Yes the lack of transparency on ARM architecture and the obfuscation of the RAM capacity with just the slot information is annoying. Not sure if it is deliberate to sell this at a premium to "mainframe" customers or whether its just their marketing folk not being on the ball. Fascinated by the fact Japan has a 2nm process node but I suspect that too is marketing. Perhaps the 2nm portion of the chip is done at TSMC and the rest of the package done in Japan or something like that. It's not like there's a plethora of 2nm Fabs around the planet to choose from.
It looks to me like it is an AI optimized HPC CPU and while HPC CPUs are actually quite fast, they are not particularly well suited for transformer based neural networks.
It's weird that they don't give out any of the interesting numbers like number of memory channels, how much SRAM they have (CPUs tend to have more of it) or what their expected performance is going to be.
No it is not at all Neoverse V2, which is obsolete and has much lower performance.
It is a custom Armv9.3-A design (same ISA like the Arm C1 CPUs from the flagship smartphones of 2026), but it has double-width execution units for SVE2 (i.e. 256-bit width, vs. 128-bit for the other Arm CPUs) and it has some ISA extensions for AI/ML, e.g. instructions for inference with FP8 (BF16 is already supported by the standard Arm ISA).
They actually do not in that article. I had to look elsewhere to find that out. Weird, one would expect that a big announcement like this would probably need to be vetted against licensing agreements with Arm that would probably want that fact to be prominently highlighted in press releases.
Yeah, searched the page for common architectures (x86,arm and Sparc since it was Fujitsu) since it'd seem silly to use a totally novel architecture but didn't find anything until noticing SVE2.
Thanks kinda how Japan works. Inventing something new is hard, but taking something that's a cool idea and making it actually usable is well within their wheelhouse.
I'm sure there's plenty of other stuff invented in Japan, but that's really beside the point.
The industry is now mature enough that nobody really wants the headache of a new architecture at this point in time, Risc-V got a pass for being a grassroots movement growing out off FPGA's since the IP situation was more or less clear or even predatory with all existing architectures, Power, Sparc,SuperH,68k,etc are buried for good reasons.
Dunno, I think Risc-V's simplicity gave it the initial popularity rush rather than being from a certain institution (the institution perhaps gave it visibility, but it does have own merits and no idea if other archs from the time had those?), don't remembere exactly but iirc it was possible to implement with a ridiculously small amount of FPGA gates. Conversly heard some rumblings that it had parts t hat were unsuitable for scaling up but it seems to have gained enough momentum for people to engineer past stuff like that.
PPC could have any number of merits, the 2 nail in the coffin imho is that the community/implementations seems to have held on to a big-endian tradition whilst the mass of users/developers firmly shifted to little-endian when Apple went to x86 and that IBM didn't push it properly open in the 2005-2008 timeframe when PS3/Xbox360 were still relevant (OpenPower seems to have been created in 2013, after the creation of Risc-V and PS4/XboxOne going to x86).
Core solo/duo was the last in the P3 heritage (but that heritage itself traced all the way back to the Pentium Pro initially) via the 2 generations of Pentium-M (even if Core2 had parts they were 64bit so more of a major upgrade).
Whoever convinced the execs at Intel to put Pentium-M into production as a "mobile" chip really saved them when NetBurst reached it's end of line, I owned a first-generation Pentium-M laptop and it was really superb (seeing the Macbook Air introduced was a big meh).
It was really a turning point though, I remember posting some compile time and people didn't believe me, but the Pentium-M chips were good at compiling (since compilers are kinda "branchy" when it comes to execution in a way P4 didn't "like") but also because I had maxed out the RAM on the machine (1.25gb when 256mb was the base for the machine and similar for many desktops) so much of the code and libraries could be retained in disk caches.
With portable performance like that, why do you need a desktop?
I had this idea that Fujitsu was one of those Japanese conglomerates that had operations in all kinds of markets, from electronics to heavy earth moving equipment to air conditioners to locomotives. But that doesn't appear to be the case so I'm wondering which company I'm confusing them with.
I am in tech (okay now I am really wondering what % of people on here are NOT in tech). I have spent years in Japan for work, and ironically right now as I type this, I am sitting in a seat on JL2 from Tokyo to SFO. I worked with all of the large Japanese companies over the years and I can say Fujitsu has always been the most impressive, and to you comment, most like golden age IBM (I worked there for a brief time also). They, like IBM are most likely to have brilliant tech, and then shoot themself in the head for non-tech reasons.
