I almost made my dream come true!

As many are planning Nehalem LN2 sessions as we speak, I decided to bring back to life my favourite setup uptill now: the Athlon XP. Ever since I had to quit this platform because of the Athlon64, I had only one goal in mind and that goal was owning the world record for K7 processors in the 1M category.

Now as the thread title may have already told you, I didn't beat the actual record, because that is currently in hands of SAE, who did 30s three years back. I did, however, break the world record on hwbot and the world record with the newer SPI version with millisecond calculation. Not entirely what I had in mind, but what the hell, I'm more than happy any way :D.

CPU: AMD Athlon XP Mobile 2600+ IQYHA 0352 RPMW
MOBO: Dfi Lanparty Nforce 2 rev.B (vcore-ovp-vdd-vdimm modded)
Cooling: LN2 (thanks to Teurorist)

http://valid.x86-secret.com/ccpuz.ph...7814&nocache=1
http://valid.x86-secret.com/show_oc.php?id=437814

http://i154.photobucket.com/albums/s...-/31703_wm.png

I hope to hit the jackpot in the near future, though, as I was able to bench at 270FSB on air cooling AND already hit 31.437s, but due to my optimism, I didn't save that score. Note that this 1M is made on a fully untweaked windows installation.

I will add pictures of the rig later this week as my buddy Geoffrey took the pictures :-).

Great score, massman !!!! Its already nice to see benching some old school hardware. ;)
Did Teurorist give you LN2 ? Was this on the AOCM in Germany this weekend ?

Teurorist brought his Socket A compatible cpu container which allowed me to do these LN2 runs.

Yes, on AOCM :D :D

That´s a really good SuperPi score :P

Needs more volts, I ran that much on air cooling lol ;)

Quote:

---

Originally Posted by STEvil

Needs more volts, I ran that much on air cooling lol ;)

---

QFT. :D

Try Win2K...it seemed to be better for 1M with AXP's for me. :shrug:

man I remember my barton with love and affection.
Great chips in their day but never saw numbers like those!:clap:

Some pictures, that's all I have Pieter. Next time make sure you bring along low ESR caps for me to add them to your board, I'm quite sure these will help stabilize the OC ;)

Quote:

---

Originally Posted by STEvil

Needs more volts, I ran that much on air cooling lol ;)

---

Don't really think it would have helped, it's an FSB issue, it would ran up to 270 ~ 275 MHz air cooled while with LN2 the machine had a hard time running at 260MHz.

witte - Actually the issue may be that the northbridge is fed by two voltage sources, vnd and vcore (assuming you were using an nForce2 board).

I never actually got around to testing the effect of decoupling them and only feeding from one source but if one half is running at vnb and the other is being fed vcore (2.0v+) then it could be causing signaling issues...


maybe i'll have to break out the nF7's..

haha great mate! socket A was my favorite but never had chance to try under ln2 :) keep pushing it !

Quote:

---

Originally Posted by Computerville

That´s a really good SuperPi score :P

---

Thanks :)

Quote:

---

Originally Posted by STEvil

Needs more volts, I ran that much on air cooling lol ;)

---

We tried up to 2.2v, but the frequency didn't go up that much. I think we'll have to add voltage stabilizing modifications to go further. Good thing is that the cpu was capable of running at -130°C. I -think- we hit the coldbug of either cpu or motherboard at -155°C, though.

Quote:

---

Originally Posted by Gautam

QFT. :D

Try Win2K...it seemed to be better for 1M with AXP's for me. :shrug:

---

Win2K faster than a tweaked XP? Got to try that before next session.

Quote:

---

Originally Posted by Movieman

man I remember my barton with love and affection.
Great chips in their day but never saw numbers like those!:clap:

---

Thank you!

Quote:

---

Originally Posted by STEvil

witte - Actually the issue may be that the northbridge is fed by two voltage sources, vnd and vcore (assuming you were using an nForce2 board).

I never actually got around to testing the effect of decoupling them and only feeding from one source but if one half is running at vnb and the other is being fed vcore (2.0v+) then it could be causing signaling issues...

---

So to keep the board stable at 2.2v vcore, we should run the Vnb at 2v?

Quote:

---

Originally Posted by hey

haha great mate! socket A was my favorite but never had chance to try under ln2 :) keep pushing it !

---

Thanks bro!! :D

True oldschool clocking, Mass, keep it up! :up: And yes, Win2k FTW all the way until the Core 2 Duos.

