Pifast Benchmark

my new score with 203*12

Oh boy..... watch out, my awesome power supply is here. Time for 2.6+ ghz!

Omg, I'm getting so frustrated. My temps are perfectly fine @ 2.5ghz--I'm doing 216x11.5 (approximately) and It goes to "Series computing time: 52.04" is what I've been getting, and then the damn thing doesn't go any further but continues to "think" but it never finishes. Have I broke the program or something? I mean I can cancel it, and then try it again and it goes to the same spot and stops! ARGH! I would have an amazingly fast score!

I just did some theoretical calculating, and it would seem that if i was getting 52.04, I would get an overall time of 65.34 approximately. >_< If you want I can screencap it no problem, but it just won't go past the "Series computing time".

Is your computer stable?

I've had that happen when I was pushing a bit too far. Just scale back 1MHz and see what happens.

I found that happen either on too extreme timings or slightly too high overclock.

I don't understand why it would hang though in the same spot, I mean I'd really like to get it to go at that speed. Going back defeats the purpose heh. I can run all sorts of stuff at this speed, but this benchmark refuses to go past that spot for some reason. :\

Because your system is not stable enough to complete the bench at that speed. Period. I found the same thing, my friend, with my own system, you have to back off a Mhz at a time until it completes the bench:)

Randi:D

Quote:

---

Originally posted by MrIcee
Because your system is not stable enough to complete the bench at that speed. Period. I found the same thing, my friend, with my own system, you have to back off a Mhz at a time until it completes the bench:)

Randi:D

---

Well then this benchmark must have a magical ability to utilize the cpu more than super_pi and seti@home combined together. I'm tellin ya, my comp. is stable at that speed. My temps are fine, but it won't complete it. When my comp isn't stable it will get to like maybe 4 or 5 and then say "pi error". That happens all the way up from 2525 up to 2600. I think the cpu needs some more juice is what I think :P

Try running a little Prime95 on your system. I've noticed that I can get SuperPi to run even when my computer isn't 100%. I think SuperPi is a good test, then this PiFast is the next step, and then Prime95 is kinda like the torture test for your processor and memory. ;)

What kind of times are you getting when you scale back 1MHz? It's got to be better than the current AMD high, right?

Quote:

---

Originally posted by Jeff
Try running a little Prime95 on your system. I've noticed that I can get SuperPi to run even when my computer isn't 100%. I think SuperPi is a good test, then this PiFast is the next step, and then Prime95 is kinda like the torture test for your processor and memory. ;)

What kind of times are you getting when you scale back 1MHz? It's got to be better than the current AMD high, right?

---

Well I did lots of benching today... and b/c I got so pissed at my 2400+/8k5a3+.... (it decided not to give me multipliers again today) I decided to teach my mobo a lesson and in windows I upped my FSB to 337mhz, and in the process fried the 85ka3+. I am not saddened at all by this loss. The motherboard was asking for it. I then decided since it was fried to try the 2400+ on my KD7 motherboard. The proc, ram, and everything else attached to the 8k5a3+ is fine, _AND_ the multipliers ALL work on the KD7. So apparently the 8k5a3+ doesn't support jack sh*t--at least w/ or w/o the 5th L3 connected.

So after I tested all that I then decided to play w/ my p4... I pin modded it, got it up to 3060, and put my 4200 on the p4, and achieved the 14th best 4200 score. *Twirls flag* (Will post more on this in the 3dmark forum later).

So anyway, out of this I have learned that the 2400+ and 8k5a3+ is not a good combo. I'm going to pick up a AT7-MAX2 instead and use that.

FYI -

The pifast benchmark stresses MUCH more than seti and super pi. You need ot be able to complete at least a 32m pi test to match the pifast bench for stability. It also pushes temps much lower than super pi. The only thing that hits the temps even more than pifast is seti at the very start of a work unit. However, the pifast test seems to hit the cpu harder despite this.

Remember as well that seti will run all the same even if your results are getting corrupted. Its only in extremely unstable circumstances that seti will cause a crash.

A few other good stability tests -

My first choice is always memtest. This proggy is awesome. Leave test 4 looping to hammer your cpu or test 5 to hammer your memory. If you can loop either for 3 hours or more youre totally stable.

Folding - Better for burning in than stability testing due ot the time involved.

Games - total annihalation to hammer the cpu. This proggy is very succeptible to crashes if your cpu is clocked too hard. Jedi Knight - Outcast - again hammers the cpu very hard.

DVD ripping - hammers the memory more than cpu, but works both hard. A very good cimbination for burning in is to loop test 4 on memtest for 3 hours, test 5 for 3 hours, rip a dvd then complete a folding unit. After that you should gain a fair amount. First run of the above youll get 1 maybe 2mhz. After that the time at which your mhz increases decreases. It may take several loops of the whole process above to gain a single mhz after a while. Generally speaking your first week gains you 5mhz, second week 3mhz, 3rd week 1mhz then every 2 weeks to a month or so thereafter for a single mhz. Dont forget that cpus are tending to show less and less increase in speed form burning in nowadays.

PiLsY.

Interesting thread. I had not heard of Pifast until I read about it in the Prometia article. I'm going to start using it for stability testing.

On the issue of stability and what to test for then I always run CPUMark99

Why cpumark99 ? Well, if it doesn't run this then it aint going to run any other benchmark successfully. It only takes 30 seconds to run and immediately gives you an indication if your BIOS / jumpers are not quite right.

Someone emailed me the other day that cpumark99 was outdated, but it can't be because it scales exactly like the cpu PCMark scores do, so if you get a 10% increase on cpumark99 then you do on PCMark !

And it takes a lot less time to run.

Regards

Andy

78.11 seconds with the specs below at 169x12 = 2032Mhz. Submitted my results. :)

Oooh a new game to play. Here's my contribution.


http://caprid.homestead.com/files/pi66.03.jpg


12x198 on a 2400+

:cool:

Quote:

---

Originally posted by Caprid
Oooh a new game to play. Here's my contribution.


[img]http://caprid.homestead.com/files/pi66.03.jpg[/


12x198 on a 2400+

:cool:

---

Which mobo?

Have it in a KD7.

Thought i'd join in on the fun :)

http://users.pandora.be/JCviggen/pifast.JPG

There is JC...and crushes everyone...again :)

Quote:

---

Originally posted by JCviggen
Thought i'd join in on the fun :)

---

Yikes, nice one JC :) Looks like < 55s is the next big target for the big boys. < 70s is still a decent target for everyone else, < 65s if you have a bit more muscle than the average overclocker and < 60s if you and your P4 are feeling particularly frisky.

Glad to see people enjoying the bench I helped cook up one boring afternoon :)

Rys

It would seem the table lags behind the results....

I posted a 62.83 a few days ago, and no sign of an update!!

/Sets new target for <58.31

Gandelf

*cough* KAIN forgot to update the table *cough*

Done now :D

One score I'm not 100% sure about right now...

Thought I'd join your fun....

Used to have an impressive score on that table, now I am just a nobody hovering above the AMD users....

Looks like I gonna have to have another go when I get me granite bay and new P4.....

The 2.26 wont last at that speed for more than 20 mins or so without a blue screen, so I leave it at 3008 for 24/7 running...

Prometeia is the daddeh tho :D

List is still being kept upto date for anyone interested :)

Andy ( Home ) is now up to 84.5 seconds with 1815 Mhz @ 173MHz RAM speed. All other settings same.

Actually, I am just waiting for the processor to cool down as I am attempting a run at 1826 but it keeps crashing near the end as it is fairly stretched.

I just managed my first ever SuperPi under 1 minute as well tonight, 58 seconds.

Regards

Andy