OCing NewCastle...

[L'enFer]

i had nothing to do so i decided to check my cpu ocing possibilities. i do it with box cooler :( so the result is only 2359 Mhz (standard is 2009).
of course the main reason is bad cooling and memory (see signature) but what can i also do to get better result (please don't offer increasing voltage or my system will burn...:( )?
i think decreasing timings will not help, but i'll try.

[BigDov]

That's about the max I get with mine too..... it's not altogether horrible for 'old' hardware

[FireDragon]

Even with better cooling u 'might' looking at 2500 maybe 2550...that is is around 1.65-1.7...but i would be happy with what u have at this point probably going to upgrade soon anyway right?

[Driver]

With my Dfi UT250GB i got 2620@1.7v watercooled

[L'enFer]

to upgrade soon anyway
it's true, but i'd like to get max with that i have now. well, i'll wait for my wc system, hope i'll get smthg better.
p.s.
Even with better cooling u 'might' looking at 2500 maybe 2550
did u see that
overclockers.ru cpu database (http://overclockers.ru/cpubase/?cpu=23&search_mark=&search_mb=&dc1=&dc2=&date=10000&items=25&showother=ON&showcoms=ON&action=results) look it attentively and u'll find that it's possible to overclock with water 3000+ to 3000 MHz :).

[DonS1mpson]

A 400MHz overclock on “Average” is nothing to be ashamed off, but I think it’s time you got a DFi and found out what your chip can really do.

[Magnj]

My Clawhammer Max's at 2.55 and the mem controller is poop aand cant run 2-2-2-X

[NickS]

i had nothing to do so i decided to check my cpu ocing possibilities. i do it with box cooler :( so the result is only 2359 Mhz (standard is 2009).
http://my-photo.ru/avatar0/b/3/3567.jpg
of course the main reason is bad cooling and memory (see signature) but what can i also do to get better result (please don't offer increasing voltage or my system will burn...:( )?
i think decreasing timings will not help, but i'll try.

Not bad for stock cooling. I got 2.2GHz semi-stable with an XP-90 and my old Newcastle s754, on a DFI LP nF3 250GB board :(.

Nick

[L'enFer]

you got a DFi and found out what your chip can really do
i think my Epox is also good or not? as you see HTT is 359 Mhz FSB/DRAM multiplier is 1/12.
and what can DFI do?

Not bad
but not so good :(.

[h@RRy]

Newcastles are rough. You actually HAVE to go 1.6v+ to get anything out of them. In fact, I hardly though it was possible to get a 20% OC with this core on air(I salute you;D)
Anybody here done any work with a newcastle 3800+?
So far she's pushin 2.6 and I'm still looking for 2.7...So please, if anybody here has any newcastle science please share it.

thanx,
h

[crackhead2k]

yes...it not very ocable
there is no stability at 2650mhz (air cooling @1.65v)
2600mhz stable for me max
It one of teh last 130mmm you can really expect them to hit 3ghz...
90nm procs are easy
You wont be able to push any further unless you have water cooling of phase change with a high voltage board...
on an a8n sli delux use the 1008 bios for high voltage settings...
1.55v is kinda stable with the latest bios...

[h@RRy]

I think my max is like 2610...(#$%(#$*)I dunno, I need to reserve so more time to ABSOLUTELY find providence that is 2700:D(cuz if I don't...)

[[ZIZI]]

I got my hands on a pretty good NewCastle a while ago, it was a 3400+ though...but it went the way of ol' yeller during a freak IHS accident.

It was a LBAZC...(quite a late NewCastle revision) and I managed to get it to 2930 (suicide), with some un-optimized pi at over 2800...it was on air, a Coolermaster Hyper-6, ambient was about 16 C.

http://www.xtremesystems.org/forums/showthread.php?t=76124

[SoF]

newcastles, specially older ones are crappy clockers...2853MHZ SS on a ss and 1.75V here...
Won't do much more so don't waste time.

[L'enFer]

now i've got a pump (Hydor L25). soon i'll continue my experiments...:)

[Kilyin]

I've been running my s754 3000 Newcastle at 2400mhz on an 8KDA3J board (air cooled, stock voltage) for several years. I'm not sure how much farther it will go, because I got tired of the vcore bug (if you change the multiplier to anything but 10 or auto in the BIOS, the board undervolts the vcore so badly that it won't boot into Windows). The only way to fix it is set the multiplier back to 10 or auto, then turn off the PC, unplug the power cord, hit the power button again, then plug the power cord back in.

http://home.comcast.net/~kilyin/cpuz.jpg

[WeakSauce]

I'm still running my 3400+ @ 2.7ghz on 1.55vcore. It's been running at this speed for about a year and a half or so. It all depends on your stepping with the newcastles. Anything past 0432 is gonna oc pretty well.

[Kilyin]

For nostalgic purposes, I just went into the BIOS and tried it again. Anything more than 244 HTT and the vcore goes bananas and starts undervolting itself. So the bug doesn't seem to be strictly tied into the multiplier, although that's one sure way to reproduce it.

