MIPS Loongson 3A Benchmarks On Debian

[phoronix]

Phoronix: MIPS Loongson 3A Benchmarks On Debian

Here are some benchmarks of the MIPS-based ICT Loongson-3A quad-core processor out of China...

http://www.phoronix.com/vr.php?view=MTEyOTE

[Tiger_Coder]

Still waiting for a day without x86 everywhere

[grotgrot]

Still waiting for a day without x86 everywhere

You don't have x86 everywhere. Look at all your mobile devices, the things connected to your TV or your car. Heck even an x86 based computer has more non-x86 processors in it than x86! For example almost all SSD drives have an ARM chip, the GPU is not x86, the network device has some sort of processor, and there are various other processors (eg power control on laptops), system management processors etc. If you just looked at numbers of processors shipped then x86 is almost irrelevant.

[kbios]

718 bogomips? My Phenom II is at 7061, with a TDP of 125W. i know bogomips are not very precise when used to compare different architectures, buy still I don't see that much difference in efficiency....

[Kivada]

Yeah, I'd like to get one of those dual CPU boards if only I could fit it with a 16x GPU as they are trying for compatibility with off the shelf PCI and PCIe boards as well as X86 emulation. The pics of the boards I saw only had closed ended 8x, 4x and 1x slots.

[mattst88]

718 bogomips? My Phenom II is at 7061, with a TDP of 125W. i know bogomips are not very precise when used to compare different architectures, buy still I don't see that much difference in efficiency....

The -bogo- in bogomips is short for bogus. It's less useful than glxgears for comparing performance.

[devius]

The -bogo- in bogomips is short for bogus. It's less useful than glxgears for comparing performance.

Still, it seems to be just about right considering the ramspeed results are also very low in the 240~300MB/s.

Unless this is a 1W CPU I'm not impressed.

[mattst88]

Still, it seems to be just about right considering the ramspeed results are also very low in the 240~300MB/s.

Unless this is a 1W CPU I'm not impressed.

No, not still. It's completely bogus.

[popper]

Still, it seems to be just about right considering the ramspeed results are also very low in the 240~300MB/s.

Unless this is a 1W CPU I'm not impressed.

well OC the ram speed results are very low , its not likely Loongson are going to licence the latest wideIO AKA 512 bit wide bus (4x128) with far lower power usage and yet far greater speed throughput (up to 17GBps),or even come up with their own compatible version soon and that's a shame as it looks to potentially become a well used lowest power usage world standard sooner rather than later if that other upstart doesn't rock the boat.

http://www.jedec.org/news/pressrelea...io-mobile-dram
JEDEC Publishes Breakthrough Standard for Wide I/O Mobile DRAM


http://www.xbitlabs.com/news/memory/...d_Servers.html
JEDEC Considers Wide I/O, Wide I/O 2 Interfaces for PCs and Servers.

DRAM Chips with Ultra Wide Memory Bus May Find Their Way to PCs, Servers[06/06/2012 08:16 PM]
by Anton Shilov

http://denalimemoryreport.wordpress....n-ic-packages/
Invensas to detail POP interconnect to rival Wide I/O with as many as 1200 interconnections between IC packages

[curaga]

718 bogomips? My Phenom II is at 7061, with a TDP of 125W. i know bogomips are not very precise when used to compare different architectures, buy still I don't see that much difference in efficiency....

On most x86, bogomips = MHz * 2
On most ARM, bogomips = MHz

Assuming MIPS is like ARM here, how the heck do you take performance out of the clock speed :P