Re: [Rpcemu] RISC OS 5.17 benchmarked against 4.02

In message <6ba362e352.old_coaster@old_coaster.yahoo.co.uk>
Tony Moore <old_coaster@yahoo.co.uk> wrote:

> On 23 Oct 2012, george greenfield <george.greenfield@tiscali.co.uk> wrote:
>
>> I recently ran Richard Spencer's benchmarking app romark 1.01 on both
>> my RPCEmu installations (0.8.8/5.17 and 0.8.9/4.02 respectively). Here
>> are the results (0.8.8/5.17 first, % advantage of 0.8.8/5.17 last):
>
> I've now carried out the same exercise on my own installation comprising
> 0.8.9/5.19 and 0.8.9/4.39 running on Win7 (32bit) on a 1.6GHz Core2 Duo
> machine. Both virtual machines have 256MB memory. My screen resolution
> is 1360x768. For ease of comparison, my results are appended to yours:
>
> George Tony
>
> 1. 572562; 592026; 97% 269685, 283301, 95%
> 2. 9139; 9380; 97% 3060, 2381, 129%
> 3 . 3665; 3053; 120% 1246, 1014, 123%
> 4. 11389; 6429; 177% 4573, 2148, 213%
> 5. 4198; 3060; 137% 1638, 1148, 143%
> 6. 4831; 3591; 135% 1898, 1526, 124%
> 7. 99020; 21340; 464% 87192, 8302, 1050%
> 8. 74777; 25683; 291% 93275, 8861, 1053%
> 9. 3525; 2322; 152% 1487, 1085, 137%
> 10. 3006; 1861; 162% 1348, 730, 185%
>
> My results confirm your conclusion that, under emulation, RO 5.xx is
> considerably faster than RO 4.xx. The differential for tests 7 and 8 (HD
> Read and HD Write) is even more striking.
>

I'm glad your results are similar to mine; I did wonder whether some
unknown factor was skewing them in favour of 5.XX. And, as you say,
the HD Read/Write stats are even more one-sided. FWIW, the 5.XX
Processor benchmark is almost exactly proportional to basic system
speed: 573k/3.4 = 168.5k; 270k/1.6 = 168.75k.

At the moment 4.02 is my primary RPCEmu setup; it has networking, a
HardDisc4 (necessary for Photodesk to run) and can run all the legacy
stuff I use (Eureka, Rhapsody). I tend to use 5.17 for
graphics-intensive stuff (just done a 24-page calendar using OvPro
which produced an 82MB file). Given these results, I can see benefit
in shifting activity further towards 5.XX, but the lack of HD4 is a
bit of a showstopper. Do you know if the memory-leak problem which
prevented HD4 installation on 5.17 has been sorted by 5.19?

George

--
george greenfield

_______________________________________________
Rpcemu mailing list
Rpcemu@riscos.info
http://www.riscos.info/cgi-bin/mailman/listinfo/rpcemu