• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

SuperPi - Does this sound about right?

agw_01

agw_01

agw_01

Soldato
Joined
11 Apr 2004
Posts
19,859
Hi guys,

Finally took the plunge and have started to overclock my Athlon 64 2800+ after being out of it since selling my old machine a few months ago.

It's running at 2.1GHz (233 x 9) with the RAM at CAS2.5 in sync with the FSB.

Now, I ran SuperPi 1M and it came back with a 42 second execution time. After looking at a few old screenshots, my old Barton 2500 (before I went to the Mobile) at 2.4GHz did it in something like 48 seconds on a 200MHz FSB.

I would have expected the 64 to whoop the Barton, despite the lower clock speed.

It's a Socket 754 Clawhammer.

Cheers :)
 
Ahh cool, as long as it's performing as it should.

I guess I just expected it to break into the high 30's.

Ah well, it passed 8 hours of Prime95 last night, so more tweaking tonight. 300MHz FSB... yes please ;)

Thanks for the reply.
 
My x2 3800+ @ 2.5GHz does 1m in 36 seconds. I don't know if mines underperforming or not, because only a 3 second difference between single and dual core? :confused:
 
H2F Scott said:
My x2 3800+ @ 2.5GHz does 1m in 36 seconds. I don't know if mines underperforming or not, because only a 3 second difference between single and dual core? :confused:
Click to expand...
Somthing's not right there :eek: Can you post a screenshot of CPU Z showing cpu and memory.
 
Hmmm my lowely celeron 356 @ 4.3ghz does super pi 1m in 36 seconds so I would say the time is possibly a little slow although I am not really up on what the times should be.

EDIT: In case I need proof:

superpi.jpg
 
Last edited:
Cheers for the replies guys... been playing a little bit and have brought it down by a second, haha.

Although that is with the RAM timings VERY relaxed, and at 208MHz rather than the 234MHz it was at before... so there's defo room for improvement.

2300mhzoj6.jpg


ramtimingsmx5.jpg
 
Hey agw,

Just as a thought single channel - would that affect the time in superpi? I know timings and speed are crucual so does dual channel make a big difference?


P.S The 620 bus got lowered last night :cool:
 
Last edited:
You need an A64 to be running at close to 3Ghz to get 30secs, so those times aren't too bad.
 
H2F Scott said:
My x2 3800+ @ 2.5GHz does 1m in 36 seconds. I don't know if mines underperforming or not, because only a 3 second difference between single and dual core? :confused:
Click to expand...

It's only a single-core application - it's not multithreading aware - so single and dual-core CPUs at the same speed should get the same result.

Of course, with the dual, you can run two instances at once...
 
-Mike- said:
Hey agw,

Just as a thought single channel - would that affect the time in superpi? I know timings and speed are crucual so does dual channel make a big difference?

P.S The 620 bus got lowered last night :cool:
Click to expand...

Hi Mike,

I completely forgot about the performance difference between single and dual channel. It made naff all difference on my Socket A system so I never bothered looking into it. Will have to wait until after christmas when I can pick up another stick of RAM.

And sweet! Take it he's enjoying it then? :D

Mattus said:
At some point you're gonna need to decrease the HTT multiplier from 4x to 3x - the HT link speed shouldn't go above 800Mhz on 754.
Click to expand...

Ohhh, why's that? Does it have anything to do with the on-chip memory controller?
 
Yea, it does.

I found when using a SKT 754 cpu when getting near the 300FSB range, HT @ 3X tends to help the cpu overclock more.

What board you using? DFI? If its the DFI they are great boards for high fsb and proven to be pretty much the best SKT 754 board there was.

What ram you got?
 
Ohhh, so it's not likely to damage the memory controller by running the HT so high, it'll just help to get a higher FSB? What kind of performance hits do you get when lowering the HT from 4 to 3x?

I've got an Asus K8N-Deluxe. Not a bad board but I got it for nothing so wasn't going to complain. Certainly been impressed with it so far.

I'm currently using a single 512mb stick of Corsair XMS PC3200. Had it left over from my old rig. Can't afford to get any more at the moment, but will look into PC3500 or even PC4000 in the new year. Any recommendations?

(Sorry, I know this should be in Overclocking and Cooling now - Mods, feel free to move it)
 
There is little or no difference going down to 3x. You'll see a nice gain when you run dual channel though.. 5-10% I think?
 
agw_01 said:
Ohhh, so it's not likely to damage the memory controller by running the HT so high, it'll just help to get a higher FSB? What kind of performance hits do you get when lowering the HT from 4 to 3x?
Click to expand...

It won't damage anything... it's just that technically you're overclocking the HT bus so at some point you'll meet instability and have to lower it to 3x. The performance hit from 4x to 3x is pretty much nothing.
 
Cant run dual channel on SKT 754 boards. Single channel memory on the SKT 754 cpu's.

There is no difference really going down to 3x HT multi.

That stick of ram could be BH-5... does it had heatspeaders on it? Idealy you would need to have a couple of 512MB sticks of BH-5 to give a nice performance increase. BH-5 would give you a better super pi even more when running tighter ram timmings... 2-2-2-5
 
You must log in or register to reply here.
Back
Top Bottom