .
It would be interesting to do a similar video on VLSI Technology. They were a major player in standard cell ASICs in the 80s and 90s. You mentioned them here but more depth would be interesting.
Seeing the PT900 test station after all these years was surprising. I maintained and was lead over a system for years. Knew it inside and out. Our system had 8 bays in an L shape with an interface panel in the middle. It had a Norland digital Oscope which was a blast to use and very impressive for the day.
I miss those times even if I didn’t live them. I was born in 1971 and started my electronics engineering grad in 1988. I studied and worked with most of the tech created in those early days of the Silicon Valley. What a time to be alive that must have been.
What a great era of PCs. I hit my stride about '98, when i was a HS freshman and think the 486 to P3/AMD Athlon time-frame was when the hoby peaked for me. I sort of lost interest when in 2010ish. At that point the gains weren't as impressive, unless u dropped serious coin on a graphics card and nowadays that's even more true.
Perhaps the lingering memory from a Schlumberger "milk round" event in my time, apart from the pronunciation, was the fairly youthful representative telling a mature student that the company thought people over 35, if I remember correctly, were "past it" and not going to be considered for positions. That pretty much turned everyone in the room against them. Weird would have been a charitable word to use.
Where I am from, hard-drive RAID controllers and high-speed computer cards from LSI is commonplace. They run great, and frankly, someone have to make those chips.
There was a lot of LSI/Avago raid controllers in servers until recently, I know almost all of Dell raid controller stuff is/was actually rebranded LSI and they barely even bother to hide it. But I'm not sure what happened with the advent of nvme SSDs and whatnot because those controllers were too slow to handle them as far as I know/remember
A lot of ASICS back in the 70s were simple because you couldn't get very many transistors on a single die. They were mostly ICs with similar logic gates on them, so you could have one IC be 5 AND or OR gates, or XOR gates, and so on, and then a person would design a circuit board to use these gates, so there was a LOT of work that went into making the circuit boards and laying out the runs. So you CERTAINLY can't talk about matrix whatever back in the 70s when ICs were really just then starting to show up in numbers in the consumer world. Circuit boards would have to be big to make a full fledged computer using those 70s ICs or you needed many circuit boards to build a mainframe computer.
I love your videos! But, as I followed the war in Ukraine, I have taken a interest in the history related to defense industries and, advanced missile systems. It would be awesome if you could do a video specifically related to things like Lockheed Martin and silicon valley as well as the widespread use of FPGA & ASIC's in advanced missiles and weapon systems!
There were some recent videos dedicated to the circuit boards from the Javelin anti-tank missile. It started with one French UA-cam channel famous for teardowns of aerospace hardware. The host had acquired a guidance section of a Javelin from a US militaria auction and was showing its internals. Unfortunately most of the follow up videos on other channels which were spawned by this were just wild and mostly inaccurate speculation. There were no FPGAs in the Javelin. The missile was an old Texas Instruments project, and they simply used their own chips for image processing. At that time FPGAs just barely came into existence and were quite small, and not very useful for the purpose. Another group of videos covered circuits from Russian Tornado-S guided missiles. Many of the boards there included semi-custom chips (hard-wired gate arrays). Also no FPGAs, (except for one lonely Altera in a commercial fiber optic gyroscope.) Regarding applications of FPGAs in weapons, one can of course find general information about possible applications in the vendor's marketing materials. But beyond that it is not a subject that lends itself easily to a public lecture.
It is interesting that surface mounting has been used for special applications at least as early as 1960s in the USA and at least since 1970s in the USSR. Lots and lots of flat ceramic packages in the early space program electronics, but it was probably assembled very differently back then.
I like the concept of making half done ICs and having just the later manufacturing capabilities. All designs today are standard cell and there is power supply from the other side of the IC now. But today there is nothing a FPGA with a CPU inside could not do.