Haha loving it P-J; need to go crazy with the cap mods now ;)

Nice! :clap:

hehe that one you've benched. never saw a score of you at the weekend. :D

it was realy nice to meet you.

Quote:

---

Originally Posted by STEvil

witte - Actually the issue may be that the northbridge is fed by two voltage sources, vnd and vcore (assuming you were using an nForce2 board).

I never actually got around to testing the effect of decoupling them and only feeding from one source but if one half is running at vnb and the other is being fed vcore (2.0v+) then it could be causing signaling issues...


maybe i'll have to break out the nF7's..

---

Two?? Never heard of that, then again, I never got further then copying the available voltage modifications :) I agree, you could run them from the same power source since they require roughly the same values, but high VNB doesn't really help right so why bother? What about VCCA?

Nice 1 massman, get that OS + pi tweaked and take your record!! :D

Really nice :D Benching old hardware gives a lot of fun :)

Nice score massie :up:

Good luck on ure next run bro ;)

nice run you get there...keep pushing :up:

Nice one Pieter!
I didnt think you made it, oh, and thanks for the beer =)

Next time, I will bring some gaffatape for Oliver, so he wont be that hyper :p:

EDIT;
Oh, just read, you know.... it helps? Hehe
Was close, yea, to bad, but you did a good job :)

Nice. Its always fun to play whith cheap hardware. Looks like you had a nice time. Sub 30 seconds next then? :)

Copy Waza, service tweaks, max mem, record :D

Nice score! :toast: That AXP is giving Phenom a hard time ^_^

Quote:

---

Originally Posted by tiborrr

True oldschool clocking, Mass, keep it up! :up: And yes, Win2k FTW all the way until the Core 2 Duos.

---

Thanks for the tip, I really had no idea about the win2k being so good for 1M. I always thought win2k was only good for 3DM01

Quote:

---

Originally Posted by Johnny Bravo

Haha loving it P-J; need to go crazy with the cap mods now ;)

---

Actually, we were going to test the cap mods last weekend, but when we tried, the board didn't boot (it booted afterwards though).

Quote:

---

Originally Posted by Krelin

Nice! :clap:

---

Thanks!

Quote:

---

Originally Posted by gianni-gt

hehe that one you've benched. never saw a score of you at the weekend. :D

it was realy nice to meet you.

---

Haha, yeah ... I probably should've benched more, but the P45 setup was giving us a very hard time. Next to that, we couldn't compete in the CPU-Z competition as apparently our S478 system had failed on us as well.

It was nice to finally meet you as well. Thanks for the awesome vids, by the way, although my case throwing skills don't really look that great in the vid :p:.

Quote:

---

Originally Posted by TheKarmakazi

Nice 1 massman, get that OS + pi tweaked and take your record!! :D

---

That's the plan! :D

Quote:

---

Originally Posted by bartx

Really nice :D Benching old hardware gives a lot of fun :)

---

I have more up my sleeve once I finish benching this platform. Asrock PT880 FTW!!

Quote:

---

Originally Posted by Cranox

Nice score massie :up:

---

Thanks bro! When are you going to start benching again?

Quote:

---

Originally Posted by blind_ripper

Good luck on ure next run bro ;)

---

Thanks mate!

Quote:

---

Originally Posted by Hazzan

nice run you get there...keep pushing :up:

---

Thanks!

Quote:

---

Originally Posted by M.Beier

Nice one Pieter!
I didnt think you made it, oh, and thanks for the beer =)

Next time, I will bring some gaffatape for Oliver, so he wont be that hyper :p:

EDIT;
Oh, just read, you know.... it helps? Hehe
Was close, yea, to bad, but you did a good job :)

---

Hehe, just don't drink it all in one night, it would be a shame.

Quote:

---

Originally Posted by Razzz

Nice. Its always fun to play whith cheap hardware. Looks like you had a nice time. Sub 30 seconds next then? :)

---

Thanks!

Quote:

---

Originally Posted by Oj101

Copy Waza, service tweaks, max mem, record :D

---

Yeah ... so much to improve. I already gained 0.3s by killing all processes and setting SPI to realtime!!

Quote:

---

Originally Posted by Fr3ak

Nice score! :toast: That AXP is giving Phenom a hard time ^_^

---

:rofl: That comment made my day :rofl:

Wow. Nice one there. :up:
For me my memory clocks made the difference, I feel. I was using multiplier 12.5x and I was having a hard time getting a stable output (esp. pifast) though, cause AXP ist really scratching its limits at these kinda clocks.