About 25 years back here in Australia they were advertising Fujitsu air-conditioners, they had the promise of some like "They will get the air down to 17c in all conditions."
About a month later there was a summer heat wave and I went to a friends house who had just bought one. It was wild walking in from a 45c day into, as advertised, a 17c house. Had to give it to them, they lived up to the claim.
That they come from the telecom/server space, it makes sense they had a good eye for aircon engineering.
Sun invented SPARC, Fujitsu just happened to be the last producer since they built mainframes on it. Reading it seems like they plan to stop producing Sparc processors and machines in a near timeframe so that makes this announcement less surprising.
For many years the Fujitsu SPARC implementations were much faster than the Sun SPARC implementations.
During the decade 1995-2005, until the 64-bit AMD Opteron servers with Linux offered an alternative that crushed any kind of SPARC CPU with Solaris by a much higher performance combined with a much lower price, the SPARC Sun or Fujitsu servers dominated the market for servers used to host the CAD/EDA design tools used in electronics engineering, for the design of integrated circuits or electronic equipment. This means that all the EDA software from vendors like Mentor, Cadence, Synopsis etc. was available only for Solaris. That kind of software could not be run on Windows, due to the 32-bit memory limit. During that time, whoever got Fujitsu servers instead of the sluggish Sun servers, was very lucky.
It was an issue of fundamentally differing directions.
In the mid-2000's, Sun decided to take SPARC towards designs with many small SMT cores. In the era of single-core processors, the UltraSPARC T1 had 8 cores x 4 threads per core. This was at the same time Intel released the Pentium 4 with hyper-threading, so it was an industry trend.
This of course works great for very specific applications, particularly considering efficiency, but is awful for others. Scientific computation was especially bad because the T1 had only one FPU for 8 cores.
Fujitsu's SPARC64 didn't go in this direction, and stayed with a conventional design (2 way SMT at most). Sun realized this and started to also sell the Fujitsu SPARC64 for customers who couldn't use the thread level parallelism, an arrangement that lasted until the end.
The idea of lots of slow cores is still a thing today: Intel's Sierra Forest Xeon is 144 E-cores.
1 FPU for 8 cores is wild, but then again they were selling webservers.
Still, iirc FPU's were silicon heavy back in those days and it'd be interesting to know how far ahead the foundries Sun and Fujitsu were, maybe it was simply a factor of being too far behind in the foundry race that left Sun with few options.
Intel's E-Cores still are functionally complete for most parts though (excl Avx512?)? Todays limits seems to be memory bandwidth and power and I guess many of the 144 core customers are in it for virtualization and servers?
My understanding is the lots of small cores are for applications where you have lots of simple, independent tasks that tend to be IO-bound: web and other front-end servers, databases, transaction processing. SMT is optimal here because you can execute other threads while waiting for IO.
I haven't been closely following it, but the Intel E-cores are definitely less wide and have less execution units. The FPU/Vector hardware is 128-bits wide instead of 256. If you ask it to do 256-bit SIMD, the front-end breaks it into 2 operations.
Not mentioning the architecture is rather strange. Being Fujitsu I'd expect SPARC, but apparently not. Maybe we're reaching a point where some circles either don't care or just defaults to ARM?
Obviously they can just ignore the license in the future and continue development out of a local branch, but it's also a bit disingenuous to speak of "sovereign infrastructure" and then use licensed processor design.
It think we are now heading towards the final long term ISA setup. x86 and ARM at the top, ARM in the middle and RISC-V for smaller embedded machines that don't need the ARM toolchain.
x86 servers are like that. PC/AT memory map have the range for VRAM, but RS232C is optional, so you can't just have the RS232C and not VGA on an x86 as its debug/recovery interface.
Platform designers can make up their own memory maps, so non-x86 ones can just have UART I/O buffer and hardware interface wired up in whatever address you want and forgo VGA.
I have a 32bit SPARC machine on a shelf I got from a junk hardware shop, it has a DB25 for console. It's ok.
Fujitsu has a long history of designing their own micro but using a standard ISA for HPC. Used to be SPARC, now it is ARM. IIRC Fujitsu were the main architects behind the SVE ARM extension.