I understand this is off-topic, but I now know how to reach out to you. I recently heard about the earthquake. I sincerely hope that everyone is safe and that things can improve as much as possible
I won't ever forget LSI because of their RAID and HBA cards. They were absorbed into the monolithic Broadcom. Atmel, LSI, Brocade, Emulex, Adaptec, and now VMware turned into a wall of flesh and metal. A billion electronic screams and several million groping hands, all calling out, "COME WITH US. BE WITH US. JOIN US."
Schlumberger, the -berger is pronounced like the -berge in Faberge, like the really fancy jewelry company. Both use French pronunciation, even though Schlumberger it looks like it might want to use a German pronunciation.
Speaking of which ... Schlumberger itself might be a good topic for one of your videos? They thought that some of the ideas they used in the oil-fields would allow them to be successful in other sectors. High performance / high temp remote analog oilwell down-bore sensors meant they could get in the residential *remote* meter-reading business (Schlumberger Industries) or the *high-performance* IC business (Schlumberger Technologies).
Maybe there's time for trying going custom with silicon again, to create some sort of "PCB way" of silicon, were you can churn new designs in mere days. I have this idea of creating more advanced compilers capable of emitting masks as output, cutting out the entire process to the last step. (I'm a msc in computer science, and I'm going for a major electrical engineering) Those guys are inspiring. I probably need less than $5M in angel capital, just send me a message plz.
Psst. PROTIP: You might want to lower your mic's sensitivity and/or use a directional mic... or even use a noise-filter on whatever audio recording software you're using... because you've got a lot of traffic noise in the background of this one. 😁
In the first world, the final warp drive signatures where solved, the universe was finally cracked. All that was left to do was put the great harvester engine into supersymetric warp lock to escape the cosmos and become dweeb overlords. But the cosmos became harder than metal as the ship reached the edge of known reality. The solutions to death where the totality of finished, get to the chopper. One *_uge_* LEGO. The serpent tricked them all, "don't you know the first is the last."? Time began to slow down by the half second, "you stop that, get back here", said a green eggs, it was a great eyherial rod so long it was bi orthogonal, as you looked back behind the ship, it looked like two rods, then 16, a kylidacope formed, *_L S I_*
Initially, FPGA's were not a serious threat to LSI's gate-arrays/standard-cell designs because the per-unit cost of the FPGA devices were very high by comparison, and the turnaround time for gate arrays was very fast. When I was an LSI customer in the early 1990s (at Intel), I got silicon back in 8 days (gate array), and 17 days (standard cell with 3 metal layers). ASIC costs were quite low, too. Gate-array NRE charges were around $10K for the LMA100K designs I did in 1 micron, and I think the LCB007 standard cell which was around 0.7 micron were around $35K. From there, you went into production at costs well below $20 per chip and that was way less than FPGAs.
Now, as process technology advanced, the number of masks increased and the costs exploded. NRE costs for 0.18 microns (LSI's G11 process) was $250K and up. At the same time, FPGA costs were coming down as the number of available gates continued to increase, so LSI got squeezed at both ends: Better-faster-cheaper FPGAs luring customers away from the low-end, and the high-cost of taping-out designs at the high end.
I just retired from the chip industry last year. When I started in 1985, I literally did the entire design myself (architecture, design, simulation, debug, layout, timing analysis, test vectors, etc). Today, you have an army of engineers working on chips, and we only get to work on small bits of the design.
Now that you told us what you did, we are hoping for a nice video about those projects you were working on back in time, nicelly ilustrated showing us the good, the bad and the ugly.
Thank you for all the good work
These days FPGAs seem like an obviously great product, but surprisingly even some of Xilinx founders doubted that it was so. Bill Carter remembers that when Ross Freeman told him about his idea for an FPGA, he thought that it was "stupid" -- "the least efficient use of silicon possible. And it was going to be expensive and slow, and nobody would buy it." Not half as obvious as it seems now in retrospect.
I just started in the chip industry 2 years ago. I am designing analog components of mixed signal ASICs, and that's the only thing I do. Our main advantage is that we need very precise devices and that we can integrate multiple specific functions on a single chip, which would require multiple chips otherwise, as we're in quite a niche market. It's the only way we can justify custom silicon to our customers.
