• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Timeline of an OC effort

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

rengleca

New Member
Joined
Feb 24, 2002
Location
bay area, ca
I was wondering if any of you fellas were interested in my posting my timeline, tricks and basically the whole story of how I got my 1.8 to 2.4. I was also interested in hearing the results of Prime 95 on this board. The reason I mention it: I can run my system at 134 at 1.70 volts, and all the software runs well, I never saw any quirks in windows... but I can't get it to successfully run Prime 95 at those speeds...

Are you guys that are running 133 able to successfully run Prime95?

Are you guys interested in hearing my (long) documentary of getting this system to these speeds?
 
I'm not trying to start a flame war, but how is rengleca's OC'ing not much of an overclock? rengleca actually got a higher percentage increase in speed than you did (1.572/1.2 = 131% and 2.4/1.8 = 133%)
 
It may be a result of excessive heat at the RDRAM prolonged by a long benchmark like this. Have you tried putting a fan directly over the rdram?
 
rengleca - I'd be happy and interested to read your overclocking chronicles. This is a great forum and not representative of that rude reply by False Christian. I am sorry you got such a negative reply to your first post here.

Welcome to the board.
 
Sure, i'd be interested in hearing your story. I'll be overclocking a pentium 4 1.8a GHz in a week or so, so it may even be helpful ) Also, i'm sure there are people running prime 95 at 133 - I think if you have a problem its because of cooling, yet all chips have different limits so I guess it could be a number of things.
 
rengleca said:
I was wondering if any of you fellas were interested in my posting my timeline, tricks and basically the whole story of how I got my 1.8 to 2.4. I was also interested in hearing the results of Prime 95 on this board. The reason I mention it: I can run my system at 134 at 1.70 volts, and all the software runs well, I never saw any quirks in windows... but I can't get it to successfully run Prime 95 at those speeds...

Are you guys that are running 133 able to successfully run Prime95?

Are you guys interested in hearing my (long) documentary of getting this system to these speeds?

I would personaly be VERY interested.
AS I am sure many others would.
We could all learn alot from what you did.... and maybe try to help you in what other options you have next time.

So share the information.
look forward to reading it

oh and how could I forget.
NICE OVERCLOCK
 
The reason I offered is because I am a Field Service Engineer and a good troubleshooter unless I try to do things without taking notes. I figured if I wanted to get a good OC I would take notes and make sure that I was logging all the important changes.

I will organize my notes this weekend and you all can do with them what you like.

If the false christian wants to flame the log, that is fine... six years in the Navy makes me chuckle at civilian attempts to be rude.

For what it is worth, I am pretty sure the heat issue is not a problem for me as I have two 120mm fans in the side of my case at full speed cooling the AGP area and the RIMM/DRCG area.

I am glad some of you are interested in learning from what I learned. I will post the notes withing 72 hours.

R
 
You dont have to worry about flaming.
FLAMES have ZERO TOLERANCE in this FORUM
if you wana flame someone their are many ot[H]er overclocking website that will let you flame away.
OC's is not one of them

Looks forward to reading what you have done,
untill then

jay
 
I'm about to make that same trip you made, maybe faster if I'm lucky :D so yes, any tips, experience you have would be helpful. Please post soon though, I'm planning on ordering my parts soon and BTW, I am particularly interested in your memory setup because that right now is the only component I'm not to sure about yet. Thanks.
 
Formula for an OC

The parts:
Intel 1.8A
Abit TH7II-R
IBM GX120 40 gb HD x2
Samsung 256 RDRAM x2
Cardex/Gainward TI 450 "Golden Sample"
Creative Audigy Platinum
Pioneer 16x DVD
Lite-On 32X CDRW
Intel PCI Lan
Lian Li PC-70 USB
Enermax 430 Watt PSU
2x enermax 80mm fan (adjustable)
2x enermax 120mm fans (adjustable)
WIN XP
Office XP
Chip cooling: Swiftech MCX-478


To start, I researched all these items. I will give a quick rundown

about the choices I made and why I chose them.

Most cases were all about quality and good history OCing with good

prices.

