The Icon Bar: News and features: Alpha benchmarks, Omega tittle tattle
Posted by Phil Mellor on 00:46, 13/7/2003
| RISC OS, Castle Technology, RISCOS Ltd, Open source, Emulation, Acorn, Hardware, Graphics, MicroDigital, Google, IYONIX
Alpha owner John McCulloch has posted some benchmarks, produced using RISCOSmark, to csa.hardware. "I got mine yesterday, it is very impressive, in all sorts of ways" says John. "It is very well made, and functioned as expected right out of the box". The following table shows the results of the Alpha running RISCOSmark with Virtual Risc PC; John's 202Mhz Risc PC is included as a comparison, as well as the results provided on the RISCOSmark web site (note these are run in a less intensive screen mode, which is significant for computers without VRAM, such as the A7000+). Machine (see below) | Alpha | SA202 | SA287 | A7000+ | A4 | Iyonix | Resolution, colour depth | 1024x768, 32K | 800x600, 256 (A4: 16 cols) | Processor - Looped instructions (cache) | 41% | 99% | 141% | 12% | 5% | 261% | Memory - Multiple register transfer | 245% | 98% | 98% | 83% | 54% | 201% | Rectangle Copy - Graphics acceleration test | 169% | 1149% | 101% | 66% | 74% | 5786% | Icon Plotting - 16 colour sprite with mask | 51% | 10% | 117% | 22% | 9% | 32% | Draw Path - Stroke narrow line | 36% | 40% | 142% | 16% | 8% | 101% | Draw Fill - Plot filled shape | 50% | 7% | 139% | 20% | 10% | 75% | HD Read - Block load 1MB file (Mb/sec) | 99 | 1.8 | 3.3 | 1.9 | 1.3 | 18.0 | HD Write - Block save 1MB file (Mb/sec) | 66.3 | 0.75 | 3.1 | 2.0 | 1.2 | 13.8 | FS Read - Byte stream file in (Kb/sec) | 649 | 820 | 315 | 141 | 74 | 1101 | FS Write - Byte stream file out (Kb/sec) | 328 | 281 | 306 | 140 | 62 | 1020 | Alpha | 2.0GHz Celeron (VRPC) RISC OS 4.02 | A7000+ | 48MHz ARM7500FE, RISC OS Select 4.33 | SA202 | Risc PC SA 202Mhz, RISC OS Select | A4 | ARM3, RISC OS 3.1, 16 Colour greyscale | SA287 | Risc PC SA 287MHz, RISC OS 4.02 | Iyonix | 600MHz, RISC OS 5.03, DMA enabled |
What do the numbers mean? Well, the percentages compare the machine to a 202Mhz SA Risc PC running RISC OS 4.02 (100%). So 200% would be twice as fast as the "base machine". We haven't had chance to verify these (unofficial) figures, and it's worth remembering that benchmarks don't necessarily reflect use in the real world. Pinches of salt, and all that. When John visited his dealer he played with a production Omega, and although he didn't run any benchmarks he "would guess it is about twice as fast as RiscPC overall till you use the disc drive, then wow it zooms." While on the subject of MD and the Omega, Desk (MicroDigital Europe) reveal in their latest newsletter (translated) that MD had been secretly negotiating with Pace prior to Castle's purchase of RISC OS. Desk expect their first bulk shipments of Omegas this week. Link: Alpha benchmarks
|
Alpha benchmarks, Omega tittle tattle |
|
(08:10 13/7/2003) mavhc (09:01 13/7/2003) ams (12:26 13/7/2003) flibble (15:00 13/7/2003) Revin Kevin (19:30 13/7/2003) monkeyson2 (20:02 13/7/2003) jmb (21:20 13/7/2003) moss (21:37 13/7/2003) jmb (22:13 13/7/2003) flibble (22:31 13/7/2003) hzn (06:40 14/7/2003) andrew (10:39 14/7/2003) not_ginger_matt (13:21 14/7/2003) not_ginger_matt (13:30 14/7/2003) hzn (15:27 15/7/2003)
|
|
g0tay |
Message #92407, posted at 08:10, 13/7/2003 |
Unregistered user
|
Looks like RISCOSmark doesn't do proper harddisk benchmarking as windows is getting in the way on the VirtualAcorn box with it's caches, hence the 99&66M/sec (in memory) transfers. |
|
[ Log in to reply ] |
|
Mark Scholes |
Message #92408, posted by mavhc at 09:01, 13/7/2003, in reply to message #92407 |
Member
Posts: 660
|
Why is the SA202 much slower than the base machine, a SA202, for Icon Plotting - 16 colour sprite with mask, Draw Path - Stroke narrow line, Draw Fill - Plot filled shape? But much faster for Rectangle Copy - Graphics acceleration test. Viewfinder? |
