Submitted by “A Chinese company called YellowSheepRiver wants to make affordable budget computing a reality with its new $150 Linux Municator, a highly compact and innovative PC built with inexpensive Chinese hardware components. Although MIT’s much touted $100 Linux laptop has yet to transcend its status as vaporware, YellowSheepRiver already has a working product which could potentially be available for purchase within the next three months.” It runs on a custom 64 bit processor, derived from the MIPS processor, and as a result won’t run Windows.
I WANT ONE!!!!
I’ve been looking for a low-cost SBC type system to run Linux / embedded systems for a resonable price (less than $200 USD) and haven’t been able to find much in that price range that has what I need: 40pin ATA & RJ45 ethernet.
Just the one, dear? Smells like compile farm fodder to me.
This is a quite powerful machine. Governments would be more interested in this offer than MIT’s one. The $100 laptop encourages imperialism and gives the third world an offer that will never make it as advanced as the empires’ IT knowledge. The $150 can be used for many things, like any other desktop, considered that there will be enough packages built against its architecture.
@youcha
The $100 laptop encourages imperialism
Ah come on. Grow up.
Try to open your mind, think deeper. It is sure that you will react this way, because it is not something you learn at school. You just can’t assimilate it.
I agree that this is an offer that many governments should consider taking up. But…
“Empires”? (seriously, who?)
And for those who have barely ever (if at all) used a computer before, some experience is vastly better than absolutely none at all.
Leave it to the Chinese to do this…
I’d happily buy one (although it doesn’t seem to be intended for international markets), heck – it even has an S-video output! Many video cards with an S-Vid output cost almost this much! One does wonder what the estimated lifespan of such a cheap machine could possibly be (cheapo hard drive comes to mind).
Unfortunately, the price is this low mostly because their workers get treated (or at least paid) like slaves. I hope nobody finds a fingertip in their pc!
As most computer (or computer parts) are build in china, taiwan…, I do suspect that most computer (or computer parts) are build with low paid and treated as slaves workers.
– external HD seems odd…is the OS on firmware? Could this work as a diskless “terminal”? Why external? Seems this would complicate manufacturing and packaging logistics (and increase costs).
– does it come with a keyboard/mouse or without (a la Mac mini)…I assume to get to this price target, you are squeezing every possible nickle out of it.
– optional optical drive is fine. Most (budget-minded) users probably don’t really need it. What will the cost of this option be?
– Comes with a “Support CD (Driver, Utility, Manual)” but CD drive is optional. Funny joke.
– If this becomes available in the U.S. shipping, marketing and distribution $ will likely increase the cost. Suddenly something like Mac mini won’t seems so bad in comparison (faster CPU, tri-boot: OS X, Linux, Windows XP, larger INTERNAL HD, combo drive standard, etc.)
Edited 2006-04-26 16:05
– external HD seems odd
For me that’s actually better. I got a 300 GB disk I want to carry around, to hook it up to a laptop, my home computer or a friend’s computer (it’s cheaper than buying a 300 GB disk for your laptop). Would be annoying if I had to install an OS for an oddball processor architecture on it however. They mention ‘2/4Mbit Flash ROM’, and 512 kb would be a bit little for all the stuff you’re supposed to be getting.
– does it come with a keyboard/mouse or without
The product page doesn’t mention it being shipped, so one can assume without.
– optional optical drive is fine. Most (budget-minded) users probably don’t really need it. What will the cost of this option be?
As someone said, not available outside of China, so moot question
ONE mac mini costs like FOUR of these Chinese machines ($599). It’s an unfair comparison.
If you were to bring this thing closer to the Mac mini (internal and larger HD, more memory, internal combo drive, tri-bootable, available in the U.S. through various resellers)…I imagine the differential wouldn’t be nearly so much. That was my point.
I never heard of those CPUs that they use, how good are they ? I read an article about it but there is not enough info on Internet about it.
MIPS processors are among the top processors when it comes to power/watt. These ones do not use as low power as ARM processors or MIP designed for embedded, but they come with floatingpoint capabilities (ARM not), and can be pretty powerful. These ones should use about 20 Watt, half what Merom uses I think..