I would imagine it took quite a while before one could do complex designs on a single FPGA that were possible on ASIC's at the time. Didn't they have much lower density?
FPGA's and CPLDs will never have the same logic density as a custom IC, then, now, or in the future because there is a lot of overhead for the routing of signals and configuration of the logic. ASICs became popular in the 1980's because they were able to replace a lot of discrete ICs for a lower cost and less PCB area. I think the tipping point for FPGAs was around 100K usable gates; the per-unit cost of FPGA's is higher than an ASIC, but there is no NRE charge for FPGAs so they are common in lower-volume designs. At higher volumes, ASICs become cheaper and there were even conversion paths from FPGA to ASIC. Speed is another issue; you will never as high of a speed on an FPGA that you can get with an ASIC/SoC.
Another factor is development time; today it takes almost 3 months from tapeout to silicon for an SoC in a 10nm or smaller process, not to mention many millions of $ for a set of masks. FPGAs can be recompiled in minutes, basically for free.
I couldn't count the number of systems and expansion cards I've encountered over the years with an LSI logic chip on them. They were everywhere, especially in storage.
I worked at LSI Logic/LSI/Avago for 17 years, from 1997 to 2014. It was interesting to hear some of the company history from before I joined.
Hi Ian - I remember you were in Milpitas; I had the design center up in Roseville.
so you left around the time they merged with Broadcom?
@@gregebert5544 Hey Greg! I definitely remember visiting the Roseville DC, and meeting you, when I was working on the HP account and before the many rounds of DC consolidation.
@@davidgavin5740 yep - I think it might have been announced, but I left a month or two before that merger.
Still have joint venture fab with GF. Rumors coming in...
I definitely remember seeing LSI Logic chips on all sorts of cards wondering what they did in the early 90s. I was perplexed because when I got into the industry in the mid 90s they were known for storage solutions. It's pretty clear now.
I still have two LSI-made HBAs in my home file server. They’re still the gold standard for home NAS use. It helps that bulk storage for home use is still entirely SATA, a standard that has not been meaningfully updated in almost two decades.
I wish you could do a piece on the history of SGI. So many of their proprietary products had these ASIC chips in them.
I worked on several LSI Logic designs when I was employed at Hughes Aircraft back in the day. It was an 8000-gate (6400 usable) chip.
Were the rest of the gates redundancies to reduce defective chips?
@@KibitoAkuyaProbably. To allow space for internal testability .
@@KibitoAkuya The answer is murky. First of all, gate arrays did not allow you to pick different transistors to get around defects on a die-by-die basis. In other words, gate arrays did not have fuses to select redundant cells the way on-die memory does today. However, not all defects are visible to a given customer because their design might not use certain available gates, whereas another customer design would.
I did a project with them back in the late 1980's. Flew off to their site for a few days and cranked out a custom design we then used in our products at National Instruments.
I only knew LSI as a expensive RAID card manifacturer. Pretty intresting videos like always, keep up the work.
Like seriously how do you constantly put out such good content!
I'll have a guess and say curiosity, imagination and a love of history ... and mr Aaron, we're in concert here, it's beautiful to watch.
“Now forgotten” - mate, as a sysadmin, LSI Logic is still the defacto name today for SAS controllers and storage HBAs. If you make a VM, it’s probably emulating an LSI storage controller even today
Even if you buy a different brand, the first port of call is to flash it to LSI.
Forgotten can be synonymous with “mostly unknown”.
Hey, as someone who dabbled the last year into virtualizing truenas and getting to know proxmox (I am still pretty young), can I just ask: why?
@@Skukkix23 It's because of its nearly universal compatibility out of the box. As someone with a Proxmox homelab myself, I have encountered a strange behavior where emulating the storage controller is multiple times faster than VirtIO in Windows VMs. Seeing 11GB/s read speed in the days of PCIe 3 is a beautiful sight.