Fun fact, Fujitsu used to run their own fabs in Japan until they were sold to UMC and now they're another UMC fab although they are stuck at like the 20nm node. So this is likely made at JASM.
Their CPUs are faster at inference than many GPUs. These are server CPUs with 144 cores and wide memory interface.
Fujitsu has added some ISA extensions for AI inference to the standard Arm ISA, e.g. instructions for inference with FP8.
Even their previous CPU generation was for some time the champion in energy efficiency, with better performance per watt than the NVIDIA GPUs, until a newer generation of NVIDIA GPUs has leapfrogged them.
There are good chances that this new Fujitsu CPU might be again for some time the CPU with the best energy efficiency, but it remains to be seen how it compares with the recent GPUs.
The two made-in-Japan, premium grade, Fujitsu laptops I had to work with were by far the worse pieces of electronic I have ever used. Both exhibited identical defects: fans running continuously at maximum speed, and the batteries would completely die within hours of the devices being powered off. Windows or Linux. And don't talk about that tiny unresponsive trackpad to me. Never again.
Reaching further back, the Sylistic line was mostly quite good/durable, and I still use my Stylstic ST-4110 when I need to use an old scanner or control a CNC machine on my back deck when cutting tropical hardwoods.
Really miss the transflective display and wish that there were newer devices with such technology
What year? The ones I had were the best. Very expensive, but service at the office & at home, very sturdy, very long battery life. That was somewhere early 2000s.
I just have Fujitsu Lifebooks at Home since 2007. I think they are good, came with a bloat free Windows. I bought even one for my Mom over 10 years ago. I just changed once the HDD with a SSD. Now all my Lifebook run with Linux just flawless.
Their software is so poor that it caused probably the biggest miscarriage of justice in history. Well that and Fujitsu staff colluding with post office executives to put innocent people in prison rather than admit there was a problem
The 'Fujitsu' that made the shitty software was a British consulting company that was acquired by Fujitsu [1] to farm public contracts. This is made by the Japanese parent company.
It doesn't seem that way to me. I'm inclined to think that the British consulting company created the problem, seeing as I've seen great things come from Fujitsu, and only just learned of this company. If some execs at Fujitsu were willfully involved, it hardly raises it to my standard of boycotting a megacorp with all the other corporate wrongs I hear about.
In a very specific legal sense, it isn't, we are talking about two entirely separate legal entities (and ultimately groups of people) which have a relation of ownership, not directorship.
I think what's most notable about this for me is the reduced environmental impact. The specifications list liquid-cooling as optional and it highlights the CPU as the main processor instead of a power hungry GPU. My takeaway is that the next generation of AI HPC's will focus more heavily on sustainable operations.
One problem we’re going to have with AI hardware is coming up with a standard set of specifications that are comparable. I don’t really care about the CPU GHz and the memory bandwidth, at least not directly. What I really want to know is how many tokens per second this will deliver, but that also depends on the model. We need a standard metric for that. Perhaps we agree on a specific open weight model (e.g. GLM 5.3 Flash or Qwen vWhatever) and then measure TPS on the hardware of interest.
I always heard of GB/s as most important number ... AI told me their 8800 MT/s on 8 Byte, 12 DDR5-Channels means 845 GB/s. A Nvidia RTX 4090 has 1008 GB/s. Nvidia B200 has 8000 GB/s.
Is this the right way of looking at it?
I fail to see how they will take a significant market share or even break even on this venture.
It's an overcrowded market. Far better would be to focus on semiconductor supply chain, which Japan already supplies some elements, to sell to fabs.
Because the bottleneck is the fabs. If a new 2 nm fab came online today, it would immediately sell all its capacity to 2030 no problem, without Fujitsu trying this gambit.
It's not 'sovereign', the architecture is not designed in Japan, the silicon is not fabbed in Japan. This whole thing is sideways.
Previous generation of this was a monstrosity based on a custom SPARC with ARM front end grafted on top, supporting the version of SPARC extensions they created backported to ARM as a new extension.
So there's a good chance that it's still an optimized half-SPARC inside.
Every developed country is going to push for their own homemade chips. Japan used to make Sparc CPU's back in the day. Probably many others I'm not aware of.