They are work at around 600MHZ, and are their second generation of CPU, the only problem for the international market is: they use patented technologies, without paying for the licence (and we’re talking about hardware, which costs a lot to develop). Which is directly the reason why it is so cheap. They didn’t have to do any research and got a good processor in return, the fact that they use linux makes it cheaper too. No wonder they have better hardware than the $100 dollar laptop, but well, the company’s not able to distribute it outside china, because china doesn’t have any patent laws, so all Americans, Europeans and Japanese people can forget getting one, the rest I don’t know.
Well, don’t count Europe out. We do have hardware patents to some extent, but nowhere near what is seen in USA.
This lamenting about Patents is the usual FUD which is based on some out of context information from a Microprocessor-Report article. Before spreading fear, uncertainity and doubt like this again, you should read this, especially the last paragraph from someone who should really know about this:
http://jonahprobell.com/lexra.html
Then this from one of the “plagiators” himself:
http://en-1.ce.cn/National/sci&edu/200507/30/t20050730_4308700.shtm…
And this which is the root-cause of all the patent-discussion:
http://www.in-stat.com/press/05/MDR_7_25_05.asp
Oh, not to forget, said patent which soon will be invalid:
http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO1&Sect2=HITOFF&d…
Some good links there.
Lexra was just down the road from here, I never really thought they would be forced to cave in.
From the patent link, I am pretty sure I have violated this patent in at least a dozen chips, although it wasn’t a cpu though, and the idea is just plain common sense and has been known since the wheel. In a nutshell its a mux dmux parallel to serial process idea with cpu words on top.
I really hate these types of patents because they force one to speak stupidly about what one is doing and to beat around the bush, engineers prefer to be blunt and upfront. The Lexra & BLX approach to doing this in SW always stinks in performance and lawyers can twist that around.
The only thing BLX can and must do is too fill up a patent portfolio of its own (US & EU) as quickly as possible, fight stupid with stupid or clever with clever. Companies with large patent portfolios are much less attractive to sue. Then unfortunately they have to hire western patent lawyers, a hard lesson for Chinese companies to learn if they want to fight with western dragons. So that means engineering jobs go to Asia, lawyer jobs stay in the US, yuck twice.
The chips do not contain the patented MIPS instructions, and should not infringe MIPS patents (in theory) if they were released abroad. However, I don’t think the Chinese want to bother to test this theory, since the chip really is intended for the internal market.
That said, the ISA is not the architecture! Just because the MIPS ISA was already designed doesn’t mean that they didn’t have to do any research. Designing the ISA is just a fraction of the job of designing a CPU.
As can be seen here,
http://en.wikinews.org/wiki/Yellow_Sheep_River_develops_%E2~*~@…
and here,
http://en.wikipedia.org/wiki/Godson_II
and here,
http://en.wikipedia.org/wiki/Longmeng
There seemed to be much disinformation about this and assorted Projects, f.e. regarding an ominous “Laptop” with similar specifications. There is no Laptop, there is
only this.
I can’t say I am that keen on either the MIT $100 or this $150 machine for the following reasons.
Compatibility of the MIT machine allows the HW to run any old OS ever written for x86 but then again the PC design is a very old dog that really needs to be gutted (IRQs,DMA garbage). I can see why some would see it as a “colonial” thing though, giving out bread crumbs to the poor world but then what can you expect for so little $. I think MIT should release it without the SW and let the locals put their own distro on it which would help the SW development take off locally.
This MIPs design is much nicer, much more like a workstation with all the PC parts needed in SFF and external 45W PSU (no ATX), but the cpu really means Linux only. Only by building this in China can the cost be driven to the floor but I would rather have the HW design done in the west in concert with the OS development, easier to market a design specified by the same market it is being sold in.
Theres really no reason why any group of EEs couldn’t put something similar together with off the shelf parts and assembled in China for similar cost.
OT but I wonder what Bill got out of his hosting with the Chinese leader?
my 2c
“I think MIT should release it without the SW and let the locals put their own distro on it which would help the SW development take off locally.”