I'm just going to let it be at 240 since I'm going to be doing a new build this coming week. Other than this quirk, this board/cpu has been stable and solid as a rock.

[Asazman]

looks good mang :) i was able to take mine up to 3ghz with my prometeia..but from what i remember i was topped out around the same place you are

[kutujambak]

mine 2800+ newcastle do 3,1ghz(344x9,1,375v*133%)on DFI nf3 250gb with TCCD ran on around 260mhz 2.5-4-3-6 1T..with alu DICE -24'c on BIOS

benchable around 3020mhz run spi 2m -->58s,3dm01 around 18500 with 9550card

is that good?
suggestion welcomed..

[Kilyin]

mine 2800+ newcastle do 3,1ghz(344x9,1,375v*133%)on DFI nf3 250gb with TCCD ran on around 260mhz 2.5-4-3-6 1T..with alu DICE -24'c on BIOS

benchable around 3020mhz run spi 2m -->58s,3dm01 around 18500 with 9550card

is that good?
suggestion welcomed..

:stick:

[kutujambak]

:stick:

ummmm...sorry dumb question,but what does it means?

[DaWaN]

My newcastle did 2700mhz on air and 2730mhz on water :)
http://nap.fol.nl/45.23.jpg :)

[kutujambak]

My newcastle did 2700mhz on air and 2730mhz on water :)
http://nap.fol.nl/45.23.jpg :)

do u ever attempt on DICE/LN2?
and what memory do u used?what vdimm?

[L'enFer]

mine 2800+ newcastle do 3,1ghz(344x9,1,375v*133%)on DFI nf3 250gb with TCCD ran on around 260mhz 2.5-4-3-6 1T..with alu DICE -24'c on BIOS
2800+? it means that the standard frequency was 1800 MHz? and u made it work on 3100?

is that good?
in my opinion it's very good :).

[Raptor One]

I managed 2420 MHz out of a 3500+ (2211 MHz stock) on air cooling. That's with 1.66 Vcore as measured by a DMM. Anything higher Vcore than that and the temp readings go above 70 C when running Prime95 torture test #2. Anybody run that hot for a while with Prime95 torture test 2? I'm 100% stable for over 16 hours with my current 10% OC.

[beast-usa]

Hi L'enFer,

New here but here's my 2 cents.

Have you tried dropping your fsb & increasing your multiplier?

I have had a couple amds that would go a little higher by doing that.

[kutujambak]

it means that the standard frequency was 1800 MHz? and u made it work on 3100?

yupz the stock freq is 1,8ghz..btw 3,1 is only screen,for bench(spi 2m,3dm01) is around 3020mhz..

[L'enFer]

beast-usa, hi. i didn't increase the multiplier because as far as i know it is blocked for increasing, i can only drop the multiplier and increase the HTT. but if u know how to do that (to unlock multiplier) please tell me and i'll try it :).
but i think i need to change the memory, because my Kingmax DDR 400 is not the best for ocing.

[h@RRy]

I shoulda got a 3000+...
I know I can hit 2700(somehow)it can suicide at 2700 and not reboot/freeze. Although with these cores it seems that you can't run(or I can't)high FSB over a 10% OC, even with a divider.

[L'enFer]

is it better to overclock on nF3 or nF4? or it more depends on the motherboard?

[L'enFer]

i tried to overclock my NewCastle 3000+ on Asus K8N again. for better cooling i opened 2 windows so the cpu temp was not higher than 43-44 degrees (due to CoreTemp). the max cpu freq was 2400 MHz. when i set 2500 MHz in ClockGen 3dmark03 (which i used as stability test) didn't load. i decided to increase the voltage to 1.575 V (due to BIOS) winXP booted but Catalyst Control Center showed me an error and refused to start. after this i had to finish :(... it's a pity, because it think i could do more...
what could it be, i mean CCC error?

[L'enFer]

it's me again with my old 3000+ NC :)...
as i said i bought Scythe Infinity and today i tested it with my cpu. let me tell you what i could do.
so, at first i attached two 80 mm fans to the heatsink:
http://img254.imageshack.us/img254/180/img01059jb.jpg
i used standard clips for it. i heard Infinity has a bad mounting system, but i didn't improve it, because i had no ideas how to do it. Infinity on Asus K8N:
http://img265.imageshack.us/img265/8756/img01027fg.jpg
and the result is:
http://img254.imageshack.us/img254/3241/414521012wh.png
as a stability test i used SuperPi (8 Mb calculation):
http://img154.imageshack.us/img154/2613/2500scytheinfinity8mb807ru.jpg
not bad, 2500 MHz is freq that i couldn't reach for a long time, but it's not a WR :D...
thnx for attention. have a good day and results.