We all know about the MB. This thing just rocks for OCers.

The chip I chose because I was reasonably certain that I could get

it to 123 FSB and that would be almost $400 cheaper than buying the

damn 2.2. That was all I really wanted.

I have a long history with IBM drives and I was a great admirer of

the DTLA line. THat was what I asked for on my order and the reason

I chose to buy from ATACOM was because they said they carried them.

When I saw the drives were 120GXP I was miffed and told them their

website was hosed. They apologized and took another $10 bucks off

each drive. That settled it for me. IBM drives just OC better. I

didn't know if I would need that kind of flexibility given the MB's

ability to fix the PCI bus but there you have it. I am pleased with

their performance.

As to the RAID choice, I am glad I got it and will test setups with

it on my next two drives, but I decided the first 80 gb would be

program data and I just don't think RAID 0 is a good idea unless you

also mirror and I didn't want to do that. For the data drives I

will try to RAID 0 them.

I haven't seen any input on any sight about any company that makes

more reliable RIMMS than Samsung, regardless of what brand name they

have. Central Computer had them on sale, no questions.

I have watched Cardex for a while and I was really impressed with

some of the write-ups they have gotten. The utility they come with

is sweet. BTW, if any of you want to explain to me how to set up

the QIII ARENA demo on a loop and showing frame-rates I will let you

know what they are.

I have over 20gb just in music that I cherish so the sound card

option was important to me. This card sounds good and doesn't use a

lot of CPU to play MP3's so that is why I went with them. My

girlfriend has a keyboard that we sample into my system and the

front live drive set up kicks *** for that. It also provided for

the fire-wire ports that I may play with soon.

My last two systems both had Pioneer DVD drives that were cheap and

worked great, lasted long time... for 75 bucks you can't beat that.

LiteOn was all about the drive's abilities to write original errors.

This means you can copy more discs and defeat more copy protection

than with any other CDRW. Besides that, the price is unbelievable

and 32X kicks major butt.

I have a long history with Intel and I trust them over others when

it comes to LAN. The only other source I like is Netgear. This one

was the right price.

The power supply for me was crucial. I wanted to have the power

available for a decent peltier if I ever went that way and this case

has a funny setup for the PSU. It has to have a fan on the top, not

the bottom.

The case was a no brainer. It is big, it is beautiful, it is tight,

it is well built, it is light, it has USB up front, it has the space

for water cooling or vapochill mods, it comes with great ADDA

fans... and I couldn't wait to hack into it. Two extra enermax 80's

are in the back top exhausting and turned down. You have to do

something here, at least tape over the holes if you don't want fans,

because if you don't, you will short circuit the flow that cools the

PSU and it will run hot and have a shorter life. These are turned

down to minimum flow, which I will discuss later. The 120s... well

here is where you all call me crazy. I cut the case on the side

panel and aimed one at full throttle at the vid card and one at full

throttle at the RIMMs/northbridge. This total combo keeps the case

under positive pressure, so air only comes in where I have input

fans: two up front 80 mm ADDAs at full throttle and the two monsters

on the side. This is good becuase those spots are filtered. So my

insides stay cleaner. I would estimate that there is over 200 cfm

going through my case. I kept the top exhaust fans turned down to

suck most of the flow out the bottom. That is where the real heat

is. The case comes with 2 more ADDA 80s here at full throttle and

three little exhaust holes.

I had trouble in the beginning because my first supplier for the

swiftech ran out on my order so I had my system ready to build

without the kick *** cooler. Oh well, I went ahead.

Things went great with all the assembly. I don't really like the

INTEL stock cooler, but it supposedly guarantees perfect even

pressure. Oh well. Scrape that black goop off the HS and put some

Arctic Silver on there. Take the fan off the Vid Card and put some

AS there too. It has something there but I didn't buy the AS to

look at... before putting it back on I placed the external thermal

lead onto the vid card and into the MB. I had good placement there

on the chip's edge without disturbing the HSF contact. The temp has

never been higher than 44c.