Well then you would have the locals putting a pirated copy of Windows on it.
Well whats to stop them from taking off what ever Linux MIT puts on it anyway and putting on the same pirated Windows.
The point is that the locals should define their own distro rather than some head in MIT. Anyway I would have thought Windows would perform pretty badly on the Geode and the small flash store, what version of Windows do you think will work so well?
Well whats to stop them from taking off what ever Linux MIT puts on it anyway and putting on the same pirated Windows.
Is Windows Rommable? It requires swap space and won’t let you turn it off last I knew. The MIT box will use Flash memory instead of a hard drive.
Is Windows Rommable? It requires swap space and won’t let you turn it off last I knew.
You can turn off swap. I used to do so often.
Ordinarily I don’t think your typical pirate would be able to do anything with a Flash only system but..
A few years back MS was touting NT3.5 for embedded use in ROM form at the Embedded Show, still I doubt even Windows geeks could fix up a modern Windows that could fit in such a small flash, kinda moot. CE though a different matter.
As for swap, I usually run W2K with no or a tiny 20M swap.
For the BLX machine, I wonder if the early MIPs version of NT could boot, I still have my Alpha/PPC/MIPs/x86 cds. I don’t recall anys apps though beyond x86.
>> A few years back MS was touting NT3.5 for embedded use in ROM form at the Embedded Show, still I doubt even Windows geeks could fix up a modern Windows that could fit in such a small flash, kinda moot.
Does a 9 meg flash bootable Win98 count?
http://www.litepc.com
Oh, btw, that’s what runs on lottery machines here in New England. A lite-PC stripped version of 98.
If the board has an IDE controller nothing is going to stop a local pirate from selling the MIT laptop packaged in a desktop cabinet with a cheap harddrive making it a $150 Windows running machine!
I haven’t seen the MIT mobo so I have no idea if it has actual or temporary ATX, 4pin power or IDE connectors, any pics anywhere?
If has no IDE and ATX connector you would be up the creek except maybe through a USB port perhaps.
I doubt it has them, if it did that would invite exactly the upgrades they said shouldn’t happen. Also picture how much power these laptops use v the same with an external running HD. A cheap regular 3.5 inch HD needs about 10W (guessing) esp peak startup current. The 2.5inch HDs need far less but cost several times as much. Would that work off a cranking system, doubt it. If one is prepared to go to the trouble of jigging an ATX supply with HD and “cabinet”, one might as well get a decent budget mobo & cpu which can be special offer <<$100 or so. Back to square one.
It reminds me 20yrs ago of the BBCs at 400quid v the poor little Electron for 100quid. Well there were these dodgy upgrades to try and boost the Electron into a Beeb, and the Timex/Sinclair upgrades too.
AMD’s PIC
http://www.amdboard.com/pic.html
Which Radio Shack was selling a version of last year priced at $199, you can still pick them up on clearance for well under $150 these days.
Came with Windows CE, but I’ve seen them boot Linux off a flash drive since it is just another x86 box by way of the Geode CPU.
Again though, thin clients; an idea that sounds GREAT on paper yet NEVER takes off in any real numbers. When the big ‘success’ in this department is Web TV, you really have to question thinking of these as anything more than a glorified money pit.
Which is a shame – for IT environments where you cannot trust the clientel, (public libraries, school computer labs) these types of devices (running CE or Linux) are a ‘tech managers wet dream compared to a more conventional PC.
… uses a unique 64-bit CPU with an instruction set based on a subset of the MIPS architecture.
Well, the PowerPC didn’t compete well against the x86 on the desktop and now Genesi and YellowDog are about the only ones pushing it. Most people never even heard of MIPS and fewer have used it.
Still, I wish them success. It seems unlikely that Microsoft will port Windows to it, so every sale advances the acceptance of Linux and makes open source software more visible.
I want one to mess with
Pretty cool.
edit: nm.
Edited 2006-04-26 17:11
I wonder how usable the hardware documentation would be to a non-Chinese programmer, given the focus on their domestic market. If the CPU and friends are fully documented, preferably in English, I would *love* to have one.