[Kaffebord]

This is my record. It's on water with a lousy radiator. Was pretty lucky with this one.

http://www.home.no/kaffebord/270FSB.JPG

[anibal4083]

http://img260.imageshack.us/my.php?image=1mcon5128hn.jpg

http://img267.imageshack.us/my.php?image=2mcon5122jj.jpg


mi old newcastle: Tforce6100-m7, newcastle @ 2650, thermaltake polo 735, 1.5 gb ddr400 @ 225, powercooler 500s, x1300xt 500/800 @ 648/820

[L'enFer]

Kaffebord, cool results. have you reach 2.7 GHz on MSI K8N or on an other motherboard?

[L'enFer]

my new "record" :D (it seems like 2.5 GHz is a limit for my cpu):
http://img516.imageshack.us/img516/2388/cpu2520ux0.jpg

[freakBYnature]

i had nothing to do so i decided to check my cpu ocing possibilities. i do it with box cooler :( so the result is only 2359 Mhz (standard is 2009).
of course the main reason is bad cooling and memory (see signature) but what can i also do to get better result (please don't offer increasing voltage or my system will burn...:( )?
i think decreasing timings will not help, but i'll try.

I don't think it's that bad, I prefer clocking with the logic steps rather against them... like 2 GHz, 2.2, 2.4, 2.6, 2.8 ,3.0 and so on. I got my socket 939 venice (1.8 GHz) up to 2.8 GHz with using stock 9x multi, I used 1.5v. Even the memory went better with single core rather than dual core, so if your memory is over DDR400 you should try tightening timings now rather than try clocking higher. I always look for the highest clock I can use 24/7 rather than the highest clock reachable. My Opteron 170 boots into windows on 3 GHz, but it's no use when it miscalculates all the time or I get bluescreens whenever :) But if you get a better cooler you might get it higher, because the voltage you are using is quite high for the stock cooler, if it's the aluminium sh*it :shrug:

[L'enFer]

freakBYnature, thanx for all advices. actually, i use Scythe Infinity. now my cpu works on 2,52 GHz and it's stable enough. i don't think it can go higher :(...

[L'enFer]

CandymanCan, yes, you should be luckier :). my doesn't want to go higher than 2.5 +/- 50 MHz... i tried 10x255 and 9x285 and the result was the same: it refused to be stable during the tests.

[freakBYnature]

freakBYnature, thanx for all advices. actually, i use Scythe Infinity. now my cpu works on 2,52 GHz and it's stable enough. i don't think it can go higher :(...

I repeat, if your cpu won't budge a nanometer you should try raising voltage for RAM and tighten the timings, it always has a huge effect on performance! Trust me! :) AMD works better on tighter timings than Intel... Intel more prefers enourmous bandwith (more mhz). And always try out command rate 1T first if you are using one or two memory sticks, if you are using four sticks you must set command rate (CR) 2T.

[freakBYnature]

2800+? it means that the standard frequency was 1800 MHz? and u made it work on 3100?


I think thought 2800+ was 1.6 GHz? The 3000+ for s939 runs on 1.8 GHz stock. They're probably different on s754 and s939.

[L'enFer]

freakBYnature, these are my timings:
http://img514.imageshack.us/img514/118/memory2520hd3.jpg
when i tried other timings system wasn't stable...

[TMM]

2800+ Clawhammer (512K L2) socket 754 is 1.8ghz
2800+ Newcastle (512K L2) socket 754 is 1.8ghz
2800+ Clawhammer (1MB L2) socket 754 is 1.6ghz
;)

[freakBYnature]

The timings seem fine on first look, what volt are you putting through them? Is it PC2700 (333mhz) or PC3200 (400mhz) RAM? If it's PC2700 you have already overclocked them pretty good, but if its PC3200 you still have some mhz to stock speed.

And what chips do you have on the RAM stick? UCCC? TCCD? MICRON? BH5?

UCCC and TCCD likes low volt, even when overclocked. 2.5~2.65V is more than enough.

MICRON likes around 2.7~2.85V

BH5 can go 24/7 with over 3V, don't remember actually where, but ONLY WITH ACTIVE COOLING!! Like a fan or two :)



The main rule I go by with regular "old" DDR RAM is; never go over 2.9V and only for investigating purposes! But the different chips has different tolerant levels, and with active cooling the threshold for burning them up is higher.

[L'enFer]

freakBYnature, thanx you a lot, but i don't think some more Mhz could rise perfomance of my hardware.

[freakBYnature]

freakBYnature, thanx you a lot, but i don't think some more Mhz could rise perfomance of my hardware.


Practically you are right, 50 more mhz won't do you any good, but
I think you principal-wise are wrong, cause every hardware has it's "sweet-spot", and if you find that exactly spot where everything goes as fast as possible - together ... you will feel that your computer is working faster.

RAM speed is often more essential than what you would think. My RAM to use that as an example has 52.2ns latency when going DDR466 3-4-4-10 1T.... but if I have it on DDR520 3-4-4-8 1T the latency lowers to 42ns... aka "sweetspot"... but when overclocking my cpu maximum the ram isn't capable (atleast not on any AMD board I had) of running more than 260 MHz FSB without the use of a divider.