But I would not have ever thought there's ne1 out there still fiddling with AXP. :D

My memory timings were pretty tight ... well, really tight. If I could've switched to the 12,5x MP, I'd been pretty close, I reckon.

What hardware modification helped you the most in terms of stability?

OCP and Vdroop IMHO. Do not remember exactly anymore since it's been a long time... ;)
You also have to cool everything getting hot on board's surface - mosfets, vreg ics for ram etc... Iirc I had a waterblock on my NB and NB was lapped. Dunno if that really helped though. My board and cpu were pretty good, my ram was not the best (max 270 2-2-2-11). My timings also were the tightest in terms of performance.

very good job massman :up: keep pushing!

Hmm Xp-m

Hello SAE long time no seen :)

http://img444.imageshack.us/img444/1972/mhzcpuzxf0.jpg

That was about max i did with mine on air :)

Fast super pi was tricky like SAE did say... max fsb super pi was about 265mhz so i think he was on a good sweet spot there :) , if you have very luck you can do about 268 if i remember right.


My stuff is still live and kicking my brother is using that stuff Xp-m 2600 0352rpmw, dfi nf2

LastViking, your NF2 did over 290FSB, no? If yes, please revive the setup and go LN2 ... if you're able to bench even at 285FSB, you could be very near breaking that 30s barrier!

Quote:

---

Originally Posted by massman

LastViking, your NF2 did over 290FSB, no? If yes, please revive the setup and go LN2 ... if you're able to bench even at 285FSB, you could be very near breaking that 30s barrier!

---

291Fsb DC but only screen.... But the big trouble was to get super pi stability, sceen was easy and sisoft bench also worked to about 282mhz. Super pi was hard to get to around 270fsb. I didnt find what was holding the stabilty back...but the vdimm and the sticks should not hade any problem to do about 280+ super pi becuse they did that on Nf4 with same vdimm.

And it can also be harder to get higher fsb on the 13 multi, did you check how high you could do on 12.5?

Definitly not higher than what I did on 13x. The weird thing was that I had no problems hitting 272FSB memset stable on air cooling (1M at 275FSB), but 260FSB was very unstable on LN2. If I'd been able to run around 265-270FSB on LN2, I'd been under 30s, I think. At that CPU frequency, the FSB is the most important factor for the performance.

Btw, using a high multiplier has a bad effect on overclocking. Higher than 15x made the system for more unstable on the same frequency as 13x. 14x multi didn't even work :).

Regarding the memory timings, I used your guide on XS to tune them. They're a bit more tight, I think, but when I have the time, I'll make a photo of my bios.

Quote:

---

Originally Posted by massman

Definitly not higher than what I did on 13x. The weird thing was that I had no problems hitting 272FSB memset stable on air cooling (1M at 275FSB), but 260FSB was very unstable on LN2. If I'd been able to run around 265-270FSB on LN2, I'd been under 30s, I think. At that CPU frequency, the FSB is the most important factor for the performance.

Btw, using a high multiplier has a bad effect on overclocking. Higher than 15x made the system for more unstable on the same frequency as 13x. 14x multi didn't even work :).

Regarding the memory timings, I used your guide on XS to tune them. They're a bit more tight, I think, but when I have the time, I'll make a photo of my bios.

---

If i dont remember 100% wrong... memset was easyer than super pi with high fsb.

So you did say 12.5x265 did not work for you?

exellent massman :clap:

now with I7 here it's time to empty the bottom shelf and try to rekill all the old hardware :D nice clocking man:up:

Hey Lastviking, pal! Nice to see you're also here (again?, still?) :D

FSB/RAM clk-wise I think there's possibly some kind of vdroop as well, as nearly everybody with great ram could only bench at those average 265MHz clock limit. Everything exceeding that would just throw you errors. Remember, those boards did neither have much (any?!) solid caps nor many phases for CPU/Dram. I had at least 4 DFI LP Nf2 and each performed different - we were trying hard to find any Taiwan-made DFIs (those were the best by far). So luck was one of the mattering factors in ocing back then.

Quote:

---

Originally Posted by wittekakker

Two?? Never heard of that, then again, I never got further then copying the available voltage modifications :) I agree, you could run them from the same power source since they require roughly the same values, but high VNB doesn't really help right so why bother? What about VCCA?

