Linux.com

Feature: News

China puts hopes in Loongson CPU

By Chen Nan Yang on October 31, 2007 (3:00:00 PM)

Share    Print    Comments   

China, which has long wished to develop its own computer industry, has chosen to go with Linux on the software side. Loongson is its hope for the hardware side.

The Institute of Computing Technology (ICT), a department of the Chinese Academy of Sciences (CAS), has been developing the Loongson processor since 2002. Rather than use the x86 instruction set used in chips from Intel and AMD, Loongson uses MIPS, an instruction set patented by MIPS Technologies. This means that Loongson chips are unable to run the full version of Microsoft Windows, so they run Linux instead.

Loongson-2F, the latest version of China's homegrown processor, will be released in November, according to CAS. Longsoon-2F will operate at a frequency of 1.2 to 1.5GHz, matching the speed of the low-end Pentium 4 processor while consuming only around 5 watts of power. Unlike its predecessor, Loongson-2E, Loongson-2F will support DDR2 SDRAM and USB 2.0.

CAS also says that Loongson Box, a computer with the Loongson processor but without a monitor, keyboard, or mouse, will begin mass production in early 2008. ICT has authorized STMicroelectronics to produce Loongson chips and Loongson Box. STMicroelectronics will cooperate with Mandriva and use Mandriva Linux as the operating system for Loongson Box, which also features GNOME, Wallpapoz, and gDesklets, and supports virtual desktop.

Behind-the-scenes controversy

Originally, ICT claimed that Loongson has completely independent intellectual property rights. As it turned out, however, ICT had been using the MIPS instruction set without obtaining a license. ICT asked the government to pay for the license at $200,000 per generation of Loongson chip but was refused. After failing to come to an agreement with MIPS Technologies, ICT had to stop referring to the Loongson chips as MIPS-compatible processors.

To earn back the trust of the government and the Chinese people, ICT launched many ambitious yet ultimately fruitless plans over the past two years. For example, it promised to produce a 1,998-yuan ($250) laptop for Chinese students in 2007, but finally declared that the price would be at 5,000 to 6,000 yuan ($660 to $800) when it might go on the market next year.

ICT had to resort to international cooperation to settle its trust problem. It worked with STMicroelectronics to get a proper license; STMicroelectronics had signed a licensing agreement with MIPS Technologies for the MIPS64 architecture. ICT hopes that STMicroelectronics, one of the 10 largest chip makers in the world, can help improve Loongson's production technology. In exchange, STMicroelectronics has a five-year agreement to produce and sell Loongson processors and computers in markets around the world. According to the agreement, STMicroelectronics will pay ICT $2 for each Loongson chip sold.

Loongson Box won't make inroads into the market soon because of its poor price-performance ratio. Last year, ICT trial-produced 1,000 Loongson Box products and sold them at 1,599 yuan ($210), but the price of the mass-produced Loongson Box will be higher than the trial-produced version. With display, keyboard, mouse, and CD-ROM, its price may reach $400, which is no less than many better-performing computers with Intel or AMD CPUs. Loongson's lone advantage seems to be its low power consumption.

A plus for ICT is that the Chinese government will support Loongson chips regardless of the price-performance ratio. Once Loongson chips can meet basic demand, China plans buy them for its army, government offices, and public education. In addition, some local governments have been purchasing computers for China's rural areas to demonstrate the achievement of the "new country construction." It's estimated that China's rural areas will utilize at least 6 million computers in 2007 and 2008, giving Loongson a big boost in this arena.

Chen Nan Yang is a Chinese freelance journalist and former IT director in the Chinese government.

Share    Print    Comments   

Comments

on China puts hopes in Loongson CPU

Note: Comments are owned by the poster. We are not responsible for their content.

Wonderful news!

Posted by: Anonymous [ip: 169.233.25.226] on October 18, 2007 08:28 PM
MIPS > x86. Finally, some mass produce of MIPS processors will hit the "average Joe [or Zhang]" desktop / laptop market.

#

Debian, Gentoo and Slackware.

Posted by: Anonymous [ip: 77.250.107.247] on October 21, 2007 09:32 PM
It already runs Debian and Gentoo. I have ported Slackware to this chip over the last few months.

#

Re: Debian, Gentoo and Slackware.

Posted by: Anonymous [ip: 78.86.121.24] on November 01, 2007 08:26 AM
Whoever you are, nice work!

Just how well dose the Debian MIPS port work? fully?

#

Re: Debian, Gentoo and Slackware.

Posted by: Anonymous [ip: 219.128.204.141] on November 01, 2007 04:21 PM
good work!

#

China puts hopes in Loongson CPU

Posted by: Anonymous [ip: 122.164.228.124] on October 31, 2007 03:48 PM
I've seen someone run a compositor on X on a Lemote. You can see the video on Youtube.

- George

#

Performance doesn't matter

Posted by: Anonymous [ip: 148.87.1.172] on October 31, 2007 05:23 PM
If the CPU does the job, it performs well enough. Most modern desktops are overspec-ed for the work they are used for.

#

Re: Performance doesn't matter

Posted by: Anonymous [ip: 78.86.121.24] on November 01, 2007 08:35 AM
indeed, my current processor runs at 800MHz, and it doesn't run on a delicious 5v

#

A poorly-informed article

Posted by: Anonymous [ip: 82.192.250.149] on October 31, 2007 08:15 PM
The power-consumption issue is much more important than the author realizes. There is a trade-off between clock speed and power consumption. The speed of the current Intel chips is limited by power consumption. Increasing the speed of the Loongson chips will be straightforward, because power consumption is not yet a problem.