On with the show. Regardless of how I tried I couldn't get the Win

XP installer to partition and format the HD while it was plugged

into the RAID header. After moving the HD to the regular IDE0

header it went well... I couldn't believe how fast it was. After

checking the boot and activating I shut down and moved the HD to the

RAID header and rebooted.............
.
.
.
BINGO! No worries, it didn't care.

I started with just the vid card, mem, 1 HD, and FD... Shutdown,

add hardware, restart, install drivers... test, test, test....

no problems.

Windows XP kicks ***. WOW. Networking, new hardware, file

transfers from my old system all went well...

After I had installed all the hardware and all the programs I

started to see what the chip temps were. Keep in mind the TI 200 I

overclocked from the start with their utility so those temps only

varied with ambient.

With all my fans and the stock cooler I had:
Ambient: 25
CPU:35 idle, 38 loaded (I will follow this standard from here down)
All the Sandra numbers will be the same format:
Drystone/Whetstone: 3385/3140
Multimedia: 7156-8761 it/s
Memory: 2486-2480

I waited to clock it up for when I got the cooler. I was thrilled

when it came but I was not looking forward to strippping my system

back down. Let me add some things here. I don't use rounded cables

for my drives as the shielding on the cheap ones is inadequate and

will slow you down. For this I got cables that were long enough and

routed them flatly along my case, always crossing each other and

other cables at right angles. This takes a lot of patience. But in

the end you get just as good air flow and faster drives. If you

want to spend the money on the really expensive platinum rounded

cables, be my guest.

So everything comes out and I pull the MB. When swiftech recommends

glueing the nylon nuts to the stand offs and MB, I say: DON'T EVEN

THINK ABOUT NOT DOING IT!!! YOU WILL STRIP NUTS, YOU WILL GO

CRAZY!!! YOU WILL WORRY ABOUT YOUR CHIP!!! Just take the time,

pull the MB all the way out, put on the stand offs and nylon nuts

and superglue them together and to the back of the MB. Trust me, it

will save you headaches galore. If you end up switching coolers,

the stand-offs may not have to be removed. After my 2nd stripped

nylon nut I got new nuts and superglued it all down. Everything

back together. I spent maybe 30 minutes cleaning the CPU. I spent

another 15 working AS into the HS according to AS's web page. Then

on with the layer of AS and on with the cooler. I have to interject

that while this is a great cooler, I would change a few things. Too

many of the little parts are cheap, like the fan hold down bracket

and screws. Put some pride into it, man! I ended up getting nuts

onto these too. Keep in mind I am a part time machinist/mechanic so

I tend to put real torque on fasteners.

All back together:
24
31/33

You can see that this cooler really does the trick!
At stock speed 33 degrees C fully loaded with P95!

Time to clock this baby!

On to MB settings that matter!

First of all I spent days looking for an explanation of differential

current. It aint out there. I left it x6.

Tip: Keep the memory setting at 400! Not AUTO! I couldn't run 105

FSB with the setting in auto. set it to 400 and I hit 102 and 105

no problems.

Keep in mind that I tested windows, office apps, browser, Norton

Antivirus, Sandra, P95, and WMP on every setting.

105*18 33 (fixed) 66(fixed)=1890 420/840(bridge speed/memory speed)
1.5 V HWD: 1.49 down to 1.44 during P95.
23 case temp
32/36 chip unloaded/loaded
3560/3294 sandra cpu
7530/9182 sandra multimedia
2610/2606 sandra memory

This cooler is great! This BUS FIXING makes OC'ing too easy!

110*18 33f 66f =1980 440/880
1.5 V 1.49-1.44
24
32/38
3736/3448
7886/9628
2735/2729

Wow, my numbers were already killing the Sandra 2.0 benches. I am

sure that is based on the heat and the L2 cache, not to mention the

memory and the fact that this MB is faster than others at any speed.

time to bring up the voltage. I did this first step on instinct.

115*18 33f 66f = 2070 460/920
1.55 V 1.53-1.49 voltage drop during P95...
23
31/38
3913/3602
8237/10070
2863/2860

Look at that! No increase in temp! 15% overclocked and no increase

in temp! That is 38 C running p95 for 1 hour.
I stopped for the night here and ran p95 over night. It went great.

