First A9home benchmarks surfacePublished: 2nd Jul 2006, 01:27:19 | Permalink | Printable
Standard health and safety warnings applyThe first set of crude benchmarks of AdvantageSix's A9home are doing the rounds showing how the little blue ARM9 machine fares against the Castle Xscale powered Iyonix. Although the A9home's memory bandwidth - the amount of data it can shunt between the processor, RAM and other components - edges past the Iyonix, it can't quite keep up with its 600Mhz rival and its Nvidia graphics card. The Iyonix's UDMA support overtakes the A9home's PIO mode grade speed for hard disc access, although the A9home appears to give the Iyonix a run for its money in the unaccelerated filesystem accesses. The A9home also comes out tops on rendering 2D shapes with the Draw module.
Richard Spencer's RISCOSmark and Michael Kübel's Firebench programs were used in the tests after Michael, wondering how fast his software ran on the new A9home, encouraged users to post their results. Firebench, pictured above, plots an animated fire-like effect by modifying individual pixels and recording the number of full screen frames plotted per second. The Iyonix is known to suffer in this sort of test because the video RAM is on the far end of the PCI bus on the Nvidia graphics card. An Omega managed to score 440 frames per second in this test, beating the Iyonix.
The other package, RISCOSmark, performs a number of tests on different sub-systems within the computer. It stresses the processor by running a small program in its cache, and measuring how many times this is executed within 100 centiseconds. It also times how fast the processor can read and write information to the system's RAM.
As always, these benchmarks are barely scientific, and are only good at measuring fairly abstract definitions of performance. They give a rough, rule-of-thumb, roundabout feel of the overall standing of the computers, as most users will prefer real benchmarks such as the speed at which images are loaded from disc, how quickly webpages are rendered in dinosaur Oregano 2, and the transfer rates over USB and Ethernet. Nevertheless, below are the figures, but please do take them with a pinch of salt.
|600MHz Intel IOP321, RISC OS 5.11, No Geminus||400MHz Samsung ARM9, RISC OS 4.42, No Geminus||233Mhz Intel StrongARM, RISC OS 4.03, Viewfinder 2|
|Processor||256%||174%||116%||Measures CPU performance, higher the better|
|Memory||185%||374%||158%||Measures memory bandwidth, higher the better|
|Rectangle copy||4433%||1405%||2251%||Measures 2D rectangle acceleration, higher the better|
|Icon plotting||28%||76%||19%||Creating and plotting a simple icon, higher the better|
|Draw path||92%||102%||46%||Plot a 2D line, higher the better|
|Draw fill||64%||157%||13%||Fill in a 2D shape, higher the better|
|Hard disc read||36MB/s||9MB/s||1.9MB/s||Read in a 1MB file in one block, higher the better|
|Hard disc write||14MB/s||7MB/s||1.7MB/s||Write out a 1MB file in one block, higher the better|
|File system read||1146KB/s||616KB/s||713KB/s||Read in a large file in one byte at a time, higher the better|
|File system write||358KB/s||493KB/s||506KB/s||Write out a large file in one byte at a time, higher the better|
|Frames per second||409.83||355.23||259.06 (RPC Turbo 287MHz)||Plotting a moving fire visual effect, higher the better|
The baseline machine for the above tests, ie: the '100%' machine, is a vanilla 202MHz StrongARM RiscPC running RISC OS 4.02, and all the tests were run in 800x600x256 colours. An A9home review by drobe.co.uk is currently in the works.
Firebench (zip download)
Previous: NetSurf conquers Wikipedia
Next: Resisting change is short-sighted
DiscussionViewing threaded comments | View comments unthreaded, listed by date | Skip to the end
Please login before posting a comment. Use the form on the right to do so or create a free account.
Search the archives
Today's featured article
Pace and Sega announce Dreamcast joint venture
Discuss this. Published: 29 Jan 2001
RISCOS Ltd's head licence with E14 leaked
A partial scanned copy of what appears to be the original 1999 licence between ROL and E14 - the company formed from the break-up of Acorn that kept the RISC OS technology - has been anonymously leaked to several RISC OS magazines and websites this afternoon. The disclosure, which arrived by email from a gmail account, is believed to have come about following the recently revived licensing dispute between ROL and Castle.
11 comments, latest by DaveLane on 22/12/08 5:27PM. Published: 20 Dec 2008
News and media:
RISCOS Ltd •
RISC OS Open •
MW Software •
Advantage Six •
CJE Micros •
Liquid Silicon •
Chris Why's Acorn/RISC OS collection •
The Register •
The Inquirer •
Apple Insider •
BBC News •
Sky News •
Google News •