I did 4 instances of Super Pi and it took me 21s xD
Printable View
I did 4 instances of Super Pi and it took me 21s xD
Well Crenn you do own a Quad Core... So you should do 8 instances ;)
Two instances of Pi to 1 Mil on mine takes 43secs :'(
Not telling how long it takes to do 4 instances of it.
But then again it's not bad for a Mobile Core Duo, not even Core 2 Duo.
What Everest Home ED. has to say about my box.
http://i163.photobucket.com/albums/t...snapshot16.png
http://i163.photobucket.com/albums/t...snapshot17.png
I am definitely going to re-visit this thread when I build my next box.
This is where we need definite rules...
My 5 instances were to 4M, not 1M......
So let's have Crenns 20 instances of SuperPi to 4M....... (I did 5 on a single core... and 5 x 4 for quad core... it's only fair....).
Hehheheeeee
lol, I'll run it this afternoon. It's going to be quite slow xD
I just noticed, this thread and my overclocking thread has been featured :eek:
Interesting point this. SuperPi is a single threaded application (correct me if I am wrong). For peeps with multiple core processors try running it as standard accross both/all four of your cores and then limit it to one.
In theory the time difference will not be that great. In practice that is exactly what I find. Running it on one core does not take twice as long as running it on two. It takes about five seconds more perhaps on a three minute run. I put that down to some sort of overhead I am creating on the system by limiting it to one core.
Thoughts?
Matthew
Proves my point I think. One of my two cores at 2214 Mhz only just beat helix's one core at 2200 Mhz and both cores wouldn't do much better. Because its a single thread it doesn't make a big difference however many processors it's shared amongst.
Scotty: half the processor speed and almost exactly double the time of crenn.
Me: 2Ghz crenn: 3Ghz = 150% processing power + a 50% boost on memory speed. his time 98 seconds mine 201.
Key factors in this test are raw Ghz (Crenn beats s&*% out of us all) and memory efficiency (ditto).
dgr and drew are using older processors. Well OC'd in dgr's case. Maybe the maths processing is less efficient or the overhead of running the OS slowing it (vista in dgr's case). Be interesting to see a result from dgr under XP or win2k for comparison.
CB
I'll show you examples of it running over multiple cores. There is a reason I set it to run on one core, makes it slightly better (running it over 2 cores can be better than one).
For example, when running UT2k4 over 4 cores, I got an average FPS of 3-7FPS. Over 2 cores I got around 105FPS avg, and on one core, I got around 225FPS avg. I can repeat that also if desired.
I'll also mention that I got this computer 2-3 weeks ago, after Intel dropped the Q6600 prices. And the overclock added to the 'value'. Just wait when I get a new HSF!
As for this test being based on Raw MHz..... not quite. Find someone with a Pent D with a higher clock speed than me, they won't beat my time. This test is based on the efficiency of the processor. Performance per clock, also the bandwidth between the CPU and RAM. Also lower timings can yield better times.
I'll show you soon that Vista isn't too bad. It is slower, but only by a few seconds.
I am xp 32, but skinned.. it is updated somewhere, in the tables...Quote:
5. dgrmkrp - 4m 21.343s | Athlon XP 1700+ @ 2105MHz | 1GB 402MHz 3-3-3-8 | Vista 32-bit
I can't really help myself so I'm gonna swap the keyboard/mouse/monitor/lan to the X2 pc :) Let's see what that can do :) Gonna see what the best settings I can get for the cpu and the memory, separate, can do to the speed. :)
I wasn't sure if it was skinned or not. Thanks for letting us know.
Well running it over two cores my time went to about 3mins 13 secs but i now can't find the screenshot.
Just a note, don't expect results tonight.... I need sleep now...
That was my point earlier though I worded it to make the opposite point :dead:
The extra few seconds is the overhead of splitting the single thread amongst the cores.
In some ways it would be a better test therefore of overall efficiency to insist the app is run over all cores. A poor splitting of the workload will show in the results. This tests the raw power of asingle core of the CPU (including its efficiency) and memory speed/efficiency as I was saying earlier.
Matthew
Well, I'm back.. the tweaking took a few hours, but I found 2 system settings that I like.. Didn't have time to fully test their abilities and I probably won't until I get myself a nice video board. I'm looking now for a cheap quadro board, but I can't find cheap ones that can be delivered all away here.. Plus, I will have the money only next month, which isn't that far... but still..
Back to the pi stuff:
No optimizations, 2GHz, one core.. This is stock, after bios reset and with a new windows xp install :)
..and the times for the test I did..
now, time to bump speeds up. I tried first the max fsb and I got 310 stable, with 2xHT, DDR2 400 settings and CL5, processor multiplier set to 5x. Then I went up with the DDR clock, then I up-ed the cpu multiplier. I got 2,79GHz, ddr@620, HT@1240MHZ, but if I raised the ddr clock, no boot. Voltages were already set to the maximum I like.