This speed is solid! You could prolly do this stock if your case is

well cooled.

So I was watching those precipitous voltage drops when the system

was running the p95 torture test and wondered if that could help me

guage when to increase the voltage without having to take steps

forward, fail, and then raise the voltage to succeed... I would say

that maybe it made little difference... I thought that if I could

minimize the voltage drop at say 115*18 by going to 1.60 (see only a

drop to 1.57) then maybe I could be assured that it would run 118*18

at 1.60. It was part of my thinking anyways.

The next day I decided to do the VPIN mod. My girlfriend talked me

out of using the hacked BIOS because she thinks I am brilliant and

that I should trust myself with the pin mod and not some crazy

hackers. She must love me a lot.

So I ran some higher numbers:
117*18 and 119*18 at 1.55 and settled into the following:

120*18 33f 66f = 2160 480/960
1.575 1.56-1.52
23
33/39
4071/3759
8594/10514
2986/2983

obviously before the VPIN mod, BUT ROCK STEADY!! ANYONE UNCOMFY

WITH THE MOD CAN DO THIS!

Here I did the mod. Here is a funny story. If you short the pins

to the wrong ones your system will not POST. The display will read

00. Recheck your pins and make sure that you are only shorting the

intended pins. Here was my little trick after failing:

Lay the chip on its back, pins up, with pins 4,3,2,1 at the bottom

right.
Cut your speaker wire, and strip out one strand about 3 inches long.
Use a needle (sewing) to make a loop and carefully lay the loop

aroung pin 1, pull the wire tight and lay your needle over the top

of the wire between pins 1 and 2, fully in the channel created by

all the parallel pins so that the pins are much higher than your

needle. This way you can easily twist the wire once, hook pin 2,

hold the wire down, and transfer your needle to the next valley to

hold down again and repeat. I have a good photo of this!!

Everything back together, clear the CMOS, power up, enter in all my

BIOS settings:

125*18 33f 66f = 2250 500/1000
1.675 volts 1.66-1.62
25
34/40
4331/3933
8924/10926
3114/3108

This is a really great temp. If you stop here, you have beaten the

2.2 in raw speed, but now you have the memory speed and quad pumped

bridge speed killing every machine on the block, and you saved

almost $400 minus all the damn fans + the really loud swiftech...

BTW, yes, it is loud... REAL LOUD.

Onward and upward

Tinkering one megahertz at a time found me wishing the voltage

adjustments were in much smaller increments... like maybe 1.675 to

1.680 to 1.685 etc... It got harder to find sweet combos up at this

speed. More about this later as it really kicked my *** at the end.

130*18 33f 66f = 2340 520/1040
1.70 V 1.69-1.63 (notice, still a .06 drop...)
26
35/41
4694/4089
9273/11365
3240/3234

Here I stopped for the night again and noticed that after a few

hours the case temp was up. That cooler is really doing it's job

and I am moving an *** ton of air through the system (*** ton is a

mechanical term). The system also failed P95 after 6 hours. This

is another good stopping point because your memory is only 72%

efficient and if you run this speed, you are running the memory at

advertised speed: 3.2 gb/sec.

131,132 were touchy and 133 was just ****ing me off...

133*18 33f 66f = 2394 533/1066
1.70v 1.69-1.63 (same drop)
26
36/42
4449/4183
9484/11641
3315/3310

Here I had quirky problems and P95 would fail after 3 minutes. I

imagine the chip temp would have gotten hotter but I coulnd't heat

it with p95 and the STABILITY TEST.EXE software just doesn't load

the chip the way p95 does.

Check this out and you will understand why I want finer voltage

adjustments.

At 1.725 volts P95 failed instantly. IE failed and Norton (NAV)

failed.

At 1.675 volts it failed to boot win xp.

SOOOOOO.... what to do. Screw it, let's go higher.
You have to find the right FSB, one the chip can handle, where the

voltage is going to be right....

BINGO!