Every developed country except Australia, perhaps. This government is atrocious. The record for highest number of bureaucrats per capita in the world is owned by Australia. All they know how to do is tax people. All the wealthy people have left, or are leaving.
> Oh god you're one of those. People who complain about taxes have historically been on the wrong side of arguments.
Oh you're one of those. The arrogance is astounding. You never been a country which overly taxes it's citizens and squanders the gains on bureaucracy and monetary black holes? Grow up, your ideological blind spots are making you look like a fool.
Perfectly describes Australia, indeed. Note how I'm not saying that taxes should be higher and should be "squandered on bureaucracy and monetary black holes", so I'm not sure if I really am one of those.
The world is not black and white, contrary to popular belief.
I've been calling for high-multicore CPUs (at least 100 cores) with local memories for a quarter century now. No winners so far.
A Pentium 4 hit 3.8 GHz longer than that ago in 2004, so that's not special. And the estimated price will be $7,000-10,000, which isn't special either since that's about 10x more than it should be.
Hot take: GPUs disrupted the CPU industry to such a degree that CPUs never recovered, and like the k-shaped economy, the current status quo only serves a small fraction of customers. We can and should do better, but sadly we won't. Still, it's good that Fujitsu did this, for the competition if nothing else.
Back in 2008 Fujitsu has one of the best performing 10Gbps Switches. We were building 40 Gbps packet sniffers at Google (4x 10 Gbps NICs) and needed switches that could do things like mirror traffic across ports at line rate. Fujitsu was way ahead of the pack. I always wondered what held them back from building a meaningful networking business in the US.
This is the first time ever I am hearing of Fujitsu being involved in any bribing/lobbying controversy in the US (whether as a victim or an offender), and I doubt I am the only user in this thread in this position.
Unless it is a very well-known controversy, I feel like there is a need for extra context here. I am not even asking for anything of the "proof beyond any reasonable doubt" nature, I simply want to understand what you are trying to reference.
> I always wondered what held them back from building a meaningful networking business in the US.
It was just timing. Their networking gear was built on some very impressive ASICs which became commodity chips a few years later. Arista et al ate their lunch using other vendors’ IP.
We have alliances with Japan and S Korea which predispose us to believing we might be able to get a clean supply chain from them if a new Cold War starts. Out other options don’t even pass a giggle test and that’s the state of the world until some manufacturing is brought back onshore for strategic reasons.
The same way Congress has been subsidizing Boeing and its competitors since WW2 to make sure we can build things that fly if another war ever starts. Or with drones, that we can build ones that will be flown BY US instead of defecting the moment they get in range of the other side.
> The same way Congress has been subsidizing Boeing and its competitors
Funny you should mention that. Boeing deliberately buys 35% of a 787's airframe from Japan (along with other equipment for the 787 and other models).
That's a deliberate play: we buy a lot of parts from Japanese companies, your airlines buy mostly Boeing airplanes. Even today despite some diversification Boeing has over 70% marketshare in Japan.
If you ever wondered why Airbus setup factories in the USA that's a similar deal. The US Gov and Mil want to ensure factories and workforce on home soil should the need arise. In some unrecorded meetings somewhere strong hints were no doubt dropped that it would be in Airbus' best interest to play ball. They obliged.
That said the USA is the largest supplier of parts for Airbus airliners so there's also a benefit to having aerospace manufacturing and R&D here.
Boeing proposes sth called the “Boeing Honeywell Uninterruptible Autopilot” to add drone-like remote control capabilities to its airliners (for security reasons, obviously). I'm not aware of a similar offering for Airbus but I have no doubt such a capability has existed in Airbus as well for perhaps two decades. I expected development of such a sensitive system for Airbus to be based in France, but your comment about “strong hints” and “play ball” has me thinking…
>We were building 40 Gbps packet sniffers at Google (4x 10 Gbps NICs) and needed switches that could do things like mirror traffic across ports at line rate
I'm sure there were good reaasons for this, but man this just sounds bad. Like that's the spec I think NSA must give to their contractors outfitting 611 Folsom Street's Room 641A
There are companies that mirror their traffic to storage for a few days to be able to look back and troubleshoot issues. ALL their traffic, proactively.