@@indignasmr7379holy f
@Asianometry Schlumberger is not pronounced like hamburger. It is pronounced SchlumberJAY, as it is a family name in French. Source, I used to work for Schlumberger. Strangely, the French do not pronounce hamburger, hamburJAY. Source, I live in France.
I was scrolling to see if someone else picked that up. I work for an electricity/water/gas metering company and upon starting there, I saw Schlumberger is thought "burger" due to the last two syllables. I quickly heard people refer to it by its French pronunciation. Found that interesting.
I had to do some searching… Why does the name Hock Tan ring a bell?… Oh yes, now I remember… he was at Commodore International!
I used some kind of ASIC design tool on a computer to design LS IC back in 1990 or 91. We wrote the rules, ran simulations, and it supposedly laid out the IC layers. Mine never worked properly but I still got an A, somehow, as neither the professor or I could figure out what was wrong.
Supposedly, the best designed would be built into a chip, but that was probably an exaggeration. Obviously, it would have taken months to fab, and we o my had 12 week courses.
I never knew if LSI was a partner for the software or just short for Large Scale Integration.
The E-Mu Emulator and Emulator II (the Ferris Bueller synthesizer) were big hits in the early '80s, but by the time E-Mu was making the Emulator III (which was about $10K) they found themselves competing with new cheaper synths. Dave Rossum at E-Mu designed the G-chip - an LSI Logic chip that was basically a single-chip DSP synth/sampler. In 1988 it was completely revolutionary. It also used DMA to play samples so it didn't need extra RAM to have 32 voices. They put it in the Proteus/1 rack synth and sold it for $995 with the best samples they could fit into 4MB. The next year they sold an orchestral version, the Proteus/2. A huge number of film/TV composers bought both. If you watched any TV in the early '90s you've heard the Proteus/1 and Proteus/2. The N64 GoldenEye pause screen is nothing but a trap beat with the Proteus Infinite One sound. The X-Files theme is the Proteus Whistlin' Joe patch. All the Thomas the Tank Engine music was done on a Proteus.
I just thought maybe some of the nerds watching this would find that interesting!
I don't know any other synth manufacturers using custom chips before the Xylinx era. Roland made a lot of synths to compete with E-Mu, but they used Hitachi RISC chips.
LSI Logic techs were not enthusiastic about Daisy Systems hardware in the mid-80s, when I taped out a graphics chip with them. They really believed the Sun workstation was much better. However they ported my netlist and test vectors without too much trouble. I have the graphics card from that project. One minor event not mentioned is LSI Logics joint venture with Video Seven, which they called G2. The graphics card business quickly became a money loser and some executives had to wear ankle bracelets. But That's a story for another video.
DO TELL DO TELL👀👀👀👀
😭😂😭😂😭😂😭
🙏 about 'ankle bracelets' part... 🙏🙏🙏
Hock Tan's closing comment sounds like something acquired VMware would say with expected results.
uh ok, you could say this when acquiring pretty much any company. It is so high level speak that it means virtually nothing.
Micromosaic reminds of Ferranti's ULA technology which was used as the glue/video logic in a lot of 80s UK microcomputers. E.g. the ZX Spectrum ULAs.
Ferranti history videos would be interesting
Ferranti seems to have been a significant player, at least in Europe, for quite some time in the 1970s and early 1980s, and as you note, their ULA products got into some high-volume consumer products. There are various well-rehearsed stories about Ferranti's ULAs in the context of Acorn Computers and, I believe, Sinclair Research, with contentious accusations about Ferranti not knowing their own technology well enough and such. Acorn went with VLSI Technology for at least one component, and other gate array vendors seem to have benefited from the reputation that Ferranti got, perhaps unfairly, in the press.
But Ferranti did seem to have an ongoing product line of design tools that may not have been used by these relatively small companies, or the evidence for their use is thin on the ground. I think the culture in companies like Acorn and Sinclair was to roll your own solutions, as opposed to pay serious money for the vendor's tools. Later on, Acorn did buy in solutions and hire domain experts to help their ARM design work along.