Well, the OS being Linux should translate into much of their source code being available in some form, except perhaps for some proprietary drivers and programs they write themselves. Any mods of GPL software in that machine should be made public. It will certainly be interesting to see how they and other “upstart” Chinese manufacturers deal with THAT licence…
I don’t think documentation is going to be that bad.
If anyone care to notice, this is not developed in mainland China, instead, it’s developed in Macau. Macau SAR is similar to Hong Kong SAR in that it’s ran by a European country until the handover back to China control in the late ’90s (in the case of Macau it’s Portugal, in the case of Hong Kong it was the UK).
The general English competency in Macau and HK is much better than the typical stuff you’ll find in most Chinese products. Considering that they have a fully-English website I think they’re planning this device for the internationalmarket
It may not run Windows but you will have a very powerfull Thinclient solution for only $150!
If the processor is based on the MIPS, didn’t Microsoft have NT 4 running on MIPS. With the potential of another billion customers, I’d bet that they are updating that code as we speak.
As the usual “Made in China”, they’re cheap, but don’t expect they have the best quality.
Most of electronic appliance that some of my relatives/friends bought usually breaks after a year or 1.5 years.
Sometime price defines quality
That rushing noise you heard was the sound of the whole point of not just the article, but the entire project its about going completely over your head…
PS) Lot’s of pricy stuff is made in China. My PowerMac was made in China. My iPod was made in China. The $7,000 amp Mark Levinson sells under his Red Rose label was not only made in China, but designed by a Chinese company.
Your points are correct, some of the stuff we used these days are Made in China. Sure iPod are Made in China, they shipped it to me directly from Shanghai factory.
(But let’s not forget iPod got recalled 🙂 )
But price… sometime Price does not lie 🙂
I love it how Chinese companies come up with their names by mashing together english words at random.
“YellowSheepRiver”, that’s a good one.
What is that supposed to mean?
Probably a translation of the local name of a river close to their office.
MIPS derivative = uncertain whether it’s compatible with MIPS-based repositories (i.e., anyone else’s distro)
512Mb memory ceiling.
400-800MHz CPU.
Web surfing and word processing, looks like to me. I can go to my campus surplus department and plop down $125 on a used P3 with replaceable/upgradeable commodity components, and that includes keyboard, monitor and mouse.
This looks less like a revolution and more like the second coming of MSX.
The MIPS architecture is old, tired, and left behind. In 1995 when Byte benchmarked Windows NT on X86, PowerPC, Alpha, and MIPS all running the same applications, MIPS trailed the whole pack.
MIPS it may be low power, but it’s also low speed. This particular architecture shows no inspiration or innovation. Unless someone in China Central Planning pours a whole lot of architecture development resources into the next generation, the only benefit of this product is cheapness.
But even cheapness has trouble competing when people want to do word processing, surfing, and MP3, MPEG, email with decent features.
Why reinvent the wheel?
MIPS it may be low power, but it’s also low speed.
Architectures are neither low speed nor low power. Implementations of them may or may not be. Reports are that this CPU is an R1000-like 4-issue design, which means its per-clock performance should be quite competitive in its market class, even if its clockspeeds are limited by China’s less than SOTA fabrication processes.
This particular architecture shows no inspiration or innovation.
Inspiron and innovation are not necessary prerequisites for utility.
Why reinvent the wheel?
Because there is a market of over one billion people in China, many of whom cannot afford even the cheap $400 computers you find in the US. Because China, with its massive number of engineers and cheap manufacturing capacity, thinks it might be able to compete in the low-end CPU market? Because they realize that depending on a potentially hostile foreign power for a need as basic, in this computerized world, as CPUs is not necessarily in their national interest? You pick the reason.
“But even cheapness has trouble competing when people want to do word processing, surfing, and MP3, MPEG, email with decent features.”
A 500Mhz PIII can easily cope with all that, if this MIPS based computer has similar CPU power then it should be equally competent.