For US federal government, "Office of Management and Budget" says to store 72 hours of packet capture, since 2021 (partially in response to SolarWinds attack). (M-21-31)
It does sound bad the way OP described it, but packet or HTTP mirroring is a standard feature of A/B testing and blue-green deployments of a very critical code. Mirror traffic between version 1 and 2, then compare response body, response time and return response A to the end users.
I used to operate several dozen 10G/25G/40G/100G taps which fed a series of tools in the DC at a regional healthcare back in the day.
80% of the relevant outcome was typically to get a true-to-the-wire sniffer capture (switch mirrors won't always mirror 100% of packets for various reasons) for troubleshooting performance of whatever the complaint of the day from the server team was.
19% was for feeding a security monitoring systems which looked for abnormal flow patterns to let us know a server was compromised.
1% was for the call recording system compliance requirement for the emergency department.
Port mirroring is the network engineer's equivalent of using breakpoints in a debugger for a programmer. It allows you to inspect what happens on the wire to get a clue for what's wrong.
Maybe I'm being naive, but as someone who's spent a lot of time capturing and reading PCAP files, 40GB/s worth of mirror is gonzo nuts! You may already be looking for a needle in a haystack, but this time the haystack is 1/60th of the entire bandwidth of the internet in the year 2000!
These tools are very common in non-nefarious ways for troubleshooting networking, and while vendors set up to serve this space solely as troubleshooting tools, I've also built my own that can sit on a network link and monitor for problems.
In my case it was for mobile wireless signaling traffic (all the coordination for creating a mobile internet connection, handing the connection off between towers, etc), and I'd credit it as one of the reasons you're mobile internet connection is so stable. When LTE first came out, myself and many others solved all sorts of bugs in the equipment and protocols by using or building these sorts of tools.
AFAIK those sold relatively well, considering the TCO of both sides (the NICs were very expensive and there were like a dozen companies vying for position once upon a time). They were also interesting because they were cut-through. CX4 was noticeably lower latency than SFP+. So a very fast, but feature limited architecture which is probably why other players like Cisco and then Arista (aka Broadcom) took over.
The XG series switches? Force10 sold them until Dell bought Force10 and continued to sell them (although the firmware changed a lot along the way). They were, AFAIK, pretty popular in the US under the Dell brand.
My home network is XG2000 10G core and a few XG0224 1G/10G dist. Very nice switches.
Ubiquity proved you can bust into the market from the bottom end if you want. Nothing was stopping them from doing the same but like many companies they prefer to focus on how much money they can siphon from the Fortune 1000 so I can't say I feel sorry for them.
I'm not sure why that's relevent, particularly since Ubiquity's approach wouldn't have remotely competed with Fujitsus custom ASICs if they were actual contemporaries, but you clearly have an agenda you're in love with, so...ok, sure.
No I was just saying Fujitsu never really made a play for any US marketshare. They had ways to go about it but they stuck to what they knew: high-touch sales, locked down support, focus on the Fortune 1000.
Fujitsu sold to Force10/Dell and did quite well in the US market. Which is completely tangential anything to do with Ubiquity, which never competed in the Fujitsu space, which even you seem to understand (maybe? vaguely?) is a different market. Which they were successful in.
Conglomerates in general, and especially Japanese ones, are heterogeneous enough that those are basically two different companies. The British "Fujitsu" was an acquisition that I'm certain they now deeply, deeply regret.
Press releases are sadly like that. Fujitsu's main page has it in the first line:
> FUJITSU-MONAKA is a Next-Gen Arm-based processor, set for release in 2027, designed to address the challenges of next-generation data centers with its unmatched performance and power efficiency.
For anyone not familiar with Fujitsu's CPUs, worth noting their overall long history with HPC, and more recently, ARM. FugakuNEXT is "NEXT" because the current Fugaku (#1 on TOP500 for a time, and still in the top 10) also used an ARM CPU from Fujitsu.
Man this looks amazing, more competition is always welcome! The issue for Ai inference is always firmware/software support. Up until recently, that rocm hardware became usable, let alone optimized. Nvidia had a big start, now everyone is trying to catch up
Could someone knowledgeable please explain why an AI-focused CPU is superior to a GPU-based solution with an ARM CPU driving the high-level operations, please, especially when the RAM isn't on-chip?
I know that for GPU's the model weights have to be transferred over the bus initially, but that only has to occur once for inference use cases, so is the Fujitsu system more about training scenarios? Or is the focus more about efficiency, as these are ARM-based cores with AI additions?