Related to Ferranti, there was also a company called Qudos that did electron beam lithography in conjunction with Ferranti's ULAs, but I think it proved uneconomical. That would be an interesting footnote in the story, I suppose.
FPGA and custom logic was the first anti-copy protection used on a lot of older arcade game boards. The downside to using all off the shelf components at the time was it was really easy for someone to take your design. This was usually augmented with some microcontroller with fused lock preventing readout alongside the general CPU such as a 68000 .
I can send you 6 million Zimbabwe Dollar. They were also high volume at cheap scale.
9:18 - Come for the history but we stay for memes. 🥰🥰🥰🥰🥰🥰🥰🥰
Bro. Settle down.
It would be interesting to do a similar video on VLSI Technology. They were a major player in standard cell ASICs in the 80s and 90s. You mentioned them here but more depth would be interesting.
I "remember" seeing graphics cards from them. Well that was before my time but I've seen a few.
20:45 i knew one person who worked at LSI Logic. It was in the early 90's, and she was in PR.
Another great video. Thank you
Dang two months ago? I thought early access was like a week before the video goes public. Had no idea you released videos 2 months before for Patreon.
Man, I love your video essays. What a story.
I kinda wonder if the emphasis on data center stuff came from buying Symbios. Those engineers were doing a lot of stuff like Fiberchannel.
Great video. Thank you Jon!
Greetings from the UK,
Anthony
Seeing the PT900 test station after all these years was surprising. I maintained and was lead over a system for years. Knew it inside and out.
Our system had 8 bays in an L shape with an interface panel in the middle.
It had a Norland digital Oscope which was a blast to use and very impressive for the day.
Thanks Jon for yet another interesting and at times entertaining video! 🙏
Saw the news today about the earthquake in Taiwan. I hope that you and your friends & family are all ok & safe.
I miss those times even if I didn’t live them. I was born in 1971 and started my electronics engineering grad in 1988. I studied and worked with most of the tech created in those early days of the Silicon Valley. What a time to be alive that must have been.
Always great content on this channel. I had heard of LSI but knew zilch. Thanks for creating this vid.
Holy sheet.... Fairchild was OP AF!!!! Every final boss in one company!!
The amount of history you pack into this is great! Thank you!
I remember my dad working there back in the day 😞
I remember a lot of semicustom ASIC manufacturers out there, including analogue component arrays. Whatever happened to Ferranti Interdesign?
Oh, man, this brings back memories. Well done.😊
5:53 Early breadboard were made out the woods!? that's crazy
The CEO of Avago sounded like the borg assimilating another species.
Thanks! I’m contributing as I have watched many episodes.
What a great era of PCs. I hit my stride about '98, when i was a HS freshman and think the 486 to P3/AMD Athlon time-frame was when the hoby peaked for me. I sort of lost interest when in 2010ish. At that point the gains weren't as impressive, unless u dropped serious coin on a graphics card and nowadays that's even more true.
Schlumberger visited my college and were really, really, really insistent that we pronounce it "schlumber-zhey". Weird group of people.
yeah, not that it matters but that is how it is pronounced
Perhaps the lingering memory from a Schlumberger "milk round" event in my time, apart from the pronunciation, was the fairly youthful representative telling a mature student that the company thought people over 35, if I remember correctly, were "past it" and not going to be considered for positions. That pretty much turned everyone in the room against them. Weird would have been a charitable word to use.
@@mitchtickets it's also how everyone who was at that lecture (...12 geologists) now pronounce hamburger
Where I am from, hard-drive RAID controllers and high-speed computer cards from LSI is commonplace. They run great, and frankly, someone have to make those chips.
"Wilf" is about as English/British a name as is possible to imagine. Maybe even more than Nigel.