#

China puts hopes in Loongson CPU

Posted by: Anonymous [ip: 72.91.22.163] on November 01, 2007 12:32 AM
Nice ... and I'm sure all the engineering was homegrown, instead of build via corporate espionage (courtesy of Intel/AMD/etc). Gotta love those wacky Chinese. They have no qualms about stealing IP.

#

Re: China puts hopes in Loongson CPU

Posted by: Anonymous [ip: 78.86.121.24] on November 01, 2007 08:29 AM
Neither did the US, back in the beginning they kickstarted their industry by ignoring European patents.

mind you this is MIPS not x86, so mabey it was home developed, China wants a home developed IT industry and might have decided that not copying was an investment in skills.

#

Re: China puts hopes in Loongson CPU

Posted by: Anonymous [ip: 91.106.245.41] on December 11, 2007 09:34 PM
You mean like that that computer developed at Bletchly park to crack the enigma code that Americans copied after WWII..

#

x86 vs MIPS

Posted by: Anonymous [ip: 78.86.121.24] on November 01, 2007 08:34 AM
The article says "Longsoon-2F will operate at a frequency of 1.2 to 1.5GHz, matching the speed of the low-end Pentium 4 processor"

Just because it runs at the same Hertz dosn't mean they're the same speed, there are diffrence in the instruction sets, and a different compiler backend will be used both of witch have an impact on the final speed of a program.

Anyone know how 1.2GHz on MIPS compares to 1.2GHz on X86?

#

Re: x86 vs MIPS

Posted by: Anonymous [ip: 131.111.113.191] on November 01, 2007 01:41 PM
The same comment once more - I didn't click Replay :/
From the orginal paper (check link on Wiki) L-2E 1.0 GHz seems be a bit faster then P4 1.4 GHz. So L-2F can be around 1.6-1.8GHz - not bad from avarage user.

#

Re: x86 vs MIPS

Posted by: Anonymous [ip: 194.74.144.130] on November 01, 2007 02:12 PM
Clockspeed bears any useful relation when comparing different microarchitectures. Here's an extreme example from the MIPS history:



o R4400 250MHz ~ 130 SpecInt92, ~ 138 SpecFP92


o R8000 75MHz ~ 135 SpecInt92, ~ 320 SpecFP92.



An R10000 delivers about twice the performance per MHz than a Sibyte SB1 or SB1 CPU core.

As for the microarchitecture of the Loongson, it has about every feature you can find in a book. It's mostly handicaped by the manufacturing process available to the ICT.

#

China puts hopes in Loongson CPU

Posted by: Anonymous [ip: 124.106.223.178] on November 01, 2007 10:55 AM

China puts hopes in Loongson CPU

Posted by: Anonymous [ip: 195.216.45.226] on November 01, 2007 01:11 PM
Fat chance!
When Intel releases its 45nm Menlow (Silverthorne CPU) platform next year (Q2 or Q3) ICT - and ARM - can only hope to compete internationally with an extremely low price for the CPU + chipset + mobo.
Menlow additionally also run all forms of Windows.... (+ maybe OSX)

REF:
http://www.engadget.com/2007/06/09/intel-ceo-compares-silverthorne-criticality-with-original-pentiu/
http://blogs.intel.com/tag/Menlow
http://www.engadget.com/tag/menlow
http://www.google.com/search?num=20&hl=en&q=Silverthorne+Intel
http://www.google.com/search?hl=en&q=Menlow+Intel

#

China puts hopes in Loongson CPU

Posted by: Anonymous [ip: 131.111.113.191] on November 01, 2007 01:39 PM
From the orginal paper (check link on Wiki) L-2E 1.0 GHz seems be a bit faster then P4 1.4 GHz. So L-2F can be around 1.6-1.8GHz - not bad from avarage user.

#

China puts hopes in Loongson CPU

Posted by: Anonymous [ip: 203.116.22.138] on November 01, 2007 05:03 PM
pirate

#

China puts hopes in Loongson CPU

Posted by: Anonymous [ip: 219.95.16.236] on November 05, 2007 01:39 AM
This may sound weird but what is actually developed in China? If the architecture is MIPS and the licensee is STMicro, where does ICT (or China Academy of Sciences) play a part? It sounds to me like it's a collaboration between MIPS and STMicro to penetrate the Chinese market through asking ICT to design a chip that's MIPS compatible? Smart move by MIPS and STMicro :-)

#

China puts hopes in Loongson CPU

Posted by: Anonymous [ip: 222.166.160.180] on November 13, 2007 06:02 AM
Most part of MIPS are like copyleft. Anyone can develope one if he has enough resources. I guess that's why the Chinese choose the MIPS instead of x86: to avoid patent issue.

#

China puts hopes in Loongson CPU

Posted by: Anonymous [ip: 128.195.52.116] on December 18, 2007 07:20 AM
i have been keeping track of the loongson chip for some time. the license issues in the article is unclear and somehow misleading. the patents of mips16 have already expired, so it doesn't make sense for ICT to pay for it anyway. also ICT added several instructions to enhance the mips' ability in multimedia processing. after cross-licensing with ST, ICT officially now implemented several other patented mips32 instructions and claim the chip to be "mips compatible". they are all within the boundary of rules and no foul play or whatsoever.

#

This story has been archived. Comments can no longer be posted.



 
Tableless layout Validate XHTML 1.0 Strict Validate CSS Powered by Xaraya