Honestly I think this is an HPC CPU that has been AI-washed. Then you might ask why not use GPUs for HPC and I think the simple answer is that Fujitsu just doesn't want to take on the effort of building a GPU.
Yeah, this very much gives the vibe that the original plan was to design a CPU, then suddenly AI AI AI happened, and they squeezed some matmul units into the design at the last minute in order to please the higher ups.
Fujitsu MONAKA Server will be
made broadly available from
November 2026 to data center
operators, enterprises, and
the academic and HPC sectors
in Japan and Europe, as well
as to the defense sector,
contributing to national
security.
So, not in America. A sign of the geopolitical times? If so, a very unusual one to come out of Japan.
206 comments
[ 24.9 ms ] story [ 431 ms ] threadIt's weird that they don't give out any of the interesting numbers like number of memory channels, how much SRAM they have (CPUs tend to have more of it) or what their expected performance is going to be.
See e.g.:
https://chipsandcheese.com/p/hot-chips-2026-fujitsus-monaka-...
They have provided far more details than companies like Apple or Qualcomm.
Edit: it’s funny because this article from 2021 mentioned Fujisu precisely https://siliconangle.com/2021/03/30/arm-unveils-armv9-archit...
It is a custom Armv9.3-A design (same ISA like the Arm C1 CPUs from the flagship smartphones of 2026), but it has double-width execution units for SVE2 (i.e. 256-bit width, vs. 128-bit for the other Arm CPUs) and it has some ISA extensions for AI/ML, e.g. instructions for inference with FP8 (BF16 is already supported by the standard Arm ISA).
Seems they have decided they don’t want to create their own architecture, but take the popular architecture and build on it.
See: - Blue LEDs - Quartz Watches - Lithium-ion Batteries - Bidets
The industry is now mature enough that nobody really wants the headache of a new architecture at this point in time, Risc-V got a pass for being a grassroots movement growing out off FPGA's since the IP situation was more or less clear or even predatory with all existing architectures, Power, Sparc,SuperH,68k,etc are buried for good reasons.
POWER/PowerPC is what happens when you take a brilliant design to execute it poorly. x86 is when you take a poor design and execute is brilliantly.
Whoever convinced the execs at Intel to put Pentium-M into production as a "mobile" chip really saved them when NetBurst reached it's end of line, I owned a first-generation Pentium-M laptop and it was really superb (seeing the Macbook Air introduced was a big meh).
It was really a turning point though, I remember posting some compile time and people didn't believe me, but the Pentium-M chips were good at compiling (since compilers are kinda "branchy" when it comes to execution in a way P4 didn't "like") but also because I had maxed out the RAM on the machine (1.25gb when 256mb was the base for the machine and similar for many desktops) so much of the code and libraries could be retained in disk caches.
With portable performance like that, why do you need a desktop?
About a month later there was a summer heat wave and I went to a friends house who had just bought one. It was wild walking in from a 45c day into, as advertised, a 17c house. Had to give it to them, they lived up to the claim.
That they come from the telecom/server space, it makes sense they had a good eye for aircon engineering.
During the decade 1995-2005, until the 64-bit AMD Opteron servers with Linux offered an alternative that crushed any kind of SPARC CPU with Solaris by a much higher performance combined with a much lower price, the SPARC Sun or Fujitsu servers dominated the market for servers used to host the CAD/EDA design tools used in electronics engineering, for the design of integrated circuits or electronic equipment. This means that all the EDA software from vendors like Mentor, Cadence, Synopsis etc. was available only for Solaris. That kind of software could not be run on Windows, due to the 32-bit memory limit. During that time, whoever got Fujitsu servers instead of the sluggish Sun servers, was very lucky.
In the mid-2000's, Sun decided to take SPARC towards designs with many small SMT cores. In the era of single-core processors, the UltraSPARC T1 had 8 cores x 4 threads per core. This was at the same time Intel released the Pentium 4 with hyper-threading, so it was an industry trend.
This of course works great for very specific applications, particularly considering efficiency, but is awful for others. Scientific computation was especially bad because the T1 had only one FPU for 8 cores.
Fujitsu's SPARC64 didn't go in this direction, and stayed with a conventional design (2 way SMT at most). Sun realized this and started to also sell the Fujitsu SPARC64 for customers who couldn't use the thread level parallelism, an arrangement that lasted until the end.