There was a lot of LSI/Avago raid controllers in servers until recently, I know almost all of Dell raid controller stuff is/was actually rebranded LSI and they barely even bother to hide it. But I'm not sure what happened with the advent of nvme SSDs and whatnot because those controllers were too slow to handle them as far as I know/remember
A lot of ASICS back in the 70s were simple because you couldn't get very many transistors on a single die. They were mostly ICs with similar logic gates on them, so you could have one IC be 5 AND or OR gates, or XOR gates, and so on, and then a person would design a circuit board to use these gates, so there was a LOT of work that went into making the circuit boards and laying out the runs. So you CERTAINLY can't talk about matrix whatever back in the 70s when ICs were really just then starting to show up in numbers in the consumer world.
Circuit boards would have to be big to make a full fledged computer using those 70s ICs or you needed many circuit boards to build a mainframe computer.
I remember the Valid Scald eda system for use with LSI Logic and VLSI TECH.
I'm still using LSI SAS controller chips/cards today in a number of my home-lab servers. Interesting story! Thanks.
Imagine if Schlumberger didn't destroy Fairchild?
DIdn’t know that Jensen Huang worked at LSI Logic. Now, everything makes sense.
And his cousin, Lisa Su, worked at Texas Instruments, IBM & Freescale Semiconductor
Excellent; thank you!
Love your topics
I love your videos! But, as I followed the war in Ukraine, I have taken a interest in the history related to defense industries and, advanced missile systems.
It would be awesome if you could do a video specifically related to things like Lockheed Martin and silicon valley as well as the widespread use of FPGA & ASIC's in advanced missiles and weapon systems!
Are you a fellow Perun fan?
@@Grak70 Of Course! HELL YEAH!!
FYI I'm the real Sgt CONSCRIPTOVITCH!
🤣🤣🤣
@@justinhealey-htcohio3798 General Oligarkov salutes you!
There were some recent videos dedicated to the circuit boards from the Javelin anti-tank missile. It started with one French UA-cam channel famous for teardowns of aerospace hardware. The host had acquired a guidance section of a Javelin from a US militaria auction and was showing its internals.
Unfortunately most of the follow up videos on other channels which were spawned by this were just wild and mostly inaccurate speculation. There were no FPGAs in the Javelin. The missile was an old Texas Instruments project, and they simply used their own chips for image processing. At that time FPGAs just barely came into existence and were quite small, and not very useful for the purpose.
Another group of videos covered circuits from Russian Tornado-S guided missiles. Many of the boards there included semi-custom chips (hard-wired gate arrays). Also no FPGAs, (except for one lonely Altera in a commercial fiber optic gyroscope.)
Regarding applications of FPGAs in weapons, one can of course find general information about possible applications in the vendor's marketing materials. But beyond that it is not a subject that lends itself easily to a public lecture.
Is there away to find out who invented the solder paste and as such smd mounting or was there already a video about it?
It is interesting that surface mounting has been used for special applications at least as early as 1960s in the USA and at least since 1970s in the USSR. Lots and lots of flat ceramic packages in the early space program electronics, but it was probably assembled very differently back then.
Jensen huang worked there and his boss referred him to the vc and got his funding
Awesome !....cheers.
2:10 former CEO of AMD of course haha. Long retired.
Hi i hope you are fine! After that terrible earthquake 😢
Slum Burger. 😆
Hope you're doing okay with the earthquake
I like the concept of making half done ICs and having just the later manufacturing capabilities.
All designs today are standard cell and there is power supply from the other side of the IC now.
But today there is nothing a FPGA with a CPU inside could not do.
6:42 Circuits, not chips.
I work gor Schlumberger/SLB. Ironically they now have several custom ASICs :P
SLUM-BERG-JEY, not SLUM-BURGER. geez! (No worries friend, I worked there for a few years, everyone mispronounces it)
You should make a video about Broadcom and it's founders. Some of it's history is . . . interesting.
How so?
Loved this video
I understand this is off-topic, but I now know how to reach out to you. I recently heard about the earthquake. I sincerely hope that everyone is safe and that things can improve as much as possible
Don’t touch my chips
I won't ever forget LSI because of their RAID and HBA cards. They were absorbed into the monolithic Broadcom.
Atmel, LSI, Brocade, Emulex, Adaptec, and now VMware turned into a wall of flesh and metal. A billion electronic screams and several million groping hands, all calling out, "COME WITH US. BE WITH US. JOIN US."