<em>
The MIPS architecture is old, tired, and left behind. In 1995 when Byte benchmarked Windows NT on X86, PowerPC, Alpha, and MIPS all running the same applications, MIPS trailed the whole pack.
</em>
And yet, 3 years earlier, MIPS R4000 was humiliating the Pentium x86 of the time, especially in FP.
MIPS fell behind because lack of volume = lack of investment = lack of development.
I can’t find a reference to the article, BTW. This article:
http://www.byte.com/art/9503/sec11/art1.htm
Indicates the performance went Alpha > MIPS > x86, but doesn’t include any PowerPC machines.
I had a question without an answer, which grow with me since the dawn of computing, Why there is no one nation in this planet who can produce CPUs and OSs that works together that could be marketed to the public? I know that some countries like japan, china, … produce some CPUs but they are for purposes other than desktop or workstation jobs.
So, maybe this one will be the answer to my question.
The basic problem is that fabricating CPUs is ridiculously expensive, and the technology is held by a few major (mostly American) companies. There are basically two or three absolutely cutting-edge manufacturing processes out there: Intel’s, IBM’s, and their derivatives (eg: AMD’s IBM-derived process, Sony’s IBM-derived process, etc). The majority of the big Japanese and American manufacturers are one step behind (eg: Texas Instruments, Fujitsu), while the Chinese and Taiwanese founderies are a couple of steps behind (eg: TSMC). To give an example, a chip fabricated for Intel’s latest manufacturing process may run at triple the clockspeed of a chip fabricated for TSMC’s latest process.
Basically, if you can’t spend a few billion a year on fabrication technology, or can get access to such technology from someone who can and is willing to license it out (eg: IBM), then you cannot produce chips that compete with the POWERs and Opterons and Itaniums of the world. Moreover, if you don’t produce a mainstream architecture (eg: x86 if your marketing to the public, ARM or PPC if you’re marketing to embedded manufacturers), then your volumes will be tiny, and your slower chips will actually cost a lot more to produce then more mainstream CPUs.
Edited 2006-04-26 23:54
All that is true but there is far more to the story why Intel x86 & clones run 3x faster on SOTA US processes than TSMCs generic processes. The way Intel, IBM, AMD design cpus is much more at a transistor level for all the critical circuits while the other companies are designing at the gate level or even block IP level using far more abstraction and synthesis.
Alot gets left on the table by standard cell guys, transistor level design easily uses less die area, less switching power, can have less leakage, but takes orders more manpower, time and resources to do. It also takes time and many product generations to master state of the art circuit design. It wasn’t that long ago that the Pentium was in the same pedestrian ballpark of chip design as many other chips (100MHz days), then Intel got on a rush and the cpus went much further out than other chips. When you design at the transistor level on std cell processes like TSMC, you should be able to get near similar performance for critical functions, but TSMC, UMC are more set up for std cell library design and rapid IP turnaround.
As for Chinese fabs, much of the investment is coming in from Taiwan mostly from the TSMCs and UMCs to lower their own costs and lock in the next market. These are all joint ventures with Chinese gov consortia, that lets China get a free ride to state of the art foundries that I suspect are really only 1 year behind US fabs. Either they will fill up with western designed job lots or some of that capacity will fire up local innovation.
One thing that China is doing is graduating EEs at a rate that is not just 4x the US but another order more, while the US graduates more lawyers. Remember those ACM contests, Russia & China are aceing those. If you put enough EEs together with some well seasoned expat veterans coming back from the US, you may well see some true state of the art designs coming out.
The real issue is whether Linux can fly as a consumer OS when offered on lower cost HW, if it does, then it doesn’t matter what the cpu ISA is. But that brings up the whole issue of DRM control, movies, more IP patents, lawyers.
A 500Mhz PIII can easily cope with all that, if this MIPS based computer has similar CPU power then it should be equally competent. Maybe for running applications made in 2000. I find it hard to believe that the Chinese, who currently fabricate and assemble the world’s laptops and desktops would really settle for something with so little capability.
I also find it hard to believe that some centrally planned device would ever be as cost effective as simply licensing any one of many better CPU designs. Lets face the fact that this design is the result of either ethnocentrism or national pride, and not the result of excellent engineering.