The idea of lots of slow cores is still a thing today: Intel's Sierra Forest Xeon is 144 E-cores.
Still, iirc FPU's were silicon heavy back in those days and it'd be interesting to know how far ahead the foundries Sun and Fujitsu were, maybe it was simply a factor of being too far behind in the foundry race that left Sun with few options.
Intel's E-Cores still are functionally complete for most parts though (excl Avx512?)? Todays limits seems to be memory bandwidth and power and I guess many of the 144 core customers are in it for virtualization and servers?
I haven't been closely following it, but the Intel E-cores are definitely less wide and have less execution units. The FPU/Vector hardware is 128-bits wide instead of 256. If you ask it to do 256-bit SIMD, the front-end breaks it into 2 operations.
Obviously they can just ignore the license in the future and continue development out of a local branch, but it's also a bit disingenuous to speak of "sovereign infrastructure" and then use licensed processor design.
What are the GPU capabilities?
Platform designers can make up their own memory maps, so non-x86 ones can just have UART I/O buffer and hardware interface wired up in whatever address you want and forgo VGA.
I have a 32bit SPARC machine on a shelf I got from a junk hardware shop, it has a DB25 for console. It's ok.
(I think Japanese sovereignty is the main point of the article.)
Also odd how big an emphasis they put on AI inference when they don't build the GPU?
Fujitsu has added some ISA extensions for AI inference to the standard Arm ISA, e.g. instructions for inference with FP8.
Even their previous CPU generation was for some time the champion in energy efficiency, with better performance per watt than the NVIDIA GPUs, until a newer generation of NVIDIA GPUs has leapfrogged them.
There are good chances that this new Fujitsu CPU might be again for some time the CPU with the best energy efficiency, but it remains to be seen how it compares with the recent GPUs.
so its as sovereign as you can get
I do find it odd that the 2U model has less storage capacity than the 1U on their chart. That doesn't make much sense to me.
Really miss the transflective display and wish that there were newer devices with such technology
Their software is so poor that it caused probably the biggest miscarriage of justice in history. Well that and Fujitsu staff colluding with post office executives to put innocent people in prison rather than admit there was a problem
[1] https://en.wikipedia.org/wiki/International_Computers_Limite...
https://en.wikipedia.org/wiki/British_Post_Office_scandal
There is a fairly direct link between the two numbers. You can predict the latter from former reasonably well
like:
- home.kpmg - global.honda
and probably more I forgot about
but as you can see, they're pretty terrible for replacing .com domains
It's an overcrowded market. Far better would be to focus on semiconductor supply chain, which Japan already supplies some elements, to sell to fabs.
Because the bottleneck is the fabs. If a new 2 nm fab came online today, it would immediately sell all its capacity to 2030 no problem, without Fujitsu trying this gambit.
It's not 'sovereign', the architecture is not designed in Japan, the silicon is not fabbed in Japan. This whole thing is sideways.
Doesn't matter. You have to start somewhere and this is a good start.
So there's a good chance that it's still an optimized half-SPARC inside.
The text says "next-generation CPU, FUJITSU-MONAKA, designed and developed in Japan".
You have been convinced by fake news, please recheck this claim, its false.
> wealthy people
*rich people
> All they know how to do is tax people
Oh god you're one of those. People who complain about taxes have historically been on the wrong side of arguments.
Oh you're one of those. The arrogance is astounding. You never been a country which overly taxes it's citizens and squanders the gains on bureaucracy and monetary black holes? Grow up, your ideological blind spots are making you look like a fool.
The world is not black and white, contrary to popular belief.
A Pentium 4 hit 3.8 GHz longer than that ago in 2004, so that's not special. And the estimated price will be $7,000-10,000, which isn't special either since that's about 10x more than it should be.
Hot take: GPUs disrupted the CPU industry to such a degree that CPUs never recovered, and like the k-shaped economy, the current status quo only serves a small fraction of customers. We can and should do better, but sadly we won't. Still, it's good that Fujitsu did this, for the competition if nothing else.
This is the first time ever I am hearing of Fujitsu being involved in any bribing/lobbying controversy in the US (whether as a victim or an offender), and I doubt I am the only user in this thread in this position.