134*18 33f 66f = 2412 536/1072
1.725 1.71-1.66
25
36/43
4638/4217
9592/11725
3324/3321

That was the trick...

And then, uhoh, I had Lan problems. Things weren't connecting.

hmm.....

135*18 showed memory errors

132 at 1.70 volts showed the same 95 problems.

BINGO!

It is a good thing I take all those notes.

Have you noticed I kept the f's in the PCI and AGP settings?
I had them fixed at 134, where there is no need...

I wonder....

134*18 33.5 67 (2/4 setting under AGP) = 2412 536/1072
1.725 1.71-1.66
26
37/44
4639/4215
9594/11731

All the internet programs ran really great!

If only everything was as simple to do as overclocking a northwood

with this mb.

BTW, P95 is still not very stable at this speed. It will shutdown

after 3 mins (show an error). The thing is: I never see a boot

problem, I never see a POST problem, I never have a software issue

and STABILITY TEST.EXE runs perfect for 12 hours.

I am happy.

Why this speed?

So I can DIVX rip DVD movies in real time at 29.97 fps into cdrw

size files and quit paying hollywood all that extra money. (I

collect coupons for blockbuster now and rip the movies).

Total cost: $2144 with tax.

Old monitor: Viewsonic pf790 (two years old, looks great).

And for my music: Central had a sale on the last box of Altec

Lansing 400w 641 system for $199.

DAMN, THE MUPPETS NEVER SOUNDED SO GOOD.

(singing) you can't live with 'em, you can't live without 'em...
there's something irresistable-ish about 'em...
we grin and bear it cuz the nights are long...
I hope that somethin' better comes along...

and the matrix on DVD on this system will ROCK YOUR WORLD!

Try not to flame me too much, I am posting this to help people reach

stable speeds, not take over small island countries with left wing

politics.

hehe

And all you guys that think 14 fans (12 system plus two in the back

of the desk) is too loud? Just turn up the speakers baby!

Check the adjusted signature for corrections...
 
i must say nice OC, and that was very informative....

you might want to edit your post a bit to get it into paragraph forum so its a bit easier to read... other than that...

Once again Nice OC, and i like the temps
 
Nice OC.

Thanks for sharing the info, the bit about not using PCI/AGP Fix at 134 FSB is very useful, and I will keep it in mind when I come to that stage.

I am currently in the process of OCing my 2.0a and I am hopping to get to 133 FSB.

Last night I went to 125 FSB and I thought it was stable but it crashed in the middle of the night during stress testing. Interestingly I was using a vcore of 1.65v and I have now upped it to 1.675v which is the same as what you used on that frequency. It would be nice to have a database correlating FSB and vcore on stable systems. I will know if it's stable when I go home later today.

I noticed you left the RDRAM at x4 all the time. I switched to x3 as early as 115 FSB because I thought it could add to unstability. Hence why I am hoping to reach 133 FSB, it will then run at the intended speed.

Could you post your picture of the vpin mod? There has not been any picture yet of someone using a needle to keep the wires down ;-)

.aR
 
That story could be turned into a major motion picture. Nice job.

Here is a funny story. If you short the pins to the wrong ones your system will not POST. The display will read 00. Recheck your pins and make sure that you are only shorting the intended pins.

I recall a couple frantic messages from him when the wire trick was first not working and he couldn't boot up. Unfortunately, I was unable to offer much assistance other than to tell him to double check everything and make sure the wire was not touching the wrong pins (which turned out to be the case). I bet his neighbors two blocks away heard his sigh of relief when he fixed the problem and it still worked. I think he was fearing the CPU got fried.

Rengleca, you mentioned having a good photo, if you want, we could post it in my "sticky" VID pinning guide. I would give you full credit for the pic of course. Let me know.

Arman68, You should be able to leave the memory setting at "400" until at least 125 FSB. Lots of us are getting 133+ FSB out of Samsung RDRAM using the higher setting, if you have good case cooling and air circulation.
 
Last edited:
As soon as I remember to get batteries for this camera I will take a look at all the pictures.

Probably I will end up posting them on my web page and giving you all the links.
 
Back