|
[ Log in to reply ] |
|
Annraoi |
Message #92409, posted by ams at 12:26, 13/7/2003, in reply to message #92408 |
Member
Posts: 56
|
I was thinking the harddisk figures looked a bit odd alright (as the VA uses the underlying OS (host) to do writes then windows would step in copy the write to it's own memory cache - hence the exaggerated performance). I don't suppose that will stop MD from claiming the harddisk access of the Alpha is faster than the Iyonix ;) Regards Annraoi |
|
[ Log in to reply ] |
|
Peter Howkins |
Message #92410, posted by flibble at 15:00, 13/7/2003, in reply to message #92409 |
Posts: 892
|
I wonder how you're getting the A4 to do 800x600 in 256, I thought the limit was 800x600 in 16 (or 640x480 in 256). |
|
[ Log in to reply ] |
|
Kevin Wells |
Message #92411, posted by Revin Kevin at 19:30, 13/7/2003, in reply to message #92410 |
Member
Posts: 644
|
It does state the A4 16 colours. |
|
[ Log in to reply ] |
|
Phil Mellor |
Message #92412, posted by monkeyson2 at 20:02, 13/7/2003, in reply to message #92411 |
Please don't let them make me be a monkey butler
Posts: 12380
|
Well, it does more clearly now :) |
|
[ Log in to reply ] |
|
JMB |
Message #92413, posted by jmb at 21:20, 13/7/2003, in reply to message #92412 |
Member
Posts: 467
|
Is there any reason why the Iyonix was tested in 800x600@256colours? |
|
[ Log in to reply ] |
|
John Hoare |
Message #92414, posted by moss at 21:37, 13/7/2003, in reply to message #92413 |
Posts: 9348
|
Indeed; isn't the IYONIX supposed to be *faster* in higher colour depths? Or have I misremembered? |
|
[ Log in to reply ] |
|
JMB |
Message #92415, posted by jmb at 22:13, 13/7/2003, in reply to message #92414 |
Member
Posts: 467
|
RISCOSmark 1.01 (14 May 2003) OS/Machine/Processor: RO5.03(update27)/Iyonix/80321@600MHz Graphics Resolution: 1024x768, 16M colours Processor: 258% Memory: 209% Rectangle Copy: 1544% Icon Plotting: 217% (no, that's not a typo) Draw Path: 104% Draw Fill: 65% HD Read: 18.1 HD Write: 13.5 FS Read: 1136 FS Write: 524 Food for thought ;) |
|
[ Log in to reply ] |
|
Peter Howkins |
Message #92416, posted by flibble at 22:31, 13/7/2003, in reply to message #92415 |
Posts: 892
|
Contractual Obligation Comment: The Evil that is benchmarks It burns my eyes. |
|
[ Log in to reply ] |
|
Herbert zur Nedden |
Message #92417, posted by hzn at 06:40, 14/7/2003, in reply to message #92416 |
Member
Posts: 11
|
Wow, how did you do those IYONIX pc benchmarks? The harddisc values suggest that you accidently did them on a floppy disc... I got HD Read - Block load 1MB file 18249 = 611% HD Wirte - Block save 1MB file 14054 = 462% This is with RISC OS 5.03 and thus UDMA up and running. |
|
[ Log in to reply ] |
|
Andrew |
Message #92418, posted by andrew at 10:39, 14/7/2003, in reply to message #92417 |
Handbag Boi
Posts: 3439
|
Did you say 'bullshit 'or 'bulkshipment'? ;-) |
|
[ Log in to reply ] |
|
Richard Wilson |
Message #92419, posted by not_ginger_matt at 13:21, 14/7/2003, in reply to message #92418 |
Member
Posts: 63
|
jmb: You've missed the differentiation between Kb/second and Mb/second. ;-) |
|
[ Log in to reply ] |
|
Richard Wilson |
Message #92420, posted by not_ginger_matt at 13:30, 14/7/2003, in reply to message #92419 |
Member
Posts: 63
|
And I've missed the difference between jmb and hzn. Oh well... |
|
[ Log in to reply ] |
|
Herbert zur Nedden |
Message #92421, posted by hzn at 15:27, 15/7/2003, in reply to message #92420 |
Member
Posts: 11
|
Thanks for the clarification. I should get myself a bigger monitor since the on-screen display is pretty small and thus (or just by being illiterate to a certain extent) I didn't see the MB/s note. Error was thus on my side. Sorry. |
|
[ Log in to reply ] |
|
|
The Icon Bar: News and features: Alpha benchmarks, Omega tittle tattle |