Unless it is a very well-known controversy, I feel like there is a need for extra context here. I am not even asking for anything of the "proof beyond any reasonable doubt" nature, I simply want to understand what you are trying to reference.
Its a genuinely terrible place to do business, and its full of genuinely terrible people to do business with.
Oh and Trump made it a bit worse with Tariffs I guess.
Hope this helps.
It was just timing. Their networking gear was built on some very impressive ASICs which became commodity chips a few years later. Arista et al ate their lunch using other vendors’ IP.
The same way Congress has been subsidizing Boeing and its competitors since WW2 to make sure we can build things that fly if another war ever starts. Or with drones, that we can build ones that will be flown BY US instead of defecting the moment they get in range of the other side.
Funny you should mention that. Boeing deliberately buys 35% of a 787's airframe from Japan (along with other equipment for the 787 and other models).
That's a deliberate play: we buy a lot of parts from Japanese companies, your airlines buy mostly Boeing airplanes. Even today despite some diversification Boeing has over 70% marketshare in Japan.
If you ever wondered why Airbus setup factories in the USA that's a similar deal. The US Gov and Mil want to ensure factories and workforce on home soil should the need arise. In some unrecorded meetings somewhere strong hints were no doubt dropped that it would be in Airbus' best interest to play ball. They obliged.
That said the USA is the largest supplier of parts for Airbus airliners so there's also a benefit to having aerospace manufacturing and R&D here.
I'm sure there were good reaasons for this, but man this just sounds bad. Like that's the spec I think NSA must give to their contractors outfitting 611 Folsom Street's Room 641A
https://www.cisa.gov/sites/default/files/2023-02/TLP%20CLEAR...
80% of the relevant outcome was typically to get a true-to-the-wire sniffer capture (switch mirrors won't always mirror 100% of packets for various reasons) for troubleshooting performance of whatever the complaint of the day from the server team was.
19% was for feeding a security monitoring systems which looked for abnormal flow patterns to let us know a server was compromised.
1% was for the call recording system compliance requirement for the emergency department.
0% was because I was a cool superspy.
When hitting a breakpoint in your debugger you also don't print the whole 32 GB of RAM. You only look at a few locations.
In my case it was for mobile wireless signaling traffic (all the coordination for creating a mobile internet connection, handing the connection off between towers, etc), and I'd credit it as one of the reasons you're mobile internet connection is so stable. When LTE first came out, myself and many others solved all sorts of bugs in the equipment and protocols by using or building these sorts of tools.
My home network is XG2000 10G core and a few XG0224 1G/10G dist. Very nice switches.
That's too bad.
Seriously...do you have a point?
https://en.wikipedia.org/wiki/British_Post_Office_scandal
(Which in no way excuses their utter evil.)
>Combined with SVE2 vector operations and software optimization,
It’s ARMv9.
> FUJITSU-MONAKA is a Next-Gen Arm-based processor, set for release in 2027, designed to address the challenges of next-generation data centers with its unmatched performance and power efficiency.
https://global.fujitsu/en-global/technology/research/fujitsu...
It is no secret it is Arm; MONAKA was announced in late 2024.
https://www.techpowerup.com/329761/fujitsu-previews-monaka-1...
Nowadays if you have the source code the ISA is practically irrelevant.
wccftech summary: https://wccftech.com/fujitsus-monaka-chip-3d-stacks-2nm-cpu-...
2026: https://global.fujitsu/-/media/Project/Fujitsu/Fujitsu-HQ/te...
2023: https://global.fujitsu/-/media/Project/Fujitsu/Fujitsu-HQ/te...
FugakuNEXT, the supercomputer it will be used in, also some details about the next-gen Monaka-X: https://global.fujitsu/-/media/Project/Fujitsu/Fujitsu-HQ/te...
For a single CPU: 844.8 GB/s memory bandwidth (12 channels of DDR5 RDIMM, 8800MT/s). 4.3-6 TFLOPS.
Basically comparable to a modern GPU, so good for HPC & AI workloads. (though only 2 CPUs per node, with GPUs 4-8 is more common)
I know that for GPU's the model weights have to be transferred over the bus initially, but that only has to occur once for inference use cases, so is the Fujitsu system more about training scenarios? Or is the focus more about efficiency, as these are ARM-based cores with AI additions?