I almost made my dream come true!

[massman]

True oldschool clocking, Mass, keep it 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

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).

Nice!

Thanks!

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

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 .

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

That's the plan!

Really nice Benching old hardware gives a lot of fun

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

Nice score massie

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

Good luck on ure next run bro

Thanks mate!

nice run you get there...keep pushing

Thanks!

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

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.

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

Thanks!

Copy Waza, service tweaks, max mem, record

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

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

That comment made my day

[SAE]

Wow. Nice one there.
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.

[massman]

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?

[SAE]

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.

[Beleno]

very good job massman keep pushing!

[Lastviking]

Hmm Xp-m

Hello SAE long time no seen



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

[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!

[Lastviking]

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?

[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.

[Lastviking]

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?

[HDCHOPPER]

exellent massman

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

[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.

[STEvil]

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

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

[Lastviking]

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.

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.

[massman]

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.

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.

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.

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

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

[fragmasterMax]

i have been doing the same thing!!!!
(with an x2 3800 on a 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.

[MidgetMariachi]

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

Midget Mariachi

[mongoled]

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



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



Now I get my turn to post a screenie



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

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

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



Thanks massman, for this thread



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