The second optimization run lead me to this:
Why am I such a dunce? I forgot to click the memory tab.. still, it is the same, but the RAS pre-charge is now 6, not 5, and the speed is less than 1GHz, because it gets really hot and unstable for the synchronous DDR800 mode, so I bumped it to the DDR667 divider... This is all because I have no cooling on the RAM and because it is fairly hot in here.. At least I'm overheating ;)
The weird tweaking payed off I think, because the time now is much smaller... down to 2:30 :) from 3:44 :| vs the 4:21 for the older processor/system.
^^^: the times for all the tests.
All runs were made on one core, setting the processor affinity in the task manager processes pane.
To end it, this is my table for data comparison :)
nice job.
and now im 3rd :(
I'm sorry drg, but we can only accept your time of 2m 30.828s.
EDIT: Added a small rule.
Oh, no problem..
(Also, just for the fun of it, I might try this again, with some optimizations on other levels.. Deactivating north-bridge components, using a discrete video adapter.. Stuff like that. Also, I know this RAM can work at 1,2GHZ, so now it isn't of much help to me.. Still, the time got by this type of processor, clocked at this frequency, shows that the core architecture is better :( .. mm.. and I wanted to be an AMD fan boy :( Hopefully the k10 is better and compatible with the AM2 socket (not only with AM+).. If it is, then I'll go quad core, peltier and who knows what more ;) )
If I wanted to, I could make mine go faster as well, but I'm worried about the heat output.
The Core 2 Duos are the leaders currently, followed by the X2s. I'm fairly sure AMD will make a beautiful come back. And the new processors, if they're AM2+, they're backwards compatible with AM2. As for going quad core, make sure you have a use for it ;).
I had a very good reason to go for quad core...... Supreme Commander and UT3!
Errmm... :) I think I do :) First, it will be used to generate more heat ;) then, I like the speed it gives in final rendering ;) Well, the second mostly.
Even now, I'm looking for a professional video board and simultaneously doing a major redesign on my antec case.. in 3ds :) So, I need power, cause right now I'm using the old XP system cause it has more graphics horse power. Well, at least the front half is redesigned.. In order to get better, quieter cooling and a place for an 8800 series card.. and more storage.. and more fans.. :)
I think you only need 2 case fans, but that's just me xD
What do you mean by "I'm using the old XP system cause it has more graphics horse power"?
Well, I have an older agp 6800 that I bought for the socket A before I ended up upgrading/assembling a new PC, with the x2... I had do some corner cutting, so I bought a mobo with integrated graphics.. Now, I have an XP 2600+ with a 6800 and an overclocked X2 with integrated graphics.. The problem: whenever a scene I make has too many poligons, the integrated 6150 gpu goes to "sleep" on the job... So, all my modeling is done on the 6800, which usually requires a hefty hi-poly object to cripple it... 12 pipe lines vs 2 pipelines, dedicated VRAM, more speed, more bits crunched. Therefore, I suck :(
This is the family XP box (affectionately known (by me) as sh!tbox.)
Before:
http://i163.photobucket.com/albums/t...k/****box1.png
After:
http://i163.photobucket.com/albums/t...k/****box2.jpg
(Now you know why I call it sh!tbox. ;))
Run time: 5 mins, 36.016secs.
checksum: 40BCBB67 (or was it 4OBCBB67, I can't read the damned font. Stupid monitor only going up to 1024x768. ;) )
Updating now.
Exley any reason why CPUz is showing your E6600 as underclocked to 1.8?
@L2M - Fine job dude. sweet overclock too. Post in the overclocking thread too, gain a top spot there too ;)
If you want to go into the OVERCLOCKING thing I can break out the Ln2 :banana:
Ln2 :o sweet.
Damn you guy's are keeping me busy, i update it and then there's another that appears. Last update then im off to bed 2:30am here.
yes, in my bios my CPU is set to throttle when there is no load on it. I took the screenshot as fast as i could but i guess i didnt catch it at full throttle. if you want i can load it and post a screenshot to prove my clock speed.Quote:
Exley any reason why CPUz is showing your E6600 as underclocked to 1.8?
It's ok when i update again ill copy what you gave in the Overclocking thread. I take it that's the right speed?
yes, thats the speed it was running at when i ran super pi
No need for others 4Million is what they're all going too :)
Updated.
Front updated now.
2m 57.14 s Woot !! :eek: :eek: Woot !! haven't even started playing with my memory timings or chipset voltages yet and im running stable AT 2.5G ON AIR LOL
Knocked 24 seconds off my time and Scotty down a place HAHAHAHHAHA (evil laugh)
CrazyB
ps my memory runs at 400Mhz Scotty its DDR1 mate !!! yr chart upfront be in error, though admittedly its running at 458.6Mhz now I think.