Page 1 of 2
Speed...
Posted: 19 Sep 2010, 15:01
by petermcleland
I noted these few lines in the Messages file in BOINC manager:-
I haven't a clue how fast that is but I assume that it is pretty quick!
In fact I am amazed to discover that I can run FS9 at McLeland Field in Real Weather and 100% Traffic with Einstein@Home running its business at the same time...All CPUs continue at 100% and FS9 behaves very respectably with only about a third knocked off the frame rates

Re: Speed...
Posted: 20 Sep 2010, 11:34
by jonesey2k
A good test of your CPU speed is to download a program called SuperPi and have it compute pi to the millionth decimal point 20 times. Back in the day of Thoroughbred Athlon XP's under 1 minute was good but now anything under 15 seconds is super fast

Re: Speed...
Posted: 20 Sep 2010, 11:54
by petermcleland
Thanks Mark...I'll try that but I expect that as with BOINC I will have to run it in "Motor Car" mode (CPU is throttled by setting the Vcore voltage at 0.96 Volt)...If I let it go in "Aeroplane" mode that voltage is set up at 1.26 volts and it will start blowing horns at me because of CPU temperature. However, it will be interesting to run that Pi benchmark

Re: Speed...
Posted: 20 Sep 2010, 12:22
by DarrenL
Just ran SuperPi on my laptop, I'll run it on my Desktop later when I turn it on.
But here's my result and the laptop spec.
1M Calculation - 20 iterations = 52.413s
Laptop
Dell Inspiron 1546
Windows 7 64bit
AMD Turion X2 Dual core 2.20ghz
4GB RAM
I might have to do that again later as I was still using it when it was running
Re: Speed...
Posted: 20 Sep 2010, 12:39
by petermcleland
Yes I just had a go and got these results:-
Motor Car Mode=27.906 secs
Aeroplane Mode=22.188 secs
Rocket Mode = 15.547 secs
However...It was set for 19 iterations and I couldn't find a way to change that...All were for 1 Million decimal places. The CPU usage only went up to 22-24% and there was no problem with temperature. Interesting Mark...Thanks for the pointer

Re: Speed...
Posted: 20 Sep 2010, 18:21
by ianhind
I thought I'd check out how far my ancient CPU (3.0 Ghz E6850) fell behind Peter's super computer.
19.859s for 19 iterations of 1M dec places suggests that this test is past its sell by date if it can't show bigger differences than that!
Possible explanation from overclock.net:
Super Pi is a single threaded application. It will only use 1 core and relies heavily on clock speed.
So my dual core CPU at 3.0Ghz benefits from that.
Ian
Re: Speed...
Posted: 20 Sep 2010, 18:35
by jonesey2k
There is a version called HyperPi that is multithreaded but it makes little difference. It basically just runs 4 instances of this instead of a joint computation.
Re: Speed...
Posted: 20 Sep 2010, 19:07
by DarrenL
ianhind wrote:I thought I'd check out how far my ancient CPU (3.0 Ghz E6850) fell behind Peter's super computer.
19.859s for 19 iterations of 1M dec places suggests that this test is past its sell by date if it can't show bigger differences than that!
Possible explanation from overclock.net:
Super Pi is a single threaded application. It will only use 1 core and relies heavily on clock speed.
So my dual core CPU at 3.0Ghz benefits from that.
Ian
Single core performance sounds about right, I have a quad core 3.4ghz CPU and it gave me similar results so it can't use it fully. I don't overclock mine though.

Re: Speed...
Posted: 20 Sep 2010, 19:38
by DaveG
20.5s looks a bit slow for a 3.4GHz quad
Mine did 17.875s on a 3.0Ghz Q6600

Re: Speed...
Posted: 20 Sep 2010, 19:44
by DarrenL
DaveG wrote:20.5s looks a bit slow for a 3.4GHz quad
Mine did 17.875s on a 3.0Ghz Q6600

Maybe for some reason it handles single core better. In the coding of the test.
This is my PC spec.
And this is the performance rating on Windows 7 (max 7.9) You see it gets 7.4 out of 7.9 for calculations per second. and 7.5 or memory operations only the HDD speed lets it down (it's a large capacity HDD) but I'll be changing it soon anyway.