---

It may not help because one source or the other may be putting out "dirty" power or the voltage differential is creating an internal signal difference.

My old Infinity would run 270mhz with BH-6 as well.. but then I dropped the board :shakes:

Anyone tried making the northbridge run hotter (around 40-50c or something) while the CPU is LN2 cooled? :shrug:

Quote:

---

Originally Posted by SAE

Hey Lastviking, pal! Nice to see you're also here (again?, still?) :D

FSB/RAM clk-wise I think there's possibly some kind of vdroop as well, as nearly everybody with great ram could only bench at those average 265MHz clock limit. Everything exceeding that would just throw you errors. Remember, those boards did neither have much (any?!) solid caps nor many phases for CPU/Dram. I had at least 4 DFI LP Nf2 and each performed different - we were trying hard to find any Taiwan-made DFIs (those were the best by far). So luck was one of the mattering factors in ocing back then.

---

Yes the Taiwan was abit beter than the China made board, i used both of them...the China board died and then i got a Taiwan board instead.

Quote:

---

Originally Posted by Lastviking

If i dont remember 100% wrong... memset was easyer than super pi with high fsb.

So you did say 12.5x265 did not work for you?

---

Nope, I couldn't. With the 12,5x multiplier I was limited to around 255FSB, just like with the 13x multiplier. The same day, I had tested 270x8 and it worked perfectly in memtest; I'm not sure anymore whether I tested it in superpi as well.

Quote:

---

Originally Posted by SAE

Hey Lastviking, pal! Nice to see you're also here (again?, still?) :D

FSB/RAM clk-wise I think there's possibly some kind of vdroop as well, as nearly everybody with great ram could only bench at those average 265MHz clock limit. Everything exceeding that would just throw you errors. Remember, those boards did neither have much (any?!) solid caps nor many phases for CPU/Dram. I had at least 4 DFI LP Nf2 and each performed different - we were trying hard to find any Taiwan-made DFIs (those were the best by far). So luck was one of the mattering factors in ocing back then.

---

That would make sense, as my previous DFI NF2 could run over 260MHz FSB quite easily as well. I'll try to add caps to the back of the motherboard to stabilize the power.

Quote:

---

Originally Posted by STEvil

Anyone tried making the northbridge run hotter (around 40-50c or something) while the CPU is LN2 cooled? :shrug:

---

I did. When on LN2, I was using the stock heatsink on the northbridge. Temperatures were not high at all.

i have been doing the same thing!!!!
(with an x2 3800 on a :banana::banana::banana::banana:ttttttey tforce 4 550, and aircooling but still. )
gotta love old school benching. It really makes you value the extra mhz. Today oc'ing is so easy, it's like bada bing bada boom, and you have a 40% increase.
Man the things we take for granted.
and i sound old.

Good job! Congrats! I like the pic of the cooling with the fog everywhere :)

Midget Mariachi http://photo-shack.com/img/c930eecd0...cc445f708f.gif

Great to see some of the old skool XS members here in this thread

:cool:

You know who you are, those were the days, ahhh the nostalga

:D

Now I get my turn to post a screenie

http://www.erraserver.com/images/29xx.gif

Never did break 3Ghz on this CPU, only went as far as water.

Quote:

---

Originally Posted by Lastviking

My stuff is still live and kicking my brother is using that stuff Xp-m 2600 0352rpmw, dfi nf2

---

Same here dude, it looks like that system is going to get retired, so maybe just maybe I will give it one last try to get 3Ghz with ice water

Quote:

---

Originally Posted by SAE

Hey Lastviking, pal! Nice to see you're also here (again?, still?)

FSB/RAM clk-wise I think there's possibly some kind of vdroop as well, as nearly everybody with great ram could only bench at those average 265MHz clock limit. Everything exceeding that would just throw you errors. Remember, those boards did neither have much (any?!) solid caps nor many phases for CPU/Dram. I had at least 4 DFI LP Nf2 and each performed different - we were trying hard to find any Taiwan-made DFIs (those were the best by far). So luck was one of the mattering factors in ocing back then.

---

Hey dude, never did get close to that 1M superPI time of yours. Good to see you around :)

:ROTF:

Thanks massman, for this thread

:up:

Bringing back the days when XS was full of members just wishing to help each other regardless of what components were being used

:clap: