SuperPi32m 5Ghz Ivy All Out Challenge!

[Splave]

what kit is this Allen?

2300 cas8 pi's

[gungstar]

can someone upload pics from this post made by youngpro?

[bullant]

Some more from Gigabyte UD3 Memory LN2 2.1v 7-10-7-25-80@1300 6.17.297

[l0ud_sil3nc3]

^that is sick Bullant, your giving me something to push for

Can't wait to see 16's from you

[bullant]

Thanks Loud,another from Gigabyte UD3 bit of bad start,Memory LN2 2.1v 7-10-6-25-80@1300 6.17.156
Memory LN2 2.1v 7-10-6-25-80@1300 6.17.094
Last one little improvement Memory LN2 2.1v 7-10-6-25-80@1300 6.17.078

[SF3D]

My memory kits are not that good after all This better one does not like high voltage and it need to be around 1.94V. Temp does not help after some certain point, so this was best I was able to achieve with loose subtimings.
OS should be new one as well, but I am too lazy to reinstall



Next week I will try some BBSE 4 x 2GB.

[bullant]

Will have a play around with Twcl today and see if any improvement,@ angoholic really nice time man

[hiwa]

Why not new thread guys ?
Maybe with category of ics this time ? So we know the fastest sams, fastes psc and bbse ?
And 4500 so 80% of cpus can do it ?

[I.nfraR.ed]

6:18:500 - 46x108.7 - 1304MHz - 8-11-7-22-72-1 - Asus Maximus V Formula - G.Skill RipjawsX 2133C8 2x2GB/BBSE - 1.89V Air/CPU water

[angoholic]

@Splave, SF3D: looking forward for your next cold runs!

@bullant: thx

@hiwa: Hm i think 5Ghz are ok because of lots of reference values at this point of time. and i do not know if people care about ics split..the idea "most efficient ivy setup" is better i think. few weeks ago i thought no one cares about Ivy 5 Jigga anymore, but luckily i was wrong

@I.nfraR.ed nice bbse action

[Alex-Ro]

Thanks Loud,another from Gigabyte UD3 bit of bad start,Memory LN2 2.1v 7-10-6-25-80@1300 6.17.156
Memory LN2 2.1v 7-10-6-25-80@1300 6.17.094
Last one little improvement Memory LN2 2.1v 7-10-6-25-80@1300 6.17.078

Good job,try to use 100:133 dividers,you'll get better eff

[bullant]

Good job,try to use 100:133 dividers,you'll get better eff

Thanks Alex,will try it next time,thanks for the help

[TaPaKaH]

a couple of random ideas about this thread:
1) Why not make result list in first a bit like this thread so that it's easier to read? Different ICs could be had in different colours, for example.
2) Since lots of people are using this thread for comparison purposes, why not include ALL results (not only best result per user) in the first post so that people don't have to look for close results throughout the whole 50 pages? Best results per user could be marked, say, in bold?

[angoholic]

Good idea with the different colors, Sam! But with different scores per user i would limit that to the best result per kit?

[I.nfraR.ed]

We have a very good "database" at 5GHz now, so everyone is comparing his performance against that frequency.
Changing it to 4.5GHz doesn't seem good to me.
Yeah, probably limits access for some people, but I think most people interested in this kind of low-clock challenge have a capable cpu/cooling.

I like the idea with different colors and limiting the results to best result with a mem kit for a given user. Bold for the top result of a user is good too.

Btw, does anyone else has problems running Pi32M after wazza/cdt and killed explorer?
No problem in other cases, but when I kill explorer prior to the run it errors out at 2-3rd loop.

Haven't tried with automatic wazza, I only use manual. Tried various voltages without any real difference.

Good job,try to use 100:133 dividers,you'll get better eff

Have tried both dividers before and 100:133 doesn't give me better efficieny, only a faster initial and first loops.

[Leeghoofd]

There's a thread about the waza issue on Hwbot, if I recall correctly perform the wazaaaaaaaaaaa a few multies lower, then up and run or it could be related to the memsize set...

[I.nfraR.ed]

Yeah I remember the thread and reported that I didn't have this issue. Now I have the same settings, but on a Formula instead of Gene.
Gonna read that thread again, thanks for the reminder.

PS: Actually the cpu is different...it was long time ago and I forgot that fact.

[Dumo]

4X4 1.975V air



Uploaded with ImageShack.us

[Raja@ASUS]

Yeah I remember the thread and reported that I didn't have this issue. Now I have the same settings, but on a Formula instead of Gene.
Gonna read that thread again, thanks for the reminder.

PS: Actually the cpu is different...it was long time ago and I forgot that fact.

It's nothing more than the system not being stable enough to handle the increased IO of the waza and maxmem combo. You can try fiddling with voltages to improve stability, or change the maxmem/waza size a bit and see if it helps. All waza does is give you contiguous memory to write to. The more contiguous memory 32M has available, the faster the time. If the data is somewhat scattered you get a latency penalty from tRP, and tRCD because you need to open and close more pages.

Faster time = increased IO over the bus, so more chance of the system crashing.

[hersounds]

narrowly

4.9 Ghz - 1.28 Vcore

http://valid.canardpc.com/show_oc.php?id=2663928

[bullant]

Little improvement on Gigabyte UD3H different Memory G skill 2133 C7 Memory LN2, 2.15v 7-10-6-25-85@1300 6.17.062

[angoholic]

nice one! how cold did you have your mems?

[bullant]

nice one! how cold did you have your mems?

Thanks ,memory's -100

[hiwa]

Awesome run bullant

[bullant]

Awesome run bullant

Thanks hiwa