I remember LSI Logic...!
Please do a video on Atomera's MST
Don't forget Tom Longo!
And here this lowly sysadmin thought they mainly sold RAID adapters back in the day...
Hope you and yours are all okay we just had a tiny 4.8 today
Prayers for the people of Taiwan, and hoping that you and yours are safe.
What surprises me in their History is them not having been the fathers of the FPGA...
Schlumberger, the -berger is pronounced like the -berge in Faberge, like the really fancy jewelry company. Both use French pronunciation, even though Schlumberger it looks like it might want to use a German pronunciation.
Its NOT pronounced SHLUMBURGER, it's SHLUMBER-JAY. just FYI
Capital letters aren't phonetics - you just sound a bit angry.
Jon's quirky pronunciations of things like "DRAM" are one of the things we've learned to love about him
It's 'Shlumburger' and its pronounced 'shlum burger' as in 'cheese burger' I should know I bought a £5000 voltmeter of them and I like cheeseburgers.
Yeah, it's pronounced a la Francaise...
Speaking of which ... Schlumberger itself might be a good topic for one of your videos? They thought that some of the ideas they used in the oil-fields would allow them to be successful in other sectors. High performance / high temp remote analog oilwell down-bore sensors meant they could get in the residential *remote* meter-reading business (Schlumberger Industries) or the *high-performance* IC business (Schlumberger Technologies).
My takeaway from this video is pay close attention if/when people from nvidia leave the company to start their own thing.
* where's your theme song
* how are you not on nebula??
Oh boy, wait til you hear about MOS.
there is only one destination: reality
micro mosaics is program once gate arrays...
Maybe there's time for trying going custom with silicon again, to create some sort of "PCB way" of silicon, were you can churn new designs in mere days. I have this idea of creating more advanced compilers capable of emitting masks as output, cutting out the entire process to the last step. (I'm a msc in computer science, and I'm going for a major electrical engineering)
Those guys are inspiring. I probably need less than $5M in angel capital, just send me a message plz.
they have this now.. numerous companies allow ASIC tapeout.. even google's in a partnership
9211-8i
they were the chosen ones to start the GPU & failed.
Wake up babe Asianometry just dropped a new video
Cringe...
@@JohnCompton1I was thinking the same. People are so strange.
what does wilf stand for?
It is pronounced "Slumber-jay"
Bongkrekiccc
Slum-burger lol
Wilf. Sporck. These names have to be made up.
Psst. PROTIP: You might want to lower your mic's sensitivity and/or use a directional mic... or even use a noise-filter on whatever audio recording software you're using... because you've got a lot of traffic noise in the background of this one. 😁
Pro tip: you might not want to listen to audio at such high levels. You could hurt your ears.
not using google translate for pronunciation strikes again. In almost every video.
Google gives "incorrect" pronunciation for Schlumberger in both English and French.
Should be called Moores Theory ..... NOT Moores Law. The guy actually went my school 40 years before me ofcourse.
Naw, it's a law. Like don't speed. Not gravity.
We'll take that under advisement.
ua-cam.com/video/iBWdaDu2nFM/v-deo.html
Schlumberger should be pronounced much in French way
Man what a time to be alive.
Uhm, "niche" is pronounced "neesh", not "nitch".
Some of these pronunciations continue to give me brain damage. I’m starting to roll with it though.
In the first world, the final warp drive signatures where solved, the universe was finally cracked. All that was left to do was put the great harvester engine into supersymetric warp lock to escape the cosmos and become dweeb overlords.
But the cosmos became harder than metal as the ship reached the edge of known reality. The solutions to death where the totality of finished, get to the chopper. One *_uge_* LEGO. The serpent tricked them all, "don't you know the first is the last."?
Time began to slow down by the half second, "you stop that, get back here", said a green eggs, it was a great eyherial rod so long it was bi orthogonal, as you looked back behind the ship, it looked like two rods, then 16, a kylidacope formed, *_L S I_*