I seriously doubt that this design will be desirable to the Chinese who can afford to purchase X86 based laptops standard software (OSS or Windows). This will have all the appeal of the Soviet designed PC’s of the 80’s. Very sad boxes they were.
I think you are putting down potential creativity that hasn’t had much chance to show itself yet.
How is this chip anymore centrally planned than the Intel chips. You think CP party hacks do or organize chip design. This was done by EE professors and their student help. Comparing to Soviet era PCs is a poor joke (remember chips popping their sockets). If a country can manufacture much of the lower cost electronic consumer products we all buy, it can certainly learn to design them, esp for local captive market. Russia never had any serious manufacturing quality.
When you own the worlds oil, the last thing you want is to just pipe it out to foriegners who add all the value at their own refining plants.
When you own a good chunk of the worlds fab capacity, China+Taiwan combined, do you really want to just make ASIC product for foriegners. No you want to move up the economic ladder and develop as much of the IP as you want to consume. Again the issue comes back to what do PCs do, and DRM and lawyers etc.
MIPs had its heydey with SGI workstations while PCs were harmless by comparison. PCs gradually acquired the same graphic and compute power then more, they had volume markets to really help. There’s no reason why MIPs cpu with more grunt and graphics support couldn’t do same as x86 except of course Windows.
Someone suggested this was a better sell for the countries that the $100 laptop is aimed for. I don’t see how.
Give a person the laptop, and if they are willing to spend the time to learn then they have all they need short of an internet hot-spot.
Give the same person this box and it looks like they need to buy a monitor (how many cheap take S-Video? Yes, I am asking, I don’t know myself), a keyboard and a mouse to get started. Wireless is extra, so extra hardware is needed if a number of people want to share the internet access point. I think the laptop is better outside China.
As for making needed volume inside China, if the government mandates the design to the base machine for schools the sales volume will be more than high enough to drive down prices.
“Unfortunately, the price is this low mostly because their workers get treated (or at least paid) like slaves. I hope nobody finds a fingertip in their pc!”
Oh, you mean just like the workers in Intel and AMD’s SEA chip plants?
You realize of course that pretty much every single electronic appliance you have is made entirely or partly in China?
The schools in my city (Chicago) need textbooks and computers. They just don’t have enough like the suburbs and the state doesn’t want to move money from the suburbs to the city (property taxes). Today there are much more expenses due to tech unlike the old days when my parents went to school so why not make tech free. If every child had this it would be completely helpful and stop allot of crime. Kids would stay in school off the streets and the college attrition rate is like only 3 percent graduation from Chicago. This could stop the need for most textbooks and help allot with CRIME.
Maybe tech van be very cheep and free to create a utopia. i think it should be since that’s why we invented it not to overburden ourselves.
Personally speaking, the idea of computers in schools as a substitute for teachers & books is a terrible idea esp when it means states like Maine forcing families to buy an iBook for every kid.
Computers were invented and developed to help people solve ever larger scientific & business problems, not to allow ordinary people to unburden themselves with the necessity of having to learn anything. If you want to have an iBook next to you in class and exams as a crutch, you might as well become a google borg.
Most of the people who design computer chips (self included) all went to school at a time when there were no PCs or computers in school or home except for occasional dialup through a terminal. Since computers have entered schools, I can only see decline in standards. They don’t do this in schools in Asia or parts of Europe where their kids whip US kids in the world tests.
Books & good teachers, eating well, exercises, no computers please. I could go on but would likely only inflame the I wanna an iBook too students. Students who want to learn should be separated from those that don’t, that will solve the CRIME problem right away.
I agree with the last staement and I dont think I said it should replace teachers. I was just saying that they could use a simple cheap laptop to replace books but maybe that would be just a s expensive. To me the largest problem is affording to pay teachers. I wouldnt want kids on computers all the time.
It smacks of multi-culturalism by homoginizing cultures into a borg. Most people don’t trust the Internet and I dont blame them. It’s not a perfect utopia.
