Report on 1st OC attempt

[vejita-sama]

First, thanks to everyone in these forums for all their help, support and suggestions. This is the first time I've every tried to OC a system and all things being equal it wasn't that hard.

I bought everything from newegg.com:

Abit NF7-S rev 2.0
AMD Barton 2500+
Thermaltake Xaser III V1000+ case
Fortron 530W PSU
Zalman NB HS
Thermaltake Smart Case Fan 2
2x Buffalo PC3700 DDR

I put the Zalman HS on the SB, and left the stock NB HSF alone (save for putting some AS3 on it).

Hooking up all the fans and the hardcano 9 for the Xaser case was a pain as the documentation is pi&& poor, but once I figured it out went fairly quickly.

I transferred all my old compotents over to the new system, tried to repair winXP (had to reinstall actually for some reason), and then got started on the OC.

I set the memory to 2.5-4-4-8, and the CPU to 5x200. I was able to raise the FSB to 215 before winXP refused to boot. I raised the Vcore to 1.9 and P95 et.al. still wasn't stable at 215 (not sure why). So I backed down to 210FSB.

Then I raised the mulitpler to 9, and started to increase it by 0.5 until winXP complained. I got up to 11.5, and raised the Vcore to 1.775 to get into winXP. But P95 wasn't stable, and my temps at load where ~52-54C, which made me (newbie OCer) nervous. So I backed off to 11x210 which was stable at 1.675.

I then changed my memory to 2-2-2-6 which took 2.8V to get me into winXP. I ran memtest o/n and the ram seems ok. I can loop sandra or fold for 16hrs+, but P95 seems to fail from 5-8hrs in. I'm calling that stable enough :) If games/software starts failing I'll back off some more...

And that's where I stand now (see sig):

http://students.washington.edu/scousin/rv/Trunks/oc.jpg

Temps @ Idol - 32 System & 44 CPU, @ load - 37 system, 49 CPU

I'm not sure why I can't a FSB over 215 (I see posts that say 225) and prehaps with better cooling I could get a multipler of 11.5 instead of 11.

But when I started all this my goal was to get a 3200+ system on a 2500+ CPU (11x200) I got that and 110Mhz more so I can not complain.

The whole process has been a lot of fun and I learned a LOT more about how my PC works (esp memory, had no idea about timings prior to this project). So thanks again, and maybe in the future I'll try to squeeze more out of the system.

[fiji]

real nice OC man, is it stable?

[hitechjb1]

Nice overclocking.

The nf7-s 2.0 though has the best potential to overclock its FSB, sometime it is very tricky and needs lots of time to do it right. I find that there is no sinlge recipe to tune it. It depends on many things like
bios setting: CPU interface enable, bios version, ... CPU FSB_sense setting. My impression is that it averaged around 220 MHz.

Summary on overclocking the NF7-S rev 2.0 (with Tbred B 1700+ DLT3C and ...) (http://forum.oc-forums.com/vb/showthread.php?s=&postid=2154286#post2154286) (page 15)


From your list of hardware, I am not familar w/ your heat sink. If it is not giving you the low temperature that you are hoping for, a SLK-800U or SLK-947 should perform better with a variable speed high CFM fan. HSF is the first important thing for CPU overclocking, besides the CPU of course.

[vejita-sama]

fiji - yeah it's stable. That is I ran memtest overnight, it lopped a few times no problems. I ran Sandra for ~16hrs and I folded for about 16hrs (ie. started before bed, turned off after work). However, P95 fails 5-8hrs in. All the software I run (TMPGenc, CCE, DVD Maestro, Premiere, Photoshop, etc) is running fine. So I'm calling it stable :)

hitechjb1 - I'm using a copper based HS from my old HSF, maybe it's not enough. Looking at newegg.com the SLK-800U is a better HS than the SLK-947, as it's only $39 I'll give it a try.

My main concern is that I wanted the temps at load to be <= 50C, and I assume that when summer comes around I might have to back off the OC some. As a newbie I don't want to push things to much. At the same time I want the best numbers I can get. That's why I raised Vdimm 0.2V to get 2-2-2-6 (instead of 2.5-3-3-7 which ran at stock voltage).

I've got CPU interfaced enabled, ACPI enabled and video shadow disabled.

Question - does lapping the HS really help? I some how I hadn't even heard of this process/step until today... there's always more to learn :)

[hitechjb1]

I think in general, the SLK-947U is a little bit better, especially you want to get highest level of overclocking, for short term benchmark and fun. The SLK-947U allows using a 92 mm fan (e.g. Vantec 92 mm Tornado delivering 110 CFM for short term testing). For 24/7, you can to add a VR of about 10 ohm to adjust its speed, or buy a fan speed control. I think the SLK-947U is around $32 now at svc.

For the NF7-S 2.0, bios setting w/ APIC enabled is good for IRQ and I/O handling, and CPU interface enabled is good for highest memory bandwidth. Though some people complain about not getting highest FSB with them enabled. I would suggest to use them whenever possible.

[vejita-sama]

Thanks, actually the Xaser III case has build in fan speed controlls. I have my TT SF2 connected to one (normally run at 3500rpm because at full speed it's just to damn loud :))

I'll checkout the Vantec Tornado, I like the high CFM but hope it's not 'too loud'. Based on other treads I think if I can keep the temps in line going to a Vcore of 1.9V should be safe.

[edit/update]: ok looking at the various websites it appears that the Vantec Tornado runs at 4800rpm and 56.4 dB (which is pretty loud).

I know these numbers aren't linear but:

dB RPM CFM
=============================
TT SF2 17 1300 20.55
48 4800 75.7

Tornado 56.4 4800 119

Right now I'm running the TT SF2 at 3500 rpm (that's ok on my ears). Assuming the numbers are linear that's ~ 55 CFM and ~ 36.5 dB.

The Tornado therefore could be ran at 3100 rpm to produce the same noise level (~ 36.5 dB) but would produce 76.8 CFM. Which is equal to the TT SF2 at full blast.

Again I know these numbers are not linear but I suspect they are in the right ball park. So I'll order the Thermalright SKU-947U + Vantec Tornado (92mm) at the end of the week (payday :))

Thanks again for all the help/support. Going to try to read the lapping guides again and figure out what's the best way to go (seems like only going one direction is best).

[sQUiRLy]

Does lappingHS really help? I some how I hadn't even heard of this process/step until today... there's always more to learn :)

After lapping my $13 Tt TR2M2 I got a 2°C drop in load temperature but this hs/f desperately needed it. The copper insert has milling bands from the center outward creating less surface contact. I used 1000 and 1500 grit wet/dry metalworking paper w/ water and a touch of soap for lubrication. If you try this make sure you sand evenly, rotating the sink each time you clean and prep will help ensure a flush lap.

I think wall-mart automotive has the fine grit paper.