How is this chip anymore centrally planned than the Intel chips. You think CP party hacks do or organize chip design. This was done by EE professors and their student help. Comparing to Soviet era PCs is a poor joke (remember chips popping their sockets). If a country can manufacture much of the lower cost electronic consumer products we all buy
These are good questions, the answers to this are the same reason why MIPS, Alpha, and SPARC all fell behind. Economies of scale make each chip cost less and less, and sell to a larger market. This makes money for the the R&D of the next generation. As Intel and AMD make more money, they put more money into the design and fabrication of the next generation. More change occurs between each generation, and the time between each release is shorter.
The R&D costs to design and fabricate a new generation of competitive CPU is very high these days despite the CAD tool available.
I would also say that a centrally planned CPU is worse (speculation here, I have no hard facts) because the government will most likely appoint a favored engineer team based on political reasons. This is how it has worked in the past. These appointees will generally keep their jobs whether or not they do a good job.
I do have inlaws in Taiwan and some also live & work on the mainland so the info I get is something like this. If you are good at what you do and keep your nose out of politics, business is business and not so different from Taiwan. I expect the professors still have some nominal party membership too but I doubt that gets in the way of interesting work. To make money is glorious, and that was said 20 some years ago.
You are really repeating what I said before on the practice of perfecting each generation. So on the MIPs, Alpha, SPARC case, lack of market share killed their high end designs, same should be true for Itanium too. Commoditizing the x86 made it possible to put every high end feature of the other chips into it and the burden of the horrid origins became less relevant. Most of the x86 chip is cache and superscaler related rather than x86 decodes. So we won’t see much resurgence in western markets for the pure RISCs except in embedded. That also means the x86 is just as “not exciting” as the MIPs if not more so, they both end up with the same OoO engines, one with more baggage on the side. There’s nothing magic about instruction sets, the implementation is more interesting.
Now in China I really don’t expect to see any credible x86 clones either (see Via) so whats left to do but make the best of whats possible without getting into patent problems. If any x86 clone were attempted, it would be a patent disaster all over. Using MIPs ISA does kill 2 birds with one stone as it makes cheap Linux boxes more possible and keeps Windows, Intel out. Might even work if the Linux is widely accepted. That also means Sparc & PPC might yet get more licenses.
Now when I hear all this nonsense about pricing of proprietary designs. SGI never could have sold more than what 100K?? MIPs machines and at such prices they could afford alot of very expensive EEs. In China we might see a market for many million of these boxes with R/D costs and fab costs orders less than those for x86. A well designed MIPs core is still going to be a fraction of the size of comparable x86 core. So what do x86 chips cost, anywhere from $40 (Semprons) to $800 dual 64b cores. What does a 600MHz Mips core cost, probably about $1 if its embedded. A bit more chip area and one could throw in most of the peripheral logic and it will still be $1. Much of the NRE cost will be obtaining all the IP or developing equivalent IP to complete ever more integrated boxes.
The past is the past, today China will have access to zillions of EEs, appointees will fade away, expats are returning from the US to take the lead. That kinda of scares me to see some of the action moving away from the US but then they have alot of issues to deal with that we don’t have to worry about.
To be fair, the Godson-2 isn’t an embedded MIPS core. It’s a 4-way superscaler design along the lines of the R10k. It’ll be small, but not $1 small
I can’t find a reference to the article, BTW. This article:
http://www.byte.com/art/9503/sec11/art1.htm
Indicates the performance went Alpha > MIPS > x86, but doesn’t include any PowerPC machines
There was another benchmark article about the same time with PowerPC. One of the companies listed in that article was mine, and I helped develop the software used. I was actually in the Byte labs. PowerPC was faster than MIPS, but yes, at that time the MIPS was faster than the Pentium 90 machines used. My mistake on that.
But today it would take a very large amount of R&D to rev the CPU. Only to be slower and more expensive than the X86 architecture. But the Chinese government has made it a priority to have it’s own CPU. This is a costly decision and we will see how these device can compete in the marketplace against the capitalist powerhouse IT companies.
In this instance, I